Compare commits

...

9 Commits

30 changed files with 1230 additions and 352 deletions

110
Cargo.lock generated
View File

@@ -493,11 +493,12 @@ checksum = "d71b6127be86fdcfddb610f7182ac57211d4b18a3e9c82eb2d17662f2227ad6a"
[[package]]
name = "canvas"
version = "0.5.0"
version = "0.5.10"
dependencies = [
"anyhow",
"async-trait",
"crossterm",
"derivative",
"once_cell",
"ratatui",
"regex",
@@ -584,7 +585,7 @@ dependencies = [
[[package]]
name = "client"
version = "0.5.0"
version = "0.5.10"
dependencies = [
"anyhow",
"async-trait",
@@ -601,6 +602,8 @@ dependencies = [
"rstest",
"serde",
"serde_json",
"strum 0.27.2",
"strum_macros 0.27.2",
"time",
"tokio",
"tokio-test",
@@ -635,7 +638,7 @@ dependencies = [
[[package]]
name = "common"
version = "0.5.0"
version = "0.5.10"
dependencies = [
"prost 0.13.5",
"prost-build 0.14.1",
@@ -936,6 +939,17 @@ dependencies = [
"serde",
]
[[package]]
name = "derivative"
version = "2.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fcc3dd5e9e9c0b295d6e1e4d811fb6f157d5ffd784b8d202fc62eac8035a770b"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
name = "digest"
version = "0.10.7"
@@ -1959,11 +1973,11 @@ checksum = "08ab2867e3eeeca90e844d1940eab391c9dc5228783db2ed999acbc0a9ed375a"
[[package]]
name = "matchers"
version = "0.1.0"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8263075bb86c5a1b1427b5ae862e8889656f126e9f77c484496e8b47cf5c5558"
checksum = "d1525a2a28c7f4fa0fc98bb91ae755d1e2d1505079e05539e35bc876b5d65ae9"
dependencies = [
"regex-automata 0.1.10",
"regex-automata",
]
[[package]]
@@ -2080,12 +2094,11 @@ dependencies = [
[[package]]
name = "nu-ansi-term"
version = "0.46.0"
version = "0.50.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "77a8165726e8236064dbb45459242600304b42a5ea24ee2948e18e023bf7ba84"
checksum = "7957b9740744892f114936ab4a57b3f487491bbeafaf8083688b16841a4240e5"
dependencies = [
"overload",
"winapi",
"windows-sys 0.60.2",
]
[[package]]
@@ -2243,12 +2256,6 @@ version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "04744f49eae99ab78e0d5c0b603ab218f515ea8cfe5a456d7629ad883a3b6e7d"
[[package]]
name = "overload"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b15813163c1d831bf4a13c3610c05c0d03b39feb07f7e09fa234dac9b15aaf39"
[[package]]
name = "ownedbytes"
version = "0.9.0"
@@ -2756,7 +2763,7 @@ dependencies = [
"itertools 0.13.0",
"lru",
"paste",
"strum",
"strum 0.26.3",
"unicode-segmentation",
"unicode-truncate",
"unicode-width 0.2.0",
@@ -2810,17 +2817,8 @@ checksum = "b544ef1b4eac5dc2db33ea63606ae9ffcfac26c1416a2806ae0bf5f56b201191"
dependencies = [
"aho-corasick",
"memchr",
"regex-automata 0.4.9",
"regex-syntax 0.8.5",
]
[[package]]
name = "regex-automata"
version = "0.1.10"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6c230d73fb8d8c1b9c0b3135c5142a8acee3a0558fb8db5cf1cb65f8d7862132"
dependencies = [
"regex-syntax 0.6.29",
"regex-automata",
"regex-syntax",
]
[[package]]
@@ -2831,15 +2829,9 @@ checksum = "809e8dc61f6de73b46c85f4c96486310fe304c434cfa43669d7b40f711150908"
dependencies = [
"aho-corasick",
"memchr",
"regex-syntax 0.8.5",
"regex-syntax",
]
[[package]]
name = "regex-syntax"
version = "0.6.29"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f162c6dd7b008981e4d40210aca20b4bd0f9b60ca9271061b07f78537722f2e1"
[[package]]
name = "regex-syntax"
version = "0.8.5"
@@ -3100,7 +3092,7 @@ checksum = "1c107b6f4780854c8b126e228ea8869f4d7b71260f962fefb57b996b8959ba6b"
[[package]]
name = "search"
version = "0.5.0"
version = "0.5.10"
dependencies = [
"anyhow",
"common",
@@ -3199,7 +3191,7 @@ dependencies = [
[[package]]
name = "server"
version = "0.5.0"
version = "0.5.10"
dependencies = [
"anyhow",
"bcrypt",
@@ -3210,6 +3202,7 @@ dependencies = [
"futures",
"jsonwebtoken",
"lazy_static",
"once_cell",
"prost 0.13.5",
"prost-build 0.14.1",
"prost-types 0.13.5",
@@ -3234,6 +3227,7 @@ dependencies = [
"tonic",
"tonic-reflection",
"tracing",
"tracing-subscriber",
"uuid",
"validator",
]
@@ -3756,9 +3750,15 @@ version = "0.26.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8fec0f0aef304996cf250b31b5a10dee7980c85da9d759361292b8bca5a18f06"
dependencies = [
"strum_macros",
"strum_macros 0.26.4",
]
[[package]]
name = "strum"
version = "0.27.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "af23d6f6c1a224baef9d3f61e287d2761385a5b88fdab4eb4c6f11aeb54c4bcf"
[[package]]
name = "strum_macros"
version = "0.26.4"
@@ -3772,6 +3772,18 @@ dependencies = [
"syn 2.0.104",
]
[[package]]
name = "strum_macros"
version = "0.27.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7695ce3845ea4b33927c055a39dc438a45b059f7c1b3d91d38d10355fb8cbca7"
dependencies = [
"heck",
"proc-macro2",
"quote",
"syn 2.0.104",
]
[[package]]
name = "subtle"
version = "2.6.1"
@@ -3830,7 +3842,7 @@ dependencies = [
"fnv",
"once_cell",
"plist",
"regex-syntax 0.8.5",
"regex-syntax",
"serde",
"serde_derive",
"serde_json",
@@ -3936,7 +3948,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d60769b80ad7953d8a7b2c70cdfe722bbcdcac6bccc8ac934c40c034d866fc18"
dependencies = [
"byteorder",
"regex-syntax 0.8.5",
"regex-syntax",
"utf8-ranges",
]
@@ -4318,9 +4330,9 @@ checksum = "8df9b6e13f2d32c91b9bd719c00d1958837bc7dec474d94952798cc8e69eeec3"
[[package]]
name = "tracing"
version = "0.1.41"
version = "0.1.44"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "784e0ac535deb450455cbfa28a6f0df145ea1bb7ae51b821cf5e7927fdcfbdd0"
checksum = "63e71662fa4b2a2c3a26f570f037eb95bb1f85397f3cd8076caed2f026a6d100"
dependencies = [
"log",
"pin-project-lite",
@@ -4330,9 +4342,9 @@ dependencies = [
[[package]]
name = "tracing-attributes"
version = "0.1.30"
version = "0.1.31"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "81383ab64e72a7a8b8e13130c49e3dab29def6d0c7d76a03087b3cf71c5c6903"
checksum = "7490cfa5ec963746568740651ac6781f701c9c5ea257c58e057f3ba8cf69e8da"
dependencies = [
"proc-macro2",
"quote",
@@ -4341,9 +4353,9 @@ dependencies = [
[[package]]
name = "tracing-core"
version = "0.1.34"
version = "0.1.36"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b9d12581f227e93f094d3af2ae690a574abb8a2b9b7a96e7cfe9647b2b617678"
checksum = "db97caf9d906fbde555dd62fa95ddba9eecfd14cb388e4f491a66d74cd5fb79a"
dependencies = [
"once_cell",
"valuable",
@@ -4362,14 +4374,14 @@ dependencies = [
[[package]]
name = "tracing-subscriber"
version = "0.3.19"
version = "0.3.22"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e8189decb5ac0fa7bc8b96b7cb9b2701d60d48805aca84a238004d665fcc4008"
checksum = "2f30143827ddab0d256fd843b7a66d164e9f271cfa0dde49142c5ca0ca291f1e"
dependencies = [
"matchers",
"nu-ansi-term",
"once_cell",
"regex",
"regex-automata",
"sharded-slab",
"smallvec",
"thread_local",

View File

@@ -5,7 +5,7 @@ resolver = "2"
[workspace.package]
# TODO: idk how to do the name, fix later
# name = "komp_ac"
version = "0.5.0"
version = "0.5.10"
edition = "2021"
license = "GPL-3.0-or-later"
authors = ["Filip Priečinský <filippriec@gmail.com>"]

View File

@@ -16,10 +16,5 @@ cargo watch -x 'run --package client -- client'
Client with tracing:
```
ENABLE_TRACING=1 RUST_LOG=client=debug cargo watch -x 'run --package client -- client'
```
Client with debug that cant be traced
```
cargo run --package client --features ui-debug -- client
```

2
canvas

Submodule canvas updated: 29fdc5a6c7...2c03fc4814

2
client

Submodule client updated: c1839bd960...6cba369adb

1
common/.gitignore vendored Normal file
View File

@@ -0,0 +1 @@
docs/

11
common/Makefile Normal file
View File

@@ -0,0 +1,11 @@
DOC_OUT := docs/grpc_reference.html
.PHONY: docs
docs:
@echo "Generating gRPC documentation..."
mkdir -p $(dir $(DOC_OUT))
protoc \
--doc_out=html,index.html:$(dir $(DOC_OUT)) \
--proto_path=proto proto/*.proto
@echo "✅ Docs written to $(DOC_OUT)"

View File

@@ -20,6 +20,18 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
".komp_ac.table_validation.TableValidationResponse",
"#[derive(serde::Serialize, serde::Deserialize)]",
)
.type_attribute(
".komp_ac.table_validation.PatternRule",
"#[derive(serde::Serialize, serde::Deserialize)]",
)
.type_attribute(
".komp_ac.table_validation.PatternRules",
"#[derive(serde::Serialize, serde::Deserialize)]",
)
.type_attribute(
".komp_ac.table_validation.CustomFormatter",
"#[derive(serde::Serialize, serde::Deserialize)]",
)
.type_attribute(
".komp_ac.table_validation.UpdateFieldValidationRequest",
"#[derive(serde::Serialize, serde::Deserialize)]",

View File

@@ -3,5 +3,9 @@ syntax = "proto3";
package komp_ac.common;
message Empty {}
message CountResponse { int64 count = 1; }
message PositionRequest { int64 position = 1; }
message CountResponse {
int64 count = 1;
}
message PositionRequest {
int64 position = 1;
}

View File

@@ -4,56 +4,136 @@ package komp_ac.table_definition;
import "common.proto";
// The TableDefinition service manages the entire lifecycle of user-defined
// tables (stored as both metadata and physical PostgreSQL tables) inside
// logical "profiles" (schemas). Each table has stored structure, links, and
// validation rules.
service TableDefinition {
// Creates a new table (and schema if missing) with system columns,
// linked-table foreign keys, user-defined columns, and optional indexes.
// Also inserts metadata and default validation rules. Entirely transactional.
rpc PostTableDefinition(PostTableDefinitionRequest) returns (TableDefinitionResponse);
// Lists all profiles (schemas) and their tables with declared dependencies.
// This provides a tree-like overview of table relationships.
rpc GetProfileTree(komp_ac.common.Empty) returns (ProfileTreeResponse);
// Drops a table and its metadata, then deletes the profile if it becomes empty.
rpc DeleteTable(DeleteTableRequest) returns (DeleteTableResponse);
}
// A single link to another table within the same profile (schema).
message TableLink {
// Name of an existing table within the same profile to link to.
// For each link, a "<linked>_id" column is created on the new table.
// That column references "<linked>"(id) and adds an index automatically.
string linked_table_name = 1;
// If true, the generated foreign key column is NOT NULL.
// Otherwise the column allows NULL.
// Duplicate links to the same target table in one request are rejected.
bool required = 2;
}
// Defines the input for creating a new table definition.
message PostTableDefinitionRequest {
// Table name to create inside the target profile.
// Must be lowercase, alphanumeric with underscores,
// start with a letter, and be <= 63 chars.
// Forbidden names: "id", "deleted", "created_at", or ending in "_id".
string table_name = 1;
// List of links (foreign keys) to existing tables in the same profile.
// Each will automatically get a "<linked>_id" column and an index.
repeated TableLink links = 2;
// List of user-defined columns (adds to system/id/fk columns).
repeated ColumnDefinition columns = 3;
// List of column names to be indexed (must match existing user-defined columns).
// Indexes can target only user-defined columns; system columns ("id", "deleted",
// "created_at") and automatically generated foreign key ("*_id") columns already
// have indexes. Requests trying to index those columns are rejected.
repeated string indexes = 4;
// Name of profile (Postgres schema) where the table will be created.
// Same naming rules as table_name; cannot collide with reserved schemas
// like "public", "information_schema", or ones starting with "pg_".
string profile_name = 5;
}
// Describes one user-defined column for a table.
message ColumnDefinition {
// Column name that follows the same validation rules as table_name.
// Must be lowercase, start with a letter, no uppercase characters,
// and cannot be "id", "deleted", "created_at", or end with "_id".
string name = 1;
// Logical column type. Supported values (case-insensitive):
// TEXT / STRING
// BOOLEAN
// TIMESTAMP / TIMESTAMPTZ / TIME
// MONEY (= NUMERIC(14,4))
// INTEGER / INT
// BIGINTEGER / BIGINT
// DATE
// DECIMAL(p,s) → NUMERIC(p,s)
// DECIMAL args must be integers (no sign, no dot, no leading zeros);
// s ≤ p and p ≥ 1.
string field_type = 2;
}
// Response after table creation (success + DDL preview).
message TableDefinitionResponse {
// True if all DB changes and metadata inserts succeeded.
bool success = 1;
// The actual SQL executed: CREATE TABLE + CREATE INDEX statements.
string sql = 2;
}
// Describes the tree of all profiles and their tables.
message ProfileTreeResponse {
// Table entry in a profile.
message Table {
// Internal ID from table_definitions.id (metadata record).
int64 id = 1;
// Table name within the profile (schema).
string name = 2;
// Other tables this one references (based on link definitions only).
repeated string depends_on = 3;
}
// Profile (schema) entry.
message Profile {
// Name of the schema/profile (as stored in `schemas.name`).
string name = 1;
// All tables in that schema and their dependencies.
repeated Table tables = 2;
}
// All profiles in the system.
repeated Profile profiles = 1;
}
// Request to delete one table definition entirely.
message DeleteTableRequest {
// Profile (schema) name owning the table (must exist).
string profile_name = 1;
// Table to drop (must exist in the profile).
// Executes DROP TABLE "profile"."table" CASCADE and then removes metadata.
string table_name = 2;
}
// Response after table deletion.
message DeleteTableResponse {
// True if table and metadata were successfully deleted in one transaction.
bool success = 1;
// Human-readable summary of what was removed.
string message = 2;
}

View File

@@ -1,18 +1,101 @@
// common/proto/table_script.proto
syntax = "proto3";
package komp_ac.table_script;
// Manages column-computation scripts for user-defined tables.
// Each script belongs to a single table (table_definition_id) and populates
// exactly one target column in that table. The server:
// - Validates script syntax (non-empty, balanced parentheses, starts with '(')
// - Validates the target column (exists, not a system column, allowed type)
// - Validates column/type usage inside math expressions
// - Validates referenced tables/columns against the schema
// - Enforces link constraints for structured access (see notes below)
// - Analyzes dependencies and prevents cycles across the schema
// - Transforms the script to decimal-safe math (steel_decimal)
// - Upserts into table_scripts and records dependencies in script_dependencies
// The whole operation is transactional.
service TableScript {
// Create or update a script for a specific table and target column.
//
// Behavior:
// - Fetches the table by table_definition_id (must exist)
// - Validates "script" (syntax), "target_column" (exists and type rules),
// and all referenced tables/columns (must exist in same schema)
// - Validates math operations: prohibits using certain data types in math
// - Enforces link constraints for structured table access:
// • Allowed always: self-references (same table)
// • Structured access via steel_get_column / steel_get_column_with_index
// requires an explicit link in table_definition_links
// • Raw SQL access via steel_query_sql is permitted (still validated)
// - Detects and rejects circular dependencies across all scripts in the schema
// (self-references are allowed and not treated as cycles)
// - Transforms the script to decimal-safe operations (steel_decimal)
// - UPSERTS into table_scripts on (table_definitions_id, target_column)
// and saves a normalized dependency list into script_dependencies
rpc PostTableScript(PostTableScriptRequest) returns (TableScriptResponse);
}
// Request to create or update a script bound to a specific table and column.
message PostTableScriptRequest {
// Required. The metadata ID from table_definitions.id that identifies the
// table this script belongs to. The table must exist; its schema determines
// where referenced tables/columns are validated and where dependencies are stored.
int64 table_definition_id = 1;
// Required. The target column in the target table that this script computes.
// Must be an existing user-defined column in that table (not a system column).
// System columns are reserved: "id", "deleted", "created_at".
// The column's data type must NOT be one of the prohibited target types:
// BIGINT, DATE, TIMESTAMPTZ
// Note: BOOLEAN targets are allowed (values are converted to Steel #true/#false).
string target_column = 2;
// Required. The script in the Steel DSL (S-expression style).
// Syntax requirements:
// - Non-empty, must start with '('
// - Balanced parentheses
//
// Referencing data:
// - Structured table/column access (enforces link constraints):
// (steel_get_column "table_name" "column_name")
// (steel_get_column_with_index "table_name" index "column_name")
// • index must be a non-negative integer literal
// • self-references are allowed without links
// • other tables require an explicit link from the source table
// (table_definition_links) or the request fails
// - Raw SQL access (no link required, but still validated):
// (steel_query_sql "SELECT ...")
// • Basic checks disallow operations that imply prohibited types,
// e.g., EXTRACT(…), DATE_PART(…), ::DATE, ::TIMESTAMPTZ, ::BIGINT, CAST(…)
// - Self variable access in transformed scripts:
// (get-var "column_name") is treated as referencing the current table
//
// Math operations:
// - The script is transformed by steel_decimal; supported math forms include:
// +, -, *, /, ^, **, pow, sqrt, >, <, =, >=, <=, min, max, abs, round,
// ln, log, log10, exp, sin, cos, tan
// - Columns of the following types CANNOT be used inside math expressions:
// BIGINT, TEXT, BOOLEAN, DATE, TIMESTAMPTZ
//
// Dependency tracking and cycles:
// - Dependencies are extracted from steel_get_column(_with_index), get-var,
// and steel_query_sql and stored in script_dependencies with context
// - Cycles across tables are rejected (self-dependency is allowed)
string script = 3;
// Optional. Free-text description stored alongside the script (no functional effect).
string description = 4;
}
// Response after creating or updating a script.
message TableScriptResponse {
// The ID of the script record in table_scripts (new or existing on upsert).
int64 id = 1;
// Human-readable warnings concatenated into a single string. Possible messages:
// - Warning if the script references itself (may affect first population)
// - Count of raw SQL queries present
// - Info about number of structured linked-table accesses
// - Warning if many dependencies may affect performance
string warnings = 2;
}

View File

@@ -1,25 +1,69 @@
// proto/table_structure.proto
// common/proto/table_structure.proto
syntax = "proto3";
package komp_ac.table_structure;
import "common.proto";
message GetTableStructureRequest {
string profile_name = 1; // e.g., "default"
string table_name = 2; // e.g., "2025_adresar6"
// Introspects the physical PostgreSQL table for a given logical table
// (defined in table_definitions) and returns its column structure.
// 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.
service TableStructureService {
// Return the physical column list (name, normalized data_type,
// nullability, primary key flag) for a table 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`
// - 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);
}
// Request identifying the profile (schema) and table 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;
}
// Response with the ordered list of columns (by ordinal position).
message TableStructureResponse {
// Columns of the physical table, including system columns (id, deleted,
// created_at), user-defined columns, and any foreign-key columns such as
// "<linked_table>_id". May be empty if the physical table is missing.
repeated TableColumn columns = 1;
}
// One physical column entry as reported by information_schema.
message TableColumn {
// Column name exactly as defined in PostgreSQL.
string name = 1;
string data_type = 2; // e.g., "TEXT", "BIGINT", "VARCHAR(255)", "TIMESTAMPTZ"
// Normalized data type string derived from information_schema:
// - VARCHAR(n) when udt_name='varchar' with character_maximum_length
// - CHAR(n) when udt_name='bpchar' with character_maximum_length
// - NUMERIC(p,s) when udt_name='numeric' with precision and scale
// - NUMERIC(p) when udt_name='numeric' with precision only
// - <TYPE>[] for array types (udt_name starting with '_', e.g., INT[] )
// - Otherwise UPPER(udt_name), e.g., TEXT, BIGINT, TIMESTAMPTZ
// Examples: "TEXT", "BIGINT", "VARCHAR(255)", "TIMESTAMPTZ", "NUMERIC(14,4)"
string data_type = 2;
// True if information_schema reports the column as nullable.
bool is_nullable = 3;
// True if the column is part of the table's PRIMARY KEY.
// Typically true for the "id" column created by the system.
bool is_primary_key = 4;
}
service TableStructureService {
rpc GetTableStructure (GetTableStructureRequest) returns (TableStructureResponse);
}

View File

@@ -22,7 +22,8 @@ message FieldValidation {
// Current: only CharacterLimits. More rules can be added later.
CharacterLimits limits = 10;
// Future expansion:
// PatternRules pattern = 11;
PatternRules pattern = 11; // Validation 2
optional CustomFormatter formatter = 14; // Validation 4 custom formatting logic
DisplayMask mask = 3;
// ExternalValidation external = 13;
// CustomFormatter formatter = 14;
@@ -57,13 +58,45 @@ message DisplayMask {
optional string template_char = 3; // e.g., "_"
}
// One positionbased 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, caseinsensitive keywords:
// "ALPHABETIC", "NUMERIC", "ALPHANUMERIC",
// "ONEOF(<chars>)", "EXACT(:)", "CUSTOM(<name>)"
string filter = 2;
}
message CustomFormatter {
// Formatter type identifier; handled clientside.
// Examples: "PSCFormatter", "PhoneFormatter", "CreditCardFormatter", "DateFormatter"
string type = 1;
// Optional freetext note or parameters (e.g. locale, pattern)
optional string description = 2;
}
// Collection of pattern rules for one field
message PatternRules {
// All rules that make up the validation logic
repeated PatternRule rules = 1;
// Optional humanreadable description for UI/debug purposes
optional string description = 2;
}
// Service to fetch validations for a table
service TableValidationService {
rpc GetTableValidation(GetTableValidationRequest)
returns (TableValidationResponse);
rpc GetTableValidation(GetTableValidationRequest) returns (TableValidationResponse);
rpc UpdateFieldValidation(UpdateFieldValidationRequest)
returns (UpdateFieldValidationResponse);
rpc UpdateFieldValidation(UpdateFieldValidationRequest) returns (UpdateFieldValidationResponse);
}
message UpdateFieldValidationRequest {

View File

@@ -5,67 +5,220 @@ package komp_ac.tables_data;
import "common.proto";
import "google/protobuf/struct.proto";
// Read and write row data for user-defined tables inside profiles (schemas).
// Operations are performed against the physical PostgreSQL table that
// corresponds to the logical table definition and are scoped by profile
// (schema). Deletions are soft (set deleted = true). Typed binding and
// script-based validation are enforced consistently.
service TablesData {
// Insert a new row into a table with strict type binding and script validation.
//
// Behavior:
// - Validates that profile (schema) exists and table is defined for it
// - Validates provided columns exist (user-defined or allowed system/FK columns)
// - For columns targeted by scripts in this table, the client MUST provide the
// value, and it MUST equal the scripts calculated value (compared type-safely)
// - Binds values with correct SQL types, rejects invalid formats/ranges
// - Inserts the row and returns the new id; queues search indexing (best effort)
// - If the physical table is missing but the definition exists, returns INTERNAL
rpc PostTableData(PostTableDataRequest) returns (PostTableDataResponse);
// Update existing row data with strict type binding and script validation.
//
// Behavior:
// - Validates profile and table, and that the record exists
// - If request data is empty, returns success without changing the row
// - For columns targeted by scripts:
// • If included in update, provided value must equal the script result
// • If not included, update must not cause the script result to differ
// from the current stored value; otherwise FAILED_PRECONDITION is returned
// - Binds values with correct SQL types; rejects invalid formats/ranges
// - Updates the row and returns the id; queues search indexing (best effort)
rpc PutTableData(PutTableDataRequest) returns (PutTableDataResponse);
// Soft-delete a single record (sets deleted = true) if it exists and is not already deleted.
//
// Behavior:
// - Validates profile and table definition
// - Updates only rows with deleted = false
// - success = true means a row was actually changed; false means nothing to delete
// - If the physical table is missing but the definition exists, returns INTERNAL
rpc DeleteTableData(DeleteTableDataRequest) returns (DeleteTableDataResponse);
// Fetch a single non-deleted row by id as textified values.
//
// Behavior:
// - Validates profile and table definition
// - Returns all columns as strings (COALESCE(col::TEXT, '') AS col)
// including: id, deleted, all user-defined columns, and FK columns
// named "<linked_table>_id" for each table link
// - Fails with NOT_FOUND if record does not exist or is soft-deleted
// - If the physical table is missing but the definition exists, returns INTERNAL
rpc GetTableData(GetTableDataRequest) returns (GetTableDataResponse);
// Count non-deleted rows in a table.
//
// Behavior:
// - Validates profile and table definition
// - Returns komp_ac.common.CountResponse.count with rows where deleted = FALSE
// - If the physical table is missing but the definition exists, returns INTERNAL
rpc GetTableDataCount(GetTableDataCountRequest) returns (komp_ac.common.CountResponse);
// Fetch the N-th non-deleted row by id order (1-based), then return its full data.
//
// Behavior:
// - position is 1-based (position = 1 → first row by id ASC with deleted = FALSE)
// - Returns NOT_FOUND if position is out of bounds
// - Otherwise identical to GetTableData for the selected id
rpc GetTableDataByPosition(GetTableDataByPositionRequest) returns (GetTableDataResponse);
}
// Insert a new row.
message PostTableDataRequest {
// Required. Profile (PostgreSQL schema) name that owns the table.
// Must exist in the schemas table.
string profile_name = 1;
// Required. Logical table (definition) name within the profile.
// Must exist in table_definitions for the given profile.
string table_name = 2;
// Required. Key-value data for columns to insert.
//
// Allowed keys:
// - User-defined columns from the table definition
// - System/FK columns:
// • "deleted" (BOOLEAN), optional; default FALSE if not provided
// • "<linked_table>_id" (BIGINT) for each table link
//
// Type expectations by SQL type:
// - TEXT: string value; empty string is treated as NULL
// - BOOLEAN: bool value
// - TIMESTAMPTZ: ISO 8601/RFC 3339 string (parsed to TIMESTAMPTZ)
// - INTEGER: number with no fractional part and within i32 range
// - BIGINT: number with no fractional part and within i64 range
// - NUMERIC(p,s): string representation only; empty string becomes NULL
// (numbers for NUMERIC are rejected to avoid precision loss)
//
// Script validation rules:
// - If a script exists for a target column, that column MUST be present here,
// and its provided value MUST equal the scripts computed value (type-aware
// comparison, e.g., decimals are compared numerically).
//
// Notes:
// - Unknown/invalid column names are rejected
// - Some application-specific validations may apply (e.g., max length for
// certain fields like "telefon")
map<string, google.protobuf.Value> data = 3;
}
// Insert response.
message PostTableDataResponse {
// True if the insert succeeded.
bool success = 1;
// Human-readable message.
string message = 2;
// The id of the inserted row.
int64 inserted_id = 3;
}
// Update an existing row.
message PutTableDataRequest {
// Required. Profile (schema) name.
string profile_name = 1;
// Required. Table name within the profile.
string table_name = 2;
// Required. Id of the row to update.
int64 id = 3;
// Required. Columns to update (same typing rules as PostTableDataRequest.data).
//
// Special script rules:
// - If a script targets column X and X is included here, the value for X must
// equal the scripts result (type-aware).
// - If X is not included here but the update would cause the scripts result
// to change compared to the current stored value, the update is rejected with
// FAILED_PRECONDITION, instructing the caller to include X explicitly.
//
// Passing an empty map results in a no-op success response.
map<string, google.protobuf.Value> data = 4;
}
// Update response.
message PutTableDataResponse {
// True if the update succeeded (or no-op on empty data).
bool success = 1;
// Human-readable message.
string message = 2;
// The id of the updated row.
int64 updated_id = 3;
}
// Soft-delete a single row.
message DeleteTableDataRequest {
// Required. Profile (schema) name.
string profile_name = 1;
// Required. Table name within the profile.
string table_name = 2;
// Required. Row id to soft-delete.
int64 record_id = 3;
}
// Soft-delete response.
message DeleteTableDataResponse {
// True if a row was marked deleted (id existed and was not already deleted).
bool success = 1;
}
// Fetch a single non-deleted row by id.
message GetTableDataRequest {
// Required. Profile (schema) name.
string profile_name = 1;
// Required. Table name within the profile.
string table_name = 2;
// Required. Id of the row to fetch.
int64 id = 3;
}
// Row payload: all columns returned as strings.
message GetTableDataResponse {
// Map of column_name → stringified value for:
// - id, deleted
// - all user-defined columns from the table definition
// - FK columns named "<linked_table>_id" for each table link
//
// All values are returned as TEXT via col::TEXT and COALESCEed to empty string
// (NULL becomes ""). The row is returned only if deleted = FALSE.
map<string, string> data = 1;
}
// Count non-deleted rows.
message GetTableDataCountRequest {
// Required. Profile (schema) name.
string profile_name = 1;
// Required. Table name within the profile.
string table_name = 2;
}
// Fetch by ordinal position among non-deleted rows (1-based).
message GetTableDataByPositionRequest {
// Required. Profile (schema) name.
string profile_name = 1;
// Required. Table name within the profile.
string table_name = 2;
// Required. 1-based position by id ascending among rows with deleted = FALSE.
int32 position = 3;
}

View File

@@ -37,7 +37,6 @@ pub mod proto {
pub mod table_validation {
include!("proto/komp_ac.table_validation.rs");
}
pub const FILE_DESCRIPTOR_SET: &[u8] =
include_bytes!("proto/descriptor.bin");
pub const FILE_DESCRIPTOR_SET: &[u8] = include_bytes!("proto/descriptor.bin");
}
}

Binary file not shown.

View File

@@ -1,77 +1,133 @@
// This file is @generated by prost-build.
/// A single link to another table within the same profile (schema).
#[derive(serde::Serialize, serde::Deserialize)]
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct TableLink {
/// Name of an existing table within the same profile to link to.
/// For each link, a "<linked>_id" column is created on the new table.
/// That column references "<linked>"(id) and adds an index automatically.
#[prost(string, tag = "1")]
pub linked_table_name: ::prost::alloc::string::String,
/// If true, the generated foreign key column is NOT NULL.
/// Otherwise the column allows NULL.
/// Duplicate links to the same target table in one request are rejected.
#[prost(bool, tag = "2")]
pub required: bool,
}
/// Defines the input for creating a new table definition.
#[derive(serde::Serialize, serde::Deserialize)]
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct PostTableDefinitionRequest {
/// Table name to create inside the target profile.
/// Must be lowercase, alphanumeric with underscores,
/// start with a letter, and be <= 63 chars.
/// Forbidden names: "id", "deleted", "created_at", or ending in "_id".
#[prost(string, tag = "1")]
pub table_name: ::prost::alloc::string::String,
/// List of links (foreign keys) to existing tables in the same profile.
/// Each will automatically get a "<linked>_id" column and an index.
#[prost(message, repeated, tag = "2")]
pub links: ::prost::alloc::vec::Vec<TableLink>,
/// List of user-defined columns (adds to system/id/fk columns).
#[prost(message, repeated, tag = "3")]
pub columns: ::prost::alloc::vec::Vec<ColumnDefinition>,
/// List of column names to be indexed (must match existing user-defined columns).
/// Indexes can target only user-defined columns; system columns ("id", "deleted",
/// "created_at") and automatically generated foreign key ("*_id") columns already
/// have indexes. Requests trying to index those columns are rejected.
#[prost(string, repeated, tag = "4")]
pub indexes: ::prost::alloc::vec::Vec<::prost::alloc::string::String>,
/// Name of profile (Postgres schema) where the table will be created.
/// Same naming rules as table_name; cannot collide with reserved schemas
/// like "public", "information_schema", or ones starting with "pg_".
#[prost(string, tag = "5")]
pub profile_name: ::prost::alloc::string::String,
}
/// Describes one user-defined column for a table.
#[derive(serde::Serialize, serde::Deserialize)]
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct ColumnDefinition {
/// Column name that follows the same validation rules as table_name.
/// Must be lowercase, start with a letter, no uppercase characters,
/// and cannot be "id", "deleted", "created_at", or end with "_id".
#[prost(string, tag = "1")]
pub name: ::prost::alloc::string::String,
/// Logical column type. Supported values (case-insensitive):
/// TEXT / STRING
/// BOOLEAN
/// TIMESTAMP / TIMESTAMPTZ / TIME
/// MONEY (= NUMERIC(14,4))
/// INTEGER / INT
/// BIGINTEGER / BIGINT
/// DATE
/// DECIMAL(p,s) → NUMERIC(p,s)
/// DECIMAL args must be integers (no sign, no dot, no leading zeros);
/// s ≤ p and p ≥ 1.
#[prost(string, tag = "2")]
pub field_type: ::prost::alloc::string::String,
}
/// Response after table creation (success + DDL preview).
#[derive(serde::Serialize, serde::Deserialize)]
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct TableDefinitionResponse {
/// True if all DB changes and metadata inserts succeeded.
#[prost(bool, tag = "1")]
pub success: bool,
/// The actual SQL executed: CREATE TABLE + CREATE INDEX statements.
#[prost(string, tag = "2")]
pub sql: ::prost::alloc::string::String,
}
/// Describes the tree of all profiles and their tables.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct ProfileTreeResponse {
/// All profiles in the system.
#[prost(message, repeated, tag = "1")]
pub profiles: ::prost::alloc::vec::Vec<profile_tree_response::Profile>,
}
/// Nested message and enum types in `ProfileTreeResponse`.
pub mod profile_tree_response {
/// Table entry in a profile.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct Table {
/// Internal ID from table_definitions.id (metadata record).
#[prost(int64, tag = "1")]
pub id: i64,
/// Table name within the profile (schema).
#[prost(string, tag = "2")]
pub name: ::prost::alloc::string::String,
/// Other tables this one references (based on link definitions only).
#[prost(string, repeated, tag = "3")]
pub depends_on: ::prost::alloc::vec::Vec<::prost::alloc::string::String>,
}
/// Profile (schema) entry.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct Profile {
/// Name of the schema/profile (as stored in `schemas.name`).
#[prost(string, tag = "1")]
pub name: ::prost::alloc::string::String,
/// All tables in that schema and their dependencies.
#[prost(message, repeated, tag = "2")]
pub tables: ::prost::alloc::vec::Vec<Table>,
}
}
/// Request to delete one table definition entirely.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct DeleteTableRequest {
/// Profile (schema) name owning the table (must exist).
#[prost(string, tag = "1")]
pub profile_name: ::prost::alloc::string::String,
/// Table to drop (must exist in the profile).
/// Executes DROP TABLE "profile"."table" CASCADE and then removes metadata.
#[prost(string, tag = "2")]
pub table_name: ::prost::alloc::string::String,
}
/// Response after table deletion.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct DeleteTableResponse {
/// True if table and metadata were successfully deleted in one transaction.
#[prost(bool, tag = "1")]
pub success: bool,
/// Human-readable summary of what was removed.
#[prost(string, tag = "2")]
pub message: ::prost::alloc::string::String,
}
@@ -86,6 +142,10 @@ pub mod table_definition_client {
)]
use tonic::codegen::*;
use tonic::codegen::http::Uri;
/// The TableDefinition service manages the entire lifecycle of user-defined
/// tables (stored as both metadata and physical PostgreSQL tables) inside
/// logical "profiles" (schemas). Each table has stored structure, links, and
/// validation rules.
#[derive(Debug, Clone)]
pub struct TableDefinitionClient<T> {
inner: tonic::client::Grpc<T>,
@@ -166,6 +226,9 @@ pub mod table_definition_client {
self.inner = self.inner.max_encoding_message_size(limit);
self
}
/// Creates a new table (and schema if missing) with system columns,
/// linked-table foreign keys, user-defined columns, and optional indexes.
/// Also inserts metadata and default validation rules. Entirely transactional.
pub async fn post_table_definition(
&mut self,
request: impl tonic::IntoRequest<super::PostTableDefinitionRequest>,
@@ -195,6 +258,8 @@ pub mod table_definition_client {
);
self.inner.unary(req, path, codec).await
}
/// Lists all profiles (schemas) and their tables with declared dependencies.
/// This provides a tree-like overview of table relationships.
pub async fn get_profile_tree(
&mut self,
request: impl tonic::IntoRequest<super::super::common::Empty>,
@@ -224,6 +289,7 @@ pub mod table_definition_client {
);
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,
request: impl tonic::IntoRequest<super::DeleteTableRequest>,
@@ -268,6 +334,9 @@ pub mod table_definition_server {
/// Generated trait containing gRPC methods that should be implemented for use with TableDefinitionServer.
#[async_trait]
pub trait TableDefinition: std::marker::Send + std::marker::Sync + 'static {
/// Creates a new table (and schema if missing) with system columns,
/// linked-table foreign keys, user-defined columns, and optional indexes.
/// Also inserts metadata and default validation rules. Entirely transactional.
async fn post_table_definition(
&self,
request: tonic::Request<super::PostTableDefinitionRequest>,
@@ -275,6 +344,8 @@ pub mod table_definition_server {
tonic::Response<super::TableDefinitionResponse>,
tonic::Status,
>;
/// Lists all profiles (schemas) and their tables with declared dependencies.
/// This provides a tree-like overview of table relationships.
async fn get_profile_tree(
&self,
request: tonic::Request<super::super::common::Empty>,
@@ -282,6 +353,7 @@ pub mod table_definition_server {
tonic::Response<super::ProfileTreeResponse>,
tonic::Status,
>;
/// Drops a table and its metadata, then deletes the profile if it becomes empty.
async fn delete_table(
&self,
request: tonic::Request<super::DeleteTableRequest>,
@@ -290,6 +362,10 @@ pub mod table_definition_server {
tonic::Status,
>;
}
/// The TableDefinition service manages the entire lifecycle of user-defined
/// tables (stored as both metadata and physical PostgreSQL tables) inside
/// logical "profiles" (schemas). Each table has stored structure, links, and
/// validation rules.
#[derive(Debug)]
pub struct TableDefinitionServer<T> {
inner: Arc<T>,

View File

@@ -1,21 +1,70 @@
// This file is @generated by prost-build.
/// Request to create or update a script bound to a specific table and column.
#[derive(serde::Serialize, serde::Deserialize)]
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct PostTableScriptRequest {
/// Required. The metadata ID from table_definitions.id that identifies the
/// table this script belongs to. The table must exist; its schema determines
/// where referenced tables/columns are validated and where dependencies are stored.
#[prost(int64, tag = "1")]
pub table_definition_id: i64,
/// Required. The target column in the target table that this script computes.
/// Must be an existing user-defined column in that table (not a system column).
/// System columns are reserved: "id", "deleted", "created_at".
/// The column's data type must NOT be one of the prohibited target types:
/// BIGINT, DATE, TIMESTAMPTZ
/// Note: BOOLEAN targets are allowed (values are converted to Steel #true/#false).
#[prost(string, tag = "2")]
pub target_column: ::prost::alloc::string::String,
/// Required. The script in the Steel DSL (S-expression style).
/// Syntax requirements:
/// - Non-empty, must start with '('
/// - Balanced parentheses
///
/// Referencing data:
/// - Structured table/column access (enforces link constraints):
/// (steel_get_column "table_name" "column_name")
/// (steel_get_column_with_index "table_name" index "column_name")
/// • index must be a non-negative integer literal
/// • self-references are allowed without links
/// • other tables require an explicit link from the source table
/// (table_definition_links) or the request fails
/// - Raw SQL access (no link required, but still validated):
/// (steel_query_sql "SELECT ...")
/// • Basic checks disallow operations that imply prohibited types,
/// e.g., EXTRACT(…), DATE_PART(…), ::DATE, ::TIMESTAMPTZ, ::BIGINT, CAST(…)
/// - Self variable access in transformed scripts:
/// (get-var "column_name") is treated as referencing the current table
///
/// Math operations:
/// - The script is transformed by steel_decimal; supported math forms include:
/// +, -, *, /, ^, **, pow, sqrt, >, <, =, >=, <=, min, max, abs, round,
/// ln, log, log10, exp, sin, cos, tan
/// - Columns of the following types CANNOT be used inside math expressions:
/// BIGINT, TEXT, BOOLEAN, DATE, TIMESTAMPTZ
///
/// Dependency tracking and cycles:
/// - Dependencies are extracted from steel_get_column(_with_index), get-var,
/// and steel_query_sql and stored in script_dependencies with context
/// - Cycles across tables are rejected (self-dependency is allowed)
#[prost(string, tag = "3")]
pub script: ::prost::alloc::string::String,
/// Optional. Free-text description stored alongside the script (no functional effect).
#[prost(string, tag = "4")]
pub description: ::prost::alloc::string::String,
}
/// Response after creating or updating a script.
#[derive(serde::Serialize, serde::Deserialize)]
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct TableScriptResponse {
/// The ID of the script record in table_scripts (new or existing on upsert).
#[prost(int64, tag = "1")]
pub id: i64,
/// Human-readable warnings concatenated into a single string. Possible messages:
/// - Warning if the script references itself (may affect first population)
/// - Count of raw SQL queries present
/// - Info about number of structured linked-table accesses
/// - Warning if many dependencies may affect performance
#[prost(string, tag = "2")]
pub warnings: ::prost::alloc::string::String,
}
@@ -30,6 +79,18 @@ pub mod table_script_client {
)]
use tonic::codegen::*;
use tonic::codegen::http::Uri;
/// Manages column-computation scripts for user-defined tables.
/// Each script belongs to a single table (table_definition_id) and populates
/// exactly one target column in that table. The server:
/// - Validates script syntax (non-empty, balanced parentheses, starts with '(')
/// - Validates the target column (exists, not a system column, allowed type)
/// - Validates column/type usage inside math expressions
/// - Validates referenced tables/columns against the schema
/// - Enforces link constraints for structured access (see notes below)
/// - Analyzes dependencies and prevents cycles across the schema
/// - Transforms the script to decimal-safe math (steel_decimal)
/// - Upserts into table_scripts and records dependencies in script_dependencies
/// The whole operation is transactional.
#[derive(Debug, Clone)]
pub struct TableScriptClient<T> {
inner: tonic::client::Grpc<T>,
@@ -110,6 +171,23 @@ pub mod table_script_client {
self.inner = self.inner.max_encoding_message_size(limit);
self
}
/// Create or update a script for a specific table and target column.
///
/// Behavior:
/// - Fetches the table by table_definition_id (must exist)
/// - Validates "script" (syntax), "target_column" (exists and type rules),
/// and all referenced tables/columns (must exist in same schema)
/// - Validates math operations: prohibits using certain data types in math
/// - Enforces link constraints for structured table access:
/// • Allowed always: self-references (same table)
/// • Structured access via steel_get_column / steel_get_column_with_index
/// requires an explicit link in table_definition_links
/// • Raw SQL access via steel_query_sql is permitted (still validated)
/// - Detects and rejects circular dependencies across all scripts in the schema
/// (self-references are allowed and not treated as cycles)
/// - Transforms the script to decimal-safe operations (steel_decimal)
/// - UPSERTS into table_scripts on (table_definitions_id, target_column)
/// and saves a normalized dependency list into script_dependencies
pub async fn post_table_script(
&mut self,
request: impl tonic::IntoRequest<super::PostTableScriptRequest>,
@@ -154,6 +232,23 @@ pub mod table_script_server {
/// Generated trait containing gRPC methods that should be implemented for use with TableScriptServer.
#[async_trait]
pub trait TableScript: std::marker::Send + std::marker::Sync + 'static {
/// Create or update a script for a specific table and target column.
///
/// Behavior:
/// - Fetches the table by table_definition_id (must exist)
/// - Validates "script" (syntax), "target_column" (exists and type rules),
/// and all referenced tables/columns (must exist in same schema)
/// - Validates math operations: prohibits using certain data types in math
/// - Enforces link constraints for structured table access:
/// • Allowed always: self-references (same table)
/// • Structured access via steel_get_column / steel_get_column_with_index
/// requires an explicit link in table_definition_links
/// • Raw SQL access via steel_query_sql is permitted (still validated)
/// - Detects and rejects circular dependencies across all scripts in the schema
/// (self-references are allowed and not treated as cycles)
/// - Transforms the script to decimal-safe operations (steel_decimal)
/// - UPSERTS into table_scripts on (table_definitions_id, target_column)
/// and saves a normalized dependency list into script_dependencies
async fn post_table_script(
&self,
request: tonic::Request<super::PostTableScriptRequest>,
@@ -162,6 +257,18 @@ pub mod table_script_server {
tonic::Status,
>;
}
/// Manages column-computation scripts for user-defined tables.
/// Each script belongs to a single table (table_definition_id) and populates
/// exactly one target column in that table. The server:
/// - Validates script syntax (non-empty, balanced parentheses, starts with '(')
/// - Validates the target column (exists, not a system column, allowed type)
/// - Validates column/type usage inside math expressions
/// - Validates referenced tables/columns against the schema
/// - Enforces link constraints for structured access (see notes below)
/// - Analyzes dependencies and prevents cycles across the schema
/// - Transforms the script to decimal-safe math (steel_decimal)
/// - Upserts into table_scripts and records dependencies in script_dependencies
/// The whole operation is transactional.
#[derive(Debug)]
pub struct TableScriptServer<T> {
inner: Arc<T>,

View File

@@ -1,27 +1,46 @@
// This file is @generated by prost-build.
/// Request identifying the profile (schema) and table to inspect.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct GetTableStructureRequest {
/// e.g., "default"
/// Required. Profile (PostgreSQL schema) name. Must exist in `schemas`.
#[prost(string, tag = "1")]
pub profile_name: ::prost::alloc::string::String,
/// e.g., "2025_adresar6"
/// 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,
}
/// Response with the ordered list of columns (by ordinal position).
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct TableStructureResponse {
/// Columns of the physical table, including system columns (id, deleted,
/// created_at), user-defined columns, and any foreign-key columns such as
/// "<linked_table>_id". May be empty if the physical table is missing.
#[prost(message, repeated, tag = "1")]
pub columns: ::prost::alloc::vec::Vec<TableColumn>,
}
/// One physical column entry as reported by information_schema.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct TableColumn {
/// Column name exactly as defined in PostgreSQL.
#[prost(string, tag = "1")]
pub name: ::prost::alloc::string::String,
/// e.g., "TEXT", "BIGINT", "VARCHAR(255)", "TIMESTAMPTZ"
/// Normalized data type string derived from information_schema:
/// - VARCHAR(n) when udt_name='varchar' with character_maximum_length
/// - CHAR(n) when udt_name='bpchar' with character_maximum_length
/// - NUMERIC(p,s) when udt_name='numeric' with precision and scale
/// - NUMERIC(p) when udt_name='numeric' with precision only
/// - <TYPE>\[\] for array types (udt_name starting with '_', e.g., INT\[\] )
/// - Otherwise UPPER(udt_name), e.g., TEXT, BIGINT, TIMESTAMPTZ
/// Examples: "TEXT", "BIGINT", "VARCHAR(255)", "TIMESTAMPTZ", "NUMERIC(14,4)"
#[prost(string, tag = "2")]
pub data_type: ::prost::alloc::string::String,
/// True if information_schema reports the column as nullable.
#[prost(bool, tag = "3")]
pub is_nullable: bool,
/// True if the column is part of the table's PRIMARY KEY.
/// Typically true for the "id" column created by the system.
#[prost(bool, tag = "4")]
pub is_primary_key: bool,
}
@@ -36,6 +55,14 @@ 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.
/// 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.
#[derive(Debug, Clone)]
pub struct TableStructureServiceClient<T> {
inner: tonic::client::Grpc<T>,
@@ -116,6 +143,16 @@ pub mod table_structure_service_client {
self.inner = self.inner.max_encoding_message_size(limit);
self
}
/// Return the physical column list (name, normalized data_type,
/// nullability, primary key flag) for a table 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`
/// - 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)
pub async fn get_table_structure(
&mut self,
request: impl tonic::IntoRequest<super::GetTableStructureRequest>,
@@ -160,6 +197,16 @@ pub mod table_structure_service_server {
/// Generated trait containing gRPC methods that should be implemented for use with TableStructureServiceServer.
#[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.
///
/// Behavior:
/// - NOT_FOUND if profile doesn't exist in `schemas`
/// - NOT_FOUND if table 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)
async fn get_table_structure(
&self,
request: tonic::Request<super::GetTableStructureRequest>,
@@ -168,6 +215,14 @@ pub mod table_structure_service_server {
tonic::Status,
>;
}
/// Introspects the physical PostgreSQL table for a given logical table
/// (defined in table_definitions) and returns its column structure.
/// 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.
#[derive(Debug)]
pub struct TableStructureServiceServer<T> {
inner: Arc<T>,

View File

@@ -26,7 +26,13 @@ pub struct FieldValidation {
#[prost(message, optional, tag = "10")]
pub limits: ::core::option::Option<CharacterLimits>,
/// Future expansion:
/// PatternRules pattern = 11;
///
/// Validation 2
#[prost(message, optional, tag = "11")]
pub pattern: ::core::option::Option<PatternRules>,
/// Validation 4 custom formatting logic
#[prost(message, optional, tag = "14")]
pub formatter: ::core::option::Option<CustomFormatter>,
#[prost(message, optional, tag = "3")]
pub mask: ::core::option::Option<DisplayMask>,
/// ExternalValidation external = 13;
@@ -65,6 +71,46 @@ pub struct DisplayMask {
#[prost(string, optional, tag = "3")]
pub template_char: ::core::option::Option<::prost::alloc::string::String>,
}
/// One positionbased validation rule, similar to CharacterFilter + PositionRange
#[derive(serde::Serialize, serde::Deserialize)]
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct 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
#[prost(string, tag = "1")]
pub range: ::prost::alloc::string::String,
/// Character filter type, caseinsensitive keywords:
/// "ALPHABETIC", "NUMERIC", "ALPHANUMERIC",
/// "ONEOF(<chars>)", "EXACT(:)", "CUSTOM(<name>)"
#[prost(string, tag = "2")]
pub filter: ::prost::alloc::string::String,
}
#[derive(serde::Serialize, serde::Deserialize)]
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct CustomFormatter {
/// Formatter type identifier; handled clientside.
/// Examples: "PSCFormatter", "PhoneFormatter", "CreditCardFormatter", "DateFormatter"
#[prost(string, tag = "1")]
pub r#type: ::prost::alloc::string::String,
/// Optional freetext note or parameters (e.g. locale, pattern)
#[prost(string, optional, tag = "2")]
pub description: ::core::option::Option<::prost::alloc::string::String>,
}
/// Collection of pattern rules for one field
#[derive(serde::Serialize, serde::Deserialize)]
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct PatternRules {
/// All rules that make up the validation logic
#[prost(message, repeated, tag = "1")]
pub rules: ::prost::alloc::vec::Vec<PatternRule>,
/// Optional humanreadable description for UI/debug purposes
#[prost(string, optional, tag = "2")]
pub description: ::core::option::Option<::prost::alloc::string::String>,
}
#[derive(serde::Serialize, serde::Deserialize)]
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct UpdateFieldValidationRequest {

View File

@@ -1,92 +1,170 @@
// This file is @generated by prost-build.
/// Insert a new row.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct PostTableDataRequest {
/// Required. Profile (PostgreSQL schema) name that owns the table.
/// Must exist in the schemas table.
#[prost(string, tag = "1")]
pub profile_name: ::prost::alloc::string::String,
/// Required. Logical table (definition) name within the profile.
/// Must exist in table_definitions for the given profile.
#[prost(string, tag = "2")]
pub table_name: ::prost::alloc::string::String,
/// Required. Key-value data for columns to insert.
///
/// Allowed keys:
/// - User-defined columns from the table definition
/// - System/FK columns:
/// • "deleted" (BOOLEAN), optional; default FALSE if not provided
/// • "<linked_table>_id" (BIGINT) for each table link
///
/// Type expectations by SQL type:
/// - TEXT: string value; empty string is treated as NULL
/// - BOOLEAN: bool value
/// - TIMESTAMPTZ: ISO 8601/RFC 3339 string (parsed to TIMESTAMPTZ)
/// - INTEGER: number with no fractional part and within i32 range
/// - BIGINT: number with no fractional part and within i64 range
/// - NUMERIC(p,s): string representation only; empty string becomes NULL
/// (numbers for NUMERIC are rejected to avoid precision loss)
///
/// Script validation rules:
/// - If a script exists for a target column, that column MUST be present here,
/// and its provided value MUST equal the scripts computed value (type-aware
/// comparison, e.g., decimals are compared numerically).
///
/// Notes:
/// - Unknown/invalid column names are rejected
/// - Some application-specific validations may apply (e.g., max length for
/// certain fields like "telefon")
#[prost(map = "string, message", tag = "3")]
pub data: ::std::collections::HashMap<
::prost::alloc::string::String,
::prost_types::Value,
>,
}
/// Insert response.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct PostTableDataResponse {
/// True if the insert succeeded.
#[prost(bool, tag = "1")]
pub success: bool,
/// Human-readable message.
#[prost(string, tag = "2")]
pub message: ::prost::alloc::string::String,
/// The id of the inserted row.
#[prost(int64, tag = "3")]
pub inserted_id: i64,
}
/// Update an existing row.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct PutTableDataRequest {
/// Required. Profile (schema) name.
#[prost(string, tag = "1")]
pub profile_name: ::prost::alloc::string::String,
/// Required. Table name within the profile.
#[prost(string, tag = "2")]
pub table_name: ::prost::alloc::string::String,
/// Required. Id of the row to update.
#[prost(int64, tag = "3")]
pub id: i64,
/// Required. Columns to update (same typing rules as PostTableDataRequest.data).
///
/// Special script rules:
/// - If a script targets column X and X is included here, the value for X must
/// equal the scripts result (type-aware).
/// - If X is not included here but the update would cause the scripts result
/// to change compared to the current stored value, the update is rejected with
/// FAILED_PRECONDITION, instructing the caller to include X explicitly.
///
/// Passing an empty map results in a no-op success response.
#[prost(map = "string, message", tag = "4")]
pub data: ::std::collections::HashMap<
::prost::alloc::string::String,
::prost_types::Value,
>,
}
/// Update response.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct PutTableDataResponse {
/// True if the update succeeded (or no-op on empty data).
#[prost(bool, tag = "1")]
pub success: bool,
/// Human-readable message.
#[prost(string, tag = "2")]
pub message: ::prost::alloc::string::String,
/// The id of the updated row.
#[prost(int64, tag = "3")]
pub updated_id: i64,
}
/// Soft-delete a single row.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct DeleteTableDataRequest {
/// Required. Profile (schema) name.
#[prost(string, tag = "1")]
pub profile_name: ::prost::alloc::string::String,
/// Required. Table name within the profile.
#[prost(string, tag = "2")]
pub table_name: ::prost::alloc::string::String,
/// Required. Row id to soft-delete.
#[prost(int64, tag = "3")]
pub record_id: i64,
}
/// Soft-delete response.
#[derive(Clone, Copy, PartialEq, ::prost::Message)]
pub struct DeleteTableDataResponse {
/// True if a row was marked deleted (id existed and was not already deleted).
#[prost(bool, tag = "1")]
pub success: bool,
}
/// Fetch a single non-deleted row by id.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct GetTableDataRequest {
/// Required. Profile (schema) name.
#[prost(string, tag = "1")]
pub profile_name: ::prost::alloc::string::String,
/// Required. Table name within the profile.
#[prost(string, tag = "2")]
pub table_name: ::prost::alloc::string::String,
/// Required. Id of the row to fetch.
#[prost(int64, tag = "3")]
pub id: i64,
}
/// Row payload: all columns returned as strings.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct GetTableDataResponse {
/// Map of column_name → stringified value for:
/// - id, deleted
/// - all user-defined columns from the table definition
/// - FK columns named "<linked_table>_id" for each table link
///
/// All values are returned as TEXT via col::TEXT and COALESCEed to empty string
/// (NULL becomes ""). The row is returned only if deleted = FALSE.
#[prost(map = "string, string", tag = "1")]
pub data: ::std::collections::HashMap<
::prost::alloc::string::String,
::prost::alloc::string::String,
>,
}
/// Count non-deleted rows.
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct GetTableDataCountRequest {
/// Required. Profile (schema) name.
#[prost(string, tag = "1")]
pub profile_name: ::prost::alloc::string::String,
/// Required. Table name within the profile.
#[prost(string, tag = "2")]
pub table_name: ::prost::alloc::string::String,
}
/// Fetch by ordinal position among non-deleted rows (1-based).
#[derive(Clone, PartialEq, ::prost::Message)]
pub struct GetTableDataByPositionRequest {
/// Required. Profile (schema) name.
#[prost(string, tag = "1")]
pub profile_name: ::prost::alloc::string::String,
/// Required. Table name within the profile.
#[prost(string, tag = "2")]
pub table_name: ::prost::alloc::string::String,
/// Required. 1-based position by id ascending among rows with deleted = FALSE.
#[prost(int32, tag = "3")]
pub position: i32,
}
@@ -101,6 +179,11 @@ pub mod tables_data_client {
)]
use tonic::codegen::*;
use tonic::codegen::http::Uri;
/// Read and write row data for user-defined tables inside profiles (schemas).
/// Operations are performed against the physical PostgreSQL table that
/// corresponds to the logical table definition and are scoped by profile
/// (schema). Deletions are soft (set deleted = true). Typed binding and
/// script-based validation are enforced consistently.
#[derive(Debug, Clone)]
pub struct TablesDataClient<T> {
inner: tonic::client::Grpc<T>,
@@ -181,6 +264,16 @@ pub mod tables_data_client {
self.inner = self.inner.max_encoding_message_size(limit);
self
}
/// Insert a new row into a table with strict type binding and script validation.
///
/// Behavior:
/// - Validates that profile (schema) exists and table is defined for it
/// - Validates provided columns exist (user-defined or allowed system/FK columns)
/// - For columns targeted by scripts in this table, the client MUST provide the
/// value, and it MUST equal the scripts calculated value (compared type-safely)
/// - Binds values with correct SQL types, rejects invalid formats/ranges
/// - Inserts the row and returns the new id; queues search indexing (best effort)
/// - If the physical table is missing but the definition exists, returns INTERNAL
pub async fn post_table_data(
&mut self,
request: impl tonic::IntoRequest<super::PostTableDataRequest>,
@@ -207,6 +300,17 @@ pub mod tables_data_client {
);
self.inner.unary(req, path, codec).await
}
/// Update existing row data with strict type binding and script validation.
///
/// Behavior:
/// - Validates profile and table, and that the record exists
/// - If request data is empty, returns success without changing the row
/// - For columns targeted by scripts:
/// • If included in update, provided value must equal the script result
/// • If not included, update must not cause the script result to differ
/// from the current stored value; otherwise FAILED_PRECONDITION is returned
/// - Binds values with correct SQL types; rejects invalid formats/ranges
/// - Updates the row and returns the id; queues search indexing (best effort)
pub async fn put_table_data(
&mut self,
request: impl tonic::IntoRequest<super::PutTableDataRequest>,
@@ -233,6 +337,13 @@ pub mod tables_data_client {
);
self.inner.unary(req, path, codec).await
}
/// Soft-delete a single record (sets deleted = true) if it exists and is not already deleted.
///
/// Behavior:
/// - Validates profile and table definition
/// - Updates only rows with deleted = false
/// - success = true means a row was actually changed; false means nothing to delete
/// - If the physical table is missing but the definition exists, returns INTERNAL
pub async fn delete_table_data(
&mut self,
request: impl tonic::IntoRequest<super::DeleteTableDataRequest>,
@@ -259,6 +370,15 @@ pub mod tables_data_client {
);
self.inner.unary(req, path, codec).await
}
/// Fetch a single non-deleted row by id as textified values.
///
/// Behavior:
/// - Validates profile and table definition
/// - Returns all columns as strings (COALESCE(col::TEXT, '') AS col)
/// including: id, deleted, all user-defined columns, and FK columns
/// named "<linked_table>_id" for each table link
/// - Fails with NOT_FOUND if record does not exist or is soft-deleted
/// - If the physical table is missing but the definition exists, returns INTERNAL
pub async fn get_table_data(
&mut self,
request: impl tonic::IntoRequest<super::GetTableDataRequest>,
@@ -285,6 +405,12 @@ pub mod tables_data_client {
);
self.inner.unary(req, path, codec).await
}
/// Count non-deleted rows in a table.
///
/// Behavior:
/// - Validates profile and table definition
/// - Returns komp_ac.common.CountResponse.count with rows where deleted = FALSE
/// - If the physical table is missing but the definition exists, returns INTERNAL
pub async fn get_table_data_count(
&mut self,
request: impl tonic::IntoRequest<super::GetTableDataCountRequest>,
@@ -314,6 +440,12 @@ pub mod tables_data_client {
);
self.inner.unary(req, path, codec).await
}
/// Fetch the N-th non-deleted row by id order (1-based), then return its full data.
///
/// Behavior:
/// - position is 1-based (position = 1 → first row by id ASC with deleted = FALSE)
/// - Returns NOT_FOUND if position is out of bounds
/// - Otherwise identical to GetTableData for the selected id
pub async fn get_table_data_by_position(
&mut self,
request: impl tonic::IntoRequest<super::GetTableDataByPositionRequest>,
@@ -358,6 +490,16 @@ pub mod tables_data_server {
/// Generated trait containing gRPC methods that should be implemented for use with TablesDataServer.
#[async_trait]
pub trait TablesData: std::marker::Send + std::marker::Sync + 'static {
/// Insert a new row into a table with strict type binding and script validation.
///
/// Behavior:
/// - Validates that profile (schema) exists and table is defined for it
/// - Validates provided columns exist (user-defined or allowed system/FK columns)
/// - For columns targeted by scripts in this table, the client MUST provide the
/// value, and it MUST equal the scripts calculated value (compared type-safely)
/// - Binds values with correct SQL types, rejects invalid formats/ranges
/// - Inserts the row and returns the new id; queues search indexing (best effort)
/// - If the physical table is missing but the definition exists, returns INTERNAL
async fn post_table_data(
&self,
request: tonic::Request<super::PostTableDataRequest>,
@@ -365,6 +507,17 @@ pub mod tables_data_server {
tonic::Response<super::PostTableDataResponse>,
tonic::Status,
>;
/// Update existing row data with strict type binding and script validation.
///
/// Behavior:
/// - Validates profile and table, and that the record exists
/// - If request data is empty, returns success without changing the row
/// - For columns targeted by scripts:
/// • If included in update, provided value must equal the script result
/// • If not included, update must not cause the script result to differ
/// from the current stored value; otherwise FAILED_PRECONDITION is returned
/// - Binds values with correct SQL types; rejects invalid formats/ranges
/// - Updates the row and returns the id; queues search indexing (best effort)
async fn put_table_data(
&self,
request: tonic::Request<super::PutTableDataRequest>,
@@ -372,6 +525,13 @@ pub mod tables_data_server {
tonic::Response<super::PutTableDataResponse>,
tonic::Status,
>;
/// Soft-delete a single record (sets deleted = true) if it exists and is not already deleted.
///
/// Behavior:
/// - Validates profile and table definition
/// - Updates only rows with deleted = false
/// - success = true means a row was actually changed; false means nothing to delete
/// - If the physical table is missing but the definition exists, returns INTERNAL
async fn delete_table_data(
&self,
request: tonic::Request<super::DeleteTableDataRequest>,
@@ -379,6 +539,15 @@ pub mod tables_data_server {
tonic::Response<super::DeleteTableDataResponse>,
tonic::Status,
>;
/// Fetch a single non-deleted row by id as textified values.
///
/// Behavior:
/// - Validates profile and table definition
/// - Returns all columns as strings (COALESCE(col::TEXT, '') AS col)
/// including: id, deleted, all user-defined columns, and FK columns
/// named "<linked_table>_id" for each table link
/// - Fails with NOT_FOUND if record does not exist or is soft-deleted
/// - If the physical table is missing but the definition exists, returns INTERNAL
async fn get_table_data(
&self,
request: tonic::Request<super::GetTableDataRequest>,
@@ -386,6 +555,12 @@ pub mod tables_data_server {
tonic::Response<super::GetTableDataResponse>,
tonic::Status,
>;
/// Count non-deleted rows in a table.
///
/// Behavior:
/// - Validates profile and table definition
/// - Returns komp_ac.common.CountResponse.count with rows where deleted = FALSE
/// - If the physical table is missing but the definition exists, returns INTERNAL
async fn get_table_data_count(
&self,
request: tonic::Request<super::GetTableDataCountRequest>,
@@ -393,6 +568,12 @@ pub mod tables_data_server {
tonic::Response<super::super::common::CountResponse>,
tonic::Status,
>;
/// Fetch the N-th non-deleted row by id order (1-based), then return its full data.
///
/// Behavior:
/// - position is 1-based (position = 1 → first row by id ASC with deleted = FALSE)
/// - Returns NOT_FOUND if position is out of bounds
/// - Otherwise identical to GetTableData for the selected id
async fn get_table_data_by_position(
&self,
request: tonic::Request<super::GetTableDataByPositionRequest>,
@@ -401,6 +582,11 @@ pub mod tables_data_server {
tonic::Status,
>;
}
/// Read and write row data for user-defined tables inside profiles (schemas).
/// Operations are performed against the physical PostgreSQL table that
/// corresponds to the logical table definition and are scoped by profile
/// (schema). Deletions are soft (set deleted = true). Typed binding and
/// script-based validation are enforced consistently.
#[derive(Debug)]
pub struct TablesDataServer<T> {
inner: Arc<T>,

View File

@@ -48,8 +48,7 @@ pub fn register_slovak_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)?)
let edge_tokenizer = TextAnalyzer::builder(NgramTokenizer::new(1, 4, true)?)
.filter(RemoveLongFilter::limit(40))
.filter(LowerCaser)
.filter(AsciiFoldingFilter)
@@ -57,8 +56,7 @@ pub fn register_slovak_tokenizers(index: &Index) -> tantivy::Result<()> {
tokenizer_manager.register("slovak_prefix_edge", edge_tokenizer);
// TOKENIZER for `prefix_full`: Simple word tokenizer
let full_tokenizer =
TextAnalyzer::builder(SimpleTokenizer::default())
let full_tokenizer = TextAnalyzer::builder(SimpleTokenizer::default())
.filter(RemoveLongFilter::limit(40))
.filter(LowerCaser)
.filter(AsciiFoldingFilter)
@@ -66,8 +64,7 @@ pub fn register_slovak_tokenizers(index: &Index) -> tantivy::Result<()> {
tokenizer_manager.register("slovak_prefix_full", full_tokenizer);
// NGRAM TOKENIZER: For substring matching.
let ngram_tokenizer =
TextAnalyzer::builder(NgramTokenizer::new(3, 3, false)?)
let ngram_tokenizer = TextAnalyzer::builder(NgramTokenizer::new(3, 3, false)?)
.filter(RemoveLongFilter::limit(40))
.filter(LowerCaser)
.filter(AsciiFoldingFilter)

View File

@@ -14,12 +14,17 @@
{
devShells.default = pkgs.mkShell {
buildInputs = with pkgs; [
mermaid-cli
# Rust toolchain
rustc
cargo
rustfmt
clippy
cargo-watch
rust-analyzer
cargo-tarpaulin
cargo-flamegraph
rust-code-analysis
# C build tools (for your linker issue)
gcc
@@ -36,6 +41,8 @@
# Protocol Buffers compiler for gRPC
protobuf
protoc-gen-doc
buf
];
shellHook = ''

View File

@@ -4,18 +4,15 @@ use std::collections::HashMap;
use std::path::Path;
use tantivy::collector::TopDocs;
use tantivy::query::{
BooleanQuery, BoostQuery, FuzzyTermQuery, Occur, Query, QueryParser,
TermQuery,
BooleanQuery, BoostQuery, FuzzyTermQuery, Occur, Query, QueryParser, TermQuery,
};
use tantivy::schema::{IndexRecordOption, Value};
use tantivy::{Index, TantivyDocument, Term};
use tonic::{Request, Response, Status};
use common::proto::komp_ac::search::{
search_response::Hit, SearchRequest, SearchResponse,
};
pub use common::proto::komp_ac::search::searcher_server::SearcherServer;
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 sqlx::{PgPool, Row};
use tracing::info;
@@ -86,22 +83,15 @@ impl Searcher for SearcherService {
qualified_table
);
let rows = sqlx::query(&sql)
.fetch_all(&self.pool)
.await
.map_err(|e| {
Status::internal(format!(
"DB query for default results failed: {}",
e
))
let rows = sqlx::query(&sql).fetch_all(&self.pool).await.map_err(|e| {
Status::internal(format!("DB query for default results failed: {}", e))
})?;
let hits: Vec<Hit> = 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();
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
@@ -111,7 +101,10 @@ impl Searcher for SearcherService {
})
.collect();
info!("--- SERVER: Successfully processed empty query. Returning {} default hits. ---", hits.len());
info!(
"--- SERVER: Successfully processed empty query. Returning {} default hits. ---",
hits.len()
);
return Ok(Response::new(SearchResponse { hits }));
}
// --- END OF MODIFIED LOGIC ---
@@ -131,15 +124,15 @@ impl Searcher for SearcherService {
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 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.")
})?;
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();
@@ -158,25 +151,17 @@ impl Searcher for SearcherService {
{
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 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,
)),
Box::new(TermQuery::new(edge_term, IndexRecordOption::Basic)),
),
(
Occur::Should,
Box::new(TermQuery::new(
full_term,
IndexRecordOption::Basic,
)),
Box::new(TermQuery::new(full_term, IndexRecordOption::Basic)),
),
]);
must_clauses.push((Occur::Must, Box::new(per_word_query) as Box<dyn Query>));
@@ -184,8 +169,7 @@ impl Searcher for SearcherService {
if !must_clauses.is_empty() {
let prefix_query = BooleanQuery::new(must_clauses);
let boosted_query =
BoostQuery::new(Box::new(prefix_query), 4.0);
let boosted_query = BoostQuery::new(Box::new(prefix_query), 4.0);
query_layers.push((Occur::Should, Box::new(boosted_query)));
}
}
@@ -195,8 +179,7 @@ impl Searcher for SearcherService {
// ===============================
{
let last_word = words.last().unwrap();
let fuzzy_term =
Term::from_field_text(prefix_full_field, last_word);
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)));
@@ -206,8 +189,7 @@ impl Searcher for SearcherService {
// 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_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);
@@ -219,11 +201,8 @@ impl Searcher for SearcherService {
// 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 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)));
}
@@ -244,9 +223,9 @@ impl Searcher for SearcherService {
// Step 1: Extract (score, pg_id) from Tantivy results.
let mut scored_ids: Vec<(f32, u64)> = Vec::new();
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))
})?;
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));
@@ -255,8 +234,7 @@ impl Searcher for SearcherService {
}
// 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 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)",
@@ -267,9 +245,7 @@ impl Searcher for SearcherService {
.bind(&pg_ids)
.fetch_all(&self.pool)
.await
.map_err(|e| {
Status::internal(format!("Database query failed: {}", e))
})?;
.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();
@@ -284,9 +260,7 @@ impl Searcher for SearcherService {
let hits: Vec<Hit> = scored_ids
.into_iter()
.filter_map(|(score, pg_id)| {
content_map
.get(&(pg_id as i64))
.map(|content_json| Hit {
content_map.get(&(pg_id as i64)).map(|content_json| Hit {
id: pg_id as i64,
score,
content_json: content_json.clone(),
@@ -294,7 +268,10 @@ impl Searcher for SearcherService {
})
.collect();
info!("--- SERVER: Successfully processed search. Returning {} hits. ---", hits.len());
info!(
"--- SERVER: Successfully processed search. Returning {} hits. ---",
hits.len()
);
let response = SearchResponse { hits };
Ok(Response::new(response))

2
server

Submodule server updated: e497676789...515f9932f8