kordoc 4.18.11 → 4.18.13

package/dist/{chunk-PIRSOTKK.js → chunk-GSW7W7AL.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   blocksToMarkdown
-} from "./chunk-WP6KG5Z7.js";
+} from "./chunk-JKHU7HID.js";
 
 // src/chunks.ts
 var LIST_MARKER_RE = /^(?:[□■◇◆○◎●◦ㅇ•▪▸▶※-]|\d{1,3}[.)]|[가나다라마바사아자차카타파하][.)]|\([가나다라마바사아자차카타파하0-9]{1,3}\)|[①-⑳㉮-㉻㈎-㈛])\s/;
@@ -79,4 +79,4 @@ function blocksToChunks(blocks, options) {
 export {
   blocksToChunks
 };
-//# sourceMappingURL=chunk-PIRSOTKK.js.map
+//# sourceMappingURL=chunk-GSW7W7AL.js.map

package/dist/{chunk-TYIUJKUT.js → chunk-GUH6HYSZ.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   blocksToPages
-} from "./chunk-UG4GIJSH.js";
+} from "./chunk-V7LKQNC4.js";
 import {
   allLinesegRemovalSplices,
   applySplices,
@@ -15,7 +15,7 @@ import {
   paraTextPureT,
   patchZipEntries,
   scanSectionXml
-} from "./chunk-GNXDTPP6.js";
+} from "./chunk-6AYTYTFY.js";
 import {
   CONVERT_MAP,
   FLAG_COMPRESSED,
@@ -65,7 +65,7 @@ import {
   romanNumeral,
   simulateWrap,
   takeTocLeaders
-} from "./chunk-L7TALZNE.js";
+} from "./chunk-SVIOP7PE.js";
 import {
   CELL_EDGES,
   HEADING_RATIO_H1,
@@ -77,7 +77,7 @@ import {
   tidyScriptTags,
   unframeLayoutTables,
   wrapScript
-} from "./chunk-UETNJSA7.js";
+} from "./chunk-XVB3PUPJ.js";
 import {
   parsePageRange
 } from "./chunk-VQFA55IC.js";
@@ -96,7 +96,7 @@ import {
   findChildByLocalName,
   localName,
   rawTextContent
-} from "./chunk-5GVDHHHO.js";
+} from "./chunk-DHEUQXXR.js";
 import {
   MAX_COLS,
   MAX_ROWS,
@@ -111,7 +111,7 @@ import {
   hasStructuredCellContent,
   mapPuaText,
   noteSuffix
-} from "./chunk-WP6KG5Z7.js";
+} from "./chunk-JKHU7HID.js";
 import {
   KordocError,
   classifyError,
@@ -127,7 +127,7 @@ import {
   toArrayBuffer,
   unescapeHtml,
   unzipLimitBytes
-} from "./chunk-JZ3VPASI.js";
+} from "./chunk-PN5WWW56.js";
 
 // src/table/visual.ts
 import { readFile as readFile2 } from "fs/promises";
@@ -30157,7 +30157,7 @@ function scriptsOff(r, off) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-MOGXLPIS.js");
+    const { parseImageDocument } = await import("./image-ocr-PAF76DAW.js");
     const { blocks, warnings } = await parseImageDocument(buffer, options);
     stripScriptTags({ blocks });
     return {
@@ -30217,7 +30217,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-PICN5MDD.js");
+    const mod = await import("./parser-I4D3TRGU.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {
@@ -30440,4 +30440,4 @@ export {
   parseHwpml,
   fillForm
 };
-//# sourceMappingURL=chunk-TYIUJKUT.js.map
+//# sourceMappingURL=chunk-GUH6HYSZ.js.map

package/dist/{chunk-CNU4NPCC.js → chunk-INESPVEV.js}

@@ -5,11 +5,11 @@ import {
   HEADING_RATIO_H2,
   HEADING_RATIO_H3,
   tidyScriptTags
-} from "./chunk-UETNJSA7.js";
+} from "./chunk-XVB3PUPJ.js";
 import {
   safeMax,
   safeMin
-} from "./chunk-JZ3VPASI.js";
+} from "./chunk-PN5WWW56.js";
 
 // src/pdf/line-extract.ts
 import { OPS } from "pdfjs-dist/legacy/build/pdf.mjs";
@@ -2169,8 +2169,15 @@ function wrapJoiner(prevText, nextText, lex) {
   return " ";
 }
 var PARA_LAST_LINE = /* @__PURE__ */ new WeakMap();
+var PARA_FIRST_LEFT = /* @__PURE__ */ new WeakMap();
 var PAGE_REF_TAIL = /(?:\t|…|·{2}|\.{3})\s*\d{1,4}(?:\s*\/\s*\d{1,4})?\s*$/;
 var PAGE_HEAD_MARK = /^(?:[\u0000-\u001f\ue000-\uf8ff\u25a0-\u25ff\u2750-\u275f<〈《〔\[]|ㅇ\s)/;
+var PAGE_ITEM_HEAD = /^(?:\S{1,4}\t|\.\d{1,3}\s)/;
+var SENTENCE_END = /[.?!。][”"’」』)]?$/;
+var titleLike = (text) => {
+  const t = text.replace(MARKUP, "").trim();
+  return !t.includes("\n") && [...t].length <= 40 && !SENTENCE_END.test(t);
+};
 var PAGE_BREAK_JOINS = /* @__PURE__ */ new WeakMap();
 function joinPageBreakWraps(blocks, lex) {
   const pageRight = /* @__PURE__ */ new Map();
@@ -2178,14 +2185,26 @@ function joinPageBreakWraps(blocks, lex) {
     if (b.type !== "paragraph" || !b.bbox || !b.pageNumber) continue;
     pageRight.set(b.pageNumber, Math.max(pageRight.get(b.pageNumber) ?? -Infinity, b.bbox.x + b.bbox.width));
   }
+  const pageLeft = /* @__PURE__ */ new Map();
+  for (const b of blocks) {
+    if (b.type !== "paragraph" || !b.bbox || !b.pageNumber) continue;
+    pageLeft.set(b.pageNumber, Math.min(pageLeft.get(b.pageNumber) ?? Infinity, b.bbox.x));
+  }
   for (let i = blocks.length - 1; i > 0; i--) {
     const a = blocks[i - 1], b = blocks[i];
     if (!a.pageNumber || b.pageNumber !== a.pageNumber + 1 || !a.text || !b.text) continue;
     if (a.type !== "paragraph" && a.type !== "list" || b.type !== "paragraph") continue;
     const last = a.bbox && PARA_LAST_LINE.get(a.bbox), fs = last ? last.fontSize : 0;
     if (!last || fs <= 0 || (pageRight.get(a.pageNumber) ?? Infinity) - last.right >= BODY_FULL_TOL * fs) continue;
-    if (last.width < BODY_MIN_WIDTH_EM * fs || Math.round(b.style?.fontSize ?? 0) !== Math.round(fs)) continue;
-    if (startsNewItem(a.text, b.text) || PAGE_REF_TAIL.test(a.text) || PAGE_HEAD_MARK.test(b.text.replace(MARKUP, ""))) continue;
+    if (last.width < BODY_MIN_WIDTH_EM * fs || Math.abs((b.style?.fontSize ?? 0) - fs) > 0.15 * fs) continue;
+    const bt = b.text.replace(MARKUP, "");
+    if (startsNewItem(a.text, b.text) || PAGE_REF_TAIL.test(a.text) || PAGE_ITEM_HEAD.test(bt)) continue;
+    if (PAGE_HEAD_MARK.test(bt) && !(/^[\u25a0-\u25ff]/.test(bt) && a.text.includes(bt[0]))) continue;
+    const firstLeft = b.bbox && PARA_FIRST_LEFT.get(b.bbox);
+    const indented = firstLeft === void 0 || firstLeft - (pageLeft.get(b.pageNumber) ?? firstLeft) >= 0.3 * fs;
+    if (indented && SENTENCE_END.test(a.text.replace(MARKUP, "").trimEnd())) continue;
+    if (Math.round(b.style?.fontSize ?? 0) !== Math.round(fs) && titleLike(b.text)) continue;
+    if (/\t/.test(a.text.slice(a.text.lastIndexOf("\n") + 1).replace(/^\S{1,8}\t/, "")) && titleLike(bt.trim().split("\n")[0])) continue;
     const head = a.text, joiner = wrapJoiner(a.text, b.text, lex);
     a.text += joiner + b.text;
     const later = b.bbox && PAGE_BREAK_JOINS.get(b.bbox) || [];
@@ -3553,6 +3572,7 @@ function pushLineParagraphs(out, yLines, pageNum, lex) {
     joins[1] = "\n";
   }
   for (let i = 0; i < lines.length; ) {
+    const first2 = i;
     let text = lines[i].text;
     const items = [...lines[i].items];
     const srcLines = [lines[i].items];
@@ -3567,6 +3587,7 @@ function pushLineParagraphs(out, yLines, pageNum, lex) {
     for (const it of items) faces.set(it.fontName, (faces.get(it.fontName) ?? 0) + it.text.length);
     FACE_CHARS.set(block, faces);
     PARA_LAST_LINE.set(block.bbox, { right: geo[i].right, width: geo[i].right - geo[i].left, fontSize: geo[i].fontSize });
+    PARA_FIRST_LEFT.set(block.bbox, geo[first2].left);
     out.push(block);
     i++;
   }
@@ -8277,4 +8298,4 @@ export {
   removeSideTabs,
   extractPageBlocksWithLines
 };
-//# sourceMappingURL=chunk-CNU4NPCC.js.map
+//# sourceMappingURL=chunk-INESPVEV.js.map