kordoc 4.18.4 → 4.18.5

package/dist/{-DDAVBMHD.js → -3HZVXMPX.js}

@@ -51,17 +51,17 @@ import {
   unknownFontWarnings,
   usesGaejosikMunche,
   validateHwpx
-} from "./chunk-F2GDLA45.js";
+} from "./chunk-XNIVXYIZ.js";
 import {
   blocksToPages
-} from "./chunk-4IBY5LTJ.js";
+} from "./chunk-GYUV53LM.js";
 import {
   applySplices,
   buildParagraphSplices,
   buildRangeSplices,
   placeSealHwpx,
   scanSectionXml
-} from "./chunk-DUV7Y6IQ.js";
+} from "./chunk-DV7DQEYU.js";
 import {
   DEFAULT_REDACT_RULES,
   redactMarkdown,
@@ -69,7 +69,7 @@ import {
 } from "./chunk-3WSNLH4B.js";
 import {
   blocksToChunks
-} from "./chunk-YCY2R4FM.js";
+} from "./chunk-U4SBX3B5.js";
 import {
   SPACE_EM_FIXED,
   SPACE_EM_FONT,
@@ -94,20 +94,20 @@ import {
   renderSceneToHtml,
   simulateWrap,
   simulateWrapKeepWord
-} from "./chunk-V5X4LLVM.js";
+} from "./chunk-KDKPSX4V.js";
 import "./chunk-7FCNOKOR.js";
-import "./chunk-TMJVD6FW.js";
+import "./chunk-OLCBIT3I.js";
 import "./chunk-VQFA55IC.js";
-import "./chunk-LGF2IO66.js";
+import "./chunk-33YID2RV.js";
 import {
   blocksToMarkdown,
   flattenLayoutTables,
   hasStructuredCellContent
-} from "./chunk-VTVK2JMI.js";
-import "./chunk-IAXGGQGD.js";
+} from "./chunk-XSCQQ74A.js";
+import "./chunk-L2K45SOK.js";
 import {
   VERSION
-} from "./chunk-F5YVCJVF.js";
+} from "./chunk-QAE2NVVO.js";
 export {
   BUILTIN_TEMPLATES,
   DEFAULT_REDACT_RULES,
@@ -198,4 +198,4 @@ export {
   usesGaejosikMunche,
   validateHwpx
 };
-//# sourceMappingURL=-DDAVBMHD.js.map
+//# sourceMappingURL=-3HZVXMPX.js.map

package/dist/{chunk-ELOPDQES.js → chunk-235UD74F.js}

@@ -1,6 +1,6 @@
 import {
   KordocError
-} from "./chunk-7GR2GCJF.js";
+} from "./chunk-ZH5Y3JT6.js";
 
 // src/pdf/formula/models.ts
 import { createHash as createHash2 } from "crypto";
@@ -308,4 +308,4 @@ export {
   ensureFormulaModels,
   ensureSingleModel
 };
-//# sourceMappingURL=chunk-ELOPDQES.js.map
+//# sourceMappingURL=chunk-235UD74F.js.map

package/dist/{chunk-KUUSP6OS.js → chunk-2CRD4Q76.js}

@@ -1,15 +1,88 @@
-#!/usr/bin/env node
 import {
   CELL_EDGES,
   HEADING_RATIO_H1,
   HEADING_RATIO_H2,
   HEADING_RATIO_H3,
   tidyScriptTags
-} from "./chunk-TMJVD6FW.js";
+} from "./chunk-PQ2ER3VQ.js";
+import {
+  ensureModelsIn,
+  getModelsDir
+} from "./chunk-235UD74F.js";
 import {
   safeMax,
   safeMin
-} from "./chunk-F5YVCJVF.js";
+} from "./chunk-ZH5Y3JT6.js";
+
+// src/ocr/models.ts
+import { join } from "path";
+import { stat } from "fs/promises";
+var OCR_DET_MODEL = {
+  name: "PP-OCRv5 mobile det",
+  filename: "det.onnx",
+  url: "https://huggingface.co/PaddlePaddle/PP-OCRv5_mobile_det_onnx/resolve/main/inference.onnx",
+  sha256: "a431985659dc921974177a95adcfbb90fd9e51989a5e04d70d0b75f597b6e61d",
+  sizeMb: 5
+};
+var OCR_REC_MODEL = {
+  name: "PP-OCRv5 korean rec",
+  filename: "rec_korean.onnx",
+  url: "https://huggingface.co/PaddlePaddle/korean_PP-OCRv5_mobile_rec_onnx/resolve/main/inference.onnx",
+  sha256: "92f0b7785e64fc9090106a241cf4c1eb97472824558272751b88a2a4476d3a08",
+  sizeMb: 13
+};
+var OCR_REC_DICT = {
+  name: "PP-OCRv5 korean dict",
+  filename: "rec_korean.yml",
+  url: "https://huggingface.co/PaddlePaddle/korean_PP-OCRv5_mobile_rec_onnx/resolve/main/inference.yml",
+  sha256: "f757fa1c40e99edcf27e9cce879b93eb2a51fa46f5ef39095689b8c37dd75998",
+  sizeMb: 1
+};
+var ALL_OCR_MODELS = [OCR_DET_MODEL, OCR_REC_MODEL, OCR_REC_DICT];
+function getOcrModelsDir() {
+  return getModelsDir("ppocr");
+}
+async function ensureOcrModels(onProgress) {
+  return ensureModelsIn(getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
+}
+async function ocrModelsCached() {
+  const dir = getOcrModelsDir();
+  for (const spec of ALL_OCR_MODELS) {
+    try {
+      if (!(await stat(join(dir, spec.filename))).size) return false;
+    } catch {
+      return false;
+    }
+  }
+  return true;
+}
+function parseCharacterDict(yml) {
+  const lines = yml.split("\n");
+  const chars = [];
+  let inDict = false;
+  let dictIndent = -1;
+  for (const line of lines) {
+    if (!inDict) {
+      const m2 = /^(\s*)character_dict:\s*$/.exec(line);
+      if (m2) {
+        inDict = true;
+        dictIndent = m2[1].length;
+      }
+      continue;
+    }
+    const m = /^(\s*)- (.*)$/.exec(line);
+    if (m && m[1].length >= dictIndent) {
+      let v = m[2];
+      if (v.length >= 2 && (v.startsWith("'") && v.endsWith("'") || v.startsWith('"') && v.endsWith('"'))) {
+        v = v.slice(1, -1).replace(/''/g, "'");
+      }
+      chars.push(v);
+      continue;
+    }
+    if (line.trim() !== "") break;
+  }
+  return chars;
+}
 
 // src/pdf/line-extract.ts
 import { OPS } from "pdfjs-dist/legacy/build/pdf.mjs";
@@ -6828,6 +6901,55 @@ function frameLayoutBoxes(blocks) {
   }
   return boxes;
 }
+function takeFrameSpanningText(items, cutX, pageNum, lex) {
+  let minX = Infinity, maxR = -Infinity;
+  for (const item of items) {
+    minX = Math.min(minX, item.x);
+    maxR = Math.max(maxR, item.x + item.w);
+  }
+  const span = maxR - minX;
+  const rows = groupByY([...items].sort((a, b) => b.y - a.y || a.x - b.x)).map((line) => {
+    const runs = [];
+    for (const item of [...line].sort((a, b) => a.x - b.x)) {
+      const run = runs[runs.length - 1], prev = run?.[run.length - 1];
+      if (prev && item.x - (prev.x + prev.w) <= Math.max(prev.fontSize, item.fontSize) * 0.6) run.push(item);
+      else runs.push([item]);
+    }
+    return runs;
+  });
+  const taken = /* @__PURE__ */ new Set(), blocks = [];
+  for (let r = 0; r < rows.length; r++) for (const run of rows[r]) {
+    if (taken.has(run[0]) || run[0].x >= cutX || Math.max(...run.map((i) => i.x + i.w)) <= cutX) continue;
+    const lines = [run], fs = Math.max(...run.map((i) => i.fontSize));
+    const left = run[0].x, right = Math.max(...run.map((i) => i.x + i.w));
+    if (right - left < span * 0.75) continue;
+    let prevY = run[0].y;
+    for (let next = r + 1; next < rows.length && rows[next].length === 1; next++) {
+      const tail = rows[next][0], y = tail[0].y;
+      if (taken.has(tail[0]) || prevY - y > fs * 1.8 || tail[0].x < left - 1 || tail[0].x > left + fs * 2 || tail.some((i) => Math.abs(i.fontSize - fs) > fs * 0.15) || Math.max(...tail.map((i) => i.x + i.w)) > right + 1) break;
+      lines.push(tail);
+      prevY = y;
+    }
+    for (const line of lines) for (const item of line) taken.add(item);
+    pushLineParagraphs(blocks, mergeSuperscriptLines(lines), pageNum, lex);
+  }
+  let keep = 0;
+  for (const item of items) if (!taken.has(item)) items[keep++] = item;
+  items.length = keep;
+  return blocks;
+}
+function frameColumnTextBands(items, spanning, cutX) {
+  const tops = spanning.filter((b) => b.bbox).map((b) => b.bbox.y + b.bbox.height).sort((a, b) => b - a);
+  const bands = /* @__PURE__ */ new Map();
+  for (const item of items) {
+    let band = 0;
+    while (band < tops.length && tops[band] > item.y + item.h) band++;
+    const sides = bands.get(band) ?? [[], [], []];
+    sides[item.x + item.w <= cutX ? 0 : item.x >= cutX ? 2 : 1].push(item);
+    bands.set(band, sides);
+  }
+  return [...bands].sort((a, b) => a[0] - b[0]).flatMap(([, sides]) => sides.filter((side) => side.length));
+}
 function groupFrameParagraphUnits(units, frames) {
   const source = /* @__PURE__ */ new Map();
   for (const unit of units) for (const block of unit) source.set(block, unit);
@@ -7738,14 +7860,15 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
   if (remaining.length > 0) {
     if (gutterX !== null) {
       const gx = gutterX;
+      const spanning = detectListBlocks(takeFrameSpanningText(remaining, gx, pageNum, lex));
+      for (const block of spanning) {
+        finalTextBlocks.push(block);
+        groupSizes.push(1);
+      }
       const allY = remaining.map((i) => i.y);
       const pageH = safeMax(allY) - safeMin(allY);
       const gapThreshold = Math.max(15, pageH * 0.03);
-      const sides = [
-        remaining.filter((i) => i.x + i.w <= gx),
-        remaining.filter((i) => i.x < gx && i.x + i.w > gx),
-        remaining.filter((i) => i.x >= gx)
-      ];
+      const sides = frameColumnTextBands(remaining, spanning, gx);
       const textBlocks = [];
       for (const side of sides) {
         if (side.length === 0) continue;
@@ -7756,7 +7879,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
           groupSizes.push(groupBlocks.length);
         }
       }
-      finalTextBlocks = detectListBlocks(textBlocks);
+      finalTextBlocks.push(...detectListBlocks(textBlocks));
     } else {
       const allY = remaining.map((i) => i.y);
       const pageH = safeMax(allY) - safeMin(allY);
@@ -8105,6 +8228,13 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
 }
 
 export {
+  OCR_DET_MODEL,
+  OCR_REC_MODEL,
+  OCR_REC_DICT,
+  getOcrModelsDir,
+  ensureOcrModels,
+  ocrModelsCached,
+  parseCharacterDict,
   extractLines,
   extractImageRegions,
   filterHiddenText,
@@ -8155,4 +8285,4 @@ export {
   removeSideTabs,
   extractPageBlocksWithLines
 };
-//# sourceMappingURL=chunk-KUUSP6OS.js.map
+//# sourceMappingURL=chunk-2CRD4Q76.js.map