kordoc 4.21.20 → 4.21.21

package/dist/index.js

@@ -1,7 +1,7 @@
 import {
   blocksToPages,
   inlineImagesIntoMarkdown
-} from "./chunk-6HIDMA5O.js";
+} from "./chunk-T3DTTK23.js";
 import {
   parsePageRange
 } from "./chunk-V4BH5SRQ.js";
@@ -32,7 +32,7 @@ import {
   tidyScriptTags,
   unframeLayoutTables,
   wrapScript
-} from "./chunk-JDVOYDIA.js";
+} from "./chunk-KTLHADU4.js";
 import {
   KordocError,
   OPTIONAL_DEP_INSTALL_HINT,
@@ -50,7 +50,7 @@ import {
   toArrayBuffer,
   unescapeHtml,
   unzipLimitBytes
-} from "./chunk-F5PISJEI.js";
+} from "./chunk-ZZIQWSGC.js";
 
 // src/plain-markdown.ts
 function toPlainMarkdown(md2) {
@@ -42688,7 +42688,7 @@ function scriptsOff(r, off) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-RNPHQ4C2.js");
+    const { parseImageDocument } = await import("./image-ocr-3C6W3KZO.js");
     const { blocks, warnings } = await parseImageDocument(buffer, options);
     stripScriptTags({ blocks });
     return {
@@ -42748,7 +42748,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-5WGRQKPI.js");
+    const mod = await import("./parser-56I364YM.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {

package/dist/mcp.js

@@ -3,7 +3,7 @@ import {
   assertWithinRoot,
   getAccessRoot,
   isOfflineMode
-} from "./chunk-42XYVZ2B.js";
+} from "./chunk-CY4QM2KQ.js";
 import {
   APPROVAL_MAX,
   BODY_FONTS,
@@ -45,25 +45,25 @@ import {
   resolveBuiltinTemplate,
   unknownFontWarnings,
   usesGaejosikMunche
-} from "./chunk-BLKAJD2G.js";
-import "./chunk-FU3SCGE7.js";
-import "./chunk-KBL24ZCP.js";
+} from "./chunk-YM3K3LCD.js";
+import "./chunk-2ZGP7A5T.js";
+import "./chunk-HWIW3WK2.js";
 import "./chunk-3WSNLH4B.js";
-import "./chunk-G6G25HUO.js";
+import "./chunk-7YAJCXCD.js";
 import {
   detectFormat,
   detectOle2Format,
   detectZipFormat,
   extractHwp5MetadataOnly
-} from "./chunk-N47I6OS7.js";
+} from "./chunk-QHNVH4TC.js";
 import "./chunk-7FCNOKOR.js";
-import "./chunk-ZMGWKGSC.js";
+import "./chunk-PTF4NCFM.js";
 import "./chunk-VQFA55IC.js";
-import "./chunk-PALYDA3M.js";
+import "./chunk-BG6LUCQL.js";
 import {
   blocksToMarkdown
-} from "./chunk-LO5A36FU.js";
-import "./chunk-LQSF4JSV.js";
+} from "./chunk-RBRHO7WL.js";
+import "./chunk-YHPJTLCA.js";
 import {
   KordocError,
   VERSION,
@@ -71,7 +71,7 @@ import {
   routeConsoleToStderr,
   sanitizeError,
   toArrayBuffer
-} from "./chunk-7ITWGPNB.js";
+} from "./chunk-FZMGF6XU.js";
 
 // src/mcp.ts
 import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
@@ -358,7 +358,7 @@ ${markdown}`);
             break;
           case "pdf":
             try {
-              const { extractPdfMetadataOnly } = await import("./parser-2RN5WNPJ.js");
+              const { extractPdfMetadataOnly } = await import("./parser-KFCY3NIN.js");
               metadata = await extractPdfMetadataOnly(buffer);
             } catch {
               metadata = void 0;
@@ -535,7 +535,7 @@ ${tableMarkdown}` }]
         if (!parsed.success) {
           return { content: [{ type: "text", text: `\uD30C\uC2F1 \uC2E4\uD328: ${parsed.error}` }], isError: true };
         }
-        const { blocksToChunks } = await import("./chunks-BJRBKGKZ.js");
+        const { blocksToChunks } = await import("./chunks-MDMJRCJ7.js");
         const chunks = blocksToChunks(parsed.blocks, { granularity, includeTableCells: include_table_cells });
         return { content: [{ type: "text", text: capResponseText(JSON.stringify(chunks, null, 2)) }] };
       } catch (err) {
@@ -805,7 +805,7 @@ ${previewMd}` }]
         }
         const image = new Uint8Array(await readFile2(imgResolved));
         const ext = extname2(imgResolved).slice(1).toLowerCase();
-        const { placeSealHwpx } = await import("./seal-YCK6TM5I.js");
+        const { placeSealHwpx } = await import("./seal-GA7NZDTG.js");
         const result = await placeSealHwpx(buffer, [{
           anchor,
           occurrence,
@@ -922,7 +922,7 @@ ${previewMd}` }]
         if (outPath && await isSameFile(outPath, resolved)) {
           return { content: [{ type: "text", text: "output_path\uAC00 \uC785\uB825 \uD30C\uC77C\uACFC \uAC19\uC2B5\uB2C8\uB2E4 \u2014 \uC6D0\uBCF8\uC744 \uB36E\uC5B4\uC4F0\uC9C0 \uC54A\uC2B5\uB2C8\uB2E4. \uB2E4\uB978 \uACBD\uB85C\uB97C \uC9C0\uC815\uD558\uC138\uC694." }], isError: true };
         }
-        const { redactDocument } = await import("./redact-doc-CJ6W6YNR.js");
+        const { redactDocument } = await import("./redact-doc-5SMPXIGG.js");
         const r = await redactDocument(buffer, { rules, maskChar: mask_char, filePath: file_path, dryRun: dry_run });
         const MAX_LINES = 50;
         const listed = r.fileHits.length > 0 ? r.fileHits.map((h) => `  - [${h.rule}] ${h.masked} (${h.where} @ ${h.part})`) : r.markdownHits.map((h) => `  - [${h.rule}] ${h.masked}`);
@@ -996,7 +996,7 @@ function registerRenderTools(server2) {
         const { buffer, resolved } = await readValidatedFile(file_path, MAX_FILE_SIZE, /* @__PURE__ */ new Set([".hwpx", ".hwp"]));
         const isHwp5 = resolved.toLowerCase().endsWith(".hwp");
         if (!isHwp5 && !pages && (format === "png" || format === "svg")) {
-          const { renderHwpxToSvg } = await import("./render-K5UDHXRI.js");
+          const { renderHwpxToSvg } = await import("./render-FGPG5A2N.js");
           const result = await renderHwpxToSvg(buffer, { highlights, reflow: true, reflowMode: reflow_mode });
           const summary2 = [
             `\uB80C\uB354 \uC644\uB8CC: ${result.pageCount}\uD398\uC774\uC9C0, ${Math.round(result.width)}x${Math.round(result.height)}pt (\uD14D\uC2A4\uD2B8 ${result.stats.texts}\xB7\uC774\uBBF8\uC9C0 ${result.stats.images}\xB7\uD45C ${result.stats.tables})`,
@@ -1008,7 +1008,7 @@ function registerRenderTools(server2) {
             summary2.push(`\uC800\uC7A5: ${outPath}`);
             return { content: [{ type: "text", text: summary2.join("\n") }] };
           }
-          const { rasterizeSvg } = await import("./rasterize-XBCDG4MD.js");
+          const { rasterizeSvg } = await import("./rasterize-SFCG4YOI.js");
           const raster = await rasterizeSvg(result.svg, result.width, result.height, max_width_px ? { maxWidthPx: max_width_px } : void 0);
           if (outPath) {
             await mkdir2(dirname3(outPath), { recursive: true });
@@ -1023,7 +1023,7 @@ function registerRenderTools(server2) {
             ]
           };
         }
-        const { renderDocument } = await import("./render-K5UDHXRI.js");
+        const { renderDocument } = await import("./render-FGPG5A2N.js");
         const { scene, assets } = await renderDocument(buffer, { format, pages, highlights, reflow: true, reflowMode: reflow_mode, maxWidthPx: max_width_px });
         const summary = [
           `\uB80C\uB354 \uC644\uB8CC: \uBB38\uC11C ${scene.pages.length}\uD398\uC774\uC9C0 \uC911 ${assets.filter((a) => a.page !== void 0).length || 1}\uAC74 (\uD14D\uC2A4\uD2B8 ${scene.stats.texts}\xB7\uC774\uBBF8\uC9C0 ${scene.stats.images}\xB7\uD45C ${scene.stats.tables}\xB7\uB3C4\uD615 ${scene.stats.shapes})`,
@@ -1073,7 +1073,7 @@ function registerRenderTools(server2) {
       try {
         const { buffer } = await readValidatedFile(file_path, MAX_FILE_SIZE, /* @__PURE__ */ new Set([".hwpx", ".hwp"]));
         const dir = safeOutputPath(join2(output_dir, "regions.json"), /* @__PURE__ */ new Set([".json"])).replace(/[\\/]regions\.json$/, "");
-        const { extractRenderedRegions } = await import("./render-K5UDHXRI.js");
+        const { extractRenderedRegions } = await import("./render-FGPG5A2N.js");
         const regions = await extractRenderedRegions(buffer, { types: target, format, pages, paddingPt: padding_pt, reflow: true });
         await mkdir2(dir, { recursive: true });
         const ext = format === "jpeg" ? "jpg" : "png";
@@ -1109,7 +1109,7 @@ function registerRenderTools(server2) {
         }
         const { buffer } = await readValidatedFile(file_path, MAX_FILE_SIZE, /* @__PURE__ */ new Set([".hwpx", ".hwp"]));
         const dir = output_dir ? safeOutputPath(join2(output_dir, "tables.json"), /* @__PURE__ */ new Set([".json"])).replace(/[\\/]tables\.json$/, "") : void 0;
-        const { extractTables } = await import("./visual-S5ZNFWFH.js");
+        const { extractTables } = await import("./visual-YMG644YW.js");
         const tables = await extractTables(buffer, { policy: visual, format, paddingPt: padding_pt });
         if (dir) await mkdir2(dir, { recursive: true });
         const ext = format === "jpeg" ? "jpg" : "png";
@@ -1166,7 +1166,7 @@ function registerGenerateTools(server2) {
       try {
         const out = safeOutputPath(output_path, PROFILE_EXTENSIONS);
         const { buffer } = await readValidatedFile(hwpx_path);
-        const { hwpxToProfile } = await import("./-QHQ2QKM7.js");
+        const { hwpxToProfile } = await import("./-ZYF6HX4M.js");
         const profile = await hwpxToProfile(buffer);
         await mkdir3(dirname4(out), { recursive: true });
         await writeOutputFile(out, JSON.stringify(profile, null, 2));
@@ -1278,7 +1278,7 @@ function registerGenerateTools(server2) {
         }
         let profile;
         if (profile_path) {
-          const { parseFormatProfileJson } = await import("./profile-io-UP7H2J6I.js");
+          const { parseFormatProfileJson } = await import("./profile-io-OCITHUQF.js");
           profile = parseFormatProfileJson(await readFile3(safePath(profile_path, PROFILE_EXTENSIONS), "utf-8"));
         }
         const out = safeOutputPath(output_path, /* @__PURE__ */ new Set([".hwpx"]));

package/dist/{model-bundle-XP4AHFLT.js → model-bundle-LAWQH75O.js}

@@ -1,14 +1,14 @@
 #!/usr/bin/env node
 import {
   ALL_OCR_MODELS
-} from "./chunk-WINHSKFD.js";
+} from "./chunk-WGMYVRSA.js";
 import {
   ALL_FORMULA_MODELS,
   getModelStatusIn,
   getModelsDir
-} from "./chunk-TYYKK5DX.js";
-import "./chunk-42XYVZ2B.js";
-import "./chunk-7ITWGPNB.js";
+} from "./chunk-3TCF5HPP.js";
+import "./chunk-CY4QM2KQ.js";
+import "./chunk-FZMGF6XU.js";
 
 // src/shared/model-bundle.ts
 import { copyFile, mkdir } from "fs/promises";
@@ -101,4 +101,4 @@ export {
   importModels,
   modelCacheStatus
 };
-//# sourceMappingURL=model-bundle-XP4AHFLT.js.map
+//# sourceMappingURL=model-bundle-LAWQH75O.js.map

package/dist/{models-AD2KAUNC.js → models-6QTD2CFM.js}

@@ -10,10 +10,10 @@ import {
   ocrModelPath,
   ocrModelsCached,
   parseCharacterDict
-} from "./chunk-WINHSKFD.js";
-import "./chunk-TYYKK5DX.js";
-import "./chunk-42XYVZ2B.js";
-import "./chunk-7ITWGPNB.js";
+} from "./chunk-WGMYVRSA.js";
+import "./chunk-3TCF5HPP.js";
+import "./chunk-CY4QM2KQ.js";
+import "./chunk-FZMGF6XU.js";
 export {
   ALL_OCR_MODELS,
   OCR_DET_MODEL,
@@ -26,4 +26,4 @@ export {
   ocrModelsCached,
   parseCharacterDict
 };
-//# sourceMappingURL=models-AD2KAUNC.js.map
+//# sourceMappingURL=models-6QTD2CFM.js.map

package/dist/{parse-worker-v2-7UHOQ2TK.js → parse-worker-v2-FEPJG3ER.js}

@@ -1,23 +1,23 @@
 #!/usr/bin/env node
 import {
   parseWorkerFile
-} from "./chunk-DBFAQJCX.js";
-import "./chunk-BLKAJD2G.js";
-import "./chunk-FU3SCGE7.js";
-import "./chunk-KBL24ZCP.js";
+} from "./chunk-BC54LBST.js";
+import "./chunk-YM3K3LCD.js";
+import "./chunk-2ZGP7A5T.js";
+import "./chunk-HWIW3WK2.js";
 import "./chunk-3WSNLH4B.js";
-import "./chunk-G6G25HUO.js";
-import "./chunk-N47I6OS7.js";
+import "./chunk-7YAJCXCD.js";
+import "./chunk-QHNVH4TC.js";
 import "./chunk-7FCNOKOR.js";
-import "./chunk-ZMGWKGSC.js";
+import "./chunk-PTF4NCFM.js";
 import "./chunk-VQFA55IC.js";
-import "./chunk-PALYDA3M.js";
-import "./chunk-LO5A36FU.js";
-import "./chunk-LQSF4JSV.js";
+import "./chunk-BG6LUCQL.js";
+import "./chunk-RBRHO7WL.js";
+import "./chunk-YHPJTLCA.js";
 import {
   VERSION,
   sanitizeError
-} from "./chunk-7ITWGPNB.js";
+} from "./chunk-FZMGF6XU.js";
 
 // src/cli/parse-worker-v2.ts
 import { createHash } from "crypto";
@@ -299,4 +299,4 @@ export {
   runParseWorkerV2,
   toParseOptions
 };
-//# sourceMappingURL=parse-worker-v2-7UHOQ2TK.js.map
+//# sourceMappingURL=parse-worker-v2-FEPJG3ER.js.map

package/dist/{parser-5WGRQKPI.js → parser-56I364YM.js}

@@ -1,7 +1,7 @@
 import {
   blocksToPages,
   encodePng
-} from "./chunk-6HIDMA5O.js";
+} from "./chunk-T3DTTK23.js";
 import {
   hasRequestedPagesAfter,
   parsePageRange
@@ -53,11 +53,13 @@ import {
   removeHeaderFooterBlocks,
   removeSideTabs,
   restoreImageBullets,
+  runningLinePatterns,
   splitPageBreakWraps,
   startsNewItem,
+  stripAbsorbedRunningLines,
   trimTrailingEmptyTableCols,
   wrapJoiner
-} from "./chunk-F5KWC7I3.js";
+} from "./chunk-YQZOPT3A.js";
 import {
   CELL_BASELINES,
   CELL_EDGES,
@@ -70,12 +72,12 @@ import {
   escapeLiteralTags,
   stripScriptTags,
   unframeLayoutTables
-} from "./chunk-JDVOYDIA.js";
-import "./chunk-AINZICPE.js";
+} from "./chunk-KTLHADU4.js";
+import "./chunk-FBJVDBMO.js";
 import {
   KordocError,
   sanitizeHref
-} from "./chunk-F5PISJEI.js";
+} from "./chunk-ZZIQWSGC.js";
 
 // src/pdf/image-extract.ts
 import { createHash } from "crypto";
@@ -3212,7 +3214,7 @@ function mergeLinkRuns(markdown) {
 
 // src/pdf/formula-ocr.ts
 async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
-  const formulaMod = await import("./formula-XAPMN6HX.js");
+  const formulaMod = await import("./formula-6XNIXZ5L.js");
   const { FormulaPipeline, ensureFormulaModels } = formulaMod;
   await ensureFormulaModels((p) => {
     if (p.phase === "download" && p.total) {
@@ -3635,7 +3637,7 @@ async function parsePdfDocument(buffer, options) {
       }
       if (targets.size > 0) {
         try {
-          const { runPdfOcr } = await import("./pdf-ocr-KHY3P4EP.js");
+          const { runPdfOcr } = await import("./pdf-ocr-MME37BL4.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);
@@ -3704,9 +3706,15 @@ async function parsePdfDocument(buffer, options) {
     }
     if (options?.removeHeaderFooter !== false && (parsedPageCount >= 3 || headerContext.length)) {
       const removed = removeHeaderFooterBlocks(blocks, pageHeights, warnings, noteMarks, false, headerContext);
+      const running = runningLinePatterns(removed.map((ri) => blocks[ri]), pageHeights);
       for (let ri = removed.length - 1; ri >= 0; ri--) {
         blocks.splice(removed[ri], 1);
       }
+      const stripped = stripAbsorbedRunningLines(blocks, pageHeights, running);
+      if (stripped !== blocks) {
+        blocks.length = 0;
+        blocks.push(...stripped);
+      }
       const kept = removeSideTabs(blocks, pageWidths);
       if (kept !== blocks) {
         blocks.length = 0;
@@ -3875,4 +3883,4 @@ export {
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-5WGRQKPI.js.map
+//# sourceMappingURL=parser-56I364YM.js.map