kordoc 4.24.21 → 4.24.23

package/dist/{chunk-B3POEH7X.js → chunk-JDDKQD7W.js}

@@ -2,7 +2,7 @@
 import {
   escapeHtml,
   sanitizeHref
-} from "./chunk-WQC4XWVT.js";
+} from "./chunk-WTFXQGND.js";
 
 // src/shared/symbol-fonts.ts
 var WINGDINGS = [
@@ -1183,4 +1183,4 @@ export {
   visibleText,
   escapeHtmlCellText
 };
-//# sourceMappingURL=chunk-B3POEH7X.js.map
+//# sourceMappingURL=chunk-JDDKQD7W.js.map

package/dist/{chunk-IZN5BTCW.js → chunk-NBETHWJH.js}

@@ -2,7 +2,7 @@
 import {
   KordocError,
   unzipLimitBytes
-} from "./chunk-WQC4XWVT.js";
+} from "./chunk-WTFXQGND.js";
 
 // src/hwpx/parser-shared.ts
 import { DOMParser } from "@xmldom/xmldom";
@@ -495,4 +495,4 @@ export {
   hwpFaceToCssStack,
   parseRenderStyles
 };
-//# sourceMappingURL=chunk-IZN5BTCW.js.map
+//# sourceMappingURL=chunk-NBETHWJH.js.map

package/dist/{chunk-4OF7DHIT.js → chunk-NRPLZ4NO.js}

@@ -36,14 +36,14 @@ import {
   splitRowBands,
   starRun,
   tallInkCount
-} from "./chunk-BMXWY36E.js";
+} from "./chunk-XEYTPWCF.js";
 import {
   validateOcrOptions
-} from "./chunk-PDOWSDNM.js";
+} from "./chunk-2UNY7AKC.js";
 import {
   KordocError,
   OPTIONAL_DEP_INSTALL_HINT
-} from "./chunk-UNYH6MTO.js";
+} from "./chunk-XLVQUVL4.js";
 
 // src/ocr/ruling-lines.ts
 var INK_LUMA_MAX = 205;
@@ -2256,4 +2256,4 @@ export {
   groupOcrLines,
   ocrItemsToBlocks
 };
-//# sourceMappingURL=chunk-4OF7DHIT.js.map
+//# sourceMappingURL=chunk-NRPLZ4NO.js.map

package/dist/{chunk-F6OHQLN7.js → chunk-NWNL5YYK.js}

@@ -6,11 +6,11 @@ import {
   HEADING_RATIO_H3,
   tidyScriptTags,
   toTex
-} from "./chunk-F4GZFIUJ.js";
+} from "./chunk-3QWZCQI6.js";
 import {
   safeMax,
   safeMin
-} from "./chunk-WQC4XWVT.js";
+} from "./chunk-WTFXQGND.js";
 
 // src/ocr/line-split.ts
 function grayCrop(rgba, pageW, box) {
@@ -1074,7 +1074,10 @@ function extractLines(fnArray, argsArray) {
       classifyAndAdd(seg, effWidth, horizontals, verticals, fromFill, fromFill ? void 0 : { h: shortH, v: shortV });
     });
     if (from) outs.forEach((o, k) => {
-      for (let n = from[k]; n < o.length; n++) if (white || k < 2 && !thinFill.has(o[n])) nonRules.add(o[n]);
+      for (let n = from[k]; n < o.length; n++) if (white || k < 2 && !thinFill.has(o[n])) {
+        nonRules.add(o[n]);
+        if (white) WHITE_EDGES.add(o[n]);
+      }
     });
     currentPath = [];
   }
@@ -1401,8 +1404,15 @@ function preprocessLines(horizontals, verticals, nonRules) {
   let v = verticals.filter((l) => l.lineWidth <= MAX_LINE_WIDTH);
   h = dropShadingStacks(h, "h");
   v = dropShadingStacks(v, "v");
-  h = mergeParallelLines(h, "h", nonRules);
+  h = mergeParallelLines(h, "h", nonRules, v);
   v = mergeParallelLines(v, "v", nonRules);
+  if (nonRules) {
+    const visH = h.filter((l) => !nonRules.has(l)), visV = v.filter((l) => !nonRules.has(l));
+    const crossH = (w) => visV.filter((c) => c.x1 > w.x1 + 1 && c.x1 < w.x2 - 1 && Math.min(c.y1, c.y2) < w.y1 - 2 && Math.max(c.y1, c.y2) > w.y1 + 2).length;
+    const crossV = (w) => visH.filter((c) => c.y1 > Math.min(w.y1, w.y2) + 1 && c.y1 < Math.max(w.y1, w.y2) - 1 && Math.min(c.x1, c.x2) < w.x1 - 2 && Math.max(c.x1, c.x2) > w.x1 + 2).length;
+    h = h.filter((l) => !(WHITE_EDGES.has(l) && nonRules.has(l) && crossH(l) >= 2));
+    v = v.filter((l) => !(WHITE_EDGES.has(l) && nonRules.has(l) && crossV(l) >= 2));
+  }
   return { horizontals: h, verticals: v };
 }
 var STACK_GAP = 2;
@@ -1570,7 +1580,8 @@ function closeOpenTableEdges(horizontals, verticals) {
   }
   return synthesized.length ? [...verticals, ...synthesized] : verticals;
 }
-function mergeParallelLines(lines, dir, nonRules) {
+var WHITE_EDGES = /* @__PURE__ */ new WeakSet();
+function mergeParallelLines(lines, dir, nonRules, perp) {
   if (lines.length <= 1) return lines;
   const sorted = [...lines].sort((a, b) => {
     const posA = dir === "h" ? a.y1 : a.x1;
@@ -1593,7 +1604,26 @@ function mergeParallelLines(lines, dir, nonRules) {
       const overlap2 = Math.min(prevEnd, currEnd) - Math.max(prevStart, currStart);
       const minLen = Math.min(prevEnd - prevStart, currEnd - currStart);
       if (overlap2 > minLen * 0.3) {
-        if (nonRules?.has(prev) && !nonRules.has(curr)) nonRules.delete(prev);
+        const bgPrev = !!nonRules?.has(prev) && !nonRules.has(curr), bgCurr = !!nonRules?.has(curr) && !nonRules.has(prev);
+        const whPrev = bgPrev && WHITE_EDGES.has(prev), whCurr = bgCurr && WHITE_EDGES.has(curr);
+        if (bgPrev) nonRules.delete(prev);
+        const clip = (w, o) => {
+          if (dir !== "h" || !perp) return false;
+          let n = 0;
+          for (const c of perp) {
+            if (nonRules.has(c) || Math.min(c.y1, c.y2) >= w.y1 - 2 || Math.max(c.y1, c.y2) <= w.y1 + 2) continue;
+            if (c.x1 > w.x1 + 1 && c.x1 < o.x1 - 1 || c.x1 > o.x2 + 1 && c.x1 < w.x2 - 1) n++;
+          }
+          return n >= 2;
+        };
+        if (whPrev && clip(prev, curr)) {
+          prev.x1 = curr.x1;
+          prev.x2 = curr.x2;
+          prev.y1 = prev.y2 = curr.y1;
+          prev.lineWidth = curr.lineWidth;
+          continue;
+        }
+        if (whCurr && clip(curr, prev)) continue;
         if (dir === "h") {
           prev.x1 = Math.min(prev.x1, curr.x1);
           prev.x2 = Math.max(prev.x2, curr.x2);
@@ -8494,6 +8524,7 @@ function headerLineAbove(free, colXs, top) {
   if (baseY - top > fs * MAX_GAP_K) return null;
   const text = [...line].sort((a, b) => a.x - b.x).map((it) => it.text).join(" ").trim();
   if (/^(?:Table|Figure|Fig\.|Chart|Exhibit|Diagram|Source|Note|<?표|<?그림)\s*[\dIVX]/i.test(text)) return null;
+  if (/\(단위[:：]/.test(text.replace(/\s+/g, ""))) return null;
   if (line.some((it) => it.seq === void 0 || it.x < x1 - EDGE_TOL || it.x + it.w > x2 + EDGE_TOL)) return null;
   if (above.some((it) => !line.includes(it) && Math.abs(it.y - baseY) <= fs * 0.5)) return null;
   const cols = colXs.slice(1).map(() => []);
@@ -10361,4 +10392,4 @@ export {
   UNRULED_REGION_TABLES,
   mergeOcrImageRegions
 };
-//# sourceMappingURL=chunk-F6OHQLN7.js.map
+//# sourceMappingURL=chunk-NWNL5YYK.js.map