kordoc 3.8.1 → 3.8.2

package/dist/{parser-IXK5V7YG.cjs → parser-NS4ZPD7B.cjs}

@@ -6,11 +6,13 @@
 
 
 
-var _chunkLFCS3UVGcjs = require('./chunk-LFCS3UVG.cjs');
+var _chunkYBPNKFJWcjs = require('./chunk-YBPNKFJW.cjs');
 
 
 var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
-require('./chunk-GS7T56RP.cjs');
+
+
+var _chunkGS7T56RPcjs = require('./chunk-GS7T56RP.cjs');
 
 // src/pdf/line-extract.ts
 var _pdfmjs = require('pdfjs-dist/legacy/build/pdf.mjs');
@@ -21,21 +23,30 @@ function extractLines(fnArray, argsArray) {
   const horizontals = [];
   const verticals = [];
   let lineWidth = 1;
+  let ctm = [1, 0, 0, 1, 0, 0];
+  const ctmStack = [];
+  const applyCtm = (x, y) => [ctm[0] * x + ctm[2] * y + ctm[4], ctm[1] * x + ctm[3] * y + ctm[5]];
+  const ctmScale = () => (Math.hypot(ctm[0], ctm[1]) + Math.hypot(ctm[2], ctm[3])) / 2;
   let currentPath = [];
   let pathStartX = 0, pathStartY = 0;
   let curX = 0, curY = 0;
-  function pushRectangle(path, rx, ry, rw, rh) {
-    if (Math.abs(rh) < ORIENTATION_TOL * 2) {
-      path.push({ x1: rx, y1: ry + rh / 2, x2: rx + rw, y2: ry + rh / 2 });
-    } else if (Math.abs(rw) < ORIENTATION_TOL * 2) {
-      path.push({ x1: rx + rw / 2, y1: ry, x2: rx + rw / 2, y2: ry + rh });
+  function pushSeg(x1, y1, x2, y2) {
+    const [tx1, ty1] = applyCtm(x1, y1);
+    const [tx2, ty2] = applyCtm(x2, y2);
+    currentPath.push({ x1: tx1, y1: ty1, x2: tx2, y2: ty2 });
+  }
+  function pushRectangle(rx, ry, rw, rh) {
+    const effH = Math.abs(rh) * Math.hypot(ctm[2], ctm[3]);
+    const effW = Math.abs(rw) * Math.hypot(ctm[0], ctm[1]);
+    if (effH < ORIENTATION_TOL * 2) {
+      pushSeg(rx, ry + rh / 2, rx + rw, ry + rh / 2);
+    } else if (effW < ORIENTATION_TOL * 2) {
+      pushSeg(rx + rw / 2, ry, rx + rw / 2, ry + rh);
     } else {
-      path.push(
-        { x1: rx, y1: ry, x2: rx + rw, y2: ry },
-        { x1: rx + rw, y1: ry, x2: rx + rw, y2: ry + rh },
-        { x1: rx + rw, y1: ry + rh, x2: rx, y2: ry + rh },
-        { x1: rx, y1: ry + rh, x2: rx, y2: ry }
-      );
+      pushSeg(rx, ry, rx + rw, ry);
+      pushSeg(rx + rw, ry, rx + rw, ry + rh);
+      pushSeg(rx + rw, ry + rh, rx, ry + rh);
+      pushSeg(rx, ry + rh, rx, ry);
     }
   }
   function flushPath(isStroke) {
@@ -43,8 +54,9 @@ function extractLines(fnArray, argsArray) {
       currentPath = [];
       return;
     }
+    const effWidth = lineWidth * ctmScale();
     for (const seg of currentPath) {
-      classifyAndAdd(seg, lineWidth, horizontals, verticals);
+      classifyAndAdd(seg, effWidth, horizontals, verticals);
     }
     currentPath = [];
   }
@@ -55,6 +67,24 @@ function extractLines(fnArray, argsArray) {
       case _pdfmjs.OPS.setLineWidth:
         lineWidth = args[0] || 1;
         break;
+      case _pdfmjs.OPS.save:
+        ctmStack.push(ctm.slice());
+        break;
+      case _pdfmjs.OPS.restore:
+        ctm = _nullishCoalesce(ctmStack.pop(), () => ( [1, 0, 0, 1, 0, 0]));
+        break;
+      case _pdfmjs.OPS.transform: {
+        const t = args;
+        ctm = [
+          ctm[0] * t[0] + ctm[2] * t[1],
+          ctm[1] * t[0] + ctm[3] * t[1],
+          ctm[0] * t[2] + ctm[2] * t[3],
+          ctm[1] * t[2] + ctm[3] * t[3],
+          ctm[0] * t[4] + ctm[2] * t[5] + ctm[4],
+          ctm[1] * t[4] + ctm[3] * t[5] + ctm[5]
+        ];
+        break;
+      }
       case _pdfmjs.OPS.constructPath: {
         const arg0 = args[0];
         if (Array.isArray(arg0)) {
@@ -69,16 +99,16 @@ function extractLines(fnArray, argsArray) {
               pathStartY = curY;
             } else if (subOp === _pdfmjs.OPS.lineTo) {
               const x2 = coords[ci++], y2 = coords[ci++];
-              currentPath.push({ x1: curX, y1: curY, x2, y2 });
+              pushSeg(curX, curY, x2, y2);
               curX = x2;
               curY = y2;
             } else if (subOp === _pdfmjs.OPS.rectangle) {
               const rx = coords[ci++], ry = coords[ci++];
               const rw = coords[ci++], rh = coords[ci++];
-              pushRectangle(currentPath, rx, ry, rw, rh);
+              pushRectangle(rx, ry, rw, rh);
             } else if (subOp === _pdfmjs.OPS.closePath) {
               if (curX !== pathStartX || curY !== pathStartY) {
-                currentPath.push({ x1: curX, y1: curY, x2: pathStartX, y2: pathStartY });
+                pushSeg(curX, curY, pathStartX, pathStartY);
               }
               curX = pathStartX;
               curY = pathStartY;
@@ -104,7 +134,7 @@ function extractLines(fnArray, argsArray) {
                 pathStartY = curY;
               } else if (drawOp === 1 /* lineTo */) {
                 const x2 = pathData[di++], y2 = pathData[di++];
-                currentPath.push({ x1: curX, y1: curY, x2, y2 });
+                pushSeg(curX, curY, x2, y2);
                 curX = x2;
                 curY = y2;
               } else if (drawOp === 2 /* curveTo */) {
@@ -113,7 +143,7 @@ function extractLines(fnArray, argsArray) {
                 di += 4;
               } else if (drawOp === 4 /* closePath */) {
                 if (curX !== pathStartX || curY !== pathStartY) {
-                  currentPath.push({ x1: curX, y1: curY, x2: pathStartX, y2: pathStartY });
+                  pushSeg(curX, curY, pathStartX, pathStartY);
                 }
                 curX = pathStartX;
                 curY = pathStartY;
@@ -1847,14 +1877,14 @@ function isProseSpread(items) {
   for (let i = 1; i < sorted.length; i++) {
     gaps.push(sorted[i].x - (sorted[i - 1].x + sorted[i - 1].w));
   }
-  const maxGap = _chunkLFCS3UVGcjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkYBPNKFJWcjs.safeMax.call(void 0, gaps);
   const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
   return maxGap < 40 && avgLen < 5;
 }
 function detectColumns(yLines) {
   const allItems = yLines.flat();
   if (allItems.length === 0) return null;
-  const pageWidth = _chunkLFCS3UVGcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkLFCS3UVGcjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkYBPNKFJWcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkYBPNKFJWcjs.safeMin.call(void 0, allItems.map((i) => i.x));
   if (pageWidth < 100) return null;
   let bigoLineIdx = -1;
   for (let i = 0; i < yLines.length; i++) {
@@ -2076,9 +2106,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkLFCS3UVGcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkLFCS3UVGcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkLFCS3UVGcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunkYBPNKFJWcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunkYBPNKFJWcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunkYBPNKFJWcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -2597,7 +2627,7 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
     }
     if (remaining.length > 0) {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkLFCS3UVGcjs.safeMax.call(void 0, allY) - _chunkLFCS3UVGcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkYBPNKFJWcjs.safeMax.call(void 0, allY) - _chunkYBPNKFJWcjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -2726,7 +2756,7 @@ function extractPageBlocksFallback(items, pageNum) {
       blocks.push({ type: "paragraph", text: tableText, pageNumber: pageNum, bbox, style: dominantStyle(items) });
     } else {
       const allY = items.map((i) => i.y);
-      const pageHeight = _chunkLFCS3UVGcjs.safeMax.call(void 0, allY) - _chunkLFCS3UVGcjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkYBPNKFJWcjs.safeMax.call(void 0, allY) - _chunkYBPNKFJWcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageHeight * 0.03);
       const orderedGroups = xyCutOrder(items, gapThreshold);
       for (const group of orderedGroups) {
@@ -2949,16 +2979,28 @@ g.pdfjsWorker = pdfjsWorker;
 
 // src/pdf/parser.ts
 
+var _module = require('module');
+var _path = require('path');
 _pdfmjs.GlobalWorkerOptions.workerSrc = "";
 var MAX_PAGES = 5e3;
 var MAX_TOTAL_TEXT = 100 * 1024 * 1024;
 var PDF_LOAD_TIMEOUT_MS = 3e4;
+var pdfjsAssets = {};
+try {
+  const _require = _module.createRequire.call(void 0, _chunkGS7T56RPcjs.importMetaUrl);
+  const pkgDir = _path.dirname.call(void 0, _require.resolve("pdfjs-dist/package.json"));
+  pdfjsAssets.cMapUrl = _path.join.call(void 0, pkgDir, "cmaps") + "/";
+  pdfjsAssets.cMapPacked = true;
+  pdfjsAssets.standardFontDataUrl = _path.join.call(void 0, pkgDir, "standard_fonts") + "/";
+} catch (e2) {
+}
 async function loadPdfWithTimeout(buffer) {
   const loadingTask = _pdfmjs.getDocument.call(void 0, {
     data: new Uint8Array(buffer),
     useSystemFonts: true,
     disableFontFace: true,
-    isEvalSupported: false
+    isEvalSupported: false,
+    ...pdfjsAssets
   });
   let timer;
   try {
@@ -2967,7 +3009,7 @@ async function loadPdfWithTimeout(buffer) {
       new Promise((_, reject) => {
         timer = setTimeout(() => {
           loadingTask.destroy();
-          reject(new (0, _chunkLFCS3UVGcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
+          reject(new (0, _chunkYBPNKFJWcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
         }, PDF_LOAD_TIMEOUT_MS);
       })
     ]);
@@ -2980,7 +3022,7 @@ async function parsePdfDocument(buffer, options) {
   const doc = await loadPdfWithTimeout(buffer);
   try {
     const pageCount = doc.numPages;
-    if (pageCount === 0) throw new (0, _chunkLFCS3UVGcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
+    if (pageCount === 0) throw new (0, _chunkYBPNKFJWcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
     const metadata = { pageCount };
     await extractPdfMetadata(doc, metadata);
     const blocks = [];
@@ -3040,11 +3082,11 @@ async function parsePdfDocument(buffer, options) {
           pageText += pageText ? "\n" + t : t;
         }
         pageQuality.push(computePageQuality(i, pageText));
-        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkLFCS3UVGcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
+        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkYBPNKFJWcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
         parsedPages++;
         _optionalChain([options, 'optionalAccess', _27 => _27.onProgress, 'optionalCall', _28 => _28(parsedPages, totalTarget)]);
       } catch (pageErr) {
-        if (pageErr instanceof _chunkLFCS3UVGcjs.KordocError) throw pageErr;
+        if (pageErr instanceof _chunkYBPNKFJWcjs.KordocError) throw pageErr;
         warnings.push({ page: i, message: `\uD398\uC774\uC9C0 ${i} \uD30C\uC2F1 \uC2E4\uD328: ${pageErr instanceof Error ? pageErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
       }
     }
@@ -3059,7 +3101,7 @@ async function parsePdfDocument(buffer, options) {
             const ocrMarkdown = ocrBlocks.map((b) => b.text || "").filter(Boolean).join("\n\n");
             return { markdown: ocrMarkdown, blocks: ocrBlocks, metadata, warnings, isImageBased: true, pageQuality, qualitySummary: summarizeDocumentQuality(pageQuality) };
           }
-        } catch (e2) {
+        } catch (e3) {
         }
       }
       isImageBased = true;
@@ -3112,7 +3154,7 @@ async function parsePdfDocument(buffer, options) {
     detectKoreanListBlocks(blocks);
     const outline = blocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
     sanitizeBlockControlChars(blocks);
-    let markdown = cleanPdfText(_chunkLFCS3UVGcjs.blocksToMarkdown.call(void 0, blocks));
+    let markdown = cleanPdfText(_chunkYBPNKFJWcjs.blocksToMarkdown.call(void 0, blocks));
     return {
       markdown,
       blocks,
@@ -3142,7 +3184,7 @@ async function extractPdfMetadata(doc, metadata) {
     }
     if (typeof info.CreationDate === "string") metadata.createdAt = parsePdfDate(info.CreationDate);
     if (typeof info.ModDate === "string") metadata.modifiedAt = parsePdfDate(info.ModDate);
-  } catch (e3) {
+  } catch (e4) {
   }
 }
 function parsePdfDate(dateStr) {
@@ -3171,4 +3213,4 @@ async function extractPdfMetadataOnly(buffer) {
 
 
 exports.cleanPdfText = cleanPdfText; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.detectTableCaptions = detectTableCaptions; exports.extractPdfMetadataOnly = extractPdfMetadataOnly; exports.mergeCrossPageTables = mergeCrossPageTables; exports.parsePdfDocument = parsePdfDocument; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks;
-//# sourceMappingURL=parser-IXK5V7YG.cjs.map
+//# sourceMappingURL=parser-NS4ZPD7B.cjs.map