@ai-matrx/content-ir 0.14.3 → 0.15.1

package/dist/source.d.cts

@@ -390,6 +390,24 @@ type InnerFenceLine =
 declare const FENCE_WHITESPACE = " \t\n\r\f\v";
 /** Strip {@link FENCE_WHITESPACE} (and nothing else) from both ends of a line. */
 declare function trimFenceLine(line: string): string;
+/**
+ * The whitespace around a fence OPENER and inside its info string. The
+ * renderer finds openers with JavaScript's `String#trim()` and splits the info
+ * string on `/\s/`; this set IS that definition, written out so the Python
+ * twin can match it exactly — ECMAScript WhiteSpace + LineTerminator. It
+ * differs from Python's `str.split()` on U+0085 and U+001C–U+001F (not JS
+ * whitespace) and U+FEFF (JS whitespace) — verify-RC-B3 residual R4′.
+ */
+declare const FENCE_OPENER_WHITESPACE = "\t\n\v\f\r \u00A0\u1680\u2000\u2001\u2002\u2003\u2004\u2005\u2006\u2007\u2008\u2009\u200A\u2028\u2029\u202F\u205F\u3000\uFEFF";
+/**
+ * Parse a backtick fence OPENER line the way the renderer does: trim the line,
+ * count the backtick run (3+), trim the rest, and take the first
+ * whitespace-separated word as the language. null when the line opens nothing.
+ */
+declare function parseFenceOpener(line: string): {
+    ticks: number;
+    lang: string;
+} | null;
 declare function classifyInnerFenceLine(trimmed: string, openTicks: number, nests: boolean, nestedDepth: number): InnerFenceLine;
 
 /**
@@ -407,9 +425,14 @@ declare function classifyInnerFenceLine(trimmed: string, openTicks: number, nest
  * a lone `$$` (a stray closer, or an opener that never closed) never locks
  * anything. Linear time: one regex pass per code shape, one sweep.
  */
-/** The guard's `looksLikeMath`, verbatim. */
-declare function looksLikeDisplayMath(inner: string): boolean;
+/**
+ * The guard's `looksLikeMath`. `flow` = the opener is alone on its line: a
+ * flow block may span blank lines (like a fence); a `$$` with text beside it
+ * never crosses one — TeX's rule, and what keeps a stray prose `$$` from
+ * pairing with the next paragraph's formula.
+ */
+declare function looksLikeDisplayMath(inner: string, flow?: boolean): boolean;
 /** Map of `$$` opener offset → closer offset for every pair the renderer shows as math. */
 declare function pairDisplayMath(text: string): Map<number, number>;
 
-export { type BlockIslandType, CALLOUT_NAMES, type CodePointIndex, type DisturbedIsland, FENCE_WHITESPACE, type InlineIslandType, type InnerFenceLine, type IslandMeta, type IslandType, type MapOptions, type MappedPosition, type MappedRange, NESTING_FENCE_LANGUAGES, type ReadXmlTag, type SourceBlock, type SourceBlockKind, type SourceChange, type SourceEdit, type SourceIsland, type SourceSpan, SourceSpliceError, type SourceSpliceErrorCode, type SpliceIntegrity, type SpliceOptions, type SpliceResult, type TokenizeOptions, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, fenceNestsInnerFences, isSingleDollarMath, islandEdit, joinSource, listIslands, looksLikeDisplayMath, mapPosition, mapRange, pairDisplayMath, readXmlTag, singleDollarMathEnd, spliceSave, splitsSurrogatePair, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine };
+export { type BlockIslandType, CALLOUT_NAMES, type CodePointIndex, type DisturbedIsland, FENCE_OPENER_WHITESPACE, FENCE_WHITESPACE, type InlineIslandType, type InnerFenceLine, type IslandMeta, type IslandType, type MapOptions, type MappedPosition, type MappedRange, NESTING_FENCE_LANGUAGES, type ReadXmlTag, type SourceBlock, type SourceBlockKind, type SourceChange, type SourceEdit, type SourceIsland, type SourceSpan, SourceSpliceError, type SourceSpliceErrorCode, type SpliceIntegrity, type SpliceOptions, type SpliceResult, type TokenizeOptions, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, fenceNestsInnerFences, isSingleDollarMath, islandEdit, joinSource, listIslands, looksLikeDisplayMath, mapPosition, mapRange, pairDisplayMath, parseFenceOpener, readXmlTag, singleDollarMathEnd, spliceSave, splitsSurrogatePair, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine };

package/dist/source.d.ts

@@ -390,6 +390,24 @@ type InnerFenceLine =
 declare const FENCE_WHITESPACE = " \t\n\r\f\v";
 /** Strip {@link FENCE_WHITESPACE} (and nothing else) from both ends of a line. */
 declare function trimFenceLine(line: string): string;
+/**
+ * The whitespace around a fence OPENER and inside its info string. The
+ * renderer finds openers with JavaScript's `String#trim()` and splits the info
+ * string on `/\s/`; this set IS that definition, written out so the Python
+ * twin can match it exactly — ECMAScript WhiteSpace + LineTerminator. It
+ * differs from Python's `str.split()` on U+0085 and U+001C–U+001F (not JS
+ * whitespace) and U+FEFF (JS whitespace) — verify-RC-B3 residual R4′.
+ */
+declare const FENCE_OPENER_WHITESPACE = "\t\n\v\f\r \u00A0\u1680\u2000\u2001\u2002\u2003\u2004\u2005\u2006\u2007\u2008\u2009\u200A\u2028\u2029\u202F\u205F\u3000\uFEFF";
+/**
+ * Parse a backtick fence OPENER line the way the renderer does: trim the line,
+ * count the backtick run (3+), trim the rest, and take the first
+ * whitespace-separated word as the language. null when the line opens nothing.
+ */
+declare function parseFenceOpener(line: string): {
+    ticks: number;
+    lang: string;
+} | null;
 declare function classifyInnerFenceLine(trimmed: string, openTicks: number, nests: boolean, nestedDepth: number): InnerFenceLine;
 
 /**
@@ -407,9 +425,14 @@ declare function classifyInnerFenceLine(trimmed: string, openTicks: number, nest
  * a lone `$$` (a stray closer, or an opener that never closed) never locks
  * anything. Linear time: one regex pass per code shape, one sweep.
  */
-/** The guard's `looksLikeMath`, verbatim. */
-declare function looksLikeDisplayMath(inner: string): boolean;
+/**
+ * The guard's `looksLikeMath`. `flow` = the opener is alone on its line: a
+ * flow block may span blank lines (like a fence); a `$$` with text beside it
+ * never crosses one — TeX's rule, and what keeps a stray prose `$$` from
+ * pairing with the next paragraph's formula.
+ */
+declare function looksLikeDisplayMath(inner: string, flow?: boolean): boolean;
 /** Map of `$$` opener offset → closer offset for every pair the renderer shows as math. */
 declare function pairDisplayMath(text: string): Map<number, number>;
 
-export { type BlockIslandType, CALLOUT_NAMES, type CodePointIndex, type DisturbedIsland, FENCE_WHITESPACE, type InlineIslandType, type InnerFenceLine, type IslandMeta, type IslandType, type MapOptions, type MappedPosition, type MappedRange, NESTING_FENCE_LANGUAGES, type ReadXmlTag, type SourceBlock, type SourceBlockKind, type SourceChange, type SourceEdit, type SourceIsland, type SourceSpan, SourceSpliceError, type SourceSpliceErrorCode, type SpliceIntegrity, type SpliceOptions, type SpliceResult, type TokenizeOptions, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, fenceNestsInnerFences, isSingleDollarMath, islandEdit, joinSource, listIslands, looksLikeDisplayMath, mapPosition, mapRange, pairDisplayMath, readXmlTag, singleDollarMathEnd, spliceSave, splitsSurrogatePair, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine };
+export { type BlockIslandType, CALLOUT_NAMES, type CodePointIndex, type DisturbedIsland, FENCE_OPENER_WHITESPACE, FENCE_WHITESPACE, type InlineIslandType, type InnerFenceLine, type IslandMeta, type IslandType, type MapOptions, type MappedPosition, type MappedRange, NESTING_FENCE_LANGUAGES, type ReadXmlTag, type SourceBlock, type SourceBlockKind, type SourceChange, type SourceEdit, type SourceIsland, type SourceSpan, SourceSpliceError, type SourceSpliceErrorCode, type SpliceIntegrity, type SpliceOptions, type SpliceResult, type TokenizeOptions, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, fenceNestsInnerFences, isSingleDollarMath, islandEdit, joinSource, listIslands, looksLikeDisplayMath, mapPosition, mapRange, pairDisplayMath, parseFenceOpener, readXmlTag, singleDollarMathEnd, spliceSave, splitsSurrogatePair, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine };

package/dist/source.js

@@ -220,6 +220,24 @@ function trimFenceLine(line) {
   while (end > start && isFenceWhitespace(line[end - 1])) end--;
   return start === 0 && end === line.length ? line : line.slice(start, end);
 }
+var FENCE_OPENER_WHITESPACE = "	\n\v\f\r \xA0\u1680\u2000\u2001\u2002\u2003\u2004\u2005\u2006\u2007\u2008\u2009\u200A\u2028\u2029\u202F\u205F\u3000\uFEFF";
+function isOpenerWhitespace(ch) {
+  return ch !== void 0 && ch !== "" && FENCE_OPENER_WHITESPACE.includes(ch);
+}
+function parseFenceOpener(line) {
+  let start = 0;
+  let end = line.length;
+  while (start < end && isOpenerWhitespace(line[start])) start++;
+  while (end > start && isOpenerWhitespace(line[end - 1])) end--;
+  let ticks = 0;
+  while (start + ticks < end && line[start + ticks] === "`") ticks++;
+  if (ticks < 3) return null;
+  let infoStart = start + ticks;
+  while (infoStart < end && isOpenerWhitespace(line[infoStart])) infoStart++;
+  let langEnd = infoStart;
+  while (langEnd < end && !isOpenerWhitespace(line[langEnd])) langEnd++;
+  return { ticks, lang: line.slice(infoStart, langEnd) };
+}
 function classifyInnerFenceLine(trimmed, openTicks, nests, nestedDepth) {
   let ticks = 0;
   while (ticks < trimmed.length && trimmed[ticks] === "`") ticks++;
@@ -301,15 +319,20 @@ var STRUCTURAL_MARKDOWN = /\]\(|https?:\/\/|\*\*|(?:^|\n)[ \t]{0,3}#{1,6}[ \t]|(
 var LATEX_COMMAND = /\\[a-zA-Z]/;
 var PROSE_WORD = /[A-Za-z]{3,}/g;
 var PROSE_WORD_LIMIT = 6;
-function looksLikeDisplayMath(inner) {
+function opensFlowBlock(text, open) {
+  for (let k = open - 1; k >= 0 && text[k] !== "\n"; k--) if (text[k] !== " " && text[k] !== "	") return false;
+  for (let k = open + 2; k < text.length && text[k] !== "\n"; k++) if (text[k] !== " " && text[k] !== "	" && text[k] !== "\r") return false;
+  return true;
+}
+function looksLikeDisplayMath(inner, flow = false) {
   const s = inner.trim();
   if (!s) return false;
   if (STRUCTURAL_MARKDOWN.test(s)) return false;
   if (s.length > MAX_MATH_SPAN) return false;
+  if (!flow && /\n[ \t]*\n/.test(inner)) return false;
   if (LATEX_COMMAND.test(s)) {
     return (s.match(PROSE_WORD) ?? []).length < PROSE_WORD_LIMIT * 3;
   }
-  if (/\n[ \t]*\n/.test(s)) return false;
   return (s.match(PROSE_WORD) ?? []).length < PROSE_WORD_LIMIT;
 }
 function codeRanges(text) {
@@ -343,7 +366,7 @@ function pairDisplayMath(text) {
     const open = tokens[j];
     const close = tokens[j + 1];
     if (close === void 0) break;
-    if (looksLikeDisplayMath(text.slice(open + 2, close))) {
+    if (looksLikeDisplayMath(text.slice(open + 2, close), opensFlowBlock(text, open))) {
       pairs.set(open, close);
       j += 2;
       continue;
@@ -980,8 +1003,7 @@ var Tokenizer = class {
    */
   backtickFence(p, li, t) {
     const openTicks = backtickRunLength(t, 0);
-    const info = t.slice(openTicks).trim();
-    const lang = info.split(/\s/)[0] ?? "";
+    const lang = parseFenceOpener(t)?.lang ?? "";
     const isJson = lang === "json";
     const normalized = normalizeFenceLanguage(lang);
     const meta = { fence: "`", ticks: openTicks, lang };
@@ -1111,6 +1133,7 @@ var Tokenizer = class {
     const endLine = this.lineOf(close);
     const endCe = this.contentEnds[endLine];
     if (endLine !== li && !this.isBlank(end, endCe)) return null;
+    if (this.isOpenTail(endLine)) return this.island("math_block", p, this.n, false, { delimiter: "$$" });
     return this.island("math_block", p, end, true, { delimiter: "$$" });
   }
   /** `<artifact …>`, `<decision …>`, … at the start of a line. */
@@ -1440,7 +1463,7 @@ var Tokenizer = class {
           continue;
         }
         const end = singleDollarMathEnd(text, i, pe);
-        if (end > 0) {
+        if (end > 0 && !(this.streaming && end === this.n)) {
           inlines.push(this.island("math_inline", i, end, true, { delimiter: "$" }));
           i = end;
           continue;
@@ -1850,11 +1873,17 @@ function spliceSave(original, edits, options = {}) {
     else newIslandsByStart.set(island.start, [island.raw]);
   }
   const disturbed = [];
+  const carried = (island) => sorted.some((edit) => !edit.island && island.start >= edit.start && island.end <= edit.end);
+  const changedInside = [];
   for (const island of islands) {
     if (replacedIslands.has(`${island.start}:${island.end}`)) continue;
     const mappedStart = mapPosition(changes, island.start, { assoc: 1 }).pos;
     const candidates = newIslandsByStart.get(mappedStart);
     if (!candidates || !candidates.includes(island.raw)) {
+      if (carried(island)) {
+        changedInside.push(island);
+        continue;
+      }
       disturbed.push({
         islandType: island.islandType,
         inline: island.inline,
@@ -1864,6 +1893,14 @@ function spliceSave(original, edits, options = {}) {
       });
     }
   }
+  if (changedInside.length > 0) {
+    const first = changedInside[0];
+    const name = first.meta.tag ?? first.meta.name ?? first.meta.lang ?? "";
+    throw new SourceSpliceError(
+      "island_edit",
+      `the edit changes the protected ${first.islandType}${name ? ` (${String(name)})` : ""} island at [${first.start}, ${first.end}) \u2014 its bytes survive only as ordinary text${changedInside.length > 1 ? ` (and ${changedInside.length - 1} more)` : ""}; islands change only through islandEdit()`
+    );
+  }
   const integrity = {
     ok: bytesOutsideEditsIdentical && disturbed.length === 0,
     bytesOutsideEditsIdentical,
@@ -1915,6 +1952,6 @@ function mapRange(changes, start, end, unit = "utf16") {
   return { start: newStart, end: newEnd, touched, collapsed: newEnd === newStart && end > start };
 }
 
-export { CALLOUT_NAMES, FENCE_WHITESPACE, NESTING_FENCE_LANGUAGES, SourceSpliceError, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, fenceNestsInnerFences, isSingleDollarMath, islandEdit, joinSource, listIslands, looksLikeDisplayMath, mapPosition, mapRange, pairDisplayMath, readXmlTag, singleDollarMathEnd, spliceSave, splitsSurrogatePair, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine };
+export { CALLOUT_NAMES, FENCE_OPENER_WHITESPACE, FENCE_WHITESPACE, NESTING_FENCE_LANGUAGES, SourceSpliceError, XmlContainerTracker, blockAt, blockEdit, blockRangeEdit, buildCodePointIndex, classifyInnerFenceLine, fenceNestsInnerFences, isSingleDollarMath, islandEdit, joinSource, listIslands, looksLikeDisplayMath, mapPosition, mapRange, pairDisplayMath, parseFenceOpener, readXmlTag, singleDollarMathEnd, spliceSave, splitsSurrogatePair, toCodePointOffset, toUtf16Offset, tokenizeSource, trimFenceLine };
 //# sourceMappingURL=source.js.map
 //# sourceMappingURL=source.js.map