@knowvah/dot-engine 2.6.10 → 2.6.11

package/dist/common/html-entities.d.ts

@@ -6,11 +6,19 @@
  */
 export declare const HTML_ENTITIES: Readonly<Record<string, number>>;
 /**
- * Decode XML/numeric (&#NNN; / &#xHH;) and HTML 4 named (`&alpha;`) entities to
- * their characters, leaving unknown names literal — C's make_label UTF-8 branch.
- * Latin-1 inputs are decoded at the byte->string boundary by the caller (the
- * survey's render-one), so this UTF-8 branch covers them too.
- * @see lib/common/utils.c:htmlEntityUTF8 / htmlEntity
+ * Substitute entities like `&#123;` and `&amp;` with their characters: C's
+ * make_label UTF-8 branch. An `&` whose entity has no value stays (scanning
+ * resumes where htmlEntity left it). Latin-1 inputs are decoded at the
+ * byte->string boundary by the caller (the survey's render-one), so this
+ * covers them too, as does latin1ToUTF8's identical entity handling.
+ *
+ * C appends a value's bytes itself: below 0x7F one byte, below 0x7FF two,
+ * else three, truncated to `char`. For every value whose bytes are valid
+ * UTF-8 that is the character itself. 0x7F, 0x7FF, surrogates and values
+ * above 0xFFFF come out as invalid UTF-8, which a JS string cannot hold; the
+ * port keeps the character there (documented divergence).
+ * @see lib/common/utils.c:htmlEntityUTF8
+ * @see lib/common/utils.c:latin1ToUTF8
  * @see lib/common/labels.c:make_label (CHAR_LATIN1 vs default)
  */
 export declare function htmlEntityUTF8(s: string): string;

package/dist/common/html-entities.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"html-entities.d.ts","sourceRoot":"","sources":["../../src/common/html-entities.ts"],"names":[],"mappings":"AACA;;;;;GAKG;AACH,eAAO,MAAM,aAAa,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CA6P1D,CAAC;AAwBF;;;;;;;GAOG;AACH,wBAAgB,cAAc,CAAC,CAAC,EAAE,MAAM,GAAG,MAAM,CAMhD"}
+{"version":3,"file":"html-entities.d.ts","sourceRoot":"","sources":["../../src/common/html-entities.ts"],"names":[],"mappings":"AACA;;;;;GAKG;AACH,eAAO,MAAM,aAAa,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CA6P1D,CAAC;AA+DF;;;;;;;;;;;;;;;GAeG;AACH,wBAAgB,cAAc,CAAC,CAAC,EAAE,MAAM,GAAG,MAAM,CAWhD"}

package/dist/common/htmltable-lex-chardata.d.ts

@@ -30,6 +30,15 @@ export declare function keepLostText(st: CharDataState, lost: readonly Token[]):
  * @see lib/common/htmlparse.y:table
  */
 export declare function releaseXb(st: CharDataState, first: Token): void;
+/**
+ * The characters expat hands characterData for a text run: scanEntity first
+ * rewrites each entities.h name to `&#N;` (its `&` only when no `#` follows),
+ * then expat decodes character references, of any length, and its five
+ * predefined entities. Runs expat rejects never get here (textRefError).
+ * @see lib/common/htmllex.c:findNext
+ * @see lib/common/utils.c:scanEntity
+ */
+export declare function decodeTextRefs(raw: string): string;
 /** The text token of a text run, if anything is left after cleaning. */
 export declare function textTokens(raw: string): Token[];
 /** characterData on what expat decoded: a T_string when any character survives. */

package/dist/common/htmltable-lex-chardata.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"htmltable-lex-chardata.d.ts","sourceRoot":"","sources":["../../src/common/htmltable-lex-chardata.ts"],"names":[],"mappings":"AACA;;;;;;;;;;;GAWG;AAGH,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,0BAA0B,CAAC;AACzD,OAAO,KAAK,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAElD,oFAAoF;AACpF,MAAM,WAAW,aAAa;IAC5B,MAAM,EAAE,OAAO,CAAC;IAChB,EAAE,EAAE,MAAM,CAAC;CACZ;AAED,uGAAuG;AACvG,eAAO,MAAM,eAAe,QAAO,aAA2C,CAAC;AAc/E,yFAAyF;AACzF,wBAAgB,SAAS,CAAC,EAAE,EAAE,aAAa,EAAE,GAAG,EAAE,QAAQ,GAAG,IAAI,CAQhE;AAED,wEAAwE;AACxE,wBAAgB,YAAY,CAAC,EAAE,EAAE,aAAa,EAAE,IAAI,EAAE,SAAS,KAAK,EAAE,GAAG,IAAI,CAE5E;AAED;;;;;GAKG;AACH,wBAAgB,SAAS,CAAC,EAAE,EAAE,aAAa,EAAE,KAAK,EAAE,KAAK,GAAG,IAAI,CAM/D;AAYD,wEAAwE;AACxE,wBAAgB,UAAU,CAAC,GAAG,EAAE,MAAM,GAAG,KAAK,EAAE,CAQ/C;AAED,mFAAmF;AACnF,wBAAgB,cAAc,CAAC,OAAO,EAAE,MAAM,GAAG,KAAK,EAAE,CAGvD"}
+{"version":3,"file":"htmltable-lex-chardata.d.ts","sourceRoot":"","sources":["../../src/common/htmltable-lex-chardata.ts"],"names":[],"mappings":"AACA;;;;;;;;;;;GAWG;AAIH,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,0BAA0B,CAAC;AACzD,OAAO,KAAK,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAElD,oFAAoF;AACpF,MAAM,WAAW,aAAa;IAC5B,MAAM,EAAE,OAAO,CAAC;IAChB,EAAE,EAAE,MAAM,CAAC;CACZ;AAED,uGAAuG;AACvG,eAAO,MAAM,eAAe,QAAO,aAA2C,CAAC;AAc/E,yFAAyF;AACzF,wBAAgB,SAAS,CAAC,EAAE,EAAE,aAAa,EAAE,GAAG,EAAE,QAAQ,GAAG,IAAI,CAQhE;AAED,wEAAwE;AACxE,wBAAgB,YAAY,CAAC,EAAE,EAAE,aAAa,EAAE,IAAI,EAAE,SAAS,KAAK,EAAE,GAAG,IAAI,CAE5E;AAED;;;;;GAKG;AACH,wBAAgB,SAAS,CAAC,EAAE,EAAE,aAAa,EAAE,KAAK,EAAE,KAAK,GAAG,IAAI,CAM/D;AAkCD;;;;;;;GAOG;AACH,wBAAgB,cAAc,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CAElD;AAED,wEAAwE;AACxE,wBAAgB,UAAU,CAAC,GAAG,EAAE,MAAM,GAAG,KAAK,EAAE,CAO/C;AAED,mFAAmF;AACnF,wBAAgB,cAAc,CAAC,OAAO,EAAE,MAAM,GAAG,KAAK,EAAE,CAGvD"}

package/dist/index.js

@@ -10439,28 +10439,46 @@ var HTML_ENTITIES = {
   "zwj": 8205,
   "zwnj": 8204
 };
-var BASIC_ENTITIES = {
-  amp: 38,
-  lt: 60,
-  gt: 62,
-  quot: 34,
-  apos: 39
-};
-function numericEntityValue(body) {
-  const hex = body[1] === "x" || body[1] === "X";
-  const v = hex ? parseInt(body.slice(2), 16) : parseInt(body.slice(1), 10);
-  return Number.isFinite(v) && v > 0 && v <= 1114111 ? v : null;
-}
-function entityValue(body) {
-  if (body[0] === "#") return numericEntityValue(body);
-  return BASIC_ENTITIES[body] ?? HTML_ENTITIES[body] ?? null;
+var ENTITY_NAME_LENGTH_MAX = 8;
+function numericEntity(s, at) {
+  const hex = s[at + 1] === "x" || s[at + 1] === "X";
+  const digit = hex ? /^[0-9A-Fa-f]$/ : /^[0-9]$/;
+  let n = 0;
+  let i = hex ? 2 : 1;
+  for (; i < 8; i++) {
+    const c = s[at + i] ?? "";
+    if (!digit.test(c)) break;
+    n = n * (hex ? 16 : 10) + parseInt(c, 16);
+  }
+  if (i < 8 && s[at + i] === ";") return { value: n, next: at + i + 1 };
+  return { value: 0, next: at };
+}
+function namedEntity(s, at) {
+  for (let i = 0; i < ENTITY_NAME_LENGTH_MAX; i++) {
+    const c = s[at + i];
+    if (c === void 0 || c === "\0") break;
+    if (c === ";") {
+      const key = s.slice(at, at + i);
+      if (Object.hasOwn(HTML_ENTITIES, key)) return { value: HTML_ENTITIES[key], next: at + i + 1 };
+      break;
+    }
+  }
+  return { value: 0, next: at };
+}
+function htmlEntity(s, at) {
+  return s[at] === "#" ? numericEntity(s, at) : namedEntity(s, at);
 }
 function htmlEntityUTF8(s) {
-  if (s.indexOf("&") < 0) return s;
-  return s.replace(/&(#x[0-9a-fA-F]+|#[0-9]+|[A-Za-z]+);/g, (m, body) => {
-    const v = entityValue(body);
-    return v === null ? m : String.fromCodePoint(v);
-  });
+  let amp = s.indexOf("&");
+  if (amp < 0) return s;
+  let out = "";
+  let from = 0;
+  for (; amp >= 0; amp = s.indexOf("&", from)) {
+    const e = htmlEntity(s, amp + 1);
+    out += s.slice(from, amp) + (e.value ? String.fromCodePoint(e.value) : "&");
+    from = e.next;
+  }
+  return out + s.slice(from);
 }
 
 // src/common/htmltable-lex-attvalue.ts
@@ -10854,8 +10872,26 @@ function releaseXb(st, first) {
 function cleanText(decoded) {
   return decoded.replace(/[\x00-\x1f]/g, "");
 }
+var XML_PREDEFINED = { lt: "<", gt: ">", amp: "&", quot: '"', apos: "'" };
+var SCAN_ENTITY_MIN = 2;
+var SCAN_ENTITY_MAX = 8;
+var TEXT_REF = /&(#x[0-9A-Fa-f]+|#[0-9]+|[^&;]+);/g;
+function textRefValue(ref, body) {
+  if (body[0] === "#") {
+    const v = charRefNumber(ref);
+    return v < 0 ? ref : String.fromCodePoint(v);
+  }
+  const n = body.length;
+  if (n >= SCAN_ENTITY_MIN && n <= SCAN_ENTITY_MAX && Object.hasOwn(HTML_ENTITIES, body)) {
+    return String.fromCodePoint(HTML_ENTITIES[body]);
+  }
+  return Object.hasOwn(XML_PREDEFINED, body) ? XML_PREDEFINED[body] : ref;
+}
+function decodeTextRefs(raw) {
+  return raw.indexOf("&") < 0 ? raw : raw.replace(TEXT_REF, textRefValue);
+}
 function textTokens(raw) {
-  return charDataTokens(htmlEntityUTF8(raw));
+  return charDataTokens(decodeTextRefs(raw));
 }
 function charDataTokens(decoded) {
   const clean = cleanText(decoded);