kordoc 4.23.1 → 4.24.0

package/dist/{chunk-UZZGT6XI.js → chunk-G4HWEM6Q.js}

@@ -10,14 +10,14 @@ import {
   ocrSourcePath,
   parseCharacterDict,
   resolveOcrModels
-} from "./chunk-KRK65GHM.js";
+} from "./chunk-PX6X4YOR.js";
 import {
   validateOcrOptions
-} from "./chunk-QV367L5S.js";
+} from "./chunk-6YDPTTIY.js";
 import {
   KordocError,
   OPTIONAL_DEP_INSTALL_HINT
-} from "./chunk-YPTW52ZJ.js";
+} from "./chunk-XLY5ZJRP.js";
 
 // src/ocr/ruling-lines.ts
 var INK_LUMA_MAX = 205;
@@ -2212,4 +2212,4 @@ export {
   groupOcrLines,
   ocrItemsToBlocks
 };
-//# sourceMappingURL=chunk-UZZGT6XI.js.map
+//# sourceMappingURL=chunk-G4HWEM6Q.js.map

package/dist/{chunk-CU7WLAPQ.js → chunk-GEY6ZMXH.js}

@@ -51,10 +51,10 @@ import {
   resolveParaHeading,
   simulateWrap,
   takeTocLeaders
-} from "./chunk-5BGQOKZS.js";
+} from "./chunk-CKCJU42Y.js";
 import {
   blocksToPages
-} from "./chunk-SSKZ2KL2.js";
+} from "./chunk-EN3PK6OT.js";
 import {
   bmpToPng,
   inlineImagesIntoMarkdown
@@ -76,7 +76,7 @@ import {
   tidyScriptTags,
   unframeLayoutTables,
   wrapScript
-} from "./chunk-DE7ETARS.js";
+} from "./chunk-P6FNIU5P.js";
 import {
   MAX_COLS,
   MAX_ROWS,
@@ -99,7 +99,7 @@ import {
   noteTableHtml,
   sanitizeText,
   visibleText
-} from "./chunk-B4WJQMSZ.js";
+} from "./chunk-DBPDAG7N.js";
 import {
   ANNOTATION_BLANK_RE,
   ValueCursor,
@@ -132,7 +132,7 @@ import {
   resolveUnmatched,
   scanInlineSegments,
   scanSectionXml
-} from "./chunk-H7K2FTV2.js";
+} from "./chunk-Z62PHCSU.js";
 import {
   MAX_DECOMPRESS_SIZE,
   MAX_XML_DEPTH,
@@ -149,7 +149,7 @@ import {
   localName,
   rawTextContent,
   readZipEntry
-} from "./chunk-K6CTZQZW.js";
+} from "./chunk-55DOBEGD.js";
 import {
   KordocError,
   classifyError,
@@ -165,7 +165,7 @@ import {
   toArrayBuffer,
   unescapeHtml,
   unzipLimitBytes
-} from "./chunk-WG43OZYN.js";
+} from "./chunk-Z5G6QNVH.js";
 
 // src/table/visual.ts
 import { readFile as readFile2 } from "fs/promises";
@@ -19329,7 +19329,7 @@ var MIN_SIDE = 24;
 async function ocrEmbeddedImages(blocks, options, warnings) {
   const ocr = options?.ocr;
   if (ocr !== true && ocr !== "force" && typeof ocr !== "function") return false;
-  const { parseImageDocument } = await import("./image-ocr-37S4CHFI.js");
+  const { parseImageDocument } = await import("./image-ocr-CGJJCD4E.js");
   let inserted = false, unreadable = 0;
   let failed = null;
   const walk = async (list, depth) => {
@@ -29897,6 +29897,104 @@ function runLintSuite(text, opts = {}) {
   };
 }
 
+// src/hwpx/lint-blocks.ts
+var MARKDOWN_STRUCTURE = /^\s*(?:```|~~~|#|\||>)/;
+var safeLine = (text) => text.replace(/[\r\n]/g, " ");
+var safeCell = (text) => safeLine(text).replace(/\|/g, "\uFF5C");
+function blocksToLines(blocks) {
+  const lines = [];
+  const paragraph = (block) => {
+    const text = safeLine(block.text);
+    const prefix = MARKDOWN_STRUCTURE.test(text) ? "\\" : "";
+    lines.push({ text: prefix + text, segments: [{ start: prefix.length, length: text.length, block: block.id }] });
+  };
+  const tables = /* @__PURE__ */ new Map();
+  for (const b of blocks) if (b.table) (tables.get(b.table.id) ?? tables.set(b.table.id, []).get(b.table.id)).push(b);
+  const emitted = /* @__PURE__ */ new Set();
+  for (const b of blocks) {
+    if (!b.table) {
+      paragraph(b);
+      continue;
+    }
+    if (emitted.has(b.table.id)) continue;
+    emitted.add(b.table.id);
+    const cells = tables.get(b.table.id);
+    if (cells.length === 1) {
+      paragraph(cells[0]);
+      continue;
+    }
+    const rows = cells.reduce((n, c) => Math.max(n, c.table.row + 1), 0);
+    const cols = cells.reduce((n, c) => Math.max(n, c.table.col + 1), 0);
+    const at = new Map(cells.map((c) => [`${c.table.row},${c.table.col}`, c]));
+    lines.push({ text: "", segments: [] });
+    for (let r = 0; r < rows; r++) {
+      let text = "|";
+      const segments = [];
+      for (let c = 0; c < cols; c++) {
+        text += " ";
+        const cell2 = at.get(`${r},${c}`);
+        if (cell2) {
+          const t = safeCell(cell2.text);
+          segments.push({ start: text.length, length: t.length, block: cell2.id });
+          text += t;
+        }
+        text += " |";
+      }
+      lines.push({ text, segments });
+      if (r === 0) lines.push({ text: "|" + " --- |".repeat(cols), segments: [] });
+    }
+    lines.push({ text: "", segments: [] });
+  }
+  return lines;
+}
+function locate(lines, f) {
+  const line = lines[f.line - 1];
+  if (!line) return { block: null, offset: null };
+  if (f.column !== void 0) {
+    const at = f.column - 1;
+    const seg2 = line.segments.find((s) => at >= s.start && at < s.start + Math.max(1, s.length)) ?? line.segments.find((s) => line.text.indexOf(f.match, s.start) >= s.start && line.text.indexOf(f.match, s.start) < s.start + s.length);
+    if (seg2) {
+      const found = line.text.indexOf(f.match, Math.max(seg2.start, at));
+      const start = found >= seg2.start && found < seg2.start + seg2.length ? found : at;
+      return { block: seg2.block, offset: Math.min(Math.max(0, start - seg2.start), seg2.length) };
+    }
+    return { block: null, offset: null };
+  }
+  if (line.segments.length !== 1) return { block: null, offset: null };
+  const seg = line.segments[0];
+  const body = line.text.slice(seg.start, seg.start + seg.length);
+  const first = f.match ? body.indexOf(f.match) : -1;
+  const unique = first >= 0 && body.indexOf(f.match, first + 1) < 0;
+  return { block: seg.block, offset: unique ? first : null };
+}
+function lintBlocks(blocks, opts = {}) {
+  const lines = blocksToLines(blocks);
+  const suite = runLintSuite(lines.map((l) => l.text).join("\n") + "\n", opts);
+  const findings = [];
+  const take = (category, list) => {
+    for (const f of list ?? []) {
+      const { block, offset } = locate(lines, f);
+      findings.push({
+        block,
+        offset,
+        length: f.match.length,
+        category,
+        rule: f.rule,
+        severity: f.severity,
+        match: f.match,
+        message: f.message,
+        ...f.suggest ? { suggest: f.suggest } : {}
+      });
+    }
+  };
+  take("notation", suite.findings);
+  take("munche", suite.munche);
+  take("slop", suite.slop);
+  take("logic", suite.logic);
+  take("budget", suite.budget);
+  return { findings, ...suite.numbers ? { numbers: suite.numbers } : {}, summary: suite.summary };
+}
+
 // src/roundtrip/patcher.ts
 import JSZip11 from "jszip";
 
@@ -32941,7 +33039,7 @@ function scriptsOff(r, off) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-37S4CHFI.js");
+    const { parseImageDocument } = await import("./image-ocr-CGJJCD4E.js");
     const { blocks, warnings, lines } = await parseImageDocument(buffer, options);
     stripScriptTags({ blocks });
     return {
@@ -33005,7 +33103,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-KHVN5B3I.js");
+    const mod = await import("./parser-IHCFHLHN.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {
@@ -33214,6 +33312,7 @@ export {
   formatGateFinding,
   lintBudgetText,
   runLintSuite,
+  lintBlocks,
   replaceOleStream,
   wipeFreeSectors,
   patchHwp,
@@ -33237,4 +33336,4 @@ export {
   parseHwpml,
   fillForm
 };
-//# sourceMappingURL=chunk-CU7WLAPQ.js.map
+//# sourceMappingURL=chunk-GEY6ZMXH.js.map