@drzl/validation-core 3.22.1 → 3.22.3

package/dist/index.cjs

@@ -55,6 +55,7 @@ __export(index_exports, {
   canonicalMembers: () => canonicalMembers,
   canonicalNumericText: () => canonicalNumericText,
   classifyTableChecks: () => classifyTableChecks,
+  codePointCompare: () => codePointCompare,
   columnMetaFacts: () => columnMetaFacts,
   comparisonWire: () => comparisonWire,
   describeSet: () => describeSet,
@@ -67,6 +68,7 @@ __export(index_exports, {
   isProjectInstallPath: () => isProjectInstallPath,
   lengthCheckLabel: () => lengthCheckLabel,
   lengthMeasure: () => lengthMeasure,
+  measureCompare: () => measureCompare,
   measureExpression: () => measureExpression,
   moduleFileName: () => moduleFileName,
   moduleSpecifier: () => moduleSpecifier,
@@ -112,6 +114,34 @@ function lengthMeasure(column, check) {
   if (column.tsType !== "string") return void 0;
   return check.unit === "bytes" ? "utf8Bytes" : "codePoints";
 }
+function codePointCompare(variable, operator, bound) {
+  const units = `${variable}.length`;
+  const spread2 = `[...${variable}].length`;
+  switch (operator) {
+    case "<=":
+    case "<":
+      return `(${units} ${operator} ${bound} || ${spread2} ${operator} ${bound})`;
+    case ">=":
+    case ">":
+      return `(${units} ${operator} ${bound} && ${spread2} ${operator} ${bound})`;
+    case "===":
+      return `(${units} >= ${bound} && ${spread2} === ${bound})`;
+    case "!==":
+      return `(${units} < ${bound} || ${spread2} !== ${bound})`;
+  }
+}
+function measureCompare(measure, variable, operator, bound) {
+  const js = {
+    ">=": ">=",
+    ">": ">",
+    "<=": "<=",
+    "<": "<",
+    "=": "===",
+    "<>": "!=="
+  }[operator];
+  if (measure === "codePoints") return codePointCompare(variable, js, bound);
+  return `${measureExpression(measure, variable)} ${js} ${bound}`;
+}
 function measureExpression(measure, variable) {
   switch (measure) {
     case "codePoints":
@@ -1743,7 +1773,18 @@ var COLUMN_FORMATS = {
   // is refused. A pattern cannot do that arithmetic, and guessing at it turns away working rows.
   // Over 3319 probes against each of a signed and an unsigned column: zero values the server takes
   // and this refuses.
-  mysqlBigint: "^\\s*[+-]?(\\d+(\\.\\d*)?|\\.\\d+)([eE][+-]?\\d*)?\\s*$"
+  mysqlBigint: "^\\s*[+-]?(\\d+(\\.\\d*)?|\\.\\d+)([eE][+-]?\\d*)?\\s*$",
+  // A temporal column carried as text, and the only thing that can be said about one: it is not
+  // blank. Unanchored on purpose, so it means "holds at least one non-whitespace character".
+  //
+  // Everything stronger was refused for the reason the analyzer's `format` comment gives: Postgres
+  // reads 'today', 'January 8, 1999', '01/08/1999' and '20200101' as dates, so a date-shaped
+  // pattern turns away rows the server stores. This one turns away nothing: measured on Postgres,
+  // every temporal type accepts a valid value with surrounding whitespace and refuses '' and ' '
+  // alike, so the set this refuses and the set the server refuses are the same set. The analyzer
+  // decides which columns carry it, per engine and per type, since MySQL's `time` takes a blank
+  // and silently stores 00:00:00.
+  temporalText: "\\S"
 };
 var COERCIBLE_DATE_STRING = "^(?!\\s*[+-])(?!\\s*\\d*\\.?\\d*(?:[eE][+-]?\\d+)?\\s*$)";
 function parsesToADate(expr) {
@@ -1933,6 +1974,7 @@ async function formatCode(code, filePath, fmt) {
   canonicalMembers,
   canonicalNumericText,
   classifyTableChecks,
+  codePointCompare,
   columnMetaFacts,
   comparisonWire,
   describeSet,
@@ -1945,6 +1987,7 @@ async function formatCode(code, filePath, fmt) {
   isProjectInstallPath,
   lengthCheckLabel,
   lengthMeasure,
+  measureCompare,
   measureExpression,
   moduleFileName,
   moduleSpecifier,

package/dist/index.d.cts

@@ -504,6 +504,35 @@ declare function lengthMeasure(column: {
  * TypeBox reach it off the row object as `o["blob"]`. `CODEPOINT_LENGTH` is the fixed-variable
  * spelling of the first arm and stays for the call sites that already use it.
  */
+/**
+ * A code-point count compared against a bound, without spreading the string when it cannot matter.
+ *
+ * `[...v].length` allocates an array of code points, and it sits on the path every ordinary row
+ * takes. A UTF-16 unit count is free and is never *smaller* than a code-point count, since only a
+ * surrogate pair costs two units for one code point. That one fact settles every operator here:
+ *
+ *   `<=`, `<`   units within the bound means code points are too      short-circuit to ACCEPT
+ *   `>=`, `>`   units under the bound means code points are too       short-circuit to REJECT
+ *   `===`       units under the bound means code points cannot equal it   short-circuit to REJECT
+ *   `!==`       units under the bound means code points cannot equal it   short-circuit to ACCEPT
+ *
+ * So the spread runs only for a string long enough in units to be in doubt, which for a cap is the
+ * only string that was ever in doubt. Measured over a 20000-string random pool plus the awkward
+ * cases (emoji, combining marks, a lone surrogate, empty, CJK at the bound): the two forms answer
+ * identically on every one. Throughput on the check alone, cap 64: ordinary ASCII rows 23171k/s
+ * against 215029k/s, rows at the cap 4250k/s against 235657k/s.
+ *
+ * Only the code-point measurement is rewritten. A `byteLength` reads `v.length` already and has
+ * nothing to avoid, and the UTF-8 arm has its own sound short-circuits in both directions but a
+ * different allocation to weigh, so it is left for its own measurement rather than folded in here.
+ */
+declare function codePointCompare(variable: string, operator: '>=' | '>' | '<=' | '<' | '===' | '!==', bound: string | number): string;
+/**
+ * A length comparison over whichever measurement the column answers, short-circuited where that is
+ * sound. The one place an operator and a count meet, so no generator has to re-derive which
+ * rewrites are safe; see `codePointCompare`.
+ */
+declare function measureCompare(measure: LengthMeasure, variable: string, operator: LengthCheck['operator'], bound: string | number): string;
 declare function measureExpression(measure: LengthMeasure, variable: string): string;
 /**
  * A count constraint as the sentence every emitted schema attaches and the ledger keys on.
@@ -1668,4 +1697,4 @@ declare function isProjectInstallPath(manifestPath: string): boolean;
  */
 declare function formatCode(code: string, filePath: string, fmt?: FormatOptions): Promise<any>;
 
-export { AFFIX_PREFIX_PATTERN, AFFIX_PROBE_TABLE, AFFIX_SUFFIX_PATTERN, type AffixIssue, type AffixOptions, type AffixValue, type BrandAlias, type BrandPlan, type BrandingOption, type BrandingOptions, CODEPOINT_LENGTH, COERCIBLE_DATE_STRING, COLUMN_FORMATS, CONSTRAINTS_MODULE, type CardinalityCheck, type CheckPart, type CheckPartPlace, type ClassifiedCheck, type ColumnCheck, type ColumnMetaFacts, type ColumnSet, type ComparisonWire, type ConstraintFacts, type ConstraintKind, type ConstraintsModuleOptions, type ConstraintsOption, DEFAULT_IMPORT_EXTENSION, DEFAULT_MODE_PREFIX, DEFAULT_NESTED_DEPTH, DEFAULT_SCHEMA_SUFFIX, DEFAULT_TYPE_SUFFIX, type FileSink, type FormatOptions, type GeneratorFs, IMPORT_EXTENSIONS, type ImportExtension, type LengthCheck, type LengthMeasure, MAX_NESTED_DEPTH, type MetaFactOptions, NAME_MODES, NESTED_PREFIX, NUMERIC_CANON_NAME, NUMERIC_CANON_SOURCE, type NameMode, type NestedArm, type NestedMode, type NestedNode, type NullCheck, type ParsedCheck, type ResolvedAffix, type ResolvedBranding, type RowCheck, type Table, type TableCase, type TableConstraints, type TableMetaFacts, type TableMetaOptions, type UnenforcedLiteral, type UnenforcedPart, type ValidationGenerateOptions, type ValidationLibrary, type ValidationRenderer, type WireLiteralFit, type WireLiteralQuestion, applyTableCase, applyWirePolicy, buildBrandPlan, buildNestedPlan, canonicalMembers, canonicalNumericText, classifyTableChecks, columnMetaFacts, comparisonWire, describeSet, fileWriter, formatCode, importSpecifier, insertColumns, isGeneratedColumn, isIntegerColumn, isProjectInstallPath, lengthCheckLabel, lengthMeasure, measureExpression, moduleFileName, moduleSpecifier, needsNumericCanon, nestedArmNotes, nestedNodeColumns, nestedSchemaName, nestedTypeName, nonFiniteAccepted, nonFiniteRefused, parseCheck, parsesToADate, pascalCase, renderConstraintsModule, renderDuplicateFinder, resolveAffix, resolveBranding, resolveConfiguredImport, resolveConstraints, resolveNestedDepth, schemaName, selectColumns, tableConstraints, tableMetaFacts, typeName, updateColumns, validateAffix, wireLiteralFit, wireNumberLiteral };
+export { AFFIX_PREFIX_PATTERN, AFFIX_PROBE_TABLE, AFFIX_SUFFIX_PATTERN, type AffixIssue, type AffixOptions, type AffixValue, type BrandAlias, type BrandPlan, type BrandingOption, type BrandingOptions, CODEPOINT_LENGTH, COERCIBLE_DATE_STRING, COLUMN_FORMATS, CONSTRAINTS_MODULE, type CardinalityCheck, type CheckPart, type CheckPartPlace, type ClassifiedCheck, type ColumnCheck, type ColumnMetaFacts, type ColumnSet, type ComparisonWire, type ConstraintFacts, type ConstraintKind, type ConstraintsModuleOptions, type ConstraintsOption, DEFAULT_IMPORT_EXTENSION, DEFAULT_MODE_PREFIX, DEFAULT_NESTED_DEPTH, DEFAULT_SCHEMA_SUFFIX, DEFAULT_TYPE_SUFFIX, type FileSink, type FormatOptions, type GeneratorFs, IMPORT_EXTENSIONS, type ImportExtension, type LengthCheck, type LengthMeasure, MAX_NESTED_DEPTH, type MetaFactOptions, NAME_MODES, NESTED_PREFIX, NUMERIC_CANON_NAME, NUMERIC_CANON_SOURCE, type NameMode, type NestedArm, type NestedMode, type NestedNode, type NullCheck, type ParsedCheck, type ResolvedAffix, type ResolvedBranding, type RowCheck, type Table, type TableCase, type TableConstraints, type TableMetaFacts, type TableMetaOptions, type UnenforcedLiteral, type UnenforcedPart, type ValidationGenerateOptions, type ValidationLibrary, type ValidationRenderer, type WireLiteralFit, type WireLiteralQuestion, applyTableCase, applyWirePolicy, buildBrandPlan, buildNestedPlan, canonicalMembers, canonicalNumericText, classifyTableChecks, codePointCompare, columnMetaFacts, comparisonWire, describeSet, fileWriter, formatCode, importSpecifier, insertColumns, isGeneratedColumn, isIntegerColumn, isProjectInstallPath, lengthCheckLabel, lengthMeasure, measureCompare, measureExpression, moduleFileName, moduleSpecifier, needsNumericCanon, nestedArmNotes, nestedNodeColumns, nestedSchemaName, nestedTypeName, nonFiniteAccepted, nonFiniteRefused, parseCheck, parsesToADate, pascalCase, renderConstraintsModule, renderDuplicateFinder, resolveAffix, resolveBranding, resolveConfiguredImport, resolveConstraints, resolveNestedDepth, schemaName, selectColumns, tableConstraints, tableMetaFacts, typeName, updateColumns, validateAffix, wireLiteralFit, wireNumberLiteral };

package/dist/index.d.ts

@@ -504,6 +504,35 @@ declare function lengthMeasure(column: {
  * TypeBox reach it off the row object as `o["blob"]`. `CODEPOINT_LENGTH` is the fixed-variable
  * spelling of the first arm and stays for the call sites that already use it.
  */
+/**
+ * A code-point count compared against a bound, without spreading the string when it cannot matter.
+ *
+ * `[...v].length` allocates an array of code points, and it sits on the path every ordinary row
+ * takes. A UTF-16 unit count is free and is never *smaller* than a code-point count, since only a
+ * surrogate pair costs two units for one code point. That one fact settles every operator here:
+ *
+ *   `<=`, `<`   units within the bound means code points are too      short-circuit to ACCEPT
+ *   `>=`, `>`   units under the bound means code points are too       short-circuit to REJECT
+ *   `===`       units under the bound means code points cannot equal it   short-circuit to REJECT
+ *   `!==`       units under the bound means code points cannot equal it   short-circuit to ACCEPT
+ *
+ * So the spread runs only for a string long enough in units to be in doubt, which for a cap is the
+ * only string that was ever in doubt. Measured over a 20000-string random pool plus the awkward
+ * cases (emoji, combining marks, a lone surrogate, empty, CJK at the bound): the two forms answer
+ * identically on every one. Throughput on the check alone, cap 64: ordinary ASCII rows 23171k/s
+ * against 215029k/s, rows at the cap 4250k/s against 235657k/s.
+ *
+ * Only the code-point measurement is rewritten. A `byteLength` reads `v.length` already and has
+ * nothing to avoid, and the UTF-8 arm has its own sound short-circuits in both directions but a
+ * different allocation to weigh, so it is left for its own measurement rather than folded in here.
+ */
+declare function codePointCompare(variable: string, operator: '>=' | '>' | '<=' | '<' | '===' | '!==', bound: string | number): string;
+/**
+ * A length comparison over whichever measurement the column answers, short-circuited where that is
+ * sound. The one place an operator and a count meet, so no generator has to re-derive which
+ * rewrites are safe; see `codePointCompare`.
+ */
+declare function measureCompare(measure: LengthMeasure, variable: string, operator: LengthCheck['operator'], bound: string | number): string;
 declare function measureExpression(measure: LengthMeasure, variable: string): string;
 /**
  * A count constraint as the sentence every emitted schema attaches and the ledger keys on.
@@ -1668,4 +1697,4 @@ declare function isProjectInstallPath(manifestPath: string): boolean;
  */
 declare function formatCode(code: string, filePath: string, fmt?: FormatOptions): Promise<any>;
 
-export { AFFIX_PREFIX_PATTERN, AFFIX_PROBE_TABLE, AFFIX_SUFFIX_PATTERN, type AffixIssue, type AffixOptions, type AffixValue, type BrandAlias, type BrandPlan, type BrandingOption, type BrandingOptions, CODEPOINT_LENGTH, COERCIBLE_DATE_STRING, COLUMN_FORMATS, CONSTRAINTS_MODULE, type CardinalityCheck, type CheckPart, type CheckPartPlace, type ClassifiedCheck, type ColumnCheck, type ColumnMetaFacts, type ColumnSet, type ComparisonWire, type ConstraintFacts, type ConstraintKind, type ConstraintsModuleOptions, type ConstraintsOption, DEFAULT_IMPORT_EXTENSION, DEFAULT_MODE_PREFIX, DEFAULT_NESTED_DEPTH, DEFAULT_SCHEMA_SUFFIX, DEFAULT_TYPE_SUFFIX, type FileSink, type FormatOptions, type GeneratorFs, IMPORT_EXTENSIONS, type ImportExtension, type LengthCheck, type LengthMeasure, MAX_NESTED_DEPTH, type MetaFactOptions, NAME_MODES, NESTED_PREFIX, NUMERIC_CANON_NAME, NUMERIC_CANON_SOURCE, type NameMode, type NestedArm, type NestedMode, type NestedNode, type NullCheck, type ParsedCheck, type ResolvedAffix, type ResolvedBranding, type RowCheck, type Table, type TableCase, type TableConstraints, type TableMetaFacts, type TableMetaOptions, type UnenforcedLiteral, type UnenforcedPart, type ValidationGenerateOptions, type ValidationLibrary, type ValidationRenderer, type WireLiteralFit, type WireLiteralQuestion, applyTableCase, applyWirePolicy, buildBrandPlan, buildNestedPlan, canonicalMembers, canonicalNumericText, classifyTableChecks, columnMetaFacts, comparisonWire, describeSet, fileWriter, formatCode, importSpecifier, insertColumns, isGeneratedColumn, isIntegerColumn, isProjectInstallPath, lengthCheckLabel, lengthMeasure, measureExpression, moduleFileName, moduleSpecifier, needsNumericCanon, nestedArmNotes, nestedNodeColumns, nestedSchemaName, nestedTypeName, nonFiniteAccepted, nonFiniteRefused, parseCheck, parsesToADate, pascalCase, renderConstraintsModule, renderDuplicateFinder, resolveAffix, resolveBranding, resolveConfiguredImport, resolveConstraints, resolveNestedDepth, schemaName, selectColumns, tableConstraints, tableMetaFacts, typeName, updateColumns, validateAffix, wireLiteralFit, wireNumberLiteral };
+export { AFFIX_PREFIX_PATTERN, AFFIX_PROBE_TABLE, AFFIX_SUFFIX_PATTERN, type AffixIssue, type AffixOptions, type AffixValue, type BrandAlias, type BrandPlan, type BrandingOption, type BrandingOptions, CODEPOINT_LENGTH, COERCIBLE_DATE_STRING, COLUMN_FORMATS, CONSTRAINTS_MODULE, type CardinalityCheck, type CheckPart, type CheckPartPlace, type ClassifiedCheck, type ColumnCheck, type ColumnMetaFacts, type ColumnSet, type ComparisonWire, type ConstraintFacts, type ConstraintKind, type ConstraintsModuleOptions, type ConstraintsOption, DEFAULT_IMPORT_EXTENSION, DEFAULT_MODE_PREFIX, DEFAULT_NESTED_DEPTH, DEFAULT_SCHEMA_SUFFIX, DEFAULT_TYPE_SUFFIX, type FileSink, type FormatOptions, type GeneratorFs, IMPORT_EXTENSIONS, type ImportExtension, type LengthCheck, type LengthMeasure, MAX_NESTED_DEPTH, type MetaFactOptions, NAME_MODES, NESTED_PREFIX, NUMERIC_CANON_NAME, NUMERIC_CANON_SOURCE, type NameMode, type NestedArm, type NestedMode, type NestedNode, type NullCheck, type ParsedCheck, type ResolvedAffix, type ResolvedBranding, type RowCheck, type Table, type TableCase, type TableConstraints, type TableMetaFacts, type TableMetaOptions, type UnenforcedLiteral, type UnenforcedPart, type ValidationGenerateOptions, type ValidationLibrary, type ValidationRenderer, type WireLiteralFit, type WireLiteralQuestion, applyTableCase, applyWirePolicy, buildBrandPlan, buildNestedPlan, canonicalMembers, canonicalNumericText, classifyTableChecks, codePointCompare, columnMetaFacts, comparisonWire, describeSet, fileWriter, formatCode, importSpecifier, insertColumns, isGeneratedColumn, isIntegerColumn, isProjectInstallPath, lengthCheckLabel, lengthMeasure, measureCompare, measureExpression, moduleFileName, moduleSpecifier, needsNumericCanon, nestedArmNotes, nestedNodeColumns, nestedSchemaName, nestedTypeName, nonFiniteAccepted, nonFiniteRefused, parseCheck, parsesToADate, pascalCase, renderConstraintsModule, renderDuplicateFinder, resolveAffix, resolveBranding, resolveConfiguredImport, resolveConstraints, resolveNestedDepth, schemaName, selectColumns, tableConstraints, tableMetaFacts, typeName, updateColumns, validateAffix, wireLiteralFit, wireNumberLiteral };

package/dist/index.js

@@ -13,6 +13,34 @@ function lengthMeasure(column, check) {
   if (column.tsType !== "string") return void 0;
   return check.unit === "bytes" ? "utf8Bytes" : "codePoints";
 }
+function codePointCompare(variable, operator, bound) {
+  const units = `${variable}.length`;
+  const spread2 = `[...${variable}].length`;
+  switch (operator) {
+    case "<=":
+    case "<":
+      return `(${units} ${operator} ${bound} || ${spread2} ${operator} ${bound})`;
+    case ">=":
+    case ">":
+      return `(${units} ${operator} ${bound} && ${spread2} ${operator} ${bound})`;
+    case "===":
+      return `(${units} >= ${bound} && ${spread2} === ${bound})`;
+    case "!==":
+      return `(${units} < ${bound} || ${spread2} !== ${bound})`;
+  }
+}
+function measureCompare(measure, variable, operator, bound) {
+  const js = {
+    ">=": ">=",
+    ">": ">",
+    "<=": "<=",
+    "<": "<",
+    "=": "===",
+    "<>": "!=="
+  }[operator];
+  if (measure === "codePoints") return codePointCompare(variable, js, bound);
+  return `${measureExpression(measure, variable)} ${js} ${bound}`;
+}
 function measureExpression(measure, variable) {
   switch (measure) {
     case "codePoints":
@@ -1644,7 +1672,18 @@ var COLUMN_FORMATS = {
   // is refused. A pattern cannot do that arithmetic, and guessing at it turns away working rows.
   // Over 3319 probes against each of a signed and an unsigned column: zero values the server takes
   // and this refuses.
-  mysqlBigint: "^\\s*[+-]?(\\d+(\\.\\d*)?|\\.\\d+)([eE][+-]?\\d*)?\\s*$"
+  mysqlBigint: "^\\s*[+-]?(\\d+(\\.\\d*)?|\\.\\d+)([eE][+-]?\\d*)?\\s*$",
+  // A temporal column carried as text, and the only thing that can be said about one: it is not
+  // blank. Unanchored on purpose, so it means "holds at least one non-whitespace character".
+  //
+  // Everything stronger was refused for the reason the analyzer's `format` comment gives: Postgres
+  // reads 'today', 'January 8, 1999', '01/08/1999' and '20200101' as dates, so a date-shaped
+  // pattern turns away rows the server stores. This one turns away nothing: measured on Postgres,
+  // every temporal type accepts a valid value with surrounding whitespace and refuses '' and ' '
+  // alike, so the set this refuses and the set the server refuses are the same set. The analyzer
+  // decides which columns carry it, per engine and per type, since MySQL's `time` takes a blank
+  // and silently stores 00:00:00.
+  temporalText: "\\S"
 };
 var COERCIBLE_DATE_STRING = "^(?!\\s*[+-])(?!\\s*\\d*\\.?\\d*(?:[eE][+-]?\\d+)?\\s*$)";
 function parsesToADate(expr) {
@@ -1833,6 +1872,7 @@ export {
   canonicalMembers,
   canonicalNumericText,
   classifyTableChecks,
+  codePointCompare,
   columnMetaFacts,
   comparisonWire,
   describeSet,
@@ -1845,6 +1885,7 @@ export {
   isProjectInstallPath,
   lengthCheckLabel,
   lengthMeasure,
+  measureCompare,
   measureExpression,
   moduleFileName,
   moduleSpecifier,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@drzl/validation-core",
-  "version": "3.22.1",
+  "version": "3.22.3",
   "private": false,
   "license": "Apache-2.0",
   "type": "module",
@@ -24,7 +24,7 @@
   ],
   "sideEffects": false,
   "dependencies": {
-    "@drzl/analyzer": "^1.21.1"
+    "@drzl/analyzer": "^1.21.3"
   },
   "peerDependencies": {
     "prettier": ">=3"