kordoc 4.18.1 → 4.18.2

package/dist/{parser-4D2M5BKY.js → parser-KB4BYQBZ.js}

@@ -36,27 +36,29 @@ import {
   normalizeItems,
   refineDocumentStyleHeadings,
   removeHeaderFooterBlocks,
+  removeSideTabs,
   startsNewItem,
   trimTrailingEmptyTableCols,
   wrapJoiner
-} from "./chunk-WWOTHS2E.js";
+} from "./chunk-OGAYGUJF.js";
 import {
   ocrModelsCached
-} from "./chunk-R2IFWAHI.js";
-import "./chunk-NQR3VKWN.js";
-import "./chunk-SQQ4ITJ5.js";
+} from "./chunk-NZM2SSI4.js";
+import "./chunk-4B2EXELH.js";
+import "./chunk-6OKBTY72.js";
 import {
   blocksToPages
-} from "./chunk-PWRY4W6Q.js";
+} from "./chunk-GM7XIZF2.js";
 import {
   encodePng
 } from "./chunk-7FCNOKOR.js";
 import {
   CELL_EDGES,
   CONTENT_CELLS,
+  escapeLiteralTags,
   stripScriptTags,
   unframeLayoutTables
-} from "./chunk-TMC4A3FM.js";
+} from "./chunk-F5HING4Y.js";
 import {
   hasRequestedPagesAfter,
   parsePageRange
@@ -65,11 +67,11 @@ import {
   WINGDINGS,
   blocksToMarkdown,
   escapeLiteralDollar
-} from "./chunk-K35WIYBD.js";
+} from "./chunk-IA3UKDHA.js";
 import {
   KordocError,
   sanitizeHref
-} from "./chunk-I7YWXECO.js";
+} from "./chunk-OBILAVWC.js";
 
 // src/pdf/ocr-region-merge.ts
 function mergeOcrImageRegions(blocks, page, regions, ocrBlocks) {
@@ -2335,61 +2337,6 @@ function orderTwoUpPage(blocks, pageWidth, pageHeight) {
   return [...left, ...right, ...rest];
 }
 
-// src/pdf/side-tabs.ts
-var BAND = 0.12;
-var MAX_CHARS = 10;
-var MIN_REPEAT = 3;
-var RUN_GAP = 5;
-function removeSideTabs(blocks, pageWidths) {
-  const edges = /* @__PURE__ */ new Map();
-  for (const b of blocks) {
-    if (!b.bbox || b.pageNumber === void 0 || [...b.text?.trim() ?? ""].length <= MAX_CHARS) continue;
-    const e = edges.get(b.pageNumber) ?? { lefts: [], rights: [] };
-    e.lefts.push(b.bbox.x);
-    e.rights.push(b.bbox.x + b.bbox.width);
-    edges.set(b.pageNumber, e);
-  }
-  const q = (xs, p) => {
-    const v = [...xs].sort((a, b) => a - b);
-    return v[Math.min(v.length - 1, Math.floor(p * v.length))];
-  };
-  const body = /* @__PURE__ */ new Map();
-  for (const [page, e] of edges) body.set(page, { left: q(e.lefts, 0.2), right: q(e.rights, 0.8) });
-  const all = [...body.values()];
-  const fallback = all.length ? { left: q(all.map((e) => e.left), 0.5), right: q(all.map((e) => e.right), 0.5) } : void 0;
-  const key = (b) => {
-    const w = b.pageNumber !== void 0 ? pageWidths.get(b.pageNumber) : void 0;
-    const text = b.text?.trim();
-    const area = (b.pageNumber !== void 0 ? body.get(b.pageNumber) : void 0) ?? fallback;
-    if (!w || !area || !b.bbox || !text || b.type === "table" || [...text].length > MAX_CHARS) return null;
-    const side = b.bbox.x + b.bbox.width <= Math.min(w * BAND, area.left - 1) ? "L" : b.bbox.x >= Math.max(w * (1 - BAND), area.right + 1) ? "R" : null;
-    return side && text.replace(/\s+/g, "");
-  };
-  const pages = /* @__PURE__ */ new Map();
-  for (const b of blocks) {
-    const k = key(b);
-    if (k) pages.set(k, (pages.get(k) ?? /* @__PURE__ */ new Set()).add(b.pageNumber));
-  }
-  const running = /* @__PURE__ */ new Set();
-  for (const [k, ps] of pages) {
-    const list = [...ps].sort((a, b) => a - b);
-    let run = [];
-    const flush = () => {
-      if (run.length >= MIN_REPEAT && run.length >= (run[run.length - 1] - run[0] + 1) * 0.4) for (const p of run) running.add(`${p}\0${k}`);
-      run = [];
-    };
-    for (const p of list) {
-      if (run.length && p - run[run.length - 1] > RUN_GAP) flush();
-      run.push(p);
-    }
-    flush();
-  }
-  return running.size ? blocks.filter((b) => {
-    const k = key(b);
-    return !k || !running.has(`${b.pageNumber}\0${k}`);
-  }) : blocks;
-}
-
 // src/pdf/footnotes.ts
 var MARK = /^\d{1,3}\)$/;
 function footnoteSeparators(horizontals, pageWidth, pageHeight) {
@@ -2739,7 +2686,7 @@ function mergeLinkRuns(markdown) {
 
 // src/pdf/formula-ocr.ts
 async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
-  const formulaMod = await import("./formula-T4Q2UHWO.js");
+  const formulaMod = await import("./formula-E3N7KDDG.js");
   const { FormulaPipeline, ensureFormulaModels } = formulaMod;
   await ensureFormulaModels((p) => {
     if (p.phase === "download" && p.total) {
@@ -3034,6 +2981,7 @@ async function parsePdfDocument(buffer, options) {
           if (!faceNames.has(face)) faceNames.set(face, (fontObj(face)?.name ?? "").replace(/^[A-Z]{6}\+/, ""));
         }
         for (const it of visible) if (it.text.includes("$")) it.text = escapeLiteralDollar(it.text);
+        for (const it of visible) if (it.text.includes("<")) it.text = escapeLiteralTags(it.text);
         wrapEquationRuns(visible, (face) => faceNames.get(face));
         const pageArea = pageW * pageH;
         if (pageArea > 0) {
@@ -3122,7 +3070,7 @@ async function parsePdfDocument(buffer, options) {
       }
       if (targets.size > 0) {
         try {
-          const { runPdfOcr } = await import("./pdf-ocr-NJTFXBRC.js");
+          const { runPdfOcr } = await import("./pdf-ocr-YEBFWU2K.js");
           const mode = typeof options?.ocr === "function" ? options.ocr : "builtin";
           const regionPages = new Map([...ocrRegions].filter(([p]) => options?.ocr !== "force" && !isImageBased && !pageQuality.find((q) => q.page === p)?.needsOcr));
           const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, options?.onProgress, options?.tables !== false, vectorPageOps, regionPages);
@@ -3354,4 +3302,4 @@ export {
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-4D2M5BKY.js.map
+//# sourceMappingURL=parser-KB4BYQBZ.js.map