ooxml-core 0.8.0 → 0.9.0

package/dist/text/index.cjs

@@ -0,0 +1,122 @@
+'use strict';
+
+// src/text/unicode-script-detection.ts
+function detectFontScript(codePoint) {
+  if (codePoint >= 11904 && codePoint <= 40959 || // CJK radicals, unified ideographs
+  codePoint >= 44032 && codePoint <= 55215 || // Hangul syllables
+  codePoint >= 63744 && codePoint <= 64255 || // CJK compatibility ideographs
+  codePoint >= 65072 && codePoint <= 65103 || // CJK compatibility forms
+  codePoint >= 131072 && codePoint <= 195103 || // CJK extension B-F
+  codePoint >= 12288 && codePoint <= 12351 || // CJK symbols & punctuation
+  codePoint >= 12352 && codePoint <= 12447 || // Hiragana
+  codePoint >= 12448 && codePoint <= 12543 || // Katakana
+  codePoint >= 12784 && codePoint <= 12799 || // Katakana extensions
+  codePoint >= 65280 && codePoint <= 65519) {
+    return "eastAsia";
+  }
+  if (codePoint >= 1424 && codePoint <= 1535 || // Hebrew
+  codePoint >= 1536 && codePoint <= 1791 || // Arabic
+  codePoint >= 1792 && codePoint <= 1871 || // Syriac
+  codePoint >= 1872 && codePoint <= 1919 || // Arabic Supplement
+  codePoint >= 1920 && codePoint <= 1983 || // Thaana
+  codePoint >= 1984 && codePoint <= 2047 || // NKo
+  codePoint >= 2048 && codePoint <= 2111 || // Samaritan
+  codePoint >= 2304 && codePoint <= 2431 || // Devanagari
+  codePoint >= 2432 && codePoint <= 2559 || // Bengali
+  codePoint >= 2560 && codePoint <= 2687 || // Gurmukhi
+  codePoint >= 2688 && codePoint <= 2815 || // Gujarati
+  codePoint >= 2816 && codePoint <= 2943 || // Oriya
+  codePoint >= 2944 && codePoint <= 3071 || // Tamil
+  codePoint >= 3072 && codePoint <= 3199 || // Telugu
+  codePoint >= 3200 && codePoint <= 3327 || // Kannada
+  codePoint >= 3328 && codePoint <= 3455 || // Malayalam
+  codePoint >= 3456 && codePoint <= 3583 || // Sinhala
+  codePoint >= 3584 && codePoint <= 3711 || // Thai
+  codePoint >= 3712 && codePoint <= 3839 || // Lao
+  codePoint >= 3840 && codePoint <= 4095 || // Tibetan
+  codePoint >= 4096 && codePoint <= 4255 || // Myanmar
+  codePoint >= 4256 && codePoint <= 4351 || // Georgian
+  codePoint >= 6016 && codePoint <= 6143 || // Khmer
+  codePoint >= 64336 && codePoint <= 65023 || // Arabic Presentation Forms-A
+  codePoint >= 65136 && codePoint <= 65279) {
+    return "complexScript";
+  }
+  if (codePoint >= 8592 && codePoint <= 8703 || // Arrows
+  codePoint >= 8704 && codePoint <= 8959 || // Mathematical operators
+  codePoint >= 8960 && codePoint <= 9215 || // Misc technical
+  codePoint >= 9472 && codePoint <= 9599 || // Box drawing
+  codePoint >= 9600 && codePoint <= 9631 || // Block elements
+  codePoint >= 9632 && codePoint <= 9727 || // Geometric shapes
+  codePoint >= 9728 && codePoint <= 9983 || // Misc symbols
+  codePoint >= 9984 && codePoint <= 10175 || // Dingbats
+  codePoint >= 57344 && codePoint <= 63743 || // Private Use Area
+  codePoint >= 126976 && codePoint <= 129535) {
+    return "symbol";
+  }
+  return "latin";
+}
+function segmentByScript(text) {
+  if (!text) {
+    return [];
+  }
+  const runs = [];
+  let currentScript = null;
+  let currentStart = 0;
+  let currentText = "";
+  for (let i = 0; i < text.length; i++) {
+    const codePoint = text.codePointAt(i);
+    if (codePoint === void 0) {
+      continue;
+    }
+    const charLength = codePoint > 65535 ? 2 : 1;
+    const char = text.substring(i, i + charLength);
+    const isNeutral = codePoint <= 127 && !/[a-zA-Z0-9]/u.test(char);
+    const script = isNeutral ? currentScript ?? "latin" : detectFontScript(codePoint);
+    if (script !== currentScript && currentScript !== null) {
+      runs.push({
+        text: currentText,
+        script: currentScript,
+        startIndex: currentStart
+      });
+      currentStart = i;
+      currentText = "";
+    }
+    currentScript = script;
+    currentText += char;
+    if (charLength === 2) {
+      i++;
+    }
+  }
+  if (currentText && currentScript) {
+    runs.push({
+      text: currentText,
+      script: currentScript,
+      startIndex: currentStart
+    });
+  }
+  return runs;
+}
+function resolveFontForScript(script, fonts) {
+  switch (script) {
+    case "latin":
+      return fonts.latin;
+    case "eastAsia":
+      return fonts.eastAsia || fonts.latin;
+    case "complexScript":
+      return fonts.complexScript || fonts.latin;
+    case "symbol":
+      return fonts.symbol || fonts.latin;
+  }
+}
+function hasDistinctScriptFonts(fonts) {
+  const base = fonts.latin;
+  if (!base) {
+    return false;
+  }
+  return Boolean(fonts.eastAsia) && fonts.eastAsia !== base || Boolean(fonts.complexScript) && fonts.complexScript !== base || Boolean(fonts.symbol) && fonts.symbol !== base;
+}
+
+exports.detectFontScript = detectFontScript;
+exports.hasDistinctScriptFonts = hasDistinctScriptFonts;
+exports.resolveFontForScript = resolveFontForScript;
+exports.segmentByScript = segmentByScript;

package/dist/text/index.d.ts

@@ -0,0 +1,3 @@
+/** Browser-neutral text primitives shared by Office consumers. */
+export { detectFontScript, segmentByScript, resolveFontForScript, hasDistinctScriptFonts, type FontScriptCategory, type ScriptRun, } from './unicode-script-detection.js';
+//# sourceMappingURL=index.d.ts.map

package/dist/text/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/text/index.ts"],"names":[],"mappings":"AAAA,kEAAkE;AAClE,OAAO,EACN,gBAAgB,EAChB,eAAe,EACf,oBAAoB,EACpB,sBAAsB,EACtB,KAAK,kBAAkB,EACvB,KAAK,SAAS,GACd,MAAM,+BAA+B,CAAC"}

package/dist/text/index.mjs

@@ -0,0 +1,117 @@
+// src/text/unicode-script-detection.ts
+function detectFontScript(codePoint) {
+  if (codePoint >= 11904 && codePoint <= 40959 || // CJK radicals, unified ideographs
+  codePoint >= 44032 && codePoint <= 55215 || // Hangul syllables
+  codePoint >= 63744 && codePoint <= 64255 || // CJK compatibility ideographs
+  codePoint >= 65072 && codePoint <= 65103 || // CJK compatibility forms
+  codePoint >= 131072 && codePoint <= 195103 || // CJK extension B-F
+  codePoint >= 12288 && codePoint <= 12351 || // CJK symbols & punctuation
+  codePoint >= 12352 && codePoint <= 12447 || // Hiragana
+  codePoint >= 12448 && codePoint <= 12543 || // Katakana
+  codePoint >= 12784 && codePoint <= 12799 || // Katakana extensions
+  codePoint >= 65280 && codePoint <= 65519) {
+    return "eastAsia";
+  }
+  if (codePoint >= 1424 && codePoint <= 1535 || // Hebrew
+  codePoint >= 1536 && codePoint <= 1791 || // Arabic
+  codePoint >= 1792 && codePoint <= 1871 || // Syriac
+  codePoint >= 1872 && codePoint <= 1919 || // Arabic Supplement
+  codePoint >= 1920 && codePoint <= 1983 || // Thaana
+  codePoint >= 1984 && codePoint <= 2047 || // NKo
+  codePoint >= 2048 && codePoint <= 2111 || // Samaritan
+  codePoint >= 2304 && codePoint <= 2431 || // Devanagari
+  codePoint >= 2432 && codePoint <= 2559 || // Bengali
+  codePoint >= 2560 && codePoint <= 2687 || // Gurmukhi
+  codePoint >= 2688 && codePoint <= 2815 || // Gujarati
+  codePoint >= 2816 && codePoint <= 2943 || // Oriya
+  codePoint >= 2944 && codePoint <= 3071 || // Tamil
+  codePoint >= 3072 && codePoint <= 3199 || // Telugu
+  codePoint >= 3200 && codePoint <= 3327 || // Kannada
+  codePoint >= 3328 && codePoint <= 3455 || // Malayalam
+  codePoint >= 3456 && codePoint <= 3583 || // Sinhala
+  codePoint >= 3584 && codePoint <= 3711 || // Thai
+  codePoint >= 3712 && codePoint <= 3839 || // Lao
+  codePoint >= 3840 && codePoint <= 4095 || // Tibetan
+  codePoint >= 4096 && codePoint <= 4255 || // Myanmar
+  codePoint >= 4256 && codePoint <= 4351 || // Georgian
+  codePoint >= 6016 && codePoint <= 6143 || // Khmer
+  codePoint >= 64336 && codePoint <= 65023 || // Arabic Presentation Forms-A
+  codePoint >= 65136 && codePoint <= 65279) {
+    return "complexScript";
+  }
+  if (codePoint >= 8592 && codePoint <= 8703 || // Arrows
+  codePoint >= 8704 && codePoint <= 8959 || // Mathematical operators
+  codePoint >= 8960 && codePoint <= 9215 || // Misc technical
+  codePoint >= 9472 && codePoint <= 9599 || // Box drawing
+  codePoint >= 9600 && codePoint <= 9631 || // Block elements
+  codePoint >= 9632 && codePoint <= 9727 || // Geometric shapes
+  codePoint >= 9728 && codePoint <= 9983 || // Misc symbols
+  codePoint >= 9984 && codePoint <= 10175 || // Dingbats
+  codePoint >= 57344 && codePoint <= 63743 || // Private Use Area
+  codePoint >= 126976 && codePoint <= 129535) {
+    return "symbol";
+  }
+  return "latin";
+}
+function segmentByScript(text) {
+  if (!text) {
+    return [];
+  }
+  const runs = [];
+  let currentScript = null;
+  let currentStart = 0;
+  let currentText = "";
+  for (let i = 0; i < text.length; i++) {
+    const codePoint = text.codePointAt(i);
+    if (codePoint === void 0) {
+      continue;
+    }
+    const charLength = codePoint > 65535 ? 2 : 1;
+    const char = text.substring(i, i + charLength);
+    const isNeutral = codePoint <= 127 && !/[a-zA-Z0-9]/u.test(char);
+    const script = isNeutral ? currentScript ?? "latin" : detectFontScript(codePoint);
+    if (script !== currentScript && currentScript !== null) {
+      runs.push({
+        text: currentText,
+        script: currentScript,
+        startIndex: currentStart
+      });
+      currentStart = i;
+      currentText = "";
+    }
+    currentScript = script;
+    currentText += char;
+    if (charLength === 2) {
+      i++;
+    }
+  }
+  if (currentText && currentScript) {
+    runs.push({
+      text: currentText,
+      script: currentScript,
+      startIndex: currentStart
+    });
+  }
+  return runs;
+}
+function resolveFontForScript(script, fonts) {
+  switch (script) {
+    case "latin":
+      return fonts.latin;
+    case "eastAsia":
+      return fonts.eastAsia || fonts.latin;
+    case "complexScript":
+      return fonts.complexScript || fonts.latin;
+    case "symbol":
+      return fonts.symbol || fonts.latin;
+  }
+}
+function hasDistinctScriptFonts(fonts) {
+  const base = fonts.latin;
+  if (!base) {
+    return false;
+  }
+  return Boolean(fonts.eastAsia) && fonts.eastAsia !== base || Boolean(fonts.complexScript) && fonts.complexScript !== base || Boolean(fonts.symbol) && fonts.symbol !== base;
+}
+
+export { detectFontScript, hasDistinctScriptFonts, resolveFontForScript, segmentByScript };

package/dist/text/unicode-script-detection.d.ts

@@ -0,0 +1,54 @@
+/**
+ * Unicode script detection for font fallback.
+ *
+ * Classifies code points into the font groups historically used by the
+ * PowerPoint viewer. This block-based heuristic is not a Unicode Script
+ * property implementation or Word's language/hint/bidi font-slot selection.
+ * It does not choose or load fonts, shape glyphs or modify document models.
+ */
+/** OOXML font script category. */
+export type FontScriptCategory = 'latin' | 'eastAsia' | 'complexScript' | 'symbol';
+/**
+ * Detect the OOXML font script category for a single character.
+ *
+ * Uses Unicode block ranges to determine if a character belongs to:
+ * - Latin (Basic Latin, Latin Extended, Greek, Cyrillic, etc.)
+ * - East Asian (CJK Unified Ideographs, Hiragana, Katakana, Hangul, etc.)
+ * - Complex Script (Arabic, Hebrew, Devanagari, Thai, etc.)
+ * - Symbol (Dingbats, Mathematical, etc.)
+ */
+export declare function detectFontScript(codePoint: number): FontScriptCategory;
+/** A contiguous run of characters sharing the same font script category. */
+export interface ScriptRun {
+    text: string;
+    script: FontScriptCategory;
+    /** UTF-16 offset in the original string, matching JavaScript string slicing. */
+    startIndex: number;
+}
+/**
+ * Segment a string into runs of the same font script category.
+ * Adjacent characters with the same script are grouped together.
+ * Whitespace and ASCII punctuation inherit the previous run's script.
+ */
+export declare function segmentByScript(text: string): ScriptRun[];
+/**
+ * Resolve the font family for a given script category from available fonts.
+ * Falls back to latin when a script-specific font is not available.
+ */
+export declare function resolveFontForScript(script: FontScriptCategory, fonts: {
+    latin?: string;
+    eastAsia?: string;
+    complexScript?: string;
+    symbol?: string;
+}): string | undefined;
+/**
+ * Check whether per-script fonts are actually different from the base latin font.
+ * Returns `true` when at least one script-specific font differs.
+ */
+export declare function hasDistinctScriptFonts(fonts: {
+    latin?: string;
+    eastAsia?: string;
+    complexScript?: string;
+    symbol?: string;
+}): boolean;
+//# sourceMappingURL=unicode-script-detection.d.ts.map

package/dist/text/unicode-script-detection.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"unicode-script-detection.d.ts","sourceRoot":"","sources":["../../src/text/unicode-script-detection.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,kCAAkC;AAClC,MAAM,MAAM,kBAAkB,GAAG,OAAO,GAAG,UAAU,GAAG,eAAe,GAAG,QAAQ,CAAC;AAEnF;;;;;;;;GAQG;AACH,wBAAgB,gBAAgB,CAAC,SAAS,EAAE,MAAM,GAAG,kBAAkB,CAkEtE;AAED,4EAA4E;AAC5E,MAAM,WAAW,SAAS;IACzB,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,kBAAkB,CAAC;IAC3B,gFAAgF;IAChF,UAAU,EAAE,MAAM,CAAC;CACnB;AAED;;;;GAIG;AACH,wBAAgB,eAAe,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS,EAAE,CAqDzD;AAED;;;GAGG;AACH,wBAAgB,oBAAoB,CACnC,MAAM,EAAE,kBAAkB,EAC1B,KAAK,EAAE;IACN,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,MAAM,CAAC,EAAE,MAAM,CAAC;CAChB,GACC,MAAM,GAAG,SAAS,CAWpB;AAED;;;GAGG;AACH,wBAAgB,sBAAsB,CAAC,KAAK,EAAE;IAC7C,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,MAAM,CAAC,EAAE,MAAM,CAAC;CAChB,GAAG,OAAO,CAUV"}

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "ooxml-core",
-	"version": "0.8.0",
+	"version": "0.9.0",
 	"description": "Shared OOXML logic for the Word, PowerPoint and Excel viewers: everything except the UI",
 	"keywords": [
 		"charts",
@@ -66,6 +66,16 @@
 			"import": "./dist/geometry/index.mjs",
 			"require": "./dist/geometry/index.cjs"
 		},
+		"./chart": {
+			"types": "./dist/chart/index.d.ts",
+			"import": "./dist/chart/index.mjs",
+			"require": "./dist/chart/index.cjs"
+		},
+		"./text": {
+			"types": "./dist/text/index.d.ts",
+			"import": "./dist/text/index.mjs",
+			"require": "./dist/text/index.cjs"
+		},
 		"./diagram": {
 			"types": "./dist/diagram/index.d.ts",
 			"import": "./dist/diagram/index.mjs",