kordoc 4.16.2 → 4.17.0

package/dist/{chunk-QOCS7W7R.js → chunk-K2VYABZ3.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   blocksToPages
-} from "./chunk-AUWGDP4G.js";
+} from "./chunk-2KF5WZCR.js";
 import {
   allLinesegRemovalSplices,
   applySplices,
@@ -15,7 +15,7 @@ import {
   paraTextPureT,
   patchZipEntries,
   scanSectionXml
-} from "./chunk-4VODJKFL.js";
+} from "./chunk-6RGGYQQC.js";
 import {
   CONVERT_MAP,
   FLAG_COMPRESSED,
@@ -61,16 +61,18 @@ import {
   renderDocumentToScene,
   romanNumeral,
   simulateWrap
-} from "./chunk-IQGIT2VX.js";
+} from "./chunk-QJWVMPB6.js";
 import {
+  CELL_EDGES,
   HEADING_RATIO_H1,
   HEADING_RATIO_H2,
   HEADING_RATIO_H3,
   plainScripts,
   stripScriptTags,
   tidyScriptTags,
+  unframeLayoutTables,
   wrapScript
-} from "./chunk-J3IF6KGW.js";
+} from "./chunk-5CDUL6PH.js";
 import {
   parsePageRange
 } from "./chunk-VQFA55IC.js";
@@ -89,7 +91,7 @@ import {
   findChildByLocalName,
   localName,
   rawTextContent
-} from "./chunk-JFLAM3MM.js";
+} from "./chunk-I5DJOP62.js";
 import {
   MAX_COLS,
   MAX_ROWS,
@@ -104,7 +106,7 @@ import {
   hasStructuredCellContent,
   mapPuaText,
   noteSuffix
-} from "./chunk-LEIAEWZ7.js";
+} from "./chunk-BFT5IM2R.js";
 import {
   KordocError,
   classifyError,
@@ -119,7 +121,7 @@ import {
   toArrayBuffer,
   unescapeHtml,
   unzipLimitBytes
-} from "./chunk-P6ZDXTV6.js";
+} from "./chunk-ZF6KHPB7.js";
 
 // src/table/visual.ts
 import { readFile as readFile2 } from "fs/promises";
@@ -420,7 +422,8 @@ async function extractHwpxStyles(zip, decompressed) {
     numberings: /* @__PURE__ */ new Map(),
     bullets: /* @__PURE__ */ new Map(),
     paraHeadings: /* @__PURE__ */ new Map(),
-    paraIndents: /* @__PURE__ */ new Map()
+    paraIndents: /* @__PURE__ */ new Map(),
+    borderEdges: /* @__PURE__ */ new Map()
   };
   const headerPaths = ["Contents/header.xml", "header.xml", "Contents/head.xml", "head.xml"];
   for (const hp of headerPaths) {
@@ -443,6 +446,7 @@ async function extractHwpxStyles(zip, decompressed) {
       parseBullets(domDoc, result.bullets);
       parseParaHeadings(domDoc, result.paraHeadings);
       parseParaIndents(domDoc, result.paraIndents);
+      parseBorderEdges(domDoc, result.borderEdges);
       break;
     } catch {
       continue;
@@ -580,6 +584,26 @@ function parseParaIndents(doc, map) {
     if (map.size > 0) break;
   }
 }
+function parseBorderEdges(doc, map) {
+  const tagNames = ["hh:borderFill", "borderFill"];
+  for (const tagName of tagNames) {
+    const elements = doc.getElementsByTagName(tagName);
+    for (let i = 0; i < elements.length; i++) {
+      const el = elements[i];
+      const id = el.getAttribute("id") || "";
+      if (!id) continue;
+      const seen = (name) => {
+        const b = findChildByLocalName(el, name);
+        if (!b) return false;
+        const type = b.getAttribute("type") || "NONE";
+        const color = (b.getAttribute("color") || "#000000").toUpperCase();
+        return type !== "NONE" && color !== "#FFFFFF";
+      };
+      map.set(id, { t: seen("topBorder"), b: seen("bottomBorder"), l: seen("leftBorder"), r: seen("rightBorder") });
+    }
+    if (map.size > 0) break;
+  }
+}
 function parseParaHeadings(doc, map) {
   const tagNames = ["hh:paraPr", "paraPr"];
   for (const tagName of tagNames) {
@@ -744,6 +768,13 @@ function buildTableWithCellMeta(state, keepAnchoredEmptyCols, keepEmptyParagraph
   }
   const srcCount = state.rows.reduce((s, r) => s + r.length, 0);
   const ordinalReliable = anchors2.length === srcCount;
+  for (const row of state.rows) {
+    for (const src of row) {
+      if (!src.edges || src.rowAddr === void 0 || src.colAddr === void 0) continue;
+      const cand = table2.cells[src.rowAddr]?.[src.colAddr];
+      if (cand && anchors2.includes(cand)) CELL_EDGES.set(cand, src.edges);
+    }
+  }
   const claimed = /* @__PURE__ */ new Set();
   let flatIdx = -1;
   for (const row of state.rows) {
@@ -1186,6 +1217,8 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
         if (tableCtx) {
           tableCtx.cell = { text: "", colSpan: 1, rowSpan: 1 };
           if (el.getAttribute("header") === "1" || el.getAttribute("header") === "true") tableCtx.cell.isHeader = true;
+          const edges = ctx.styleMap?.borderEdges.get(el.getAttribute("borderFillIDRef") ?? "");
+          if (edges) tableCtx.cell.edges = edges;
           walkSection(el, blocks, tableCtx, tableStack, ctx, depth + 1);
           if (tableCtx.cell) {
             tableCtx.currentRow.push(tableCtx.cell);
@@ -1862,15 +1895,13 @@ function extractParagraphInfo(para2, styleMap, ctx) {
           walk(child, depth + 1);
           break;
         // 자식 순회 (tab 등 하위 요소 처리)
-        case "tab": {
-          const leader = child.getAttribute("leader");
-          if (leader && leader !== "0") {
-            text += "";
-          } else {
-            text += "	";
-          }
+        // 탭 — 채움(leader≠0, 목차 점선) 탭도 보통 탭. 채움선은 글이 아니라 내지 않고 뒤 글은 남긴다(PDF tab-leaders 가
+        // 점 채움을 탭 하나로 바꾸고 쪽 번호를 남기는 것과 같다). 종전엔 채움 탭 뒤를 쪽 번호로 보고 잘라, 목차 쪽 번호만이 아니라
+        // 일정표 "사용자 의견조사 ····· '26년 8~9월"(gate-fill 36646162)의 일정, 줄바꿈 뒤 다음 목차 항목
+        // "<참고2> 직종별사업체노동력조사 개요"(rhwp issue6044)까지 사라졌다
+        case "tab":
+          text += "	";
           break;
-        }
         case "br":
           if ((child.getAttribute("type") || "line") === "line") text += "\n";
           break;
@@ -1992,7 +2023,7 @@ function extractParagraphInfo(para2, styleMap, ctx) {
           walk(child, depth + 1);
           if (script && linkRanges.length === nLinks && openFields.length === nOpen && runHasOnlyText(child)) {
             const added = text.slice(start);
-            if (added && !/[\x1E\x1F\n$]/.test(added)) text = text.slice(0, start) + wrapScript(added, script);
+            if (added && !/[\x1E\n$]/.test(added)) text = text.slice(0, start) + wrapScript(added, script);
           }
           break;
         }
@@ -2009,14 +2040,12 @@ function extractParagraphInfo(para2, styleMap, ctx) {
     for (const r of closed) {
       if (applied.some(([s, e]) => r.start < e && r.end > s)) continue;
       const anchor = text.slice(r.start, r.end);
-      if (!anchor.trim() || /[\n\x1F\x1E\[\]]/.test(anchor)) continue;
+      if (!anchor.trim() || /[\n\x1E\[\]]/.test(anchor)) continue;
       text = text.slice(0, r.start) + `[${anchor}](${r.url})` + text.slice(r.end);
       applied.push([r.start, r.end]);
     }
     if (applied.length) href = void 0;
   }
-  const leaderIdx = text.indexOf("");
-  if (leaderIdx >= 0) text = text.substring(0, leaderIdx);
   text = tidyScriptTags(text);
   const cleanParaText = (raw) => {
     let t = raw.replace(/[ \t]+/g, " ").trim();
@@ -2529,10 +2558,11 @@ async function parseHwpxDocument(buffer, options) {
   const blocks = allBlocks;
   applyPageText(blocks, shared);
   const images = await extractImagesFromZip(zip, blocks, decompressed, warnings, !options?.pages);
-  detectHwpxHeadings(blocks, styleMap);
-  const outline = blocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
-  const markdown = blocksToMarkdown(blocks);
-  return { markdown, blocks, metadata, outline: outline.length > 0 ? outline : void 0, warnings: warnings.length > 0 ? warnings : void 0, images: images.length > 0 ? images : void 0 };
+  const shown = options?.layoutTables === "keep" ? blocks : unframeLayoutTables(blocks, !!options?.keepTrailingEmptyCols);
+  detectHwpxHeadings(shown, styleMap);
+  const outline = shown.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
+  const markdown = blocksToMarkdown(shown);
+  return { markdown, blocks: shown, metadata, outline: outline.length > 0 ? outline : void 0, warnings: warnings.length > 0 ? warnings : void 0, images: images.length > 0 ? images : void 0 };
 }
 
 // src/hwp3/parser.ts
@@ -15104,7 +15134,6 @@ var MAX_DECOMPRESS_SIZE2 = 100 * 1024 * 1024;
 var PIC_TYPE_DRAWING = 3;
 var OBJ_EQUATION = 2;
 var OPT_HYPERTEXT = 16;
-var LEADER_TAB_MARK = "";
 var OUTLINE_FORMATS = [
   { fmt: "romanUpper", before: "", after: "." },
   { fmt: "digit", before: "", after: ")" },
@@ -15244,8 +15273,6 @@ function parseParagraphList(reader, ctx, sink) {
 }
 function emitHwp3Paragraph(para2) {
   let text = para2.text;
-  const leaderAt = text.indexOf(LEADER_TAB_MARK);
-  if (leaderAt >= 0) text = text.slice(0, leaderAt);
   let headMarker = null;
   if (para2.outline) {
     const { level, numbers } = para2.outline;
@@ -15273,11 +15300,9 @@ function parseCharStream(reader, charCount, ctx, para2) {
       continue;
     }
     if (ch === 9) {
-      reader.skip(2);
-      const leader = reader.readU16();
-      reader.skip(2);
+      reader.skip(6);
       i += 3;
-      para2.text += leader ? LEADER_TAB_MARK : "	";
+      para2.text += "	";
       continue;
     }
     if (ch === 18) {
@@ -17825,8 +17850,10 @@ function collectCharText(node, parts, depth = 0) {
     if (el.nodeType !== 1) continue;
     const tag = localName(el);
     if (tag === "CHAR") {
-      const t = rawTextContent(el);
-      if (t) parts.push(t);
+      for (const c of Array.from(el.childNodes)) {
+        const t = c.nodeType === 1 ? localName(c) === "TAB" ? "	" : rawTextContent(c) : c.nodeType === 3 || c.nodeType === 4 ? c.nodeValue ?? "" : "";
+        if (t) parts.push(t);
+      }
     } else if (tag === "TABLE" || tag === "PICTURE" || tag === "SHAPEOBJECT") {
     } else if (tag === "AUTONUM") {
     } else {
@@ -22428,6 +22455,18 @@ function removeNestedTables(html) {
   if (depth === 0) result += html.slice(last);
   return result;
 }
+var LAYOUT_MODE_MISMATCH = '\uD3B8\uC9D1 \uB9C8\uD06C\uB2E4\uC6B4\uC774 \uAE30\uBCF8(\uBCF4\uC774\uB294 \uB300\uB85C) \uCD9C\uB825\uC5D0\uC11C \uB098\uC654\uC2B5\uB2C8\uB2E4. \uC774 \uBB38\uC11C\uC5D0\uB294 \uD14C\uB450\uB9AC\uAC00 \uC548 \uBCF4\uC774\uB294 \uD2C0 \uD45C\uAC00 \uC788\uC5B4 \uADF8 \uCD9C\uB825\uC73C\uB85C\uB294 \uC6D0\uBCF8 \uC704\uCE58\uB97C \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4 \u2014 `kordoc <\uC6D0\uBCF8> --keep-layout-tables`(API: parse(\uC6D0\uBCF8, { layoutTables: "keep" }))\uB85C \uBF51\uC740 \uB9C8\uD06C\uB2E4\uC6B4\uC744 \uACE0\uCCD0 \uB2E4\uC2DC \uD328\uCE58\uD558\uC138\uC694';
+function editedFromVisual(keepMarkdown, visualMarkdown, editedMarkdown) {
+  if (keepMarkdown === visualMarkdown) return false;
+  const keep = new Set(splitMarkdownUnits(keepMarkdown).map((u) => u.raw));
+  const visual = new Set(splitMarkdownUnits(visualMarkdown).map((u) => u.raw));
+  let k = 0, v = 0;
+  for (const u of splitMarkdownUnits(editedMarkdown)) {
+    if (keep.has(u.raw)) k++;
+    if (visual.has(u.raw)) v++;
+  }
+  return v > k;
+}
 
 // src/hwpx/gen-table-bf.ts
 var THIN = ["0.12 mm", "#000000"];
@@ -27885,7 +27924,9 @@ async function patchHwp(original, editedMarkdown, options) {
   const compressed = (flags & FLAG_COMPRESSED) !== 0;
   let origBlocks;
   try {
-    origBlocks = parseHwp5Document(originalBuf).blocks;
+    const kept = parseHwp5Document(originalBuf, { layoutTables: "keep" });
+    origBlocks = kept.blocks;
+    if (editedFromVisual(kept.markdown, parseHwp5Document(originalBuf).markdown, editedMarkdown)) return fail(LAYOUT_MODE_MISMATCH);
   } catch (err) {
     return fail(`\uC6D0\uBCF8 HWP \uD30C\uC2F1 \uC2E4\uD328: ${msg(err)}`);
   }
@@ -27947,7 +27988,7 @@ async function patchHwp(original, editedMarkdown, options) {
   let verification;
   if (options?.verify !== false) {
     try {
-      const reparsed = parseHwp5Document(Buffer.from(data));
+      const reparsed = parseHwp5Document(Buffer.from(data), { layoutTables: "keep" });
       const normBr = (u) => ({ ...u, raw: u.raw.replace(/<br\s*\/?\s*>/gi, "\n") });
       verification = diffUnitLists(splitMarkdownUnits(reparsed.markdown).map(normBr), editedUnits.map(normBr));
     } catch (err) {
@@ -28546,8 +28587,10 @@ async function patchHwpx(original, editedMarkdown, options) {
   }
   let origBlocks;
   try {
-    const parsed = await parseHwpxDocument(u8ToArrayBuffer(original));
+    const parsed = await parseHwpxDocument(u8ToArrayBuffer(original), { layoutTables: "keep" });
     origBlocks = parsed.blocks;
+    const visual = await parseHwpxDocument(u8ToArrayBuffer(original));
+    if (editedFromVisual(parsed.markdown, visual.markdown, editedMarkdown)) return { success: false, applied: 0, skipped, error: LAYOUT_MODE_MISMATCH };
   } catch (err) {
     return { success: false, applied: 0, skipped, error: `\uC6D0\uBCF8 HWPX \uD30C\uC2F1 \uC2E4\uD328: ${err instanceof Error ? err.message : String(err)}` };
   }
@@ -28631,7 +28674,7 @@ async function patchHwpx(original, editedMarkdown, options) {
   let verification;
   if (options?.verify !== false) {
     try {
-      const reparsed = await parseHwpxDocument(u8ToArrayBuffer(data));
+      const reparsed = await parseHwpxDocument(u8ToArrayBuffer(data), { layoutTables: "keep" });
       verification = diffUnitLists(splitMarkdownUnits(reparsed.markdown), editedUnits);
     } catch (err) {
       return { success: false, applied, skipped, error: `\uD328\uCE58\uBCF8 \uC7AC\uD30C\uC2F1 \uC2E4\uD328 \u2014 \uD328\uCE58 \uC911\uB2E8: ${err instanceof Error ? err.message : String(err)}` };
@@ -29088,7 +29131,7 @@ import JSZip11 from "jszip";
 async function buildState(bytes) {
   const format2 = detectFormat(new Uint8Array(bytes.subarray(0, 512)).buffer);
   if (format2 !== "hwpx") throw new Error(`HWPX \uC138\uC158\uC740 HWPX \uBB38\uC11C\uB9CC \uC5FD\uB2C8\uB2E4 (\uAC10\uC9C0\uB41C \uD3EC\uB9F7: ${format2}) \u2014 HWP 5.x \uB294 patchHwp \uB85C \uD328\uCE58\uD558\uC138\uC694`);
-  const parsed = await parseHwpxDocument(u8ToArrayBuffer2(bytes));
+  const parsed = await parseHwpxDocument(u8ToArrayBuffer2(bytes), { layoutTables: "keep" });
   const zip = await JSZip11.loadAsync(bytes);
   const sectionPaths = await resolveSectionEntryNames(zip);
   if (sectionPaths.length === 0) {
@@ -29519,7 +29562,7 @@ function scriptsOff(r, off) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-3WJ4QH3T.js");
+    const { parseImageDocument } = await import("./image-ocr-P6ECJCT2.js");
     const { blocks, warnings } = await parseImageDocument(buffer, options);
     stripScriptTags({ blocks });
     return {
@@ -29579,7 +29622,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-N6476XAC.js");
+    const mod = await import("./parser-ZPRQ3SSU.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {
@@ -29656,7 +29699,7 @@ async function fillForm(input, values, outputFormat = "markdown") {
       fill: { filled: hwpxResult.filled, unmatched: hwpxResult.unmatched }
     };
   }
-  const parsed = await parse(buffer);
+  const parsed = await parse(buffer, { layoutTables: "keep" });
   if (!parsed.success) {
     throw new Error(`\uC11C\uC2DD \uD30C\uC2F1 \uC2E4\uD328: ${parsed.error}`);
   }
@@ -29679,7 +29722,7 @@ function wantsCrop(kind, policy) {
 async function extractTables(input, options = {}) {
   const policy = options.policy ?? "non-tabular-and-uncertain";
   const buffer = typeof input === "string" ? toArrayBuffer(await readFile2(input)) : Buffer.isBuffer(input) ? toArrayBuffer(input) : input;
-  const parsed = await parse(buffer, { classifyTables: true });
+  const parsed = await parse(buffer, { classifyTables: true, layoutTables: "keep" });
   if (!parsed.success) throw new KordocError(`\uD30C\uC2F1 \uC2E4\uD328: ${parsed.error}`);
   const blocks = collectTableBlocks(parsed.blocks);
   const out = blocks.map((b, i) => ({
@@ -29791,4 +29834,4 @@ export {
   parseHwpml,
   fillForm
 };
-//# sourceMappingURL=chunk-QOCS7W7R.js.map
+//# sourceMappingURL=chunk-K2VYABZ3.js.map