@drzl/validation-core 3.13.0 → 3.15.0

package/dist/index.d.cts

@@ -1,4 +1,4 @@
-import { Column, Analysis } from '@drzl/analyzer';
+import { Table as Table$1, Column, Analysis } from '@drzl/analyzer';
 
 /**
  * One place that decides what a generated module is called, on disk and in an import.
@@ -284,6 +284,30 @@ declare function parseCheck(expression: string | undefined, name?: string): Pars
  */
 declare function describeSet(set: ColumnSet): string;
 
+/**
+ * A duplicate finder for a batch of rows, emitted beside the schemas.
+ *
+ * A validator checks one row at a time, so uniqueness is the one constraint it structurally
+ * cannot see: whether a value is unique is a fact about the table, not about the row. That is
+ * fine for a single insert, where the database answers immediately. It is not fine for a bulk
+ * insert, where a batch of a thousand rows fails whole on one collision and the error names a
+ * constraint rather than a row.
+ *
+ * What *is* checkable without the database is whether a batch collides with **itself**, and that
+ * is worth checking, because it is the half a user can fix before sending anything.
+ *
+ * The emitted function is plain TypeScript with no reference to any validation library, so all
+ * four generators emit the same thing. It is rendered from one place for that reason.
+ */
+
+/**
+ * `findDuplicate<Table>s` for a table with unique constraints, or nothing.
+ *
+ * `rowType` is the name of the insert type, which is what a caller has in hand before an insert.
+ * Passed in rather than derived, because each generator names its types differently.
+ */
+declare function renderDuplicateFinder(table: Table$1, fnName: string, rowType: string): string | undefined;
+
 interface Table {
     name: string;
     tsName: string;
@@ -451,14 +475,34 @@ declare const COLUMN_FORMATS: Record<string, string>;
  * Refusing a user's emoji is the failure mode this avoids, and it is the same rule applied
  * everywhere else here: never reject what the database accepts.
  *
- * `@sinclair/typebox` and ArkType cannot express it. Both state a length declaratively with no
- * predicate to hook, so they keep the UTF-16 form and are documented as approximate.
+ * All four generators count code points. `@sinclair/typebox` and ArkType cannot say it in their
+ * declarative forms, so neither uses `maxLength` or `string <= n`: TypeBox intersects a registered
+ * kind onto the field and ArkType puts a Type carrying a narrow there. Both cost something,
+ * TypeBox's cap no longer serialising into a JSON Schema, and emitting a number that means a
+ * different measurement is not a better trade.
+ *
+ * MySQL's TEXT family is a byte budget rather than a character count, carried separately as
+ * `maxBytes`. Two measurements on string columns in the same database, verified against a real
+ * MySQL 8 on utf8mb4: `varchar(10)` takes ten emoji, `tinytext` takes 63 of them and refuses 64.
  */
 declare const CODEPOINT_LENGTH = "[...v].length";
 declare function isIntegerColumn(c: Column): boolean;
 declare function insertColumns(table: Table): Column[];
 declare function updateColumns(table: Table): Column[];
 declare function selectColumns(table: Table): Column[];
+/**
+ * Pretty-print emitted code with whatever formatter the consumer already has.
+ *
+ * Both formatters are optional peers, reached at call time and never bundled. Prettier used to be
+ * bundled, because tsup resolves the specifier below statically and esbuild then inlined all of
+ * it: 11 MB per package across the three that had a copy of this function, roughly 32 MB for
+ * anyone installing @drzl/cli. It is `--external` in every build script that can reach it, and
+ * no-bundled-formatter.spec.ts builds those scripts and checks.
+ *
+ * Neither absence is an error. A consumer with no formatter gets the code as rendered, which is
+ * valid TypeScript that merely looks worse, and losing generated files at the last step would be
+ * a far worse trade than losing their whitespace.
+ */
 declare function formatCode(code: string, filePath: string, fmt?: FormatOptions): Promise<any>;
 
-export { AFFIX_PROBE_TABLE, type AffixIssue, type AffixOptions, type AffixValue, CODEPOINT_LENGTH, COLUMN_FORMATS, type CardinalityCheck, type ColumnCheck, type ColumnSet, DEFAULT_IMPORT_EXTENSION, DEFAULT_MODE_PREFIX, DEFAULT_SCHEMA_SUFFIX, DEFAULT_TYPE_SUFFIX, type FormatOptions, IMPORT_EXTENSIONS, type ImportExtension, type LengthCheck, NAME_MODES, type NameMode, type ParsedCheck, type ResolvedAffix, type RowCheck, type Table, type TableCase, type ValidationGenerateOptions, type ValidationLibrary, type ValidationRenderer, applyTableCase, describeSet, formatCode, importSpecifier, insertColumns, isGeneratedColumn, isIntegerColumn, moduleFileName, moduleSpecifier, parseCheck, pascalCase, resolveAffix, resolveConfiguredImport, schemaName, selectColumns, typeName, updateColumns, validateAffix };
+export { AFFIX_PROBE_TABLE, type AffixIssue, type AffixOptions, type AffixValue, CODEPOINT_LENGTH, COLUMN_FORMATS, type CardinalityCheck, type ColumnCheck, type ColumnSet, DEFAULT_IMPORT_EXTENSION, DEFAULT_MODE_PREFIX, DEFAULT_SCHEMA_SUFFIX, DEFAULT_TYPE_SUFFIX, type FormatOptions, IMPORT_EXTENSIONS, type ImportExtension, type LengthCheck, NAME_MODES, type NameMode, type ParsedCheck, type ResolvedAffix, type RowCheck, type Table, type TableCase, type ValidationGenerateOptions, type ValidationLibrary, type ValidationRenderer, applyTableCase, describeSet, formatCode, importSpecifier, insertColumns, isGeneratedColumn, isIntegerColumn, moduleFileName, moduleSpecifier, parseCheck, pascalCase, renderDuplicateFinder, resolveAffix, resolveConfiguredImport, schemaName, selectColumns, typeName, updateColumns, validateAffix };

package/dist/index.d.ts

@@ -1,4 +1,4 @@
-import { Column, Analysis } from '@drzl/analyzer';
+import { Table as Table$1, Column, Analysis } from '@drzl/analyzer';
 
 /**
  * One place that decides what a generated module is called, on disk and in an import.
@@ -284,6 +284,30 @@ declare function parseCheck(expression: string | undefined, name?: string): Pars
  */
 declare function describeSet(set: ColumnSet): string;
 
+/**
+ * A duplicate finder for a batch of rows, emitted beside the schemas.
+ *
+ * A validator checks one row at a time, so uniqueness is the one constraint it structurally
+ * cannot see: whether a value is unique is a fact about the table, not about the row. That is
+ * fine for a single insert, where the database answers immediately. It is not fine for a bulk
+ * insert, where a batch of a thousand rows fails whole on one collision and the error names a
+ * constraint rather than a row.
+ *
+ * What *is* checkable without the database is whether a batch collides with **itself**, and that
+ * is worth checking, because it is the half a user can fix before sending anything.
+ *
+ * The emitted function is plain TypeScript with no reference to any validation library, so all
+ * four generators emit the same thing. It is rendered from one place for that reason.
+ */
+
+/**
+ * `findDuplicate<Table>s` for a table with unique constraints, or nothing.
+ *
+ * `rowType` is the name of the insert type, which is what a caller has in hand before an insert.
+ * Passed in rather than derived, because each generator names its types differently.
+ */
+declare function renderDuplicateFinder(table: Table$1, fnName: string, rowType: string): string | undefined;
+
 interface Table {
     name: string;
     tsName: string;
@@ -451,14 +475,34 @@ declare const COLUMN_FORMATS: Record<string, string>;
  * Refusing a user's emoji is the failure mode this avoids, and it is the same rule applied
  * everywhere else here: never reject what the database accepts.
  *
- * `@sinclair/typebox` and ArkType cannot express it. Both state a length declaratively with no
- * predicate to hook, so they keep the UTF-16 form and are documented as approximate.
+ * All four generators count code points. `@sinclair/typebox` and ArkType cannot say it in their
+ * declarative forms, so neither uses `maxLength` or `string <= n`: TypeBox intersects a registered
+ * kind onto the field and ArkType puts a Type carrying a narrow there. Both cost something,
+ * TypeBox's cap no longer serialising into a JSON Schema, and emitting a number that means a
+ * different measurement is not a better trade.
+ *
+ * MySQL's TEXT family is a byte budget rather than a character count, carried separately as
+ * `maxBytes`. Two measurements on string columns in the same database, verified against a real
+ * MySQL 8 on utf8mb4: `varchar(10)` takes ten emoji, `tinytext` takes 63 of them and refuses 64.
  */
 declare const CODEPOINT_LENGTH = "[...v].length";
 declare function isIntegerColumn(c: Column): boolean;
 declare function insertColumns(table: Table): Column[];
 declare function updateColumns(table: Table): Column[];
 declare function selectColumns(table: Table): Column[];
+/**
+ * Pretty-print emitted code with whatever formatter the consumer already has.
+ *
+ * Both formatters are optional peers, reached at call time and never bundled. Prettier used to be
+ * bundled, because tsup resolves the specifier below statically and esbuild then inlined all of
+ * it: 11 MB per package across the three that had a copy of this function, roughly 32 MB for
+ * anyone installing @drzl/cli. It is `--external` in every build script that can reach it, and
+ * no-bundled-formatter.spec.ts builds those scripts and checks.
+ *
+ * Neither absence is an error. A consumer with no formatter gets the code as rendered, which is
+ * valid TypeScript that merely looks worse, and losing generated files at the last step would be
+ * a far worse trade than losing their whitespace.
+ */
 declare function formatCode(code: string, filePath: string, fmt?: FormatOptions): Promise<any>;
 
-export { AFFIX_PROBE_TABLE, type AffixIssue, type AffixOptions, type AffixValue, CODEPOINT_LENGTH, COLUMN_FORMATS, type CardinalityCheck, type ColumnCheck, type ColumnSet, DEFAULT_IMPORT_EXTENSION, DEFAULT_MODE_PREFIX, DEFAULT_SCHEMA_SUFFIX, DEFAULT_TYPE_SUFFIX, type FormatOptions, IMPORT_EXTENSIONS, type ImportExtension, type LengthCheck, NAME_MODES, type NameMode, type ParsedCheck, type ResolvedAffix, type RowCheck, type Table, type TableCase, type ValidationGenerateOptions, type ValidationLibrary, type ValidationRenderer, applyTableCase, describeSet, formatCode, importSpecifier, insertColumns, isGeneratedColumn, isIntegerColumn, moduleFileName, moduleSpecifier, parseCheck, pascalCase, resolveAffix, resolveConfiguredImport, schemaName, selectColumns, typeName, updateColumns, validateAffix };
+export { AFFIX_PROBE_TABLE, type AffixIssue, type AffixOptions, type AffixValue, CODEPOINT_LENGTH, COLUMN_FORMATS, type CardinalityCheck, type ColumnCheck, type ColumnSet, DEFAULT_IMPORT_EXTENSION, DEFAULT_MODE_PREFIX, DEFAULT_SCHEMA_SUFFIX, DEFAULT_TYPE_SUFFIX, type FormatOptions, IMPORT_EXTENSIONS, type ImportExtension, type LengthCheck, NAME_MODES, type NameMode, type ParsedCheck, type ResolvedAffix, type RowCheck, type Table, type TableCase, type ValidationGenerateOptions, type ValidationLibrary, type ValidationRenderer, applyTableCase, describeSet, formatCode, importSpecifier, insertColumns, isGeneratedColumn, isIntegerColumn, moduleFileName, moduleSpecifier, parseCheck, pascalCase, renderDuplicateFinder, resolveAffix, resolveConfiguredImport, schemaName, selectColumns, typeName, updateColumns, validateAffix };

package/dist/index.js

@@ -1,5 +1,3 @@
-import "./chunk-ZZGILH5A.js";
-
 // src/checks.ts
 var COMPARISON = /^\s*([A-Za-z_][A-Za-z0-9_]*)\s*(>=|<=|<>|!=|>|<|=)\s*(.+?)\s*$/;
 var IN_LIST = /^\s*([A-Za-z_][A-Za-z0-9_]*)\s+IN\s*\((.+)\)\s*$/i;
@@ -379,6 +377,54 @@ function validateAffix(affix, schemaSuffix) {
   return issues;
 }
 
+// src/duplicates.ts
+function usableKeys(table) {
+  return (table.unique ?? []).filter((k) => k.columns.length > 0);
+}
+function renderDuplicateFinder(table, fnName, rowType) {
+  const keys = usableKeys(table);
+  if (!keys.length) return void 0;
+  const constraints = keys.map((k, i) => {
+    const name = k.name ?? k.columns.join("_");
+    return `  { name: ${JSON.stringify(name)}, columns: ${JSON.stringify(k.columns)} }${i === keys.length - 1 ? "" : ","}`;
+  }).join("\n");
+  return `/**
+ * Rows in \`rows\` that collide with an earlier row on a unique constraint.
+ *
+ * Uniqueness is a fact about the table rather than about a row, so no schema can check it. This
+ * checks the half that needs no database: whether the batch collides with itself. A batch that
+ * passes here can still collide with rows already stored.
+ *
+ * A constraint is skipped for any row where one of its columns is null or absent, matching SQL,
+ * where NULL is not equal to NULL and a unique index therefore permits repeats.
+ */
+export function ${fnName}(
+  rows: readonly ${rowType}[]
+): Array<{ index: number; constraint: string; firstIndex: number }> {
+  const constraints = [
+${constraints}
+  ] as const;
+  const seen = constraints.map(() => new Map<string, number>());
+  const out: Array<{ index: number; constraint: string; firstIndex: number }> = [];
+
+  for (let i = 0; i < rows.length; i++) {
+    const row = rows[i] as Record<string, unknown>;
+    for (let c = 0; c < constraints.length; c++) {
+      const cols = constraints[c].columns;
+      const values = cols.map((col) => row?.[col]);
+      if (values.some((v) => v === null || v === undefined)) continue;
+      // JSON, so a composite key compares by value and \`[1, "2"]\` never collides with
+      // \`["1", 2]\`. A join on a separator would.
+      const key = JSON.stringify(values);
+      const first = seen[c].get(key);
+      if (first === undefined) seen[c].set(key, i);
+      else out.push({ index: i, constraint: constraints[c].name, firstIndex: first });
+    }
+  }
+  return out;
+}`;
+}
+
 // src/index.ts
 function isGeneratedColumn(c, _primaryKeyColumns = []) {
   return c.isGenerated;
@@ -409,7 +455,7 @@ async function formatCode(code, filePath, fmt) {
   const engine = fmt?.engine ?? "auto";
   try {
     if (engine === "prettier" || engine === "auto") {
-      const prettier = await import("./prettier-LT6G5PW7.js");
+      const prettier = await import("prettier");
       const cfgRef = fmt?.configPath ?? filePath;
       const cfg = await prettier.resolveConfig(cfgRef).catch(() => null);
       return prettier.format(code, { ...cfg ?? {}, parser: "typescript", filepath: filePath });
@@ -424,6 +470,10 @@ async function formatCode(code, filePath, fmt) {
         const res = await biome.formatContent(code, { filePath });
         return (res && (res.content || res.formatted)) ?? code;
       }
+      if (biome?.format) {
+        const res = await biome.format(code, { filePath });
+        return res ?? code;
+      }
     }
   } catch {
   }
@@ -450,6 +500,7 @@ export {
   moduleSpecifier,
   parseCheck,
   pascalCase,
+  renderDuplicateFinder,
   resolveAffix,
   resolveConfiguredImport,
   schemaName,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@drzl/validation-core",
-  "version": "3.13.0",
+  "version": "3.15.0",
   "private": false,
   "license": "Apache-2.0",
   "type": "module",
@@ -11,7 +11,15 @@
   ],
   "sideEffects": false,
   "dependencies": {
-    "@drzl/analyzer": "^1.11.0"
+    "@drzl/analyzer": "^1.14.0"
+  },
+  "peerDependencies": {
+    "prettier": ">=3"
+  },
+  "peerDependenciesMeta": {
+    "prettier": {
+      "optional": true
+    }
   },
   "devDependencies": {
     "tsup": "^8.5.1",
@@ -35,7 +43,7 @@
     "url": "https://github.com/sponsors/omar-dulaimi"
   },
   "scripts": {
-    "build": "tsup src/index.ts --dts --format esm,cjs",
-    "test": "vitest run"
+    "build": "tsup src/index.ts --dts --format esm,cjs --external prettier --clean",
+    "test": "vitest run --testTimeout=20000"
   }
 }