cargo fmt

.gitignore

@@ -6,3 +6,4 @@ steel_decimal/tests/property_tests.proptest-regressions
 .direnv/
 canvas/*.toml
 .aider*
+.codex

Cargo.lock

@@ -493,7 +493,7 @@ checksum = "d71b6127be86fdcfddb610f7182ac57211d4b18a3e9c82eb2d17662f2227ad6a"
 
 [[package]]
 name = "canvas"
-version = "0.6.0"
+version = "0.6.7"
 dependencies = [
  "anyhow",
  "async-trait",
@@ -512,6 +512,7 @@ dependencies = [
  "tracing",
  "tracing-subscriber",
  "unicode-width 0.2.0",
+ "validation-core",
 ]
 
 [[package]]
@@ -585,7 +586,7 @@ dependencies = [
 
 [[package]]
 name = "client"
-version = "0.6.0"
+version = "0.6.7"
 dependencies = [
  "anyhow",
  "async-trait",
@@ -596,6 +597,7 @@ dependencies = [
  "dotenvy",
  "futures",
  "lazy_static",
+ "nucleo",
  "prost 0.13.5",
  "prost-types 0.13.5",
  "ratatui",
@@ -640,7 +642,7 @@ dependencies = [
 
 [[package]]
 name = "common"
-version = "0.6.0"
+version = "0.6.7"
 dependencies = [
  "prost 0.13.5",
  "prost-build 0.14.1",
@@ -2103,6 +2105,27 @@ dependencies = [
  "windows-sys 0.60.2",
 ]
 
+[[package]]
+name = "nucleo"
+version = "0.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5262af4c94921c2646c5ac6ff7900c2af9cbb08dc26a797e18130a7019c039d4"
+dependencies = [
+ "nucleo-matcher",
+ "parking_lot",
+ "rayon",
+]
+
+[[package]]
+name = "nucleo-matcher"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "bf33f538733d1a5a3494b836ba913207f14d9d4a1d3cd67030c5061bdd2cac85"
+dependencies = [
+ "memchr",
+ "unicode-segmentation",
+]
+
 [[package]]
 name = "num-bigint"
 version = "0.4.6"
@@ -3094,7 +3117,7 @@ checksum = "1c107b6f4780854c8b126e228ea8869f4d7b71260f962fefb57b996b8959ba6b"
 
 [[package]]
 name = "search"
-version = "0.6.0"
+version = "0.6.7"
 dependencies = [
  "anyhow",
  "common",
@@ -3193,7 +3216,7 @@ dependencies = [
 
 [[package]]
 name = "server"
-version = "0.6.0"
+version = "0.6.7"
 dependencies = [
  "anyhow",
  "bcrypt",
@@ -3230,7 +3253,9 @@ dependencies = [
  "tonic-reflection",
  "tracing",
  "tracing-subscriber",
+ "unicode-width 0.2.0",
  "uuid",
+ "validation-core",
  "validator",
 ]
 
@@ -4522,6 +4547,16 @@ dependencies = [
  "wasm-bindgen",
 ]
 
+[[package]]
+name = "validation-core"
+version = "0.6.7"
+dependencies = [
+ "regex",
+ "serde",
+ "thiserror 2.0.12",
+ "unicode-width 0.2.0",
+]
+
 [[package]]
 name = "validator"
 version = "0.20.0"

Cargo.toml

@@ -1,11 +1,11 @@
 [workspace]
-members = ["client", "server", "common", "search", "canvas"]
+members = ["client", "server", "common", "search", "canvas", "validation-core"]
 resolver = "2"
 
 [workspace.package]
 # TODO: idk how to do the name, fix later
 # name = "komp_ac"
-version = "0.6.2"
+version = "0.6.7"
 edition = "2021"
 license = "GPL-3.0-or-later"
 authors = ["Filip Priečinský <filippriec@gmail.com>"]
@@ -53,3 +53,4 @@ toml = "0.8.20"
 unicode-width = "0.2.0"
 
 common = { path = "./common" }
+validation-core = { path = "./validation-core" }

common/build.rs

@@ -24,6 +24,14 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
             ".komp_ac.table_validation.PatternRule",
             "#[derive(serde::Serialize, serde::Deserialize)]",
         )
+        .type_attribute(
+            ".komp_ac.table_validation.PatternPosition",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.CharacterConstraint",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
         .type_attribute(
             ".komp_ac.table_validation.PatternRules",
             "#[derive(serde::Serialize, serde::Deserialize)]",
@@ -32,6 +40,14 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
             ".komp_ac.table_validation.CustomFormatter",
             "#[derive(serde::Serialize, serde::Deserialize)]",
         )
+        .type_attribute(
+            ".komp_ac.table_validation.FormatterOption",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.AllowedValues",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
         .type_attribute(
             ".komp_ac.table_validation.UpdateFieldValidationRequest",
             "#[derive(serde::Serialize, serde::Deserialize)]",
@@ -40,11 +56,91 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
             ".komp_ac.table_validation.UpdateFieldValidationResponse",
             "#[derive(serde::Serialize, serde::Deserialize)]",
         )
+        .type_attribute(
+            ".komp_ac.table_validation.ReplaceTableValidationRequest",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.ReplaceTableValidationResponse",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.ValidationRuleDefinition",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.ValidationSetDefinition",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.UpsertValidationRuleRequest",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.UpsertValidationRuleResponse",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.ListValidationRulesRequest",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.ListValidationRulesResponse",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.DeleteValidationRuleRequest",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.DeleteValidationRuleResponse",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.UpsertValidationSetRequest",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.UpsertValidationSetResponse",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.ListValidationSetsRequest",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.ListValidationSetsResponse",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.DeleteValidationSetRequest",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.DeleteValidationSetResponse",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.ApplyValidationSetRequest",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.ApplyValidationSetResponse",
+            "#[derive(serde::Serialize, serde::Deserialize)]",
+        )
         // Enum -> readable strings in JSON ("BYTES", "DISPLAY_WIDTH")
         .type_attribute(
             ".komp_ac.table_validation.CountMode",
             "#[derive(serde::Serialize, serde::Deserialize)] #[serde(rename_all = \"SCREAMING_SNAKE_CASE\")]",
         )
+        .type_attribute(
+            ".komp_ac.table_validation.PatternPositionKind",
+            "#[derive(serde::Serialize, serde::Deserialize)] #[serde(rename_all = \"SCREAMING_SNAKE_CASE\")]",
+        )
+        .type_attribute(
+            ".komp_ac.table_validation.CharacterConstraintKind",
+            "#[derive(serde::Serialize, serde::Deserialize)] #[serde(rename_all = \"SCREAMING_SNAKE_CASE\")]",
+        )
         .type_attribute(
             ".komp_ac.table_definition.ColumnDefinition",
             "#[derive(serde::Serialize, serde::Deserialize)]",
@@ -61,6 +157,26 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
             ".komp_ac.table_definition.TableDefinitionResponse",
             "#[derive(serde::Serialize, serde::Deserialize)]"
         )
+        .type_attribute(
+            ".komp_ac.table_definition.GetColumnAliasRenameHistoryRequest",
+            "#[derive(serde::Serialize, serde::Deserialize)]"
+        )
+        .type_attribute(
+            ".komp_ac.table_definition.ColumnAliasRenameHistoryEntry",
+            "#[derive(serde::Serialize, serde::Deserialize)]"
+        )
+        .type_attribute(
+            ".komp_ac.table_definition.GetColumnAliasRenameHistoryResponse",
+            "#[derive(serde::Serialize, serde::Deserialize)]"
+        )
+        .type_attribute(
+            ".komp_ac.table_definition.RenameColumnAliasRequest",
+            "#[derive(serde::Serialize, serde::Deserialize)]"
+        )
+        .type_attribute(
+            ".komp_ac.table_definition.RenameColumnAliasResponse",
+            "#[derive(serde::Serialize, serde::Deserialize)]"
+        )
         .type_attribute(
             ".komp_ac.table_script.PostTableScriptRequest",
             "#[derive(serde::Serialize, serde::Deserialize)]",

common/proto/search.proto

@@ -3,18 +3,34 @@ syntax = "proto3";
 package komp_ac.search;
 
 service Searcher {
-  rpc SearchTable(SearchRequest) returns (SearchResponse);
+  rpc Search(SearchRequest) returns (SearchResponse);
+}
+
+enum MatchMode {
+  MATCH_MODE_UNSPECIFIED = 0;
+  MATCH_MODE_FUZZY = 1;
+  MATCH_MODE_EXACT = 2;
+}
+
+message ColumnConstraint {
+  string column = 1;
+  string query = 2;
+  MatchMode mode = 3;
 }
 
 message SearchRequest {
-  string table_name = 1;
-  string query = 2;
+  string profile_name = 1;
+  optional string table_name = 2;
+  string free_query = 3;
+  repeated ColumnConstraint must = 4;
+  optional uint32 limit = 5;
 }
 message SearchResponse {
   message Hit {
     int64 id = 1; // PostgreSQL row ID
     float score = 2;
     string content_json = 3;
+    string table_name = 4;
   }
   repeated Hit hits = 1;
 }

common/proto/table_definition.proto

@@ -18,6 +18,16 @@ service TableDefinition {
   // This provides a tree-like overview of table relationships.
   rpc GetProfileTree(komp_ac.common.Empty) returns (ProfileTreeResponse);
 
+  // Fetches all tables with their columns and scripts for a specific profile.
+  // Pure data retrieval - no business logic.
+  rpc GetProfileDetails(GetProfileDetailsRequest) returns (GetProfileDetailsResponse);
+
+  // Returns the stored rename history for column aliases in one profile.
+  rpc GetColumnAliasRenameHistory(GetColumnAliasRenameHistoryRequest) returns (GetColumnAliasRenameHistoryResponse);
+
+  // Renames a user-visible column alias while keeping the physical column unchanged.
+  rpc RenameColumnAlias(RenameColumnAliasRequest) returns (RenameColumnAliasResponse);
+
   // Drops a table and its metadata, then deletes the profile if it becomes empty.
   rpc DeleteTable(DeleteTableRequest) returns (DeleteTableResponse);
 }
@@ -119,6 +129,74 @@ message ProfileTreeResponse {
   repeated Profile profiles = 1;
 }
 
+// Request to fetch all tables, columns and scripts for a profile.
+message GetProfileDetailsRequest {
+  // Profile (schema) name to fetch details for.
+  string profile_name = 1;
+}
+
+// Response with all tables, columns and scripts for a profile.
+message GetProfileDetailsResponse {
+  string profile_name = 1;
+  repeated TableDetail tables = 2;
+}
+
+// Request to fetch recorded column alias rename history for one profile.
+message GetColumnAliasRenameHistoryRequest {
+  string profile_name = 1;
+
+  // Optional filter. When omitted, returns all tables in the profile.
+  optional int64 table_definition_id = 2;
+}
+
+// One recorded column alias rename.
+message ColumnAliasRenameHistoryEntry {
+  int64 id = 1;
+  string profile_name = 2;
+  int64 table_definition_id = 3;
+  string table_name = 4;
+  string old_column_name = 5;
+  string new_column_name = 6;
+  string created_at = 7;
+}
+
+// Response with stored column alias rename history rows.
+message GetColumnAliasRenameHistoryResponse {
+  string profile_name = 1;
+  repeated ColumnAliasRenameHistoryEntry entries = 2;
+}
+
+// Describes a table with its columns and associated scripts.
+message TableDetail {
+  string name = 1;
+  int64 id = 2;
+  repeated ColumnDefinition columns = 3;
+  repeated ScriptInfo scripts = 4;
+}
+
+// A script that targets a specific column in a table.
+message ScriptInfo {
+  int64 script_id = 1;
+  string target_column = 2;
+  string target_column_type = 3;
+  string script = 4;
+  string description = 5;
+}
+
+// Request to rename one user-visible column alias in a table.
+message RenameColumnAliasRequest {
+  string profile_name = 1;
+  string table_name = 2;
+  string old_column_name = 3;
+  string new_column_name = 4;
+}
+
+// Response after renaming one column alias.
+message RenameColumnAliasResponse {
+  bool success = 1;
+  string message = 2;
+}
+
 // Request to delete one table definition entirely.
 message DeleteTableRequest {
   // Profile (schema) name owning the table (must exist).

common/proto/table_structure.proto

@@ -4,40 +4,45 @@ package komp_ac.table_structure;
 
 import "common.proto";
 
-// Introspects the physical PostgreSQL table for a given logical table
-// (defined in table_definitions) and returns its column structure.
+// Introspects the physical PostgreSQL tables for one or more logical tables
+// (defined in table_definitions) and returns their column structures.
 // The server validates that:
 // - The profile (schema) exists in `schemas`
-// - The table is defined for that profile in `table_definitions`
-// It then queries information_schema for the physical table and returns
-// normalized column metadata. If the physical table is missing despite
-// a definition, the response may contain an empty `columns` list.
+// - Every table is defined for that profile in `table_definitions`
+// It then queries information_schema for the physical tables and returns
+// normalized column metadata.
 service TableStructureService {
   // Return the physical column list (name, normalized data_type,
-  // nullability, primary key flag) for a table in a profile.
+  // nullability, primary key flag) for one or more tables in a profile.
   //
   // Behavior:
   // - NOT_FOUND if profile doesn't exist in `schemas`
-  // - NOT_FOUND if table not defined for that profile in `table_definitions`
+  // - NOT_FOUND if any table is not defined for that profile in `table_definitions`
   // - Queries information_schema.columns ordered by ordinal position
   // - Normalizes data_type text (details under TableColumn.data_type)
-  // - Returns an empty list if the table is validated but has no visible
-  //   columns in information_schema (e.g., physical table missing)
-  rpc GetTableStructure(GetTableStructureRequest) returns (TableStructureResponse);
+  // - Returns an error if any validated table has no visible columns in
+  //   information_schema (e.g., physical table missing)
+  rpc GetTableStructure(GetTableStructureRequest) returns (GetTableStructureResponse);
 }
 
-// Request identifying the profile (schema) and table to inspect.
+// Request identifying the profile (schema) and tables to inspect.
 message GetTableStructureRequest {
   // Required. Profile (PostgreSQL schema) name. Must exist in `schemas`.
   string profile_name = 1;
 
-  // Required. Table name within the profile. Must exist in `table_definitions`
-  // for the given profile. The physical table is then introspected via
-  // information_schema.
-  string table_name = 2;
+  // Required. Table names within the profile. Each must exist in
+  // `table_definitions` for the given profile. The physical tables are then
+  // introspected via information_schema.
+  repeated string table_names = 2;
 }
 
-// Response with the ordered list of columns (by ordinal position).
+// Batched response keyed by table name.
+message GetTableStructureResponse {
+  // Per-table physical column lists keyed by requested table name.
+  map<string, TableStructureResponse> table_structures = 1;
+}
+
+// Response with the ordered list of columns (by ordinal position) for one table.
 message TableStructureResponse {
   // Columns of the physical table, including system columns (id, deleted,
   // created_at), user-defined columns, and any foreign-key columns such as

common/proto/table_validation.proto

@@ -2,31 +2,55 @@
 syntax = "proto3";
 package komp_ac.table_validation;
 
+// This proto is the canonical server-side storage and distribution contract for
+// client validation configuration.
+//
+// Design goals:
+// - The server stores the entire field validation definition in one structured payload.
+// - Clients fetch the validation rules for a table in one batch and map them to
+//   their local validation/runtime system (for example canvas).
+// - Common validation must be represented as typed data, not as string mini-languages.
+//
+// Important split:
+// - limits / pattern / allowed_values / required are validation rules.
+// - mask / formatter are presentation and input-shaping metadata for clients.
+
 // Request validation rules for a table
 message GetTableValidationRequest {
   string profileName = 1;
   string tableName = 2;
 }
 
-// Response with field-level validations; if a field is omitted,
-// no validation is applied (default unspecified).
+// Response with field-level validations for the whole table.
+// If a field is omitted, no validation configuration exists for that field.
 message TableValidationResponse {
   repeated FieldValidation fields = 1;
 }
 
-// Field-level validation (extensible for future kinds)
+// Field-level validation definition stored on the server and distributed to clients.
 message FieldValidation {
   // MUST match your frontend FormState.dataKey for the column
   string dataKey = 1;
 
-  // Current: only CharacterLimits. More rules can be added later.
+  // Validation 1: length and counting rules.
   CharacterLimits limits = 10;
-  // Future expansion:
-  PatternRules pattern = 11; // Validation 2
-  optional CustomFormatter formatter = 14; // Validation 4 – custom formatting logic
+
+  // Validation 2: position-based character constraints.
+  PatternRules pattern = 11;
+
+  // Exact-value whitelist.
+  AllowedValues allowed_values = 12;
+
+  // Client-side hint that this field participates in external/asynchronous validation UI.
+  bool external_validation_enabled = 13;
+
+  // Client-side formatter metadata. This is intentionally data-only, not executable code.
+  optional CustomFormatter formatter = 14;
+
+  // Client-side display mask metadata. The server stores raw data without mask literals.
   DisplayMask mask = 3;
-  // ExternalValidation external = 13;
-  // CustomFormatter formatter = 14;
+
+  // Field must be provided / treated as required by clients and server enforcement layers.
   bool required = 4;
 }
 
@@ -38,7 +62,8 @@ enum CountMode {
   DISPLAY_WIDTH = 3;
 }
 
-// Character limit validation (Validation 1)
+// Character limit validation (Validation 1).
+// These rules map directly to canvas CharacterLimits.
 message CharacterLimits {
   // When zero, the field is considered "not set". If both min/max are zero,
   // the server should avoid sending this FieldValidation (no validation).
@@ -51,39 +76,91 @@ message CharacterLimits {
   CountMode countMode = 4; // defaults to CHARS if unspecified
 }
 
-// Mask for pretty display
+// Mask for pretty display only.
+//
+// This is not a validation rule by itself. It exists so clients can render and
+// navigate masked input while still storing raw values server-side.
 message DisplayMask {
   string pattern = 1; // e.g., "(###) ###-####" or "####-##-##"
   string input_char = 2; // e.g., "#"
   optional string template_char = 3; // e.g., "_"
 }
 
-// One position‑based validation rule, similar to CharacterFilter + PositionRange
-message PatternRule {
-  // Range descriptor: how far the rule applies
-  // Examples:
-  //  - "0"        → Single position 0
-  //  - "0-3"      → Range 0..3 inclusive
-  //  - "from:5"   → From position 5 onward
-  //  - "0,2,5"    → Multiple discrete positions
-  string range = 1;
-
-  // Character filter type, case‑insensitive keywords:
-  // "ALPHABETIC", "NUMERIC", "ALPHANUMERIC",
-  // "ONEOF(<chars>)", "EXACT(:)", "CUSTOM(<name>)"
-  string filter = 2;
+// Which positions a pattern rule applies to.
+// This exists instead of a string syntax like "0-3" so the server can validate
+// the structure directly and clients do not need to parse a DSL.
+message PatternPosition {
+  PatternPositionKind kind = 1;
+  uint32 single = 2;
+  uint32 start = 3;
+  uint32 end = 4;
+  repeated uint32 positions = 5;
 }
 
+enum PatternPositionKind {
+  PATTERN_POSITION_KIND_UNSPECIFIED = 0;
+  PATTERN_POSITION_SINGLE = 1;
+  PATTERN_POSITION_RANGE = 2;
+  PATTERN_POSITION_FROM = 3;
+  PATTERN_POSITION_MULTIPLE = 4;
+}
+
+// What type of character constraint a pattern rule applies.
+// This mirrors the typed character filters used by canvas.
+message CharacterConstraint {
+  CharacterConstraintKind kind = 1;
+
+  // Used when kind == CHARACTER_CONSTRAINT_EXACT.
+  optional string exact = 2;
+
+  // Used when kind == CHARACTER_CONSTRAINT_ONE_OF.
+  repeated string one_of = 3;
+
+  // Used when kind == CHARACTER_CONSTRAINT_REGEX.
+  optional string regex = 4;
+}
+
+enum CharacterConstraintKind {
+  CHARACTER_CONSTRAINT_KIND_UNSPECIFIED = 0;
+  CHARACTER_CONSTRAINT_ALPHABETIC = 1;
+  CHARACTER_CONSTRAINT_NUMERIC = 2;
+  CHARACTER_CONSTRAINT_ALPHANUMERIC = 3;
+  CHARACTER_CONSTRAINT_EXACT = 4;
+  CHARACTER_CONSTRAINT_ONE_OF = 5;
+  CHARACTER_CONSTRAINT_REGEX = 6;
+}
+
+// One position-based validation rule, similar to canvas PositionFilter.
+message PatternRule {
+  PatternPosition position = 1;
+  CharacterConstraint constraint = 2;
+}
+
+// Client-side formatter metadata.
+// The formatter "type" is intended to be resolved by a client-side formatter registry.
 message CustomFormatter {
   // Formatter type identifier; handled client‑side.
   // Examples: "PSCFormatter", "PhoneFormatter", "CreditCardFormatter", "DateFormatter"
   string type = 1;
 
-  // Optional free‑text note or parameters (e.g. locale, pattern)
-  optional string description = 2;
+  repeated FormatterOption options = 2;
+  optional string description = 3;
 }
 
-// Collection of pattern rules for one field
+message FormatterOption {
+  string key = 1;
+  string value = 2;
+}
+
+// Exact-value whitelist configuration.
+// This maps to canvas AllowedValues semantics.
+message AllowedValues {
+  repeated string values = 1;
+  bool allow_empty = 2;
+  bool case_insensitive = 3;
+}
+
+// Collection of pattern rules for one field.
 message PatternRules {
   // All rules that make up the validation logic
   repeated PatternRule rules = 1;
@@ -92,11 +169,28 @@ message PatternRules {
   optional string description = 2;
 }
 
-// Service to fetch validations for a table
+// Service for storing and fetching field-validation definitions.
 service TableValidationService {
   rpc GetTableValidation(GetTableValidationRequest) returns (TableValidationResponse);
 
+  // Upsert a single field validation definition.
   rpc UpdateFieldValidation(UpdateFieldValidationRequest) returns (UpdateFieldValidationResponse);
+
+  // Replace the full validation definition set for a table in one transaction.
+  rpc ReplaceTableValidation(ReplaceTableValidationRequest) returns (ReplaceTableValidationResponse);
+
+  // Reusable named rule fragments.
+  rpc UpsertValidationRule(UpsertValidationRuleRequest) returns (UpsertValidationRuleResponse);
+  rpc ListValidationRules(ListValidationRulesRequest) returns (ListValidationRulesResponse);
+  rpc DeleteValidationRule(DeleteValidationRuleRequest) returns (DeleteValidationRuleResponse);
+
+  // Reusable named sets composed from rules.
+  rpc UpsertValidationSet(UpsertValidationSetRequest) returns (UpsertValidationSetResponse);
+  rpc ListValidationSets(ListValidationSetsRequest) returns (ListValidationSetsResponse);
+  rpc DeleteValidationSet(DeleteValidationSetRequest) returns (DeleteValidationSetResponse);
+
+  // Snapshot a reusable set onto a concrete table field.
+  rpc ApplyValidationSet(ApplyValidationSetRequest) returns (ApplyValidationSetResponse);
 }
 
 message UpdateFieldValidationRequest {
@@ -110,3 +204,102 @@ message UpdateFieldValidationResponse {
   bool success = 1;
   string message = 2;
 }
+
+message ReplaceTableValidationRequest {
+  string profileName = 1;
+  string tableName = 2;
+
+  // Full replacement set. Fields omitted here are removed from the stored config.
+  repeated FieldValidation fields = 3;
+}
+
+message ReplaceTableValidationResponse {
+  bool success = 1;
+  string message = 2;
+}
+
+message ValidationRuleDefinition {
+  string name = 1;
+  optional string description = 2;
+
+  // Reusable rule fragment. dataKey is ignored by the server for reusable rules.
+  FieldValidation validation = 3;
+}
+
+message ValidationSetDefinition {
+  string name = 1;
+  optional string description = 2;
+  repeated string ruleNames = 3;
+
+  // Server-resolved snapshot of all rules in ruleNames order.
+  FieldValidation resolvedValidation = 4;
+}
+
+message UpsertValidationRuleRequest {
+  string profileName = 1;
+  ValidationRuleDefinition rule = 2;
+}
+
+message UpsertValidationRuleResponse {
+  bool success = 1;
+  string message = 2;
+}
+
+message ListValidationRulesRequest {
+  string profileName = 1;
+}
+
+message ListValidationRulesResponse {
+  repeated ValidationRuleDefinition rules = 1;
+}
+
+message DeleteValidationRuleRequest {
+  string profileName = 1;
+  string name = 2;
+}
+
+message DeleteValidationRuleResponse {
+  bool success = 1;
+  string message = 2;
+}
+
+message UpsertValidationSetRequest {
+  string profileName = 1;
+  ValidationSetDefinition set = 2;
+}
+
+message UpsertValidationSetResponse {
+  bool success = 1;
+  string message = 2;
+}
+
+message ListValidationSetsRequest {
+  string profileName = 1;
+}
+
+message ListValidationSetsResponse {
+  repeated ValidationSetDefinition sets = 1;
+}
+
+message DeleteValidationSetRequest {
+  string profileName = 1;
+  string name = 2;
+}
+
+message DeleteValidationSetResponse {
+  bool success = 1;
+  string message = 2;
+}
+
+message ApplyValidationSetRequest {
+  string profileName = 1;
+  string tableName = 2;
+  string dataKey = 3;
+  string setName = 4;
+}
+
+message ApplyValidationSetResponse {
+  bool success = 1;
+  string message = 2;
+  FieldValidation validation = 3;
+}

common/src/proto/komp_ac.search.rs

@@ -1,10 +1,25 @@
 // This file is @generated by prost-build.
 #[derive(Clone, PartialEq, ::prost::Message)]
+pub struct ColumnConstraint {
+    #[prost(string, tag = "1")]
+    pub column: ::prost::alloc::string::String,
+    #[prost(string, tag = "2")]
+    pub query: ::prost::alloc::string::String,
+    #[prost(enumeration = "MatchMode", tag = "3")]
+    pub mode: i32,
+}
+#[derive(Clone, PartialEq, ::prost::Message)]
 pub struct SearchRequest {
     #[prost(string, tag = "1")]
-    pub table_name: ::prost::alloc::string::String,
-    #[prost(string, tag = "2")]
-    pub query: ::prost::alloc::string::String,
+    pub profile_name: ::prost::alloc::string::String,
+    #[prost(string, optional, tag = "2")]
+    pub table_name: ::core::option::Option<::prost::alloc::string::String>,
+    #[prost(string, tag = "3")]
+    pub free_query: ::prost::alloc::string::String,
+    #[prost(message, repeated, tag = "4")]
+    pub must: ::prost::alloc::vec::Vec<ColumnConstraint>,
+    #[prost(uint32, optional, tag = "5")]
+    pub limit: ::core::option::Option<u32>,
 }
 #[derive(Clone, PartialEq, ::prost::Message)]
 pub struct SearchResponse {
@@ -22,6 +37,37 @@ pub mod search_response {
         pub score: f32,
         #[prost(string, tag = "3")]
         pub content_json: ::prost::alloc::string::String,
+        #[prost(string, tag = "4")]
+        pub table_name: ::prost::alloc::string::String,
+    }
+}
+#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash, PartialOrd, Ord, ::prost::Enumeration)]
+#[repr(i32)]
+pub enum MatchMode {
+    Unspecified = 0,
+    Fuzzy = 1,
+    Exact = 2,
+}
+impl MatchMode {
+    /// String value of the enum field names used in the ProtoBuf definition.
+    ///
+    /// The values are not transformed in any way and thus are considered stable
+    /// (if the ProtoBuf definition does not change) and safe for programmatic use.
+    pub fn as_str_name(&self) -> &'static str {
+        match self {
+            Self::Unspecified => "MATCH_MODE_UNSPECIFIED",
+            Self::Fuzzy => "MATCH_MODE_FUZZY",
+            Self::Exact => "MATCH_MODE_EXACT",
+        }
+    }
+    /// Creates an enum from field names used in the ProtoBuf definition.
+    pub fn from_str_name(value: &str) -> ::core::option::Option<Self> {
+        match value {
+            "MATCH_MODE_UNSPECIFIED" => Some(Self::Unspecified),
+            "MATCH_MODE_FUZZY" => Some(Self::Fuzzy),
+            "MATCH_MODE_EXACT" => Some(Self::Exact),
+            _ => None,
+        }
     }
 }
 /// Generated client implementations.
@@ -115,7 +161,7 @@ pub mod searcher_client {
             self.inner = self.inner.max_encoding_message_size(limit);
             self
         }
-        pub async fn search_table(
+        pub async fn search(
             &mut self,
             request: impl tonic::IntoRequest<super::SearchRequest>,
         ) -> std::result::Result<tonic::Response<super::SearchResponse>, tonic::Status> {
@@ -129,11 +175,11 @@ pub mod searcher_client {
                 })?;
             let codec = tonic::codec::ProstCodec::default();
             let path = http::uri::PathAndQuery::from_static(
-                "/komp_ac.search.Searcher/SearchTable",
+                "/komp_ac.search.Searcher/Search",
             );
             let mut req = request.into_request();
             req.extensions_mut()
-                .insert(GrpcMethod::new("komp_ac.search.Searcher", "SearchTable"));
+                .insert(GrpcMethod::new("komp_ac.search.Searcher", "Search"));
             self.inner.unary(req, path, codec).await
         }
     }
@@ -151,7 +197,7 @@ pub mod searcher_server {
     /// Generated trait containing gRPC methods that should be implemented for use with SearcherServer.
     #[async_trait]
     pub trait Searcher: std::marker::Send + std::marker::Sync + 'static {
-        async fn search_table(
+        async fn search(
             &self,
             request: tonic::Request<super::SearchRequest>,
         ) -> std::result::Result<tonic::Response<super::SearchResponse>, tonic::Status>;
@@ -232,11 +278,11 @@ pub mod searcher_server {
         }
         fn call(&mut self, req: http::Request<B>) -> Self::Future {
             match req.uri().path() {
-                "/komp_ac.search.Searcher/SearchTable" => {
+                "/komp_ac.search.Searcher/Search" => {
                     #[allow(non_camel_case_types)]
-                    struct SearchTableSvc<T: Searcher>(pub Arc<T>);
+                    struct SearchSvc<T: Searcher>(pub Arc<T>);
                     impl<T: Searcher> tonic::server::UnaryService<super::SearchRequest>
-                    for SearchTableSvc<T> {
+                    for SearchSvc<T> {
                         type Response = super::SearchResponse;
                         type Future = BoxFuture<
                             tonic::Response<Self::Response>,
@@ -248,7 +294,7 @@ pub mod searcher_server {
                         ) -> Self::Future {
                             let inner = Arc::clone(&self.0);
                             let fut = async move {
-                                <T as Searcher>::search_table(&inner, request).await
+                                <T as Searcher>::search(&inner, request).await
                             };
                             Box::pin(fut)
                         }
@@ -259,7 +305,7 @@ pub mod searcher_server {
                     let max_encoding_message_size = self.max_encoding_message_size;
                     let inner = self.inner.clone();
                     let fut = async move {
-                        let method = SearchTableSvc(inner);
+                        let method = SearchSvc(inner);
                         let codec = tonic::codec::ProstCodec::default();
                         let mut grpc = tonic::server::Grpc::new(codec)
                             .apply_compression_config(

common/src/proto/komp_ac.table_definition.rs

@@ -110,6 +110,107 @@ pub mod profile_tree_response {
         pub tables: ::prost::alloc::vec::Vec<Table>,
     }
 }
+/// Request to fetch all tables, columns and scripts for a profile.
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct GetProfileDetailsRequest {
+    /// Profile (schema) name to fetch details for.
+    #[prost(string, tag = "1")]
+    pub profile_name: ::prost::alloc::string::String,
+}
+/// Response with all tables, columns and scripts for a profile.
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct GetProfileDetailsResponse {
+    #[prost(string, tag = "1")]
+    pub profile_name: ::prost::alloc::string::String,
+    #[prost(message, repeated, tag = "2")]
+    pub tables: ::prost::alloc::vec::Vec<TableDetail>,
+}
+/// Request to fetch recorded column alias rename history for one profile.
+#[derive(serde::Serialize, serde::Deserialize)]
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct GetColumnAliasRenameHistoryRequest {
+    #[prost(string, tag = "1")]
+    pub profile_name: ::prost::alloc::string::String,
+    /// Optional filter. When omitted, returns all tables in the profile.
+    #[prost(int64, optional, tag = "2")]
+    pub table_definition_id: ::core::option::Option<i64>,
+}
+/// One recorded column alias rename.
+#[derive(serde::Serialize, serde::Deserialize)]
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct ColumnAliasRenameHistoryEntry {
+    #[prost(int64, tag = "1")]
+    pub id: i64,
+    #[prost(string, tag = "2")]
+    pub profile_name: ::prost::alloc::string::String,
+    #[prost(int64, tag = "3")]
+    pub table_definition_id: i64,
+    #[prost(string, tag = "4")]
+    pub table_name: ::prost::alloc::string::String,
+    #[prost(string, tag = "5")]
+    pub old_column_name: ::prost::alloc::string::String,
+    #[prost(string, tag = "6")]
+    pub new_column_name: ::prost::alloc::string::String,
+    #[prost(string, tag = "7")]
+    pub created_at: ::prost::alloc::string::String,
+}
+/// Response with stored column alias rename history rows.
+#[derive(serde::Serialize, serde::Deserialize)]
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct GetColumnAliasRenameHistoryResponse {
+    #[prost(string, tag = "1")]
+    pub profile_name: ::prost::alloc::string::String,
+    #[prost(message, repeated, tag = "2")]
+    pub entries: ::prost::alloc::vec::Vec<ColumnAliasRenameHistoryEntry>,
+}
+/// Describes a table with its columns and associated scripts.
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct TableDetail {
+    #[prost(string, tag = "1")]
+    pub name: ::prost::alloc::string::String,
+    #[prost(int64, tag = "2")]
+    pub id: i64,
+    #[prost(message, repeated, tag = "3")]
+    pub columns: ::prost::alloc::vec::Vec<ColumnDefinition>,
+    #[prost(message, repeated, tag = "4")]
+    pub scripts: ::prost::alloc::vec::Vec<ScriptInfo>,
+}
+/// A script that targets a specific column in a table.
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct ScriptInfo {
+    #[prost(int64, tag = "1")]
+    pub script_id: i64,
+    #[prost(string, tag = "2")]
+    pub target_column: ::prost::alloc::string::String,
+    #[prost(string, tag = "3")]
+    pub target_column_type: ::prost::alloc::string::String,
+    #[prost(string, tag = "4")]
+    pub script: ::prost::alloc::string::String,
+    #[prost(string, tag = "5")]
+    pub description: ::prost::alloc::string::String,
+}
+/// Request to rename one user-visible column alias in a table.
+#[derive(serde::Serialize, serde::Deserialize)]
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct RenameColumnAliasRequest {
+    #[prost(string, tag = "1")]
+    pub profile_name: ::prost::alloc::string::String,
+    #[prost(string, tag = "2")]
+    pub table_name: ::prost::alloc::string::String,
+    #[prost(string, tag = "3")]
+    pub old_column_name: ::prost::alloc::string::String,
+    #[prost(string, tag = "4")]
+    pub new_column_name: ::prost::alloc::string::String,
+}
+/// Response after renaming one column alias.
+#[derive(serde::Serialize, serde::Deserialize)]
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct RenameColumnAliasResponse {
+    #[prost(bool, tag = "1")]
+    pub success: bool,
+    #[prost(string, tag = "2")]
+    pub message: ::prost::alloc::string::String,
+}
 /// Request to delete one table definition entirely.
 #[derive(Clone, PartialEq, ::prost::Message)]
 pub struct DeleteTableRequest {
@@ -289,6 +390,97 @@ pub mod table_definition_client {
                 );
             self.inner.unary(req, path, codec).await
         }
+        /// Fetches all tables with their columns and scripts for a specific profile.
+        /// Pure data retrieval - no business logic.
+        pub async fn get_profile_details(
+            &mut self,
+            request: impl tonic::IntoRequest<super::GetProfileDetailsRequest>,
+        ) -> std::result::Result<
+            tonic::Response<super::GetProfileDetailsResponse>,
+            tonic::Status,
+        > {
+            self.inner
+                .ready()
+                .await
+                .map_err(|e| {
+                    tonic::Status::unknown(
+                        format!("Service was not ready: {}", e.into()),
+                    )
+                })?;
+            let codec = tonic::codec::ProstCodec::default();
+            let path = http::uri::PathAndQuery::from_static(
+                "/komp_ac.table_definition.TableDefinition/GetProfileDetails",
+            );
+            let mut req = request.into_request();
+            req.extensions_mut()
+                .insert(
+                    GrpcMethod::new(
+                        "komp_ac.table_definition.TableDefinition",
+                        "GetProfileDetails",
+                    ),
+                );
+            self.inner.unary(req, path, codec).await
+        }
+        /// Returns the stored rename history for column aliases in one profile.
+        pub async fn get_column_alias_rename_history(
+            &mut self,
+            request: impl tonic::IntoRequest<super::GetColumnAliasRenameHistoryRequest>,
+        ) -> std::result::Result<
+            tonic::Response<super::GetColumnAliasRenameHistoryResponse>,
+            tonic::Status,
+        > {
+            self.inner
+                .ready()
+                .await
+                .map_err(|e| {
+                    tonic::Status::unknown(
+                        format!("Service was not ready: {}", e.into()),
+                    )
+                })?;
+            let codec = tonic::codec::ProstCodec::default();
+            let path = http::uri::PathAndQuery::from_static(
+                "/komp_ac.table_definition.TableDefinition/GetColumnAliasRenameHistory",
+            );
+            let mut req = request.into_request();
+            req.extensions_mut()
+                .insert(
+                    GrpcMethod::new(
+                        "komp_ac.table_definition.TableDefinition",
+                        "GetColumnAliasRenameHistory",
+                    ),
+                );
+            self.inner.unary(req, path, codec).await
+        }
+        /// Renames a user-visible column alias while keeping the physical column unchanged.
+        pub async fn rename_column_alias(
+            &mut self,
+            request: impl tonic::IntoRequest<super::RenameColumnAliasRequest>,
+        ) -> std::result::Result<
+            tonic::Response<super::RenameColumnAliasResponse>,
+            tonic::Status,
+        > {
+            self.inner
+                .ready()
+                .await
+                .map_err(|e| {
+                    tonic::Status::unknown(
+                        format!("Service was not ready: {}", e.into()),
+                    )
+                })?;
+            let codec = tonic::codec::ProstCodec::default();
+            let path = http::uri::PathAndQuery::from_static(
+                "/komp_ac.table_definition.TableDefinition/RenameColumnAlias",
+            );
+            let mut req = request.into_request();
+            req.extensions_mut()
+                .insert(
+                    GrpcMethod::new(
+                        "komp_ac.table_definition.TableDefinition",
+                        "RenameColumnAlias",
+                    ),
+                );
+            self.inner.unary(req, path, codec).await
+        }
         /// Drops a table and its metadata, then deletes the profile if it becomes empty.
         pub async fn delete_table(
             &mut self,
@@ -353,6 +545,31 @@ pub mod table_definition_server {
             tonic::Response<super::ProfileTreeResponse>,
             tonic::Status,
         >;
+        /// Fetches all tables with their columns and scripts for a specific profile.
+        /// Pure data retrieval - no business logic.
+        async fn get_profile_details(
+            &self,
+            request: tonic::Request<super::GetProfileDetailsRequest>,
+        ) -> std::result::Result<
+            tonic::Response<super::GetProfileDetailsResponse>,
+            tonic::Status,
+        >;
+        /// Returns the stored rename history for column aliases in one profile.
+        async fn get_column_alias_rename_history(
+            &self,
+            request: tonic::Request<super::GetColumnAliasRenameHistoryRequest>,
+        ) -> std::result::Result<
+            tonic::Response<super::GetColumnAliasRenameHistoryResponse>,
+            tonic::Status,
+        >;
+        /// Renames a user-visible column alias while keeping the physical column unchanged.
+        async fn rename_column_alias(
+            &self,
+            request: tonic::Request<super::RenameColumnAliasRequest>,
+        ) -> std::result::Result<
+            tonic::Response<super::RenameColumnAliasResponse>,
+            tonic::Status,
+        >;
         /// Drops a table and its metadata, then deletes the profile if it becomes empty.
         async fn delete_table(
             &self,
@@ -537,6 +754,152 @@ pub mod table_definition_server {
                     };
                     Box::pin(fut)
                 }
+                "/komp_ac.table_definition.TableDefinition/GetProfileDetails" => {
+                    #[allow(non_camel_case_types)]
+                    struct GetProfileDetailsSvc<T: TableDefinition>(pub Arc<T>);
+                    impl<
+                        T: TableDefinition,
+                    > tonic::server::UnaryService<super::GetProfileDetailsRequest>
+                    for GetProfileDetailsSvc<T> {
+                        type Response = super::GetProfileDetailsResponse;
+                        type Future = BoxFuture<
+                            tonic::Response<Self::Response>,
+                            tonic::Status,
+                        >;
+                        fn call(
+                            &mut self,
+                            request: tonic::Request<super::GetProfileDetailsRequest>,
+                        ) -> Self::Future {
+                            let inner = Arc::clone(&self.0);
+                            let fut = async move {
+                                <T as TableDefinition>::get_profile_details(&inner, request)
+                                    .await
+                            };
+                            Box::pin(fut)
+                        }
+                    }
+                    let accept_compression_encodings = self.accept_compression_encodings;
+                    let send_compression_encodings = self.send_compression_encodings;
+                    let max_decoding_message_size = self.max_decoding_message_size;
+                    let max_encoding_message_size = self.max_encoding_message_size;
+                    let inner = self.inner.clone();
+                    let fut = async move {
+                        let method = GetProfileDetailsSvc(inner);
+                        let codec = tonic::codec::ProstCodec::default();
+                        let mut grpc = tonic::server::Grpc::new(codec)
+                            .apply_compression_config(
+                                accept_compression_encodings,
+                                send_compression_encodings,
+                            )
+                            .apply_max_message_size_config(
+                                max_decoding_message_size,
+                                max_encoding_message_size,
+                            );
+                        let res = grpc.unary(method, req).await;
+                        Ok(res)
+                    };
+                    Box::pin(fut)
+                }
+                "/komp_ac.table_definition.TableDefinition/GetColumnAliasRenameHistory" => {
+                    #[allow(non_camel_case_types)]
+                    struct GetColumnAliasRenameHistorySvc<T: TableDefinition>(
+                        pub Arc<T>,
+                    );
+                    impl<
+                        T: TableDefinition,
+                    > tonic::server::UnaryService<
+                        super::GetColumnAliasRenameHistoryRequest,
+                    > for GetColumnAliasRenameHistorySvc<T> {
+                        type Response = super::GetColumnAliasRenameHistoryResponse;
+                        type Future = BoxFuture<
+                            tonic::Response<Self::Response>,
+                            tonic::Status,
+                        >;
+                        fn call(
+                            &mut self,
+                            request: tonic::Request<
+                                super::GetColumnAliasRenameHistoryRequest,
+                            >,
+                        ) -> Self::Future {
+                            let inner = Arc::clone(&self.0);
+                            let fut = async move {
+                                <T as TableDefinition>::get_column_alias_rename_history(
+                                        &inner,
+                                        request,
+                                    )
+                                    .await
+                            };
+                            Box::pin(fut)
+                        }
+                    }
+                    let accept_compression_encodings = self.accept_compression_encodings;
+                    let send_compression_encodings = self.send_compression_encodings;
+                    let max_decoding_message_size = self.max_decoding_message_size;
+                    let max_encoding_message_size = self.max_encoding_message_size;
+                    let inner = self.inner.clone();
+                    let fut = async move {
+                        let method = GetColumnAliasRenameHistorySvc(inner);
+                        let codec = tonic::codec::ProstCodec::default();
+                        let mut grpc = tonic::server::Grpc::new(codec)
+                            .apply_compression_config(
+                                accept_compression_encodings,
+                                send_compression_encodings,
+                            )
+                            .apply_max_message_size_config(
+                                max_decoding_message_size,
+                                max_encoding_message_size,
+                            );
+                        let res = grpc.unary(method, req).await;
+                        Ok(res)
+                    };
+                    Box::pin(fut)
+                }
+                "/komp_ac.table_definition.TableDefinition/RenameColumnAlias" => {
+                    #[allow(non_camel_case_types)]
+                    struct RenameColumnAliasSvc<T: TableDefinition>(pub Arc<T>);
+                    impl<
+                        T: TableDefinition,
+                    > tonic::server::UnaryService<super::RenameColumnAliasRequest>
+                    for RenameColumnAliasSvc<T> {
+                        type Response = super::RenameColumnAliasResponse;
+                        type Future = BoxFuture<
+                            tonic::Response<Self::Response>,
+                            tonic::Status,
+                        >;
+                        fn call(
+                            &mut self,
+                            request: tonic::Request<super::RenameColumnAliasRequest>,
+                        ) -> Self::Future {
+                            let inner = Arc::clone(&self.0);
+                            let fut = async move {
+                                <T as TableDefinition>::rename_column_alias(&inner, request)
+                                    .await
+                            };
+                            Box::pin(fut)
+                        }
+                    }
+                    let accept_compression_encodings = self.accept_compression_encodings;
+                    let send_compression_encodings = self.send_compression_encodings;
+                    let max_decoding_message_size = self.max_decoding_message_size;
+                    let max_encoding_message_size = self.max_encoding_message_size;
+                    let inner = self.inner.clone();
+                    let fut = async move {
+                        let method = RenameColumnAliasSvc(inner);
+                        let codec = tonic::codec::ProstCodec::default();
+                        let mut grpc = tonic::server::Grpc::new(codec)
+                            .apply_compression_config(
+                                accept_compression_encodings,
+                                send_compression_encodings,
+                            )
+                            .apply_max_message_size_config(
+                                max_decoding_message_size,
+                                max_encoding_message_size,
+                            );
+                        let res = grpc.unary(method, req).await;
+                        Ok(res)
+                    };
+                    Box::pin(fut)
+                }
                 "/komp_ac.table_definition.TableDefinition/DeleteTable" => {
                     #[allow(non_camel_case_types)]
                     struct DeleteTableSvc<T: TableDefinition>(pub Arc<T>);

common/src/proto/komp_ac.table_structure.rs

@@ -1,17 +1,27 @@
 // This file is @generated by prost-build.
-/// Request identifying the profile (schema) and table to inspect.
+/// Request identifying the profile (schema) and tables to inspect.
 #[derive(Clone, PartialEq, ::prost::Message)]
 pub struct GetTableStructureRequest {
     /// Required. Profile (PostgreSQL schema) name. Must exist in `schemas`.
     #[prost(string, tag = "1")]
     pub profile_name: ::prost::alloc::string::String,
-    /// Required. Table name within the profile. Must exist in `table_definitions`
-    /// for the given profile. The physical table is then introspected via
-    /// information_schema.
-    #[prost(string, tag = "2")]
-    pub table_name: ::prost::alloc::string::String,
+    /// Required. Table names within the profile. Each must exist in
+    /// `table_definitions` for the given profile. The physical tables are then
+    /// introspected via information_schema.
+    #[prost(string, repeated, tag = "2")]
+    pub table_names: ::prost::alloc::vec::Vec<::prost::alloc::string::String>,
 }
-/// Response with the ordered list of columns (by ordinal position).
+/// Batched response keyed by table name.
+#[derive(Clone, PartialEq, ::prost::Message)]
+pub struct GetTableStructureResponse {
+    /// Per-table physical column lists keyed by requested table name.
+    #[prost(map = "string, message", tag = "1")]
+    pub table_structures: ::std::collections::HashMap<
+        ::prost::alloc::string::String,
+        TableStructureResponse,
+    >,
+}
+/// Response with the ordered list of columns (by ordinal position) for one table.
 #[derive(Clone, PartialEq, ::prost::Message)]
 pub struct TableStructureResponse {
     /// Columns of the physical table, including system columns (id, deleted,
@@ -55,14 +65,13 @@ pub mod table_structure_service_client {
     )]
     use tonic::codegen::*;
     use tonic::codegen::http::Uri;
-    /// Introspects the physical PostgreSQL table for a given logical table
-    /// (defined in table_definitions) and returns its column structure.
+    /// Introspects the physical PostgreSQL tables for one or more logical tables
+    /// (defined in table_definitions) and returns their column structures.
     /// The server validates that:
     /// - The profile (schema) exists in `schemas`
-    /// - The table is defined for that profile in `table_definitions`
-    /// It then queries information_schema for the physical table and returns
-    /// normalized column metadata. If the physical table is missing despite
-    /// a definition, the response may contain an empty `columns` list.
+    /// - Every table is defined for that profile in `table_definitions`
+    /// It then queries information_schema for the physical tables and returns
+    /// normalized column metadata.
     #[derive(Debug, Clone)]
     pub struct TableStructureServiceClient<T> {
         inner: tonic::client::Grpc<T>,
@@ -144,20 +153,20 @@ pub mod table_structure_service_client {
             self
         }
         /// Return the physical column list (name, normalized data_type,
-        /// nullability, primary key flag) for a table in a profile.
+        /// nullability, primary key flag) for one or more tables in a profile.
         ///
         /// Behavior:
         /// - NOT_FOUND if profile doesn't exist in `schemas`
-        /// - NOT_FOUND if table not defined for that profile in `table_definitions`
+        /// - NOT_FOUND if any table is not defined for that profile in `table_definitions`
         /// - Queries information_schema.columns ordered by ordinal position
         /// - Normalizes data_type text (details under TableColumn.data_type)
-        /// - Returns an empty list if the table is validated but has no visible
-        ///   columns in information_schema (e.g., physical table missing)
+        /// - Returns an error if any validated table has no visible columns in
+        ///   information_schema (e.g., physical table missing)
         pub async fn get_table_structure(
             &mut self,
             request: impl tonic::IntoRequest<super::GetTableStructureRequest>,
         ) -> std::result::Result<
-            tonic::Response<super::TableStructureResponse>,
+            tonic::Response<super::GetTableStructureResponse>,
             tonic::Status,
         > {
             self.inner
@@ -198,31 +207,30 @@ pub mod table_structure_service_server {
     #[async_trait]
     pub trait TableStructureService: std::marker::Send + std::marker::Sync + 'static {
         /// Return the physical column list (name, normalized data_type,
-        /// nullability, primary key flag) for a table in a profile.
+        /// nullability, primary key flag) for one or more tables in a profile.
         ///
         /// Behavior:
         /// - NOT_FOUND if profile doesn't exist in `schemas`
-        /// - NOT_FOUND if table not defined for that profile in `table_definitions`
+        /// - NOT_FOUND if any table is not defined for that profile in `table_definitions`
         /// - Queries information_schema.columns ordered by ordinal position
         /// - Normalizes data_type text (details under TableColumn.data_type)
-        /// - Returns an empty list if the table is validated but has no visible
-        ///   columns in information_schema (e.g., physical table missing)
+        /// - Returns an error if any validated table has no visible columns in
+        ///   information_schema (e.g., physical table missing)
         async fn get_table_structure(
             &self,
             request: tonic::Request<super::GetTableStructureRequest>,
         ) -> std::result::Result<
-            tonic::Response<super::TableStructureResponse>,
+            tonic::Response<super::GetTableStructureResponse>,
             tonic::Status,
         >;
     }
-    /// Introspects the physical PostgreSQL table for a given logical table
-    /// (defined in table_definitions) and returns its column structure.
+    /// Introspects the physical PostgreSQL tables for one or more logical tables
+    /// (defined in table_definitions) and returns their column structures.
     /// The server validates that:
     /// - The profile (schema) exists in `schemas`
-    /// - The table is defined for that profile in `table_definitions`
-    /// It then queries information_schema for the physical table and returns
-    /// normalized column metadata. If the physical table is missing despite
-    /// a definition, the response may contain an empty `columns` list.
+    /// - Every table is defined for that profile in `table_definitions`
+    /// It then queries information_schema for the physical tables and returns
+    /// normalized column metadata.
     #[derive(Debug)]
     pub struct TableStructureServiceServer<T> {
         inner: Arc<T>,
@@ -307,7 +315,7 @@ pub mod table_structure_service_server {
                         T: TableStructureService,
                     > tonic::server::UnaryService<super::GetTableStructureRequest>
                     for GetTableStructureSvc<T> {
-                        type Response = super::TableStructureResponse;
+                        type Response = super::GetTableStructureResponse;
                         type Future = BoxFuture<
                             tonic::Response<Self::Response>,
                             tonic::Status,

common/src/search.rs

@@ -1,75 +1,178 @@
-// common/src/search.rs
+use std::path::{Path, PathBuf};
 
-use tantivy::schema::*;
-use tantivy::tokenizer::*;
+use tantivy::schema::{
+    Field, IndexRecordOption, JsonObjectOptions, Schema, Term, TextFieldIndexing, INDEXED, STORED,
+    STRING,
+};
+use tantivy::tokenizer::{
+    AsciiFoldingFilter, LowerCaser, NgramTokenizer, RawTokenizer, RemoveLongFilter,
+    SimpleTokenizer, TextAnalyzer, TokenStream,
+};
 use tantivy::Index;
 
-/// Creates a hybrid Slovak search schema with optimized prefix fields.
+pub const F_PG_ID: &str = "pg_id";
+pub const F_TABLE_NAME: &str = "table_name";
+pub const F_ROW_KEY: &str = "row_key";
+pub const F_DATA_WORD: &str = "data_word";
+pub const F_DATA_NGRAM: &str = "data_ngram";
+pub const F_DATA_EXACT: &str = "data_exact";
+
+pub const TOK_WORD: &str = "kw_word";
+pub const TOK_NGRAM: &str = "kw_ngram";
+pub const TOK_EXACT: &str = "kw_exact";
+
+/// Returns the on-disk path for a profile search index.
+pub fn search_index_path(root: &Path, profile_name: &str) -> PathBuf {
+    root.join(profile_name)
+}
+
+/// Returns the unique index key for one table row inside a profile index.
+pub fn search_row_key(table_name: &str, row_id: i64) -> String {
+    format!("{}:{}", table_name, row_id)
+}
+
+/// Normalizes user-entered values for exact-mode terms.
+pub fn normalize_exact(input: &str) -> String {
+    let trimmed = input.trim();
+    if trimmed.is_empty() {
+        return String::new();
+    }
+
+    let mut analyzer = exact_analyzer();
+    let mut stream = analyzer.token_stream(trimmed);
+    let mut out = String::with_capacity(trimmed.len());
+    while let Some(token) = stream.next() {
+        out.push_str(&token.text);
+    }
+    out
+}
+
+/// Normalizes a column name to the JSON-key form used at index time.
+pub fn normalize_column_name(column: &str) -> String {
+    column.to_ascii_lowercase()
+}
+
+/// Creates the column-aware search schema.
 pub fn create_search_schema() -> Schema {
     let mut schema_builder = Schema::builder();
 
-    schema_builder.add_u64_field("pg_id", INDEXED | STORED);
+    schema_builder.add_u64_field(F_PG_ID, INDEXED | STORED);
+    schema_builder.add_text_field(F_TABLE_NAME, STRING | STORED);
+    schema_builder.add_text_field(F_ROW_KEY, STRING | STORED);
 
-    // FIELD 1: For prefixes (1-4 chars).
-    let short_prefix_indexing = TextFieldIndexing::default()
-        .set_tokenizer("slovak_prefix_edge")
-        .set_index_option(IndexRecordOption::WithFreqsAndPositions);
-    let short_prefix_options = TextOptions::default()
-        .set_indexing_options(short_prefix_indexing)
-        .set_stored();
-    schema_builder.add_text_field("prefix_edge", short_prefix_options);
-
-    // FIELD 2: For the full word.
-    let full_word_indexing = TextFieldIndexing::default()
-        .set_tokenizer("slovak_prefix_full")
-        .set_index_option(IndexRecordOption::WithFreqsAndPositions);
-    let full_word_options = TextOptions::default()
-        .set_indexing_options(full_word_indexing)
-        .set_stored();
-    schema_builder.add_text_field("prefix_full", full_word_options);
-
-    // NGRAM FIELD: For substring matching.
-    let ngram_field_indexing = TextFieldIndexing::default()
-        .set_tokenizer("slovak_ngram")
-        .set_index_option(IndexRecordOption::WithFreqsAndPositions);
-    let ngram_options = TextOptions::default()
-        .set_indexing_options(ngram_field_indexing)
-        .set_stored();
-    schema_builder.add_text_field("text_ngram", ngram_options);
+    schema_builder.add_json_field(F_DATA_WORD, json_options(TOK_WORD, true, false));
+    schema_builder.add_json_field(F_DATA_NGRAM, json_options(TOK_NGRAM, true, false));
+    schema_builder.add_json_field(F_DATA_EXACT, json_options(TOK_EXACT, false, false));
 
     schema_builder.build()
 }
 
-/// Registers all necessary Slovak tokenizers with the index.
-///
-/// This must be called by ANY process that opens the index
-/// to ensure the tokenizers are loaded into memory.
-pub fn register_slovak_tokenizers(index: &Index) -> tantivy::Result<()> {
+fn json_options(tokenizer_name: &str, with_positions: bool, stored: bool) -> JsonObjectOptions {
+    let index_option = if with_positions {
+        IndexRecordOption::WithFreqsAndPositions
+    } else {
+        IndexRecordOption::Basic
+    };
+
+    let indexing = TextFieldIndexing::default()
+        .set_tokenizer(tokenizer_name)
+        .set_index_option(index_option);
+
+    let mut options = JsonObjectOptions::default().set_indexing_options(indexing);
+    if stored {
+        options = options.set_stored();
+    }
+    options
+}
+
+/// Registers all required tokenizers with the index.
+pub fn register_tokenizers(index: &Index) -> tantivy::Result<()> {
     let tokenizer_manager = index.tokenizers();
 
-    // TOKENIZER for `prefix_edge`: Edge N-gram (1-4 chars)
-    let edge_tokenizer = TextAnalyzer::builder(NgramTokenizer::new(1, 4, true)?)
-        .filter(RemoveLongFilter::limit(40))
-        .filter(LowerCaser)
-        .filter(AsciiFoldingFilter)
-        .build();
-    tokenizer_manager.register("slovak_prefix_edge", edge_tokenizer);
-
-    // TOKENIZER for `prefix_full`: Simple word tokenizer
-    let full_tokenizer = TextAnalyzer::builder(SimpleTokenizer::default())
-        .filter(RemoveLongFilter::limit(40))
-        .filter(LowerCaser)
-        .filter(AsciiFoldingFilter)
-        .build();
-    tokenizer_manager.register("slovak_prefix_full", full_tokenizer);
-
-    // NGRAM TOKENIZER: For substring matching.
-    let ngram_tokenizer = TextAnalyzer::builder(NgramTokenizer::new(3, 3, false)?)
-        .filter(RemoveLongFilter::limit(40))
-        .filter(LowerCaser)
-        .filter(AsciiFoldingFilter)
-        .build();
-    tokenizer_manager.register("slovak_ngram", ngram_tokenizer);
+    tokenizer_manager.register(TOK_WORD, word_analyzer());
+    tokenizer_manager.register(TOK_NGRAM, ngram_analyzer()?);
+    tokenizer_manager.register(TOK_EXACT, exact_analyzer());
 
     Ok(())
 }
+
+fn word_analyzer() -> TextAnalyzer {
+    TextAnalyzer::builder(SimpleTokenizer::default())
+        .filter(RemoveLongFilter::limit(80))
+        .filter(LowerCaser)
+        .filter(AsciiFoldingFilter)
+        .build()
+}
+
+fn ngram_analyzer() -> tantivy::Result<TextAnalyzer> {
+    Ok(TextAnalyzer::builder(NgramTokenizer::new(3, 3, false)?)
+        .filter(RemoveLongFilter::limit(80))
+        .filter(LowerCaser)
+        .filter(AsciiFoldingFilter)
+        .build())
+}
+
+fn exact_analyzer() -> TextAnalyzer {
+    TextAnalyzer::builder(RawTokenizer::default())
+        .filter(LowerCaser)
+        .filter(AsciiFoldingFilter)
+        .build()
+}
+
+/// Tokenizes text the same way `data_word` is indexed.
+pub fn tokenize_word(text: &str) -> Vec<String> {
+    tokenize_with(word_analyzer(), text)
+}
+
+/// Tokenizes text the same way `data_ngram` is indexed.
+pub fn tokenize_ngram(text: &str) -> Vec<String> {
+    match ngram_analyzer() {
+        Ok(analyzer) => tokenize_with(analyzer, text),
+        Err(_) => Vec::new(),
+    }
+}
+
+fn tokenize_with(mut analyzer: TextAnalyzer, text: &str) -> Vec<String> {
+    let mut stream = analyzer.token_stream(text);
+    let mut out = Vec::new();
+    while let Some(token) = stream.next() {
+        out.push(token.text.clone());
+    }
+    out
+}
+
+/// Builds a term scoped to a specific JSON path within a JSON field.
+pub fn json_path_term(field: Field, column: &str, text: &str) -> Term {
+    let mut term = Term::from_field_json_path(field, column, false);
+    term.append_type_and_str(text);
+    term
+}
+
+/// Returns all required schema fields or fails loudly on mismatch.
+pub struct SchemaFields {
+    pub pg_id: Field,
+    pub table_name: Field,
+    pub row_key: Field,
+    pub data_word: Field,
+    pub data_ngram: Field,
+    pub data_exact: Field,
+}
+
+impl SchemaFields {
+    pub fn from(schema: &Schema) -> tantivy::Result<Self> {
+        Ok(Self {
+            pg_id: get_field(schema, F_PG_ID)?,
+            table_name: get_field(schema, F_TABLE_NAME)?,
+            row_key: get_field(schema, F_ROW_KEY)?,
+            data_word: get_field(schema, F_DATA_WORD)?,
+            data_ngram: get_field(schema, F_DATA_NGRAM)?,
+            data_exact: get_field(schema, F_DATA_EXACT)?,
+        })
+    }
+}
+
+fn get_field(schema: &Schema, name: &str) -> tantivy::Result<Field> {
+    schema.get_field(name).map_err(|e| {
+        tantivy::TantivyError::SchemaError(format!("schema is missing field '{name}': {e}"))
+    })
+}

search/.gitignore

@@ -0,0 +1 @@
+.codex

search/src/lib.rs

@@ -1,279 +1,436 @@
-// src/lib.rs
+mod query_builder;
 
 use std::collections::HashMap;
 use std::path::Path;
-use tantivy::collector::TopDocs;
-use tantivy::query::{
-    BooleanQuery, BoostQuery, FuzzyTermQuery, Occur, Query, QueryParser, TermQuery,
-};
-use tantivy::schema::{IndexRecordOption, Value};
-use tantivy::{Index, TantivyDocument, Term};
-use tonic::{Request, Response, Status};
+use std::sync::{Arc, Mutex};
 
 use common::proto::komp_ac::search::searcher_server::Searcher;
 pub use common::proto::komp_ac::search::searcher_server::SearcherServer;
 use common::proto::komp_ac::search::{search_response::Hit, SearchRequest, SearchResponse};
-use common::search::register_slovak_tokenizers;
+use common::search::{register_tokenizers, search_index_path, SchemaFields};
+use query_builder::{build_master_query, ConstraintMode, SearchConstraint};
 use sqlx::{PgPool, Row};
+use tantivy::collector::TopDocs;
+use tantivy::schema::Value;
+use tantivy::{Index, IndexReader, ReloadPolicy, TantivyDocument};
+use tonic::{Request, Response, Status};
 use tracing::info;
 
-// We need to hold the database pool in our service struct.
+const INDEX_ROOT: &str = "./tantivy_indexes";
+const DEFAULT_RESULT_LIMIT: usize = 25;
+const HARD_RESULT_LIMIT: usize = 200;
+const DEFAULT_LIST_LIMIT: usize = 5;
+
 pub struct SearcherService {
     pub pool: PgPool,
+    profiles: Mutex<HashMap<String, Arc<ProfileIndex>>>,
 }
 
-// normalize_slovak_text function remains unchanged...
-fn normalize_slovak_text(text: &str) -> String {
-    // ... function content is unchanged ...
-    text.chars()
-        .map(|c| match c {
-            'á' | 'à' | 'â' | 'ä' | 'ă' | 'ā' => 'a',
-            'Á' | 'À' | 'Â' | 'Ä' | 'Ă' | 'Ā' => 'A',
-            'é' | 'è' | 'ê' | 'ë' | 'ě' | 'ē' => 'e',
-            'É' | 'È' | 'Ê' | 'Ë' | 'Ě' | 'Ē' => 'E',
-            'í' | 'ì' | 'î' | 'ï' | 'ī' => 'i',
-            'Í' | 'Ì' | 'Î' | 'Ï' | 'Ī' => 'I',
-            'ó' | 'ò' | 'ô' | 'ö' | 'ō' | 'ő' => 'o',
-            'Ó' | 'Ò' | 'Ô' | 'Ö' | 'Ō' | 'Ő' => 'O',
-            'ú' | 'ù' | 'û' | 'ü' | 'ū' | 'ű' => 'u',
-            'Ú' | 'Ù' | 'Û' | 'Ü' | 'Ū' | 'Ű' => 'U',
-            'ý' | 'ỳ' | 'ŷ' | 'ÿ' => 'y',
-            'Ý' | 'Ỳ' | 'Ŷ' | 'Ÿ' => 'Y',
-            'č' => 'c',
-            'Č' => 'C',
-            'ď' => 'd',
-            'Ď' => 'D',
-            'ľ' => 'l',
-            'Ľ' => 'L',
-            'ň' => 'n',
-            'Ň' => 'N',
-            'ř' => 'r',
-            'Ř' => 'R',
-            'š' => 's',
-            'Š' => 'S',
-            'ť' => 't',
-            'Ť' => 'T',
-            'ž' => 'z',
-            'Ž' => 'Z',
-            _ => c,
-        })
-        .collect()
-}
+impl SearcherService {
+    pub fn new(pool: PgPool) -> Self {
+        Self {
+            pool,
+            profiles: Mutex::new(HashMap::new()),
+        }
+    }
 
-#[tonic::async_trait]
-impl Searcher for SearcherService {
-    async fn search_table(
+    async fn run_rpc(
         &self,
         request: Request<SearchRequest>,
     ) -> Result<Response<SearchResponse>, Status> {
         let req = request.into_inner();
-        let table_name = req.table_name;
-        let query_str = req.query;
+        let normalized = normalize_request(req)?;
+
+        if !profile_exists(&self.pool, &normalized.profile_name).await? {
+            return Err(Status::not_found(format!(
+                "Profile '{}' was not found",
+                normalized.profile_name
+            )));
+        }
+
+        if let Some(table_name) = normalized.table_name.as_deref() {
+            if !table_exists(&self.pool, &normalized.profile_name, table_name).await? {
+                return Err(Status::not_found(format!(
+                    "Table '{}' was not found in profile '{}'",
+                    table_name, normalized.profile_name
+                )));
+            }
+        }
+
+        if !normalized.has_input() {
+            let Some(table_name) = normalized.table_name.as_deref() else {
+                return Err(Status::invalid_argument(
+                    "table_name is required when query is empty",
+                ));
+            };
+
+            let hits = fetch_latest_rows(
+                &self.pool,
+                &normalized.profile_name,
+                table_name,
+                normalized.limit.unwrap_or(DEFAULT_LIST_LIMIT),
+            )
+            .await?;
+            return Ok(Response::new(SearchResponse { hits }));
+        }
+
+        let index_path = search_index_path(Path::new(INDEX_ROOT), &normalized.profile_name);
+        if !index_path.exists() {
+            return Err(Status::not_found(format!(
+                "No search index found for profile '{}'",
+                normalized.profile_name
+            )));
+        }
+
+        let profile = profile_index(&self.profiles, &normalized.profile_name, &index_path)?;
+        let mut hits = run_search(
+            &self.pool,
+            &profile,
+            &normalized.profile_name,
+            normalized.table_name.as_deref(),
+            &normalized.free_query,
+            &normalized.must,
+            normalized.limit.unwrap_or(DEFAULT_RESULT_LIMIT),
+        )
+        .await?;
+
+        hits.sort_by(|left, right| right.score.total_cmp(&left.score));
+        if let Some(limit) = normalized.limit {
+            if hits.len() > limit {
+                hits.truncate(limit);
+            }
+        }
 
-        // --- MODIFIED LOGIC ---
-        // If the query is empty, fetch the 5 most recent records.
-        if query_str.trim().is_empty() {
         info!(
-                "Empty query for table '{}'. Fetching default results.",
-                table_name
-            );
-            let qualified_table = format!("gen.\"{}\"", table_name);
-            let sql = format!(
-                "SELECT id, to_jsonb(t) AS data FROM {} t ORDER BY id DESC LIMIT 5",
-                qualified_table
+            "search: profile={} table={:?} free='{}' constraints={} hits={}",
+            normalized.profile_name,
+            normalized.table_name,
+            normalized.free_query,
+            normalized.must.len(),
+            hits.len()
         );
 
-            let rows = sqlx::query(&sql).fetch_all(&self.pool).await.map_err(|e| {
-                Status::internal(format!("DB query for default results failed: {}", e))
+        Ok(Response::new(SearchResponse { hits }))
+    }
+}
+
+struct ProfileIndex {
+    index: Index,
+    reader: IndexReader,
+    fields: SchemaFields,
+}
+
+impl ProfileIndex {
+    fn open(path: &Path) -> Result<Self, Status> {
+        let index = Index::open_in_dir(path)
+            .map_err(|e| Status::internal(format!("Failed to open index: {}", e)))?;
+        register_tokenizers(&index)
+            .map_err(|e| Status::internal(format!("Failed to register tokenizers: {}", e)))?;
+        let reader = index
+            .reader_builder()
+            .reload_policy(ReloadPolicy::OnCommitWithDelay)
+            .try_into()
+            .map_err(|e| Status::internal(format!("Failed to build index reader: {}", e)))?;
+        let fields = SchemaFields::from(&index.schema()).map_err(|e| {
+            Status::internal(format!(
+                "Search index schema mismatch. Reindex required: {}",
+                e
+            ))
         })?;
 
-            let hits: Vec<Hit> = rows
+        Ok(Self {
+            index,
+            reader,
+            fields,
+        })
+    }
+}
+
+#[derive(Debug)]
+struct NormalizedSearchRequest {
+    profile_name: String,
+    table_name: Option<String>,
+    free_query: String,
+    must: Vec<SearchConstraint>,
+    limit: Option<usize>,
+}
+
+impl NormalizedSearchRequest {
+    fn has_input(&self) -> bool {
+        !self.free_query.is_empty() || !self.must.is_empty()
+    }
+}
+
+fn profile_index(
+    cache: &Mutex<HashMap<String, Arc<ProfileIndex>>>,
+    profile_name: &str,
+    path: &Path,
+) -> Result<Arc<ProfileIndex>, Status> {
+    {
+        let cache_guard = cache
+            .lock()
+            .map_err(|_| Status::internal("Profile index cache lock poisoned"))?;
+        if let Some(index) = cache_guard.get(profile_name) {
+            return Ok(index.clone());
+        }
+    }
+
+    let opened = Arc::new(ProfileIndex::open(path)?);
+    let mut cache_guard = cache
+        .lock()
+        .map_err(|_| Status::internal("Profile index cache lock poisoned"))?;
+    if let Some(index) = cache_guard.get(profile_name) {
+        return Ok(index.clone());
+    }
+    cache_guard.insert(profile_name.to_string(), opened.clone());
+    Ok(opened)
+}
+
+fn validate_identifier(value: &str, field_name: &str) -> Result<(), Status> {
+    let mut chars = value.chars();
+    let Some(first) = chars.next() else {
+        return Err(Status::invalid_argument(format!(
+            "{field_name} must not be empty"
+        )));
+    };
+
+    if !(first.is_ascii_alphabetic() || first == '_')
+        || !chars.all(|ch| ch.is_ascii_alphanumeric() || ch == '_')
+    {
+        return Err(Status::invalid_argument(format!(
+            "{field_name} contains invalid characters"
+        )));
+    }
+
+    Ok(())
+}
+
+fn qualify_profile_table(profile_name: &str, table_name: &str) -> String {
+    format!("\"{}\".\"{}\"", profile_name, table_name)
+}
+
+async fn profile_exists(pool: &PgPool, profile_name: &str) -> Result<bool, Status> {
+    let exists =
+        sqlx::query_scalar::<_, bool>("SELECT EXISTS(SELECT 1 FROM schemas WHERE name = $1)")
+            .bind(profile_name)
+            .fetch_one(pool)
+            .await
+            .map_err(|e| Status::internal(format!("Profile lookup failed: {}", e)))?;
+    Ok(exists)
+}
+
+async fn table_exists(pool: &PgPool, profile_name: &str, table_name: &str) -> Result<bool, Status> {
+    let exists = sqlx::query_scalar::<_, bool>(
+        r#"
+        SELECT EXISTS(
+            SELECT 1
+            FROM table_definitions td
+            JOIN schemas s ON td.schema_id = s.id
+            WHERE s.name = $1 AND td.table_name = $2
+        )
+        "#,
+    )
+    .bind(profile_name)
+    .bind(table_name)
+    .fetch_one(pool)
+    .await
+    .map_err(|e| Status::internal(format!("Table lookup failed: {}", e)))?;
+    Ok(exists)
+}
+
+fn normalize_request(req: SearchRequest) -> Result<NormalizedSearchRequest, Status> {
+    let profile_name = req.profile_name.trim();
+    if profile_name.is_empty() {
+        return Err(Status::invalid_argument("profile_name is required"));
+    }
+    validate_identifier(profile_name, "profile_name")?;
+
+    let table_name = match req.table_name.as_deref().map(str::trim) {
+        Some(table_name) if !table_name.is_empty() => {
+            validate_identifier(table_name, "table_name")?;
+            Some(table_name.to_string())
+        }
+        _ => None,
+    };
+
+    let free_query = req.free_query.trim().to_string();
+    let mut must = Vec::new();
+
+    for constraint in req.must {
+        let column = constraint.column.trim();
+        if column.is_empty() {
+            return Err(Status::invalid_argument(
+                "constraint.column must not be empty",
+            ));
+        }
+        validate_identifier(column, "constraint.column")?;
+
+        let query = constraint.query.trim();
+        if query.is_empty() {
+            return Err(Status::invalid_argument(
+                "constraint.query must not be empty",
+            ));
+        }
+
+        must.push(SearchConstraint {
+            column: column.to_string(),
+            query: query.to_string(),
+            mode: constraint_mode_from_proto(constraint.mode),
+        });
+    }
+
+    let limit = req
+        .limit
+        .map(|value| (value as usize).min(HARD_RESULT_LIMIT));
+
+    Ok(NormalizedSearchRequest {
+        profile_name: profile_name.to_string(),
+        table_name,
+        free_query,
+        must,
+        limit,
+    })
+}
+
+fn constraint_mode_from_proto(raw_mode: i32) -> ConstraintMode {
+    match raw_mode {
+        2 => ConstraintMode::Exact,
+        _ => ConstraintMode::Fuzzy,
+    }
+}
+
+async fn fetch_latest_rows(
+    pool: &PgPool,
+    profile_name: &str,
+    table_name: &str,
+    limit: usize,
+) -> Result<Vec<Hit>, Status> {
+    let sql = format!(
+        "SELECT id, to_jsonb(t) AS data FROM {} t WHERE deleted = FALSE ORDER BY id DESC LIMIT $1",
+        qualify_profile_table(profile_name, table_name)
+    );
+
+    let rows = sqlx::query(&sql)
+        .bind(limit as i64)
+        .fetch_all(pool)
+        .await
+        .map_err(|e| Status::internal(format!("DB query for default results failed: {}", e)))?;
+
+    Ok(rows
         .into_iter()
         .map(|row| {
             let id: i64 = row.try_get("id").unwrap_or_default();
             let json_data: serde_json::Value = row.try_get("data").unwrap_or_default();
             Hit {
                 id,
-                        // Score is 0.0 as this is not a relevance-ranked search
                 score: 0.0,
                 content_json: json_data.to_string(),
+                table_name: table_name.to_string(),
             }
         })
-                .collect();
+        .collect())
+}
 
-            info!(
-                "--- SERVER: Successfully processed empty query. Returning {} default hits. ---",
-                hits.len()
-            );
-            return Ok(Response::new(SearchResponse { hits }));
-        }
-        // --- END OF MODIFIED LOGIC ---
-
-        let index_path = Path::new("./tantivy_indexes").join(&table_name);
-        if !index_path.exists() {
-            return Err(Status::not_found(format!(
-                "No search index found for table '{}'",
-                table_name
-            )));
-        }
-
-        let index = Index::open_in_dir(&index_path)
-            .map_err(|e| Status::internal(format!("Failed to open index: {}", e)))?;
-
-        register_slovak_tokenizers(&index).map_err(|e| {
-            Status::internal(format!("Failed to register Slovak tokenizers: {}", e))
-        })?;
-
-        let reader = index
-            .reader()
-            .map_err(|e| Status::internal(format!("Failed to create index reader: {}", e)))?;
-        let searcher = reader.searcher();
-        let schema = index.schema();
-
-        let pg_id_field = schema
-            .get_field("pg_id")
-            .map_err(|_| Status::internal("Schema is missing the 'pg_id' field."))?;
-
-        // --- Query Building Logic (no changes here) ---
-        let prefix_edge_field = schema.get_field("prefix_edge").unwrap();
-        let prefix_full_field = schema.get_field("prefix_full").unwrap();
-        let text_ngram_field = schema.get_field("text_ngram").unwrap();
-        let normalized_query = normalize_slovak_text(&query_str);
-        let words: Vec<&str> = normalized_query.split_whitespace().collect();
-        if words.is_empty() {
-            return Ok(Response::new(SearchResponse { hits: vec![] }));
-        }
-        let mut query_layers: Vec<(Occur, Box<dyn Query>)> = Vec::new();
-        // ... all your query building layers remain exactly the same ...
-        // ===============================
-        // LAYER 1: PREFIX MATCHING (HIGHEST PRIORITY, Boost: 4.0)
-        // ===============================
-        {
-            let mut must_clauses: Vec<(Occur, Box<dyn Query>)> = Vec::new();
-            for word in &words {
-                let edge_term = Term::from_field_text(prefix_edge_field, word);
-                let full_term = Term::from_field_text(prefix_full_field, word);
-
-                let per_word_query = BooleanQuery::new(vec![
-                    (
-                        Occur::Should,
-                        Box::new(TermQuery::new(edge_term, IndexRecordOption::Basic)),
-                    ),
-                    (
-                        Occur::Should,
-                        Box::new(TermQuery::new(full_term, IndexRecordOption::Basic)),
-                    ),
-                ]);
-                must_clauses.push((Occur::Must, Box::new(per_word_query) as Box<dyn Query>));
-            }
-
-            if !must_clauses.is_empty() {
-                let prefix_query = BooleanQuery::new(must_clauses);
-                let boosted_query = BoostQuery::new(Box::new(prefix_query), 4.0);
-                query_layers.push((Occur::Should, Box::new(boosted_query)));
-            }
-        }
-
-        // ===============================
-        // LAYER 2: FUZZY MATCHING (HIGH PRIORITY, Boost: 3.0)
-        // ===============================
-        {
-            let last_word = words.last().unwrap();
-            let fuzzy_term = Term::from_field_text(prefix_full_field, last_word);
-            let fuzzy_query = FuzzyTermQuery::new(fuzzy_term, 2, true);
-            let boosted_query = BoostQuery::new(Box::new(fuzzy_query), 3.0);
-            query_layers.push((Occur::Should, Box::new(boosted_query)));
-        }
-
-        // ===============================
-        // LAYER 3: PHRASE MATCHING WITH SLOP (MEDIUM PRIORITY, Boost: 2.0)
-        // ===============================
-        if words.len() > 1 {
-            let slop_parser = QueryParser::for_index(&index, vec![prefix_full_field]);
-            let slop_query_str = format!("\"{}\"~3", normalized_query);
-            if let Ok(slop_query) = slop_parser.parse_query(&slop_query_str) {
-                let boosted_query = BoostQuery::new(slop_query, 2.0);
-                query_layers.push((Occur::Should, Box::new(boosted_query)));
-            }
-        }
-
-        // ===============================
-        // LAYER 4: NGRAM SUBSTRING MATCHING (LOWEST PRIORITY, Boost: 1.0)
-        // ===============================
-        {
-            let ngram_parser = QueryParser::for_index(&index, vec![text_ngram_field]);
-            if let Ok(ngram_query) = ngram_parser.parse_query(&normalized_query) {
-                let boosted_query = BoostQuery::new(ngram_query, 1.0);
-                query_layers.push((Occur::Should, Box::new(boosted_query)));
-            }
-        }
-        let master_query = BooleanQuery::new(query_layers);
-        // --- End of Query Building Logic ---
+async fn run_search(
+    pool: &PgPool,
+    profile: &ProfileIndex,
+    profile_name: &str,
+    table_filter: Option<&str>,
+    free_query: &str,
+    must: &[SearchConstraint],
+    limit: usize,
+) -> Result<Vec<Hit>, Status> {
+    let master_query = build_master_query(
+        &profile.index,
+        &profile.fields,
+        free_query,
+        must,
+        table_filter,
+    )?;
 
+    let searcher = profile.reader.searcher();
     let top_docs = searcher
-            .search(&master_query, &TopDocs::with_limit(100))
+        .search(&*master_query, &TopDocs::with_limit(limit))
         .map_err(|e| Status::internal(format!("Search failed: {}", e)))?;
 
     if top_docs.is_empty() {
-            return Ok(Response::new(SearchResponse { hits: vec![] }));
+        return Ok(vec![]);
     }
 
-        // --- NEW LOGIC: Fetch from DB and combine results ---
-
-        // Step 1: Extract (score, pg_id) from Tantivy results.
-        let mut scored_ids: Vec<(f32, u64)> = Vec::new();
+    let mut candidates: Vec<(f32, i64, String)> = Vec::with_capacity(top_docs.len());
     for (score, doc_address) in top_docs {
         let doc: TantivyDocument = searcher
             .doc(doc_address)
             .map_err(|e| Status::internal(format!("Failed to retrieve document: {}", e)))?;
-            if let Some(pg_id_value) = doc.get_first(pg_id_field) {
-                if let Some(pg_id) = pg_id_value.as_u64() {
-                    scored_ids.push((score, pg_id));
-                }
-            }
+        let Some(pg_id) = doc
+            .get_first(profile.fields.pg_id)
+            .and_then(|value| value.as_u64())
+        else {
+            continue;
+        };
+        let Some(table_name) = doc
+            .get_first(profile.fields.table_name)
+            .and_then(|value| value.as_str())
+        else {
+            continue;
+        };
+        candidates.push((score, pg_id as i64, table_name.to_string()));
     }
 
-        // Step 2: Fetch all corresponding rows from Postgres in a single query.
-        let pg_ids: Vec<i64> = scored_ids.iter().map(|(_, id)| *id as i64).collect();
-        let qualified_table = format!("gen.\"{}\"", table_name);
-        let query_str = format!(
-            "SELECT id, to_jsonb(t) AS data FROM {} t WHERE id = ANY($1)",
-            qualified_table
+    if candidates.is_empty() {
+        return Ok(vec![]);
+    }
+
+    let mut ids_by_table: HashMap<String, Vec<i64>> = HashMap::new();
+    for (_, pg_id, table_name) in &candidates {
+        ids_by_table
+            .entry(table_name.clone())
+            .or_default()
+            .push(*pg_id);
+    }
+
+    let mut content_map: HashMap<(String, i64), String> = HashMap::new();
+    for (table_name, pg_ids) in ids_by_table {
+        validate_identifier(&table_name, "table_name")?;
+        let sql = format!(
+            "SELECT id, to_jsonb(t) AS data FROM {} t WHERE deleted = FALSE AND id = ANY($1)",
+            qualify_profile_table(profile_name, &table_name)
         );
 
-        let rows = sqlx::query(&query_str)
+        let rows = sqlx::query(&sql)
             .bind(&pg_ids)
-            .fetch_all(&self.pool)
+            .fetch_all(pool)
             .await
             .map_err(|e| Status::internal(format!("Database query failed: {}", e)))?;
 
-        // Step 3: Map the database results by ID for quick lookup.
-        let mut content_map: HashMap<i64, String> = HashMap::new();
         for row in rows {
-            let id: i64 = row.try_get("id").unwrap_or(0);
-            let json_data: serde_json::Value =
-                row.try_get("data").unwrap_or(serde_json::Value::Null);
-            content_map.insert(id, json_data.to_string());
+            let id: i64 = row.try_get("id").unwrap_or_default();
+            let json_data: serde_json::Value = row.try_get("data").unwrap_or_default();
+            content_map.insert((table_name.clone(), id), json_data.to_string());
+        }
     }
 
-        // Step 4: Build the final response, combining Tantivy scores with PG content.
-        let hits: Vec<Hit> = scored_ids
+    Ok(candidates
         .into_iter()
-            .filter_map(|(score, pg_id)| {
-                content_map.get(&(pg_id as i64)).map(|content_json| Hit {
-                    id: pg_id as i64,
+        .filter_map(|(score, pg_id, table_name)| {
+            content_map
+                .get(&(table_name.clone(), pg_id))
+                .map(|content_json| Hit {
+                    id: pg_id,
                     score,
                     content_json: content_json.clone(),
+                    table_name,
                 })
         })
-            .collect();
+        .collect())
+}
 
-        info!(
-            "--- SERVER: Successfully processed search. Returning {} hits. ---",
-            hits.len()
-        );
-
-        let response = SearchResponse { hits };
-        Ok(Response::new(response))
+#[tonic::async_trait]
+impl Searcher for SearcherService {
+    async fn search(
+        &self,
+        request: Request<SearchRequest>,
+    ) -> Result<Response<SearchResponse>, Status> {
+        self.run_rpc(request).await
     }
 }

search/src/query_builder.rs

@@ -0,0 +1,239 @@
+use common::search::{
+    json_path_term, normalize_exact, tokenize_ngram, tokenize_word, SchemaFields,
+};
+use tantivy::query::{
+    BooleanQuery, BoostQuery, EmptyQuery, FuzzyTermQuery, Occur, PhraseQuery, Query, QueryParser,
+    TermQuery,
+};
+use tantivy::schema::{IndexRecordOption, Term};
+use tantivy::Index;
+use tonic::Status;
+
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+pub enum ConstraintMode {
+    Fuzzy,
+    Exact,
+}
+
+#[derive(Clone, Debug)]
+pub struct SearchConstraint {
+    pub column: String,
+    pub query: String,
+    pub mode: ConstraintMode,
+}
+
+pub fn build_master_query(
+    index: &Index,
+    fields: &SchemaFields,
+    free_query: &str,
+    must: &[SearchConstraint],
+    table_filter: Option<&str>,
+) -> Result<Box<dyn Query>, Status> {
+    let mut clauses: Vec<(Occur, Box<dyn Query>)> = Vec::new();
+    let mut has_search_clause = false;
+
+    for constraint in must {
+        let predicate = match constraint.mode {
+            ConstraintMode::Exact => {
+                exact_predicate(fields, &constraint.column, &constraint.query)?
+            }
+            ConstraintMode::Fuzzy => {
+                fuzzy_predicate_scoped(fields, &constraint.column, &constraint.query)?
+            }
+        };
+        clauses.push((Occur::Must, predicate));
+        has_search_clause = true;
+    }
+
+    let free_words = tokenize_word(free_query);
+    if !free_words.is_empty() {
+        let predicate = fuzzy_predicate_unscoped(index, fields, &free_words)?;
+        clauses.push((Occur::Should, predicate));
+        has_search_clause = true;
+    }
+
+    if let Some(table_name) = table_filter {
+        let term = Term::from_field_text(fields.table_name, table_name);
+        clauses.push((
+            Occur::Must,
+            Box::new(TermQuery::new(term, IndexRecordOption::Basic)),
+        ));
+    }
+
+    if !has_search_clause {
+        return Ok(Box::new(EmptyQuery));
+    }
+
+    Ok(Box::new(BooleanQuery::new(clauses)))
+}
+
+fn exact_predicate(
+    fields: &SchemaFields,
+    column: &str,
+    query: &str,
+) -> Result<Box<dyn Query>, Status> {
+    let normalized_value = normalize_exact(query);
+    if normalized_value.is_empty() {
+        return Err(Status::invalid_argument(
+            "exact query is empty after normalization",
+        ));
+    }
+
+    let term = json_path_term(fields.data_exact, column, &normalized_value);
+    Ok(Box::new(TermQuery::new(term, IndexRecordOption::Basic)))
+}
+
+fn fuzzy_predicate_scoped(
+    fields: &SchemaFields,
+    column: &str,
+    query: &str,
+) -> Result<Box<dyn Query>, Status> {
+    let words = tokenize_word(query);
+    if words.is_empty() {
+        return Err(Status::invalid_argument(
+            "fuzzy query has no searchable tokens",
+        ));
+    }
+
+    let mut layers: Vec<(Occur, Box<dyn Query>)> = Vec::new();
+
+    let mut per_word_clauses: Vec<(Occur, Box<dyn Query>)> = Vec::new();
+    for word in &words {
+        let term = json_path_term(fields.data_word, column, word);
+        let mut alternates: Vec<(Occur, Box<dyn Query>)> = Vec::new();
+
+        alternates.push((
+            Occur::Should,
+            Box::new(BoostQuery::new(
+                Box::new(TermQuery::new(term.clone(), IndexRecordOption::WithFreqs)),
+                4.0,
+            )),
+        ));
+
+        alternates.push((
+            Occur::Should,
+            Box::new(BoostQuery::new(
+                Box::new(FuzzyTermQuery::new_prefix(term.clone(), 0, false)),
+                3.0,
+            )),
+        ));
+
+        if let Some(distance) = fuzzy_distance(word.chars().count()) {
+            alternates.push((
+                Occur::Should,
+                Box::new(BoostQuery::new(
+                    Box::new(FuzzyTermQuery::new(term.clone(), distance, true)),
+                    2.0,
+                )),
+            ));
+        }
+
+        per_word_clauses.push((Occur::Must, Box::new(BooleanQuery::new(alternates))));
+    }
+    layers.push((Occur::Should, Box::new(BooleanQuery::new(per_word_clauses))));
+
+    if words.len() > 1 {
+        let phrase_terms: Vec<(usize, Term)> = words
+            .iter()
+            .enumerate()
+            .map(|(offset, word)| (offset, json_path_term(fields.data_word, column, word)))
+            .collect();
+        let phrase = PhraseQuery::new_with_offset_and_slop(phrase_terms, 3);
+        layers.push((
+            Occur::Should,
+            Box::new(BoostQuery::new(Box::new(phrase), 2.0)),
+        ));
+    }
+
+    let ngrams = tokenize_ngram(query);
+    if !ngrams.is_empty() {
+        let ngram_clauses: Vec<(Occur, Box<dyn Query>)> = ngrams
+            .into_iter()
+            .map(|gram| {
+                let term = json_path_term(fields.data_ngram, column, &gram);
+                (
+                    Occur::Must,
+                    Box::new(TermQuery::new(term, IndexRecordOption::Basic)) as Box<dyn Query>,
+                )
+            })
+            .collect();
+        layers.push((
+            Occur::Should,
+            Box::new(BoostQuery::new(
+                Box::new(BooleanQuery::new(ngram_clauses)),
+                1.0,
+            )),
+        ));
+    }
+
+    Ok(Box::new(BooleanQuery::new(layers)))
+}
+
+fn fuzzy_predicate_unscoped(
+    index: &Index,
+    fields: &SchemaFields,
+    words: &[String],
+) -> Result<Box<dyn Query>, Status> {
+    let mut layers: Vec<(Occur, Box<dyn Query>)> = Vec::new();
+
+    {
+        let parser = QueryParser::for_index(index, vec![fields.data_word]);
+        let query_string = words
+            .iter()
+            .map(|word| format!("+{}*", word))
+            .collect::<Vec<_>>()
+            .join(" ");
+        if let Ok(query) = parser.parse_query(&query_string) {
+            layers.push((Occur::Should, Box::new(BoostQuery::new(query, 4.0))));
+        }
+    }
+
+    {
+        let parser = QueryParser::for_index(index, vec![fields.data_word]);
+        let query_string = words
+            .iter()
+            .map(|word| match fuzzy_distance(word.chars().count()) {
+                Some(distance) => format!("+{}~{}", word, distance),
+                None => format!("+{}", word),
+            })
+            .collect::<Vec<_>>()
+            .join(" ");
+        if let Ok(query) = parser.parse_query(&query_string) {
+            layers.push((Occur::Should, Box::new(BoostQuery::new(query, 2.0))));
+        }
+    }
+
+    if words.len() > 1 {
+        let parser = QueryParser::for_index(index, vec![fields.data_word]);
+        let query_string = format!("\"{}\"~3", words.join(" "));
+        if let Ok(query) = parser.parse_query(&query_string) {
+            layers.push((Occur::Should, Box::new(BoostQuery::new(query, 2.0))));
+        }
+    }
+
+    {
+        let parser = QueryParser::for_index(index, vec![fields.data_ngram]);
+        let query_string = words
+            .iter()
+            .map(|word| format!("+{}", word))
+            .collect::<Vec<_>>()
+            .join(" ");
+        if let Ok(query) = parser.parse_query(&query_string) {
+            layers.push((Occur::Should, Box::new(BoostQuery::new(query, 1.0))));
+        }
+    }
+
+    if layers.is_empty() {
+        return Ok(Box::new(EmptyQuery));
+    }
+
+    Ok(Box::new(BooleanQuery::new(layers)))
+}
+
+fn fuzzy_distance(word_len: usize) -> Option<u8> {
+    match word_len {
+        0..=3 => None,
+        4..=6 => Some(1),
+        _ => Some(2),
+    }
+}

validation-core/Cargo.toml

@@ -0,0 +1,18 @@
+[package]
+name = "validation-core"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+authors.workspace = true
+description = "Shared validation primitives, rules, and sets."
+repository.workspace = true
+
+[dependencies]
+serde = { workspace = true }
+thiserror = { workspace = true }
+unicode-width = { workspace = true }
+regex = { workspace = true, optional = true }
+
+[features]
+default = []
+regex = ["dep:regex"]

validation-core/docs/validation_architecture.md

@@ -0,0 +1,493 @@
+# Validation
+
+This document is the frontend guide for the validation system.
+
+The important idea: reusable validation is built from **rules** and **sets**.
+The frontend creates and manages those. When a set is applied to a table field,
+the server resolves it into the existing `FieldValidation` shape, and the form
+runtime continues to work through the normal table-validation flow.
+
+## Ownership
+
+```mermaid
+flowchart LR
+    Core[validation-core<br/>validation meaning<br/>rule/set merge rules]
+    Server[server<br/>stores rules/sets<br/>applies sets<br/>enforces writes]
+    Common[common/proto<br/>gRPC contract]
+    Client[client/frontend<br/>rule/set UI<br/>calls gRPC]
+    Canvas[canvas<br/>field editing<br/>mask display<br/>local feedback]
+
+    Server --> Core
+    Canvas --> Core
+    Client --> Common
+    Server --> Common
+    Client --> Canvas
+```
+
+`server` stores simple serializable settings. `validation-core` owns how those
+settings combine. `canvas` uses resolved field validation to guide editing.
+
+## Terms
+
+| Term | Meaning |
+| --- | --- |
+| `FieldValidation` | Existing per-column validation config from `common/proto/table_validation.proto`. This is what forms/canvas already consume. |
+| `ValidationRule` | One named reusable fragment, for example `digits-only`, `phone-length`, or `required`. Stored by the server as a `FieldValidation` fragment with no meaningful `dataKey`. |
+| `ValidationSet` | Ordered collection of rule names, for example `phone = [required, phone-length, digits-only, phone-mask]`. |
+| Applied validation | A resolved snapshot of a set written to `table_validation_rules` for a concrete `(table, dataKey)`. |
+| Snapshot | Applying a set copies the resolved config to a field. Later edits to the set do not automatically update fields that were already applied. |
+
+## What Backend Enforces
+
+Backend write validation enforces only server-relevant parts:
+
+| FieldValidation part | Backend | Canvas/frontend |
+| --- | --- | --- |
+| `required` | Yes | Yes |
+| `limits` | Yes | Yes |
+| `pattern` | Yes | Yes |
+| `allowed_values` | Yes | Yes |
+| `mask` | Partly: raw value length/literals | Yes: display/editing mask |
+| `formatter` | No | Yes |
+| `external_validation_enabled` | No | Yes/UI hint |
+
+`mask` is visual metadata, but the backend still uses it to reject incorrectly
+submitted raw values. Example: if the mask is `(###) ###-####`, the backend
+expects the stored value to be raw digits, not `(123) 456-7890`.
+
+## Main User Flow
+
+```mermaid
+sequenceDiagram
+    participant UI as Frontend UI
+    participant API as TableValidationService
+    participant DB as Server DB
+    participant Form as Existing Form Runtime
+
+    UI->>API: UpsertValidationRule(required)
+    UI->>API: UpsertValidationRule(digits-only)
+    UI->>API: UpsertValidationRule(phone-length)
+    UI->>API: UpsertValidationSet(phone: [required, phone-length, digits-only])
+    UI->>API: ApplyValidationSet(profile, table, dataKey, phone)
+    API->>DB: write resolved FieldValidation snapshot
+    Form->>API: GetTableValidation(profile, table)
+    API->>Form: resolved FieldValidation for dataKey
+```
+
+After `ApplyValidationSet`, the existing form code does not need to know that a
+set was used. It receives normal `FieldValidation`.
+
+## API
+
+All APIs live on `TableValidationService`.
+
+### Rules
+
+Create or update one reusable rule:
+
+```text
+UpsertValidationRule(UpsertValidationRuleRequest)
+```
+
+Request shape:
+
+```text
+profileName: string
+rule:
+  name: string
+  description: optional string
+  validation: FieldValidation
+```
+
+Frontend rules:
+
+- `rule.name` is required and unique inside a profile.
+- `rule.validation.dataKey` is ignored by the server.
+- A rule should usually configure one logical fragment.
+- Examples: `required`, `phone-length`, `digits-only`, `phone-mask`.
+
+List rules:
+
+```text
+ListValidationRules({ profileName })
+```
+
+Delete rule:
+
+```text
+DeleteValidationRule({ profileName, name })
+```
+
+Deleting a rule removes it from future reusable composition. Already applied
+field snapshots are not changed.
+
+### Sets
+
+Create or update one reusable set:
+
+```text
+UpsertValidationSet(UpsertValidationSetRequest)
+```
+
+Request shape:
+
+```text
+profileName: string
+set:
+  name: string
+  description: optional string
+  ruleNames: repeated string
+```
+
+Frontend rules:
+
+- `set.name` is required and unique inside a profile.
+- `ruleNames` must contain at least one rule.
+- `ruleNames` are ordered.
+- Every rule name must already exist.
+- Duplicate rule names in the same set are rejected.
+- Conflicting singleton fragments are rejected.
+
+Singleton fragments are:
+
+```text
+limits
+allowed_values
+mask
+formatter
+```
+
+That means a set cannot currently contain two rules that both define `limits`.
+Pattern rules are additive: multiple rules with `pattern` are merged into one
+combined pattern.
+
+List sets:
+
+```text
+ListValidationSets({ profileName })
+```
+
+Response includes each set plus `resolvedValidation`, so the frontend can show
+what the set expands to.
+
+Delete set:
+
+```text
+DeleteValidationSet({ profileName, name })
+```
+
+Deleting a set does not change already applied fields.
+
+### Apply Set To Field
+
+Apply a reusable set to one field:
+
+```text
+ApplyValidationSet(ApplyValidationSetRequest)
+```
+
+Request shape:
+
+```text
+profileName: string
+tableName: string
+dataKey: string
+setName: string
+```
+
+Server behavior:
+
+1. Loads the set.
+2. Loads its ordered rules.
+3. Resolves/merges them through `validation-core`.
+4. Validates that `dataKey` exists in the table definition.
+5. Writes the resolved config into existing `table_validation_rules`.
+
+This is a snapshot. If the user later edits the `phone` set, fields that already
+used `phone` keep their old resolved config until the set is applied again.
+
+## FieldValidation Guide
+
+Rules and direct field validation both use `FieldValidation`.
+
+### Required
+
+```text
+required: true
+```
+
+Backend rejects missing or empty values.
+
+### Limits
+
+```text
+limits:
+  min: 10
+  max: 10
+  warnAt: optional
+  countMode: CHARS | BYTES | DISPLAY_WIDTH
+```
+
+Backend enforces `min` and `max`. `warnAt` is mainly UI feedback.
+
+### Pattern
+
+Pattern rules validate characters at positions.
+
+Example digits-only:
+
+```text
+pattern:
+  rules:
+    - position:
+        kind: PATTERN_POSITION_FROM
+        start: 0
+      constraint:
+        kind: CHARACTER_CONSTRAINT_NUMERIC
+```
+
+Useful constraints:
+
+```text
+CHARACTER_CONSTRAINT_ALPHABETIC
+CHARACTER_CONSTRAINT_NUMERIC
+CHARACTER_CONSTRAINT_ALPHANUMERIC
+CHARACTER_CONSTRAINT_EXACT
+CHARACTER_CONSTRAINT_ONE_OF
+CHARACTER_CONSTRAINT_REGEX
+```
+
+Pattern fragments from multiple rules are merged.
+
+### Allowed Values
+
+```text
+allowed_values:
+  values: ["open", "closed"]
+  allow_empty: false
+  case_insensitive: true
+```
+
+Backend rejects values not in the list.
+
+### Mask
+
+```text
+mask:
+  pattern: "(###) ###-####"
+  input_char: "#"
+  template_char: "_"
+```
+
+Canvas uses this for display/editing. Backend expects raw values without mask
+literals.
+
+### Formatter
+
+```text
+formatter:
+  type: "PhoneFormatter"
+  options: []
+  description: optional
+```
+
+Formatter is resolved client-side. Backend stores it but does not execute it.
+
+### External Validation
+
+```text
+external_validation_enabled: true
+```
+
+This is a frontend/UI hint. Backend stores it but does not perform external
+validation.
+
+## Recommended Frontend Screens
+
+### Rule List
+
+Show all rules for a profile.
+
+Actions:
+
+```text
+create rule
+edit rule
+delete rule
+preview rule config
+```
+
+### Rule Editor
+
+Build a `ValidationRuleDefinition`.
+
+Recommended UI:
+
+```text
+name
+description
+required toggle
+limits section
+pattern section
+allowed values section
+mask section
+formatter section
+external validation toggle
+```
+
+For v1, encourage one fragment per rule. Example: create `phone-length` and
+`digits-only` separately, instead of one huge rule.
+
+### Set List
+
+Show all sets for a profile.
+
+Use `ListValidationSets`, because it returns `resolvedValidation`.
+
+Actions:
+
+```text
+create set
+edit set
+delete set
+preview resolved validation
+```
+
+### Set Editor
+
+Build a `ValidationSetDefinition`.
+
+Recommended UI:
+
+```text
+name
+description
+ordered rule picker
+resolved preview
+```
+
+When rule ordering changes, call `UpsertValidationSet` and then refresh
+`ListValidationSets`.
+
+### Apply Set
+
+On the table/field validation screen, add:
+
+```text
+Apply validation set
+```
+
+Flow:
+
+1. Load sets with `ListValidationSets`.
+2. User selects a set.
+3. Call `ApplyValidationSet(profileName, tableName, dataKey, setName)`.
+4. Refresh `GetTableValidation(profileName, tableName)`.
+
+The field should now behave exactly like a directly configured field validation.
+
+## Example: Phone
+
+Create rule `required`:
+
+```text
+validation:
+  required: true
+```
+
+Create rule `phone-length`:
+
+```text
+validation:
+  limits:
+    min: 10
+    max: 10
+    countMode: CHARS
+```
+
+Create rule `digits-only`:
+
+```text
+validation:
+  pattern:
+    rules:
+      - position:
+          kind: PATTERN_POSITION_FROM
+          start: 0
+        constraint:
+          kind: CHARACTER_CONSTRAINT_NUMERIC
+```
+
+Create rule `phone-mask`:
+
+```text
+validation:
+  mask:
+    pattern: "(###) ###-####"
+    input_char: "#"
+```
+
+Create set `phone`:
+
+```text
+ruleNames:
+  - required
+  - phone-length
+  - digits-only
+  - phone-mask
+```
+
+Apply set:
+
+```text
+profileName: "default"
+tableName: "customers"
+dataKey: "customer_phone"
+setName: "phone"
+```
+
+Then refresh:
+
+```text
+GetTableValidation(default, customers)
+```
+
+The response contains a normal `FieldValidation` for `customer_phone`.
+
+## Important UX Notes
+
+- Applying a set is not a live link.
+- Editing a rule or set does not mutate fields where it was already applied.
+- To update a field after set changes, apply the set again.
+- If a set has conflicting singleton rules, the server rejects it.
+- For now, the system does not store field metadata like `sourceSetName` on
+  applied fields. The field only stores the resolved validation snapshot.
+
+## Files
+
+Core model:
+
+```text
+validation-core/src/set.rs
+validation-core/src/config.rs
+```
+
+Wire contract:
+
+```text
+common/proto/table_validation.proto
+```
+
+Server implementation:
+
+```text
+server/src/table_validation/get/service.rs
+server/src/table_validation/post/repo.rs
+server/src/table_validation/config.rs
+```
+
+Storage:
+
+```text
+server/migrations/20260506170000_create_validation_rules_and_sets.sql
+```

validation-core/src/config.rs

@@ -0,0 +1,311 @@
+use crate::rules::{
+    CharacterFilter, CharacterLimits, DisplayMask, PatternFilters, PositionFilter, PositionRange,
+};
+use serde::{Deserialize, Serialize};
+use std::sync::Arc;
+use thiserror::Error;
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct AllowedValues {
+    pub values: Vec<String>,
+    pub allow_empty: bool,
+    pub case_insensitive: bool,
+}
+
+impl AllowedValues {
+    pub fn new(values: Vec<String>) -> Self {
+        Self {
+            values,
+            allow_empty: true,
+            case_insensitive: false,
+        }
+    }
+
+    pub fn allow_empty(mut self, allow_empty: bool) -> Self {
+        self.allow_empty = allow_empty;
+        self
+    }
+
+    pub fn case_insensitive(mut self, case_insensitive: bool) -> Self {
+        self.case_insensitive = case_insensitive;
+        self
+    }
+
+    pub fn matches(&self, text: &str) -> bool {
+        if self.case_insensitive {
+            self.values
+                .iter()
+                .any(|allowed| allowed.eq_ignore_ascii_case(text))
+        } else {
+            self.values.iter().any(|allowed| allowed == text)
+        }
+    }
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct FormatterSettings {
+    pub formatter_type: String,
+    pub options: Vec<FormatterOption>,
+    pub description: Option<String>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct FormatterOption {
+    pub key: String,
+    pub value: String,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub enum CharacterFilterSettings {
+    Alphabetic,
+    Numeric,
+    Alphanumeric,
+    Exact(char),
+    OneOf(Vec<char>),
+    Regex(String),
+}
+
+impl CharacterFilterSettings {
+    pub fn resolve(&self) -> CharacterFilter {
+        match self {
+            Self::Alphabetic => CharacterFilter::Alphabetic,
+            Self::Numeric => CharacterFilter::Numeric,
+            Self::Alphanumeric => CharacterFilter::Alphanumeric,
+            Self::Exact(ch) => CharacterFilter::Exact(*ch),
+            Self::OneOf(chars) => CharacterFilter::OneOf(chars.clone()),
+            Self::Regex(pattern) => {
+                #[cfg(feature = "regex")]
+                {
+                    match regex::Regex::new(pattern) {
+                        Ok(regex) => CharacterFilter::Custom(Arc::new(move |ch| {
+                            regex.is_match(&ch.to_string())
+                        })),
+                        Err(_) => CharacterFilter::Custom(Arc::new(|_| false)),
+                    }
+                }
+                #[cfg(not(feature = "regex"))]
+                {
+                    let _ = pattern;
+                    CharacterFilter::Custom(Arc::new(|_| false))
+                }
+            }
+        }
+    }
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct PositionFilterSettings {
+    pub positions: PositionRange,
+    pub filter: CharacterFilterSettings,
+}
+
+impl PositionFilterSettings {
+    pub fn resolve(&self) -> PositionFilter {
+        PositionFilter::new(self.positions.clone(), self.filter.resolve())
+    }
+}
+
+#[derive(Debug, Clone, Default, Serialize, Deserialize)]
+pub struct PatternSettings {
+    pub filters: Vec<PositionFilterSettings>,
+    pub description: Option<String>,
+}
+
+impl PatternSettings {
+    pub fn resolve(&self) -> PatternFilters {
+        PatternFilters::new().add_filters(
+            self.filters
+                .iter()
+                .map(PositionFilterSettings::resolve)
+                .collect(),
+        )
+    }
+}
+
+#[derive(Debug, Clone, Default, Serialize, Deserialize)]
+pub struct ValidationSettings {
+    pub required: bool,
+    pub character_limits: Option<CharacterLimits>,
+    pub pattern: Option<PatternSettings>,
+    pub allowed_values: Option<AllowedValues>,
+    pub display_mask: Option<DisplayMask>,
+    pub formatter: Option<FormatterSettings>,
+    pub external_validation_enabled: bool,
+}
+
+impl ValidationSettings {
+    pub fn resolve(&self) -> ValidationConfig {
+        ValidationConfig {
+            required: self.required,
+            character_limits: self.character_limits.clone(),
+            pattern_filters: self.pattern.as_ref().map(PatternSettings::resolve),
+            allowed_values: self.allowed_values.clone(),
+            display_mask: self.display_mask.clone(),
+            formatter: self.formatter.clone(),
+            external_validation_enabled: self.external_validation_enabled,
+        }
+    }
+
+    pub fn merge_rules<'a>(
+        rules: impl IntoIterator<Item = &'a ValidationSettings>,
+    ) -> Result<Self, ValidationMergeError> {
+        let mut merged = ValidationSettings::default();
+
+        for rule in rules {
+            merged.merge_rule(rule)?;
+        }
+
+        Ok(merged)
+    }
+
+    pub fn merge_rule(&mut self, rule: &ValidationSettings) -> Result<(), ValidationMergeError> {
+        self.required |= rule.required;
+        self.external_validation_enabled |= rule.external_validation_enabled;
+
+        merge_singleton(
+            "character_limits",
+            &mut self.character_limits,
+            &rule.character_limits,
+        )?;
+        merge_singleton(
+            "allowed_values",
+            &mut self.allowed_values,
+            &rule.allowed_values,
+        )?;
+        merge_singleton("display_mask", &mut self.display_mask, &rule.display_mask)?;
+        merge_singleton("formatter", &mut self.formatter, &rule.formatter)?;
+
+        if let Some(pattern) = &rule.pattern {
+            match &mut self.pattern {
+                Some(existing) => {
+                    existing.filters.extend(pattern.filters.clone());
+                    if existing.description.is_none() {
+                        existing.description = pattern.description.clone();
+                    }
+                }
+                None => self.pattern = Some(pattern.clone()),
+            }
+        }
+
+        Ok(())
+    }
+}
+
+fn merge_singleton<T: Clone>(
+    field_name: &'static str,
+    target: &mut Option<T>,
+    source: &Option<T>,
+) -> Result<(), ValidationMergeError> {
+    if let Some(source) = source {
+        if target.is_some() {
+            return Err(ValidationMergeError::DuplicateSingleton { field_name });
+        }
+
+        *target = Some(source.clone());
+    }
+
+    Ok(())
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Error)]
+pub enum ValidationMergeError {
+    #[error("validation set contains more than one rule configuring {field_name}")]
+    DuplicateSingleton { field_name: &'static str },
+}
+
+#[derive(Debug, Clone, Default)]
+pub struct ValidationConfig {
+    pub required: bool,
+    pub character_limits: Option<CharacterLimits>,
+    pub pattern_filters: Option<PatternFilters>,
+    pub allowed_values: Option<AllowedValues>,
+    pub display_mask: Option<DisplayMask>,
+    pub formatter: Option<FormatterSettings>,
+    pub external_validation_enabled: bool,
+}
+
+impl ValidationConfig {
+    pub fn validate_content(&self, text: &str) -> ValidationResult {
+        if text.is_empty() {
+            if self.required {
+                return ValidationResult::error("Value required");
+            }
+
+            if let Some(allowed_values) = &self.allowed_values {
+                if !allowed_values.allow_empty {
+                    return ValidationResult::error("Empty value is not allowed");
+                }
+            }
+
+            return ValidationResult::Valid;
+        }
+
+        if let Some(limits) = &self.character_limits {
+            if let Some(result) = limits.validate_content(text) {
+                if !result.is_acceptable() {
+                    return result;
+                }
+            }
+        }
+
+        if let Some(pattern_filters) = &self.pattern_filters {
+            if let Err(message) = pattern_filters.validate_text(text) {
+                return ValidationResult::error(message);
+            }
+        }
+
+        if let Some(allowed_values) = &self.allowed_values {
+            if !allowed_values.matches(text) {
+                return ValidationResult::error("Value must be one of the allowed options");
+            }
+        }
+
+        ValidationResult::Valid
+    }
+
+    pub fn has_validation(&self) -> bool {
+        self.required
+            || self.character_limits.is_some()
+            || self.pattern_filters.is_some()
+            || self.allowed_values.is_some()
+            || self.display_mask.is_some()
+            || self.formatter.is_some()
+            || self.external_validation_enabled
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum ValidationResult {
+    Valid,
+    Warning { message: String },
+    Error { message: String },
+}
+
+impl ValidationResult {
+    pub fn is_acceptable(&self) -> bool {
+        matches!(self, Self::Valid | Self::Warning { .. })
+    }
+
+    pub fn is_error(&self) -> bool {
+        matches!(self, Self::Error { .. })
+    }
+
+    pub fn message(&self) -> Option<&str> {
+        match self {
+            Self::Valid => None,
+            Self::Warning { message } | Self::Error { message } => Some(message),
+        }
+    }
+
+    pub fn warning(message: impl Into<String>) -> Self {
+        Self::Warning {
+            message: message.into(),
+        }
+    }
+
+    pub fn error(message: impl Into<String>) -> Self {
+        Self::Error {
+            message: message.into(),
+        }
+    }
+}

validation-core/src/lib.rs

@@ -0,0 +1,14 @@
+pub mod config;
+pub mod rules;
+pub mod set;
+
+pub use config::{
+    AllowedValues, CharacterFilterSettings, FormatterOption, FormatterSettings, PatternSettings,
+    PositionFilterSettings, ValidationConfig, ValidationMergeError, ValidationResult,
+    ValidationSettings,
+};
+pub use rules::{
+    count_text, CharacterFilter, CharacterLimits, CountMode, DisplayMask, LimitCheckResult,
+    MaskDisplayMode, PatternFilters, PositionFilter, PositionRange,
+};
+pub use set::{AppliedValidation, ValidationRule, ValidationSet};

validation-core/src/rules/character_limits.rs

@@ -0,0 +1,452 @@
+// src/validation/limits.rs
+//! Character limits validation implementation
+
+use crate::ValidationResult;
+use serde::{Deserialize, Serialize};
+use unicode_width::UnicodeWidthStr;
+
+/// Character limits configuration for a field
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct CharacterLimits {
+    /// Maximum number of characters allowed (None = unlimited)
+    max_length: Option<usize>,
+
+    /// Minimum number of characters required (None = no minimum)
+    min_length: Option<usize>,
+
+    /// Warning threshold (warn when approaching max limit)
+    warning_threshold: Option<usize>,
+
+    /// Count mode: characters vs display width
+    count_mode: CountMode,
+}
+
+/// How to count characters for limit checking
+#[derive(Debug, Clone, Copy, Serialize, Deserialize, Default)]
+pub enum CountMode {
+    /// Count actual characters (default)
+    #[default]
+    Characters,
+
+    /// Count display width (useful for CJK characters)
+    DisplayWidth,
+
+    /// Count bytes (rarely used, but available)
+    Bytes,
+}
+
+/// Result of a character limit check
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum LimitCheckResult {
+    /// Within limits
+    Ok,
+
+    /// Approaching limit (warning)
+    Warning { current: usize, max: usize },
+
+    /// At or exceeding limit (error)
+    Exceeded { current: usize, max: usize },
+
+    /// Below minimum length
+    TooShort { current: usize, min: usize },
+}
+
+impl CharacterLimits {
+    /// Create new character limits with just max length
+    pub fn new(max_length: usize) -> Self {
+        Self {
+            max_length: Some(max_length),
+            min_length: None,
+            warning_threshold: None,
+            count_mode: CountMode::default(),
+        }
+    }
+
+    /// Create new character limits with min and max
+    pub fn new_range(min_length: usize, max_length: usize) -> Self {
+        Self {
+            max_length: Some(max_length),
+            min_length: Some(min_length),
+            warning_threshold: None,
+            count_mode: CountMode::default(),
+        }
+    }
+
+    /// Create new character limits with just minimum length
+    pub fn new_min(min_length: usize) -> Self {
+        Self {
+            max_length: None,
+            min_length: Some(min_length),
+            warning_threshold: None,
+            count_mode: CountMode::default(),
+        }
+    }
+
+    /// Create new character limits with only a warning threshold.
+    pub fn new_warning(threshold: usize) -> Self {
+        Self {
+            max_length: None,
+            min_length: None,
+            warning_threshold: Some(threshold),
+            count_mode: CountMode::default(),
+        }
+    }
+
+    /// Set warning threshold (when to show warning before hitting limit)
+    pub fn with_warning_threshold(mut self, threshold: usize) -> Self {
+        self.warning_threshold = Some(threshold);
+        self
+    }
+
+    /// Set count mode (characters vs display width vs bytes)
+    pub fn with_count_mode(mut self, mode: CountMode) -> Self {
+        self.count_mode = mode;
+        self
+    }
+
+    /// Get maximum length
+    pub fn max_length(&self) -> Option<usize> {
+        self.max_length
+    }
+
+    /// Get minimum length
+    pub fn min_length(&self) -> Option<usize> {
+        self.min_length
+    }
+
+    /// Get warning threshold
+    pub fn warning_threshold(&self) -> Option<usize> {
+        self.warning_threshold
+    }
+
+    /// Get count mode
+    pub fn count_mode(&self) -> CountMode {
+        self.count_mode
+    }
+
+    /// Count characters/width/bytes according to the configured mode
+    fn count(&self, text: &str) -> usize {
+        match self.count_mode {
+            CountMode::Characters => text.chars().count(),
+            CountMode::DisplayWidth => text.width(),
+            CountMode::Bytes => text.len(),
+        }
+    }
+
+    /// Check if inserting a character would exceed limits
+    pub fn validate_insertion(
+        &self,
+        current_text: &str,
+        position: usize,
+        character: char,
+    ) -> Option<ValidationResult> {
+        let mut new_text = String::with_capacity(current_text.len() + character.len_utf8());
+        let mut chars = current_text.chars();
+
+        let clamped_pos = position.min(current_text.chars().count());
+        for _ in 0..clamped_pos {
+            if let Some(ch) = chars.next() {
+                new_text.push(ch);
+            }
+        }
+
+        new_text.push(character);
+
+        for ch in chars {
+            new_text.push(ch);
+        }
+
+        let new_count = self.count(&new_text);
+        let current_count = self.count(current_text);
+
+        if let Some(max) = self.max_length {
+            if new_count > max {
+                return Some(ValidationResult::error(format!(
+                    "Character limit exceeded: {new_count}/{max}"
+                )));
+            }
+
+            if let Some(warning_threshold) = self.warning_threshold {
+                if new_count >= warning_threshold && current_count < warning_threshold {
+                    return Some(ValidationResult::warning(format!(
+                        "Approaching character limit: {new_count}/{max}"
+                    )));
+                }
+            }
+        }
+
+        None // No validation issues
+    }
+
+    /// Validate the current content
+    pub fn validate_content(&self, text: &str) -> Option<ValidationResult> {
+        let count = self.count(text);
+
+        if let Some(min) = self.min_length {
+            if count < min {
+                return Some(ValidationResult::warning(format!(
+                    "Minimum length not met: {count}/{min}"
+                )));
+            }
+        }
+
+        if let Some(max) = self.max_length {
+            if count > max {
+                return Some(ValidationResult::error(format!(
+                    "Character limit exceeded: {count}/{max}"
+                )));
+            }
+
+            if let Some(warning_threshold) = self.warning_threshold {
+                if count >= warning_threshold {
+                    return Some(ValidationResult::warning(format!(
+                        "Approaching character limit: {count}/{max}"
+                    )));
+                }
+            }
+        }
+
+        None // No validation issues
+    }
+
+    /// Get the current status of the text against limits
+    pub fn check_limits(&self, text: &str) -> LimitCheckResult {
+        let count = self.count(text);
+
+        if let Some(max) = self.max_length {
+            if count > max {
+                return LimitCheckResult::Exceeded {
+                    current: count,
+                    max,
+                };
+            }
+
+            if let Some(warning_threshold) = self.warning_threshold {
+                if count >= warning_threshold {
+                    return LimitCheckResult::Warning {
+                        current: count,
+                        max,
+                    };
+                }
+            }
+        }
+
+        // Check min length
+        if let Some(min) = self.min_length {
+            if count < min {
+                return LimitCheckResult::TooShort {
+                    current: count,
+                    min,
+                };
+            }
+        }
+
+        LimitCheckResult::Ok
+    }
+
+    /// Get a human-readable status string
+    pub fn status_text(&self, text: &str) -> Option<String> {
+        match self.check_limits(text) {
+            LimitCheckResult::Ok => {
+                // Show current/max if we have a max limit
+                self.max_length
+                    .map(|max| format!("{}/{}", self.count(text), max))
+            }
+            LimitCheckResult::Warning { current, max } => {
+                Some(format!("{current}/{max} (approaching limit)"))
+            }
+            LimitCheckResult::Exceeded { current, max } => {
+                Some(format!("{current}/{max} (exceeded)"))
+            }
+            LimitCheckResult::TooShort { current, min } => Some(format!("{current}/{min} minimum")),
+        }
+    }
+    pub fn allows_field_switch(&self, text: &str) -> bool {
+        if let Some(min) = self.min_length {
+            let count = self.count(text);
+            // Allow switching if field is empty OR meets minimum requirement
+            count == 0 || count >= min
+        } else {
+            true // No minimum requirement, always allow switching
+        }
+    }
+
+    /// Get reason why field switching is not allowed (if any)
+    pub fn field_switch_block_reason(&self, text: &str) -> Option<String> {
+        if let Some(min) = self.min_length {
+            let count = self.count(text);
+            if count > 0 && count < min {
+                return Some(format!(
+                    "Field must be empty or have at least {min} characters (currently: {count})"
+                ));
+            }
+        }
+        None
+    }
+}
+
+pub fn count_text(text: &str, mode: CountMode) -> usize {
+    match mode {
+        CountMode::Characters => text.chars().count(),
+        CountMode::DisplayWidth => text.width(),
+        CountMode::Bytes => text.len(),
+    }
+}
+
+impl Default for CharacterLimits {
+    fn default() -> Self {
+        Self {
+            max_length: Some(30), // Default 30 character limit as specified
+            min_length: None,
+            warning_threshold: None,
+            count_mode: CountMode::default(),
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_character_limits_creation() {
+        let limits = CharacterLimits::new(10);
+        assert_eq!(limits.max_length(), Some(10));
+        assert_eq!(limits.min_length(), None);
+
+        let range_limits = CharacterLimits::new_range(5, 15);
+        assert_eq!(range_limits.min_length(), Some(5));
+        assert_eq!(range_limits.max_length(), Some(15));
+    }
+
+    #[test]
+    fn test_default_limits() {
+        let limits = CharacterLimits::default();
+        assert_eq!(limits.max_length(), Some(30));
+    }
+
+    #[test]
+    fn test_character_counting() {
+        let limits = CharacterLimits::new(5);
+
+        // Test character mode (default)
+        assert_eq!(limits.count("hello"), 5);
+        assert_eq!(limits.count("héllo"), 5); // Accented character counts as 1
+
+        // Test display width mode
+        let limits = limits.with_count_mode(CountMode::DisplayWidth);
+        assert_eq!(limits.count("hello"), 5);
+
+        // Test bytes mode
+        let limits = limits.with_count_mode(CountMode::Bytes);
+        assert_eq!(limits.count("hello"), 5);
+        assert_eq!(limits.count("héllo"), 6); // é takes 2 bytes in UTF-8
+    }
+
+    #[test]
+    fn test_insertion_validation() {
+        let limits = CharacterLimits::new(5);
+
+        // Valid insertion
+        let result = limits.validate_insertion("test", 4, 'x');
+        assert!(result.is_none()); // No validation issues
+
+        // Invalid insertion (would exceed limit)
+        let result = limits.validate_insertion("tests", 5, 'x');
+        assert!(result.is_some());
+        assert!(!result.unwrap().is_acceptable());
+    }
+
+    #[test]
+    fn test_content_validation() {
+        let limits = CharacterLimits::new_range(3, 10);
+
+        // Too short
+        let result = limits.validate_content("hi");
+        assert!(result.is_some());
+        assert!(result.unwrap().is_acceptable()); // Warning, not error
+
+        // Just right
+        let result = limits.validate_content("hello");
+        assert!(result.is_none());
+
+        // Too long
+        let result = limits.validate_content("hello world!");
+        assert!(result.is_some());
+        assert!(!result.unwrap().is_acceptable()); // Error
+    }
+
+    #[test]
+    fn test_warning_threshold() {
+        let limits = CharacterLimits::new(10).with_warning_threshold(8);
+
+        // Below warning threshold
+        let result = limits.validate_insertion("123456", 6, 'x');
+        assert!(result.is_none());
+
+        // At warning threshold
+        let result = limits.validate_insertion("1234567", 7, 'x');
+        assert!(result.is_some()); // This brings us to 8 chars
+        assert!(result.unwrap().is_acceptable()); // Warning, not error
+
+        let result = limits.validate_insertion("12345678", 8, 'x');
+        assert!(result.is_none());
+    }
+
+    #[test]
+    fn test_status_text() {
+        let limits = CharacterLimits::new(10);
+
+        assert_eq!(limits.status_text("hello"), Some("5/10".to_string()));
+
+        let limits = limits.with_warning_threshold(8);
+        assert_eq!(
+            limits.status_text("12345678"),
+            Some("8/10 (approaching limit)".to_string())
+        );
+        assert_eq!(
+            limits.status_text("1234567890x"),
+            Some("11/10 (exceeded)".to_string())
+        );
+    }
+
+    #[test]
+    fn test_field_switch_blocking() {
+        let limits = CharacterLimits::new_range(3, 10);
+
+        // Empty field: should allow switching
+        assert!(limits.allows_field_switch(""));
+        assert!(limits.field_switch_block_reason("").is_none());
+
+        // Field with content below minimum: should block switching
+        assert!(!limits.allows_field_switch("hi"));
+        assert!(limits.field_switch_block_reason("hi").is_some());
+        assert!(limits
+            .field_switch_block_reason("hi")
+            .unwrap()
+            .contains("at least 3 characters"));
+
+        // Field meeting minimum: should allow switching
+        assert!(limits.allows_field_switch("hello"));
+        assert!(limits.field_switch_block_reason("hello").is_none());
+
+        // Field exceeding maximum: should still allow switching (validation shows error but doesn't block)
+        assert!(limits.allows_field_switch("this is way too long"));
+        assert!(limits
+            .field_switch_block_reason("this is way too long")
+            .is_none());
+    }
+
+    #[test]
+    fn test_field_switch_no_minimum() {
+        let limits = CharacterLimits::new(10); // Only max, no minimum
+
+        // Should always allow switching when there's no minimum
+        assert!(limits.allows_field_switch(""));
+        assert!(limits.allows_field_switch("a"));
+        assert!(limits.allows_field_switch("hello"));
+
+        assert!(limits.field_switch_block_reason("").is_none());
+        assert!(limits.field_switch_block_reason("a").is_none());
+    }
+}

validation-core/src/rules/display_mask.rs

@@ -0,0 +1,348 @@
+// src/validation/mask.rs
+//! Pure display mask system - user-defined patterns only
+
+use serde::{Deserialize, Serialize};
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default)]
+pub enum MaskDisplayMode {
+    /// Only show separators as user types
+    /// Example: "" → "", "123" → "123", "12345" → "(123) 45"
+    #[default]
+    Dynamic,
+
+    /// Show full template with placeholders from start
+    /// Example: "" → "(___) ___-____", "123" → "(123) ___-____"
+    Template {
+        /// Character to use as placeholder for empty input positions
+        placeholder: char,
+    },
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct DisplayMask {
+    /// Mask pattern like "##-##-####" where # = input position, others are visual separators
+    pattern: String,
+    /// Character used to represent input positions (usually '#')
+    input_char: char,
+    /// How to display the mask (dynamic vs template)
+    display_mode: MaskDisplayMode,
+}
+
+impl DisplayMask {
+    /// Create a new display mask with dynamic mode (current behavior)
+    ///
+    /// # Arguments
+    /// * `pattern` - The mask pattern (e.g., "##-##-####", "(###) ###-####")  
+    /// * `input_char` - Character representing input positions (usually '#')
+    ///
+    /// # Examples
+    /// ```
+    /// use validation_core::DisplayMask;
+    ///
+    /// // Phone number format
+    /// let phone_mask = DisplayMask::new("(###) ###-####", '#');
+    ///
+    /// // Date format  
+    /// let date_mask = DisplayMask::new("##/##/####", '#');
+    ///
+    /// // Custom business format
+    /// let employee_id = DisplayMask::new("EMP-####-##", '#');
+    /// ```
+    pub fn new(pattern: impl Into<String>, input_char: char) -> Self {
+        Self {
+            pattern: pattern.into(),
+            input_char,
+            display_mode: MaskDisplayMode::Dynamic,
+        }
+    }
+
+    /// Set the display mode for this mask
+    ///
+    /// # Examples
+    /// ```
+    /// use validation_core::{DisplayMask, MaskDisplayMode};
+    ///
+    /// let dynamic_mask = DisplayMask::new("##-##", '#')
+    ///     .with_mode(MaskDisplayMode::Dynamic);
+    ///     
+    /// let template_mask = DisplayMask::new("##-##", '#')
+    ///     .with_mode(MaskDisplayMode::Template { placeholder: '_' });
+    /// ```
+    pub fn with_mode(mut self, mode: MaskDisplayMode) -> Self {
+        self.display_mode = mode;
+        self
+    }
+
+    /// Set template mode with custom placeholder
+    ///
+    /// # Examples
+    /// ```
+    /// use validation_core::DisplayMask;
+    ///
+    /// let phone_template = DisplayMask::new("(###) ###-####", '#')
+    ///     .with_template('_');  // Shows "(___) ___-____" when empty
+    ///     
+    /// let date_dots = DisplayMask::new("##/##/####", '#')
+    ///     .with_template('•');  // Shows "••/••/••••" when empty
+    /// ```
+    pub fn with_template(self, placeholder: char) -> Self {
+        self.with_mode(MaskDisplayMode::Template { placeholder })
+    }
+
+    /// Apply mask to raw input, showing visual separators and handling display mode
+    pub fn apply_to_display(&self, raw_input: &str) -> String {
+        match &self.display_mode {
+            MaskDisplayMode::Dynamic => self.apply_dynamic(raw_input),
+            MaskDisplayMode::Template { placeholder } => {
+                self.apply_template(raw_input, *placeholder)
+            }
+        }
+    }
+
+    /// Dynamic mode - only show separators as user types
+    fn apply_dynamic(&self, raw_input: &str) -> String {
+        if raw_input.is_empty() {
+            return String::new();
+        }
+
+        let mut result = String::new();
+        let mut raw_chars = raw_input.chars();
+
+        for pattern_char in self.pattern.chars() {
+            if pattern_char == self.input_char {
+                // Input position - take from raw input
+                if let Some(input_char) = raw_chars.next() {
+                    result.push(input_char);
+                } else {
+                    // No more input - stop here in dynamic mode
+                    break;
+                }
+            } else {
+                // Visual separator - always show
+                result.push(pattern_char);
+            }
+        }
+
+        // Append any remaining raw characters that don't fit the pattern
+        for remaining_char in raw_chars {
+            result.push(remaining_char);
+        }
+
+        result
+    }
+
+    /// Template mode - show full pattern with placeholders
+    fn apply_template(&self, raw_input: &str, placeholder: char) -> String {
+        let mut result = String::new();
+        let mut raw_chars = raw_input.chars().peekable();
+
+        for pattern_char in self.pattern.chars() {
+            if pattern_char == self.input_char {
+                // Input position - take from raw input or use placeholder
+                if let Some(input_char) = raw_chars.next() {
+                    result.push(input_char);
+                } else {
+                    // No more input - use placeholder to show template
+                    result.push(placeholder);
+                }
+            } else {
+                // Visual separator - always show in template mode
+                result.push(pattern_char);
+            }
+        }
+
+        // In template mode, we don't append extra characters beyond the pattern
+        // This keeps the template consistent
+        result
+    }
+
+    /// Check if a display position should accept cursor/input
+    pub fn is_input_position(&self, display_position: usize) -> bool {
+        self.pattern
+            .chars()
+            .nth(display_position)
+            .map(|c| c == self.input_char)
+            .unwrap_or(true) // Beyond pattern = accept input
+    }
+
+    /// Map display position to raw position
+    pub fn display_pos_to_raw_pos(&self, display_pos: usize) -> usize {
+        let mut raw_pos = 0;
+
+        for (i, pattern_char) in self.pattern.chars().enumerate() {
+            if i >= display_pos {
+                break;
+            }
+            if pattern_char == self.input_char {
+                raw_pos += 1;
+            }
+        }
+
+        raw_pos
+    }
+
+    /// Map raw position to display position
+    pub fn raw_pos_to_display_pos(&self, raw_pos: usize) -> usize {
+        let mut input_positions_seen = 0;
+
+        for (display_pos, pattern_char) in self.pattern.chars().enumerate() {
+            if pattern_char == self.input_char {
+                if input_positions_seen == raw_pos {
+                    return display_pos;
+                }
+                input_positions_seen += 1;
+            }
+        }
+
+        // Beyond pattern, return position after pattern
+        self.pattern.len() + (raw_pos - input_positions_seen)
+    }
+
+    /// Find next input position at or after the given display position
+    pub fn next_input_position(&self, display_pos: usize) -> usize {
+        for (i, pattern_char) in self.pattern.chars().enumerate().skip(display_pos) {
+            if pattern_char == self.input_char {
+                return i;
+            }
+        }
+        // Beyond pattern = all positions are input positions
+        display_pos.max(self.pattern.len())
+    }
+
+    /// Find previous input position at or before the given display position
+    pub fn prev_input_position(&self, display_pos: usize) -> Option<usize> {
+        // Collect pattern chars with indices first, then search backwards
+        let pattern_chars: Vec<(usize, char)> = self.pattern.chars().enumerate().collect();
+
+        // Search backwards from display_pos
+        for &(i, pattern_char) in pattern_chars.iter().rev() {
+            if i <= display_pos && pattern_char == self.input_char {
+                return Some(i);
+            }
+        }
+        None
+    }
+
+    /// Get the display mode
+    pub fn display_mode(&self) -> &MaskDisplayMode {
+        &self.display_mode
+    }
+
+    /// Check if this mask uses template mode
+    pub fn is_template_mode(&self) -> bool {
+        matches!(self.display_mode, MaskDisplayMode::Template { .. })
+    }
+
+    /// Get the pattern string
+    pub fn pattern(&self) -> &str {
+        &self.pattern
+    }
+
+    /// Get the input placeholder character
+    pub fn input_char(&self) -> char {
+        self.input_char
+    }
+
+    /// Get the position of the first input character in the pattern
+    pub fn first_input_position(&self) -> usize {
+        for (pos, ch) in self.pattern.chars().enumerate() {
+            if ch == self.input_char {
+                return pos;
+            }
+        }
+        0
+    }
+}
+
+impl Default for DisplayMask {
+    fn default() -> Self {
+        Self::new("", '#')
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_user_defined_phone_mask() {
+        // User creates their own phone mask
+        let dynamic = DisplayMask::new("(###) ###-####", '#');
+        let template = DisplayMask::new("(###) ###-####", '#').with_template('_');
+
+        // Dynamic mode
+        assert_eq!(dynamic.apply_to_display(""), "");
+        assert_eq!(dynamic.apply_to_display("1234567890"), "(123) 456-7890");
+
+        // Template mode
+        assert_eq!(template.apply_to_display(""), "(___) ___-____");
+        assert_eq!(template.apply_to_display("123"), "(123) ___-____");
+    }
+
+    #[test]
+    fn test_user_defined_date_mask() {
+        // User creates their own date formats
+        let us_date = DisplayMask::new("##/##/####", '#');
+        let eu_date = DisplayMask::new("##.##.####", '#');
+        let iso_date = DisplayMask::new("####-##-##", '#');
+
+        assert_eq!(us_date.apply_to_display("12252024"), "12/25/2024");
+        assert_eq!(eu_date.apply_to_display("25122024"), "25.12.2024");
+        assert_eq!(iso_date.apply_to_display("20241225"), "2024-12-25");
+    }
+
+    #[test]
+    fn test_user_defined_business_formats() {
+        // User creates custom business formats
+        let employee_id = DisplayMask::new("EMP-####-##", '#');
+        let product_code = DisplayMask::new("###-###-###", '#');
+        let invoice = DisplayMask::new("INV####/##", '#');
+
+        assert_eq!(employee_id.apply_to_display("123456"), "EMP-1234-56");
+        assert_eq!(product_code.apply_to_display("123456789"), "123-456-789");
+        assert_eq!(invoice.apply_to_display("123456"), "INV1234/56");
+    }
+
+    #[test]
+    fn test_custom_input_characters() {
+        // User can define their own input character
+        let mask_with_x = DisplayMask::new("XXX-XX-XXXX", 'X');
+        let mask_with_hash = DisplayMask::new("###-##-####", '#');
+        let mask_with_n = DisplayMask::new("NNN-NN-NNNN", 'N');
+
+        assert_eq!(mask_with_x.apply_to_display("123456789"), "123-45-6789");
+        assert_eq!(mask_with_hash.apply_to_display("123456789"), "123-45-6789");
+        assert_eq!(mask_with_n.apply_to_display("123456789"), "123-45-6789");
+    }
+
+    #[test]
+    fn test_custom_placeholders() {
+        // User can define custom placeholder characters
+        let underscores = DisplayMask::new("##-##", '#').with_template('_');
+        let dots = DisplayMask::new("##-##", '#').with_template('•');
+        let dashes = DisplayMask::new("##-##", '#').with_template('-');
+
+        assert_eq!(underscores.apply_to_display(""), "__-__");
+        assert_eq!(dots.apply_to_display(""), "••-••");
+        assert_eq!(dashes.apply_to_display(""), "-----"); // Note: dashes blend with separator
+    }
+
+    #[test]
+    fn test_position_mapping_user_patterns() {
+        let custom = DisplayMask::new("ABC-###-XYZ", '#');
+
+        // Position mapping should work correctly with any pattern
+        assert_eq!(custom.raw_pos_to_display_pos(0), 4); // First # at position 4
+        assert_eq!(custom.raw_pos_to_display_pos(1), 5); // Second # at position 5
+        assert_eq!(custom.raw_pos_to_display_pos(2), 6); // Third # at position 6
+
+        assert_eq!(custom.display_pos_to_raw_pos(4), 0); // Position 4 -> first input
+        assert_eq!(custom.display_pos_to_raw_pos(5), 1); // Position 5 -> second input
+        assert_eq!(custom.display_pos_to_raw_pos(6), 2); // Position 6 -> third input
+
+        assert!(!custom.is_input_position(0)); // A
+        assert!(!custom.is_input_position(3)); // -
+        assert!(custom.is_input_position(4)); // #
+        assert!(!custom.is_input_position(8)); // Y
+    }
+}

validation-core/src/rules/mod.rs

@@ -0,0 +1,7 @@
+pub mod character_limits;
+pub mod display_mask;
+pub mod pattern_rules;
+
+pub use character_limits::{count_text, CharacterLimits, CountMode, LimitCheckResult};
+pub use display_mask::{DisplayMask, MaskDisplayMode};
+pub use pattern_rules::{CharacterFilter, PatternFilters, PositionFilter, PositionRange};

validation-core/src/rules/pattern_rules.rs

@@ -0,0 +1,330 @@
+// src/validation/patterns.rs
+//! Position-based pattern filtering for validation
+
+use serde::{Deserialize, Serialize};
+use std::sync::Arc;
+
+/// A filter that applies to specific character positions in a field
+#[derive(Debug, Clone)]
+pub struct PositionFilter {
+    /// Which positions this filter applies to
+    pub positions: PositionRange,
+    /// What type of character filter to apply
+    pub filter: CharacterFilter,
+}
+
+/// Defines which character positions a filter applies to
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub enum PositionRange {
+    /// Single position (e.g., position 3 only)
+    Single(usize),
+    /// Range of positions (e.g., positions 0-2, inclusive)
+    Range(usize, usize),
+    /// From position onwards (e.g., position 4 and beyond)
+    From(usize),
+    /// Multiple specific positions (e.g., positions 0, 2, 5)
+    Multiple(Vec<usize>),
+}
+
+/// Types of character filters that can be applied
+pub enum CharacterFilter {
+    /// Allow only alphabetic characters (a-z, A-Z)
+    Alphabetic,
+    /// Allow only numeric characters (0-9)
+    Numeric,
+    /// Allow alphanumeric characters (a-z, A-Z, 0-9)
+    Alphanumeric,
+    /// Allow only exact character match
+    Exact(char),
+    /// Allow any character from the provided set
+    OneOf(Vec<char>),
+    /// Custom user-defined filter function
+    Custom(Arc<dyn Fn(char) -> bool + Send + Sync>),
+}
+
+// Manual implementations for Debug and Clone
+impl std::fmt::Debug for CharacterFilter {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            CharacterFilter::Alphabetic => write!(f, "Alphabetic"),
+            CharacterFilter::Numeric => write!(f, "Numeric"),
+            CharacterFilter::Alphanumeric => write!(f, "Alphanumeric"),
+            CharacterFilter::Exact(ch) => write!(f, "Exact('{ch}')"),
+            CharacterFilter::OneOf(chars) => write!(f, "OneOf({chars:?})"),
+            CharacterFilter::Custom(_) => write!(f, "Custom(<function>)"),
+        }
+    }
+}
+
+impl Clone for CharacterFilter {
+    fn clone(&self) -> Self {
+        match self {
+            CharacterFilter::Alphabetic => CharacterFilter::Alphabetic,
+            CharacterFilter::Numeric => CharacterFilter::Numeric,
+            CharacterFilter::Alphanumeric => CharacterFilter::Alphanumeric,
+            CharacterFilter::Exact(ch) => CharacterFilter::Exact(*ch),
+            CharacterFilter::OneOf(chars) => CharacterFilter::OneOf(chars.clone()),
+            CharacterFilter::Custom(func) => CharacterFilter::Custom(Arc::clone(func)),
+        }
+    }
+}
+
+impl PositionRange {
+    /// Check if a position is included in this range
+    pub fn contains(&self, position: usize) -> bool {
+        match self {
+            PositionRange::Single(pos) => position == *pos,
+            PositionRange::Range(start, end) => position >= *start && position <= *end,
+            PositionRange::From(start) => position >= *start,
+            PositionRange::Multiple(positions) => positions.contains(&position),
+        }
+    }
+
+    /// Get all positions up to a given length that this range covers
+    pub fn positions_up_to(&self, max_length: usize) -> Vec<usize> {
+        match self {
+            PositionRange::Single(pos) => {
+                if *pos < max_length {
+                    vec![*pos]
+                } else {
+                    vec![]
+                }
+            }
+            PositionRange::Range(start, end) => {
+                let actual_end = (*end).min(max_length.saturating_sub(1));
+                if *start <= actual_end {
+                    (*start..=actual_end).collect()
+                } else {
+                    vec![]
+                }
+            }
+            PositionRange::From(start) => {
+                if *start < max_length {
+                    (*start..max_length).collect()
+                } else {
+                    vec![]
+                }
+            }
+            PositionRange::Multiple(positions) => positions
+                .iter()
+                .filter(|&&pos| pos < max_length)
+                .copied()
+                .collect(),
+        }
+    }
+}
+
+impl CharacterFilter {
+    /// Test if a character passes this filter
+    pub fn accepts(&self, ch: char) -> bool {
+        match self {
+            CharacterFilter::Alphabetic => ch.is_alphabetic(),
+            CharacterFilter::Numeric => ch.is_numeric(),
+            CharacterFilter::Alphanumeric => ch.is_alphanumeric(),
+            CharacterFilter::Exact(expected) => ch == *expected,
+            CharacterFilter::OneOf(chars) => chars.contains(&ch),
+            CharacterFilter::Custom(func) => func(ch),
+        }
+    }
+
+    /// Get a human-readable description of this filter
+    pub fn description(&self) -> String {
+        match self {
+            CharacterFilter::Alphabetic => "alphabetic characters (a-z, A-Z)".to_string(),
+            CharacterFilter::Numeric => "numeric characters (0-9)".to_string(),
+            CharacterFilter::Alphanumeric => "alphanumeric characters (a-z, A-Z, 0-9)".to_string(),
+            CharacterFilter::Exact(ch) => format!("exactly '{ch}'"),
+            CharacterFilter::OneOf(chars) => {
+                let char_list: String = chars.iter().collect();
+                format!("one of: {char_list}")
+            }
+            CharacterFilter::Custom(_) => "custom filter".to_string(),
+        }
+    }
+}
+
+impl PositionFilter {
+    /// Create a new position filter
+    pub fn new(positions: PositionRange, filter: CharacterFilter) -> Self {
+        Self { positions, filter }
+    }
+
+    /// Validate a character at a specific position
+    pub fn validate_position(&self, position: usize, character: char) -> bool {
+        if self.positions.contains(position) {
+            self.filter.accepts(character)
+        } else {
+            true // Position not covered by this filter, allow any character
+        }
+    }
+
+    /// Get error message for invalid character at position
+    pub fn error_message(&self, position: usize, character: char) -> Option<String> {
+        if self.positions.contains(position) && !self.filter.accepts(character) {
+            Some(format!(
+                "Position {} requires {} but got '{}'",
+                position,
+                self.filter.description(),
+                character
+            ))
+        } else {
+            None
+        }
+    }
+}
+
+/// A collection of position filters for a field
+#[derive(Debug, Clone, Default)]
+pub struct PatternFilters {
+    filters: Vec<PositionFilter>,
+}
+
+impl PatternFilters {
+    /// Create empty pattern filters
+    pub fn new() -> Self {
+        Self::default()
+    }
+
+    /// Add a position filter
+    pub fn add_filter(mut self, filter: PositionFilter) -> Self {
+        self.filters.push(filter);
+        self
+    }
+
+    /// Add multiple filters
+    pub fn add_filters(mut self, filters: Vec<PositionFilter>) -> Self {
+        self.filters.extend(filters);
+        self
+    }
+
+    /// Validate a character at a specific position against all applicable filters
+    pub fn validate_char_at_position(
+        &self,
+        position: usize,
+        character: char,
+    ) -> Result<(), String> {
+        for filter in &self.filters {
+            if let Some(error) = filter.error_message(position, character) {
+                return Err(error);
+            }
+        }
+        Ok(())
+    }
+
+    /// Validate entire text against all filters
+    pub fn validate_text(&self, text: &str) -> Result<(), String> {
+        for (position, character) in text.char_indices() {
+            self.validate_char_at_position(position, character)?
+        }
+        Ok(())
+    }
+
+    /// Check if any filters are configured
+    pub fn has_filters(&self) -> bool {
+        !self.filters.is_empty()
+    }
+
+    /// Get all configured filters
+    pub fn filters(&self) -> &[PositionFilter] {
+        &self.filters
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_position_range_contains() {
+        assert!(PositionRange::Single(3).contains(3));
+        assert!(!PositionRange::Single(3).contains(2));
+
+        assert!(PositionRange::Range(1, 4).contains(3));
+        assert!(!PositionRange::Range(1, 4).contains(5));
+
+        assert!(PositionRange::From(2).contains(5));
+        assert!(!PositionRange::From(2).contains(1));
+
+        assert!(PositionRange::Multiple(vec![0, 2, 5]).contains(2));
+        assert!(!PositionRange::Multiple(vec![0, 2, 5]).contains(3));
+    }
+
+    #[test]
+    fn test_position_range_positions_up_to() {
+        assert_eq!(PositionRange::Single(3).positions_up_to(5), vec![3]);
+        assert_eq!(PositionRange::Single(5).positions_up_to(3), vec![]);
+
+        assert_eq!(PositionRange::Range(1, 3).positions_up_to(5), vec![1, 2, 3]);
+        assert_eq!(PositionRange::Range(1, 5).positions_up_to(3), vec![1, 2]);
+
+        assert_eq!(PositionRange::From(2).positions_up_to(5), vec![2, 3, 4]);
+
+        assert_eq!(
+            PositionRange::Multiple(vec![0, 2, 5]).positions_up_to(4),
+            vec![0, 2]
+        );
+    }
+
+    #[test]
+    fn test_character_filter_accepts() {
+        assert!(CharacterFilter::Alphabetic.accepts('a'));
+        assert!(CharacterFilter::Alphabetic.accepts('Z'));
+        assert!(!CharacterFilter::Alphabetic.accepts('1'));
+
+        assert!(CharacterFilter::Numeric.accepts('5'));
+        assert!(!CharacterFilter::Numeric.accepts('a'));
+
+        assert!(CharacterFilter::Alphanumeric.accepts('a'));
+        assert!(CharacterFilter::Alphanumeric.accepts('5'));
+        assert!(!CharacterFilter::Alphanumeric.accepts('-'));
+
+        assert!(CharacterFilter::Exact('x').accepts('x'));
+        assert!(!CharacterFilter::Exact('x').accepts('y'));
+
+        assert!(CharacterFilter::OneOf(vec!['a', 'b', 'c']).accepts('b'));
+        assert!(!CharacterFilter::OneOf(vec!['a', 'b', 'c']).accepts('d'));
+    }
+
+    #[test]
+    fn test_position_filter_validation() {
+        let filter = PositionFilter::new(PositionRange::Range(0, 1), CharacterFilter::Alphabetic);
+
+        assert!(filter.validate_position(0, 'A'));
+        assert!(filter.validate_position(1, 'b'));
+        assert!(!filter.validate_position(0, '1'));
+        assert!(filter.validate_position(2, '1')); // Position 2 not covered, allow anything
+    }
+
+    #[test]
+    fn test_pattern_filters_validation() {
+        let patterns = PatternFilters::new()
+            .add_filter(PositionFilter::new(
+                PositionRange::Range(0, 1),
+                CharacterFilter::Alphabetic,
+            ))
+            .add_filter(PositionFilter::new(
+                PositionRange::Range(2, 4),
+                CharacterFilter::Numeric,
+            ));
+
+        // Valid pattern: AB123
+        assert!(patterns.validate_text("AB123").is_ok());
+
+        // Invalid: number in alphabetic position
+        assert!(patterns.validate_text("A1123").is_err());
+
+        // Invalid: letter in numeric position
+        assert!(patterns.validate_text("AB1A3").is_err());
+    }
+
+    #[test]
+    fn test_custom_filter() {
+        let pattern = PatternFilters::new().add_filter(PositionFilter::new(
+            PositionRange::From(0),
+            CharacterFilter::Custom(Arc::new(|c| c.is_lowercase())),
+        ));
+
+        assert!(pattern.validate_text("hello").is_ok());
+        assert!(pattern.validate_text("Hello").is_err()); // Uppercase not allowed
+    }
+}

validation-core/src/set.rs

@@ -0,0 +1,118 @@
+use crate::{ValidationConfig, ValidationMergeError, ValidationSettings};
+use serde::{Deserialize, Serialize};
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ValidationRule {
+    pub name: String,
+    pub description: Option<String>,
+    pub settings: ValidationSettings,
+}
+
+impl ValidationRule {
+    pub fn resolve(&self) -> ValidationConfig {
+        self.settings.resolve()
+    }
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ValidationSet {
+    pub name: String,
+    pub description: Option<String>,
+    pub rules: Vec<ValidationRule>,
+}
+
+impl ValidationSet {
+    pub fn resolve_settings(&self) -> Result<ValidationSettings, ValidationMergeError> {
+        ValidationSettings::merge_rules(self.rules.iter().map(|rule| &rule.settings))
+    }
+
+    pub fn resolve(&self) -> Result<ValidationConfig, ValidationMergeError> {
+        Ok(self.resolve_settings()?.resolve())
+    }
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct AppliedValidation {
+    pub set_name: Option<String>,
+    pub settings: ValidationSettings,
+}
+
+impl AppliedValidation {
+    pub fn resolve(&self) -> ValidationConfig {
+        self.settings.resolve()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::{
+        CharacterFilterSettings, CharacterLimits, PatternSettings, PositionFilterSettings,
+        PositionRange,
+    };
+
+    #[test]
+    fn validation_set_merges_rule_fragments() {
+        let set = ValidationSet {
+            name: "phone".to_string(),
+            description: None,
+            rules: vec![
+                ValidationRule {
+                    name: "phone-length".to_string(),
+                    description: None,
+                    settings: ValidationSettings {
+                        character_limits: Some(CharacterLimits::new_range(10, 15)),
+                        ..ValidationSettings::default()
+                    },
+                },
+                ValidationRule {
+                    name: "digits-only".to_string(),
+                    description: None,
+                    settings: ValidationSettings {
+                        pattern: Some(PatternSettings {
+                            filters: vec![PositionFilterSettings {
+                                positions: PositionRange::From(0),
+                                filter: CharacterFilterSettings::Numeric,
+                            }],
+                            description: None,
+                        }),
+                        ..ValidationSettings::default()
+                    },
+                },
+            ],
+        };
+
+        let settings = set.resolve_settings().expect("set should resolve");
+
+        assert!(settings.character_limits.is_some());
+        assert_eq!(settings.pattern.expect("pattern").filters.len(), 1);
+    }
+
+    #[test]
+    fn validation_set_rejects_duplicate_singleton_rules() {
+        let set = ValidationSet {
+            name: "conflict".to_string(),
+            description: None,
+            rules: vec![
+                ValidationRule {
+                    name: "short".to_string(),
+                    description: None,
+                    settings: ValidationSettings {
+                        character_limits: Some(CharacterLimits::new(10)),
+                        ..ValidationSettings::default()
+                    },
+                },
+                ValidationRule {
+                    name: "long".to_string(),
+                    description: None,
+                    settings: ValidationSettings {
+                        character_limits: Some(CharacterLimits::new(20)),
+                        ..ValidationSettings::default()
+                    },
+                },
+            ],
+        };
+
+        assert!(set.resolve_settings().is_err());
+    }
+}