kordoc 4.18.1 → 4.18.3

package/dist/{parser-5G5G7HHG.js → parser-SCXRLOW7.js}

@@ -1,7 +1,7 @@
 import {
   blocksToPages,
   encodePng
-} from "./chunk-KBZ3MQJX.js";
+} from "./chunk-BGOR5GCF.js";
 import {
   hasRequestedPagesAfter,
   parsePageRange
@@ -44,24 +44,26 @@ import {
   ocrModelsCached,
   refineDocumentStyleHeadings,
   removeHeaderFooterBlocks,
+  removeSideTabs,
   startsNewItem,
   trimTrailingEmptyTableCols,
   wrapJoiner
-} from "./chunk-6V7QRQY3.js";
+} from "./chunk-BJ6RYYXV.js";
 import {
   CELL_EDGES,
   CONTENT_CELLS,
   WINGDINGS,
   blocksToMarkdown,
   escapeLiteralDollar,
+  escapeLiteralTags,
   stripScriptTags,
   unframeLayoutTables
-} from "./chunk-HUGBJVAA.js";
-import "./chunk-KXR4CTFT.js";
+} from "./chunk-PSJLB3SZ.js";
+import "./chunk-3FZGL3IL.js";
 import {
   KordocError,
   sanitizeHref
-} from "./chunk-VDIXNEWV.js";
+} from "./chunk-CAN5P6N6.js";
 
 // src/pdf/ocr-region-merge.ts
 function mergeOcrImageRegions(blocks, page, regions, ocrBlocks) {
@@ -914,6 +916,17 @@ var CARRIED_MIN_COL_W = 4;
 var LABEL_ROOM = 1.5;
 var HANGING_MIN = 0.5;
 var HANGING_TOL = 1;
+var KO_ITEMS = "\uAC00\uB098\uB2E4\uB77C\uB9C8\uBC14\uC0AC\uC544\uC790\uCC28\uCE74\uD0C0\uD30C\uD558\uAC70\uB108\uB354\uB7EC\uBA38\uBC84\uC11C\uC5B4\uC800\uCC98\uCEE4\uD130\uD37C\uD5C8\uACE0\uB178\uB3C4\uB85C\uBAA8\uBCF4\uC18C\uC624\uC870\uCD08\uCF54\uD1A0\uD3EC\uD638\uAD6C\uB204\uB450\uB8E8\uBB34\uBD80\uC218\uC6B0\uC8FC\uCD94\uCFE0\uD22C\uD478\uD6C4";
+var ITEM_MARK = /^\s*(?:([가-힣])\.|(\d{1,3})[.)]|\(([가-힣\d]{1,3})\))(?=\s)/;
+function nextItemHead(prev, next) {
+  const a = ITEM_MARK.exec(prev), b = ITEM_MARK.exec(next);
+  if (!a || !b) return false;
+  const succ = (x, y) => /^\d+$/.test(x) ? /^\d+$/.test(y) && Number(y) === Number(x) + 1 : KO_ITEMS.indexOf(x) >= 0 && KO_ITEMS.indexOf(y) === KO_ITEMS.indexOf(x) + 1;
+  if (a[1] && b[1]) return succ(a[1], b[1]);
+  if (a[2] && b[2]) return succ(a[2], b[2]) && prev.trimStart()[a[2].length] === next.trimStart()[b[2].length];
+  if (a[3] && b[3]) return succ(a[3], b[3]);
+  return false;
+}
 function continuesAcross(u, d, x1, x2) {
   const U = CELL_LINES.get(u), D = CELL_LINES.get(d);
   if (!U?.length || !D?.length) return false;
@@ -1045,6 +1058,8 @@ function mergeSplitRow(table, owner, first, colXs, textOnly = false, cut = NO_CU
   if (!carriedSplit) {
     const norm = (c) => flowText(c.text).replace(/\s+/g, "");
     if (pairs.some(([u, d]) => hasContent(d.cell) && norm(cell(d)) !== norm(cell(u)) && lineEnded(cell(u), colXs[u.c], colXs[u.c + u.cs], cell(d)))) return false;
+    const oneLine = (c) => CELL_LINES.get(c)?.length === 1;
+    if (pairs.some(([u, d]) => nextItemHead(cell(u).text, cell(d).text)) && pairs.some(([u, d]) => hasContent(u.cell) && hasContent(d.cell) && oneLine(cell(u)) && oneLine(cell(d)) && norm(cell(u)) !== norm(cell(d)))) return false;
     const flows = ([u, d]) => continuesAcross(cell(u), cell(d), colXs[u.c], colXs[u.c + u.cs]) || clauseContinues(cell(u), cell(d));
     if (!(cut.open && !newCell)) {
       if (pairs.some((p) => p[1].rs > 1 && !flows(p))) return false;
@@ -2327,61 +2342,6 @@ 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) {
@@ -2731,7 +2691,7 @@ function mergeLinkRuns(markdown) {
 
 // src/pdf/formula-ocr.ts
 async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
-  const formulaMod = await import("./formula-LICM6K27.js");
+  const formulaMod = await import("./formula-GPCOJJXN.js");
   const { FormulaPipeline, ensureFormulaModels } = formulaMod;
   await ensureFormulaModels((p) => {
     if (p.phase === "download" && p.total) {
@@ -3026,6 +2986,7 @@ 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) {
@@ -3114,7 +3075,7 @@ async function parsePdfDocument(buffer, options) {
       }
       if (targets.size > 0) {
         try {
-          const { runPdfOcr } = await import("./pdf-ocr-ZXR6Q6BK.js");
+          const { runPdfOcr } = await import("./pdf-ocr-OLII3WAG.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);
@@ -3346,4 +3307,4 @@ export {
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-5G5G7HHG.js.map
+//# sourceMappingURL=parser-SCXRLOW7.js.map