kordoc 4.17.2 → 4.17.4

package/dist/{parser-VH3F4EQU.js → parser-T4FBG2FR.js}

@@ -1,10 +1,7 @@
 import {
-  WINGDINGS,
-  blocksToMarkdown,
   blocksToPages,
-  encodePng,
-  escapeLiteralDollar
-} from "./chunk-SDRKHEVT.js";
+  encodePng
+} from "./chunk-PNXJYEOR.js";
 import {
   hasRequestedPagesAfter,
   parsePageRange
@@ -50,18 +47,21 @@ import {
   startsNewItem,
   trimTrailingEmptyTableCols,
   wrapJoiner
-} from "./chunk-3N272C5L.js";
+} from "./chunk-YBIWPCUX.js";
 import {
   CELL_EDGES,
   CONTENT_CELLS,
+  WINGDINGS,
+  blocksToMarkdown,
+  escapeLiteralDollar,
   stripScriptTags,
   unframeLayoutTables
-} from "./chunk-UX3HHAWH.js";
-import "./chunk-5JZEY5IL.js";
+} from "./chunk-NNNRN6BY.js";
+import "./chunk-VJP7O6BL.js";
 import {
   KordocError,
   sanitizeHref
-} from "./chunk-YGKC6E6A.js";
+} from "./chunk-PTMYCK7R.js";
 
 // src/pdf/ocr-region-merge.ts
 function mergeOcrImageRegions(blocks, page, regions, ocrBlocks) {
@@ -1072,6 +1072,15 @@ function besideOwn(b, p, c) {
   const bx1 = b.bbox.x, bx2 = b.bbox.x + b.bbox.width;
   return bx2 <= t.x + 1 || bx1 >= t.x + t.width - 1;
 }
+var PAGE_NUMBER_TEXT = /^[-–—]?\s*\d{1,4}\s*[-–—]?$/;
+function pageNumberBetween(b, p, c, pageHeights) {
+  if (b.type !== "paragraph" || !b.bbox || !PAGE_NUMBER_TEXT.test(b.text?.trim() ?? "")) return false;
+  const h = b.pageNumber ? pageHeights?.get(b.pageNumber) : void 0;
+  if (!h) return false;
+  if (b.pageNumber === p.pageNumber) return b.bbox.y + b.bbox.height <= Math.min(p.bbox.y, h * PAGE_EDGE_BAND) + 1;
+  if (b.pageNumber === c.pageNumber) return b.bbox.y >= Math.max(c.bbox.y + c.bbox.height, h * (1 - PAGE_EDGE_BAND)) - 1;
+  return false;
+}
 function leftColumnOf(b, t) {
   return !!b.bbox && b.pageNumber === t.pageNumber && b.bbox.x + b.bbox.width <= t.bbox.x + 1;
 }
@@ -1113,7 +1122,7 @@ function mergeCrossPageTables(blocks, pageHeights, lex) {
     if (!prev.table || !prev.bbox || !prev.pageNumber) continue;
     if (!curr || curr.type !== "table" || !curr.table || !curr.bbox || curr.pageNumber !== prev.pageNumber + 1) continue;
     if (startsWithUnitRow(prev.table) && startsWithUnitRow(curr.table)) continue;
-    const joined = j === i + 1 || blocks.slice(i + 1, j).every((b) => besideOwn(b, prev, curr) || insideTable(b, prev)) ? looksContinued(prev, curr, pageHeights) && !restartsTable(blocks, i, curr.table, pageHeights) ? joinClipParts(prev, curr, pageHeights, lex) ?? false : null : null;
+    const joined = j === i + 1 || blocks.slice(i + 1, j).every((b) => besideOwn(b, prev, curr) || insideTable(b, prev) || pageNumberBetween(b, prev, curr, pageHeights)) ? looksContinued(prev, curr, pageHeights) && !restartsTable(blocks, i, curr.table, pageHeights) ? joinClipParts(prev, curr, pageHeights, lex) ?? false : null : null;
     if (joined) {
       placeJoined(blocks, i, j, joined);
       continue;
@@ -1327,7 +1336,7 @@ function chainHead(blocks, i, pageHeights) {
     }
     const pb = tailOf(blocks[p]);
     if (pb.pageNumber !== (cur.pageNumber ?? 0) - 1) break;
-    if (!blocks.slice(p + 1, k).every((b) => besideOwn(b, pb, cur) || insideTable(b, pb))) break;
+    if (!blocks.slice(p + 1, k).every((b) => besideOwn(b, pb, cur) || insideTable(b, pb) || pageNumberBetween(b, pb, cur, pageHeights))) break;
     if (!looksContinued(pb, cur, pageHeights)) break;
     const px = TABLE_COLXS.get(pb.table), cx = TABLE_COLXS.get(cur.table);
     if (!px || !cx) break;
@@ -2704,7 +2713,7 @@ function mergeLinkRuns(markdown) {
 
 // src/pdf/formula-ocr.ts
 async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
-  const formulaMod = await import("./formula-QZZDW5GW.js");
+  const formulaMod = await import("./formula-WZH5CMA3.js");
   const { FormulaPipeline, ensureFormulaModels } = formulaMod;
   await ensureFormulaModels((p) => {
     if (p.phase === "download" && p.total) {
@@ -3087,7 +3096,7 @@ async function parsePdfDocument(buffer, options) {
       }
       if (targets.size > 0) {
         try {
-          const { runPdfOcr } = await import("./pdf-ocr-XC33L43D.js");
+          const { runPdfOcr } = await import("./pdf-ocr-RG2MFWBX.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);
@@ -3319,4 +3328,4 @@ export {
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-VH3F4EQU.js.map
+//# sourceMappingURL=parser-T4FBG2FR.js.map