kordoc 4.0.4 → 4.0.6

package/dist/{chunk-MLNHZN32.js → chunk-HC2SEH6I.js}

@@ -11,13 +11,13 @@ import {
   dedupeRunningHeaders,
   flattenLayoutTables,
   mapPuaText
-} from "./chunk-ZJXR4PWY.js";
+} from "./chunk-UVSOSFER.js";
 import {
   detectFormat,
   detectOle2Format,
   detectZipFormat,
   parseLenientCfb
-} from "./chunk-MEPHGCPQ.js";
+} from "./chunk-5CJGKKMZ.js";
 import {
   parsePageRange
 } from "./chunk-MOL7MDBG.js";
@@ -32,14 +32,14 @@ import {
   paraTextPureT,
   patchZipEntries,
   scanSectionXml
-} from "./chunk-TXNRIVWA.js";
+} from "./chunk-63ASYC4N.js";
 import {
   SPACE_EM_FIXED,
   charWidthEm1000,
   fitRatioForFewerLines,
   measureTextWidth,
   simulateWrap
-} from "./chunk-S3KGEEBQ.js";
+} from "./chunk-5SOZXPQY.js";
 import {
   MAX_DECOMPRESS_SIZE,
   MAX_XML_DEPTH,
@@ -50,7 +50,7 @@ import {
   createXmlParser,
   extractTextFromNode,
   findChildByLocalName
-} from "./chunk-DKMKJMR6.js";
+} from "./chunk-4C7SBZNW.js";
 import {
   KordocError,
   classifyError,
@@ -61,7 +61,7 @@ import {
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-KA5AN73R.js";
+} from "./chunk-L3I4TIRP.js";
 
 // src/index.ts
 import { readFile } from "fs/promises";
@@ -1557,8 +1557,9 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
             if (footnote) text += ` (\uC8FC: ${footnote})`;
             cell2.text += (cell2.text ? "\n" : "") + text;
             const cellBlock = { type: "paragraph", text, pageNumber: ctx.sectionNum };
-            if (ctx.shared.kordocLayout === "default" && !ph?.prefix) {
-              const spans = extractRunSpans(el, ctx);
+            const cellSpanMode = spanModeOf(ctx.shared.kordocLayout);
+            if (cellSpanMode && !ph?.prefix) {
+              const spans = extractRunSpans(el, ctx, cellSpanMode, cellSpanMode !== "kordoc");
               if (spans && spans.map((s) => s.text).join("").replace(/[ \t]+/g, " ").trim() === text) {
                 cellBlock.spans = spans;
               }
@@ -1581,14 +1582,19 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
               const eff = ind.left + Math.max(0, ind.intent);
               if (eff > 0) block.indent = eff;
             }
-            if (!headingLevel && ctx.shared.kordocLayout === "default") {
+            if (ctx.shared.kordocLayout === "gongmun" && block.indent && /^(?:-|\d{1,3}\)) /.test(text)) {
+              const depth2 = gongmunDepthFromIndent(el, block.indent, ctx);
+              if (depth2) block.listDepth = depth2;
+            }
+            const spanMode = spanModeOf(ctx.shared.kordocLayout);
+            if (!headingLevel && spanMode) {
               if (!ph?.prefix) {
-                const spans = extractRunSpans(el, ctx);
+                const spans = extractRunSpans(el, ctx, spanMode, spanMode === "gongmun");
                 if (spans && spans.map((s) => s.text).join("").replace(/[ \t]+/g, " ").trim() === text) {
                   block.spans = spans;
                 }
               }
-              if (el.getAttribute("paraPrIDRef") === KORDOC_PARA_QUOTE) block.quote = true;
+              if (spanMode !== "foreign" && el.getAttribute("paraPrIDRef") === KORDOC_PARA_QUOTE) block.quote = true;
             }
             blocks.push(block);
           } else {
@@ -1711,8 +1717,15 @@ function collectSubListText(el, ctx, depth = 0) {
     if (tag === "p" || tag === "para") {
       const t = extractParagraphInfo(ch, ctx.styleMap, ctx).text;
       if (t) parts.push(t);
+      const tbls = [];
+      findTopLevelTbls(ch, tbls);
+      for (const tbl2 of tbls) {
+        const flat = flattenSubListTable(tbl2, ctx, depth);
+        if (flat) parts.push(flat);
+      }
     } else if (tag === "tbl") {
-      continue;
+      const flat = flattenSubListTable(ch, ctx, depth);
+      if (flat) parts.push(flat);
     } else {
       const t = collectSubListText(ch, ctx, depth + 1);
       if (t) parts.push(t);
@@ -1720,6 +1733,29 @@ function collectSubListText(el, ctx, depth = 0) {
   }
   return parts.join("\n").trim();
 }
+function flattenSubListTable(el, ctx, depth) {
+  const sink = [];
+  const st = { rows: [], currentRow: [], cell: null };
+  walkSection(el, sink, st, [], ctx, depth + 1);
+  let flat = convertTableToText(st.rows);
+  if (st.caption) flat = st.caption + (flat ? "\n" + flat : "");
+  return flat;
+}
+function findTopLevelTbls(el, out, depth = 0) {
+  if (depth > MAX_XML_DEPTH) return;
+  const kids2 = el.childNodes;
+  if (!kids2) return;
+  for (let i = 0; i < kids2.length; i++) {
+    const ch = kids2[i];
+    if (ch.nodeType !== 1) continue;
+    const tag = (ch.tagName || ch.localName || "").replace(/^[^:]+:/, "");
+    if (tag === "tbl") {
+      out.push(ch);
+      continue;
+    }
+    findTopLevelTbls(ch, out, depth + 1);
+  }
+}
 function walkParagraphChildren(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
   if (depth > MAX_XML_DEPTH) return tableCtx;
   const children = node.childNodes;
@@ -2078,7 +2114,13 @@ function extractParagraphInfo(para2, styleMap, ctx) {
 }
 var KORDOC_CHAR_CODE = "4";
 var KORDOC_PARA_QUOTE = "6";
-function extractRunSpans(para2, ctx) {
+function spanModeOf(layout) {
+  if (layout === "default") return "kordoc";
+  if (layout === "gongmun") return "gongmun";
+  if (!layout) return "foreign";
+  return null;
+}
+function extractRunSpans(para2, ctx, mode, requireMixed) {
   const styleMap = ctx.styleMap;
   if (!styleMap) return null;
   const spans = [];
@@ -2115,7 +2157,7 @@ function extractRunSpans(para2, ctx) {
     const prId = child.getAttribute("charPrIDRef") ?? "";
     const cp = styleMap.charProperties.get(prId);
     const span = { text };
-    if (prId === KORDOC_CHAR_CODE) span.code = true;
+    if (mode !== "foreign" && prId === KORDOC_CHAR_CODE) span.code = true;
     else {
       if (cp?.bold) span.bold = true;
       if (cp?.italic) span.italic = true;
@@ -2123,7 +2165,38 @@ function extractRunSpans(para2, ctx) {
     if (span.bold || span.italic || span.code) styled = true;
     spans.push(span);
   }
-  return styled && spans.length > 0 ? spans : null;
+  if (!styled || spans.length === 0) return null;
+  const merged = [];
+  for (const s of spans) {
+    const last = merged[merged.length - 1];
+    if (last && !!last.bold === !!s.bold && !!last.italic === !!s.italic && !!last.code === !!s.code) {
+      last.text += s.text;
+    } else {
+      merged.push(s);
+    }
+  }
+  if (requireMixed && !merged.some((s) => !(s.bold || s.italic || s.code))) return null;
+  return merged;
+}
+function gongmunDepthFromIndent(para2, left, ctx) {
+  const styleMap = ctx.styleMap;
+  if (!styleMap) return null;
+  let unit = 0;
+  const kids2 = para2.childNodes;
+  for (let i = 0; i < (kids2?.length ?? 0); i++) {
+    const child = kids2[i];
+    if (child.nodeType !== 1) continue;
+    const tag = (child.tagName || child.localName || "").replace(/^[^:]+:/, "");
+    if (tag !== "run" && tag !== "r") continue;
+    const size = styleMap.charProperties.get(child.getAttribute("charPrIDRef") ?? "")?.fontSize;
+    if (size) unit = size * 100;
+    break;
+  }
+  if (!unit) return null;
+  const r = left / unit;
+  const nearInt = Math.round(r);
+  const depth = Math.abs(r - nearInt) < 0.2 ? nearInt : Math.round(2 * r - 1);
+  return depth >= 1 && depth < 8 ? depth : null;
 }
 
 // src/hwpx/images.ts
@@ -27030,7 +27103,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-46JGSTPR.js");
+    const mod = await import("./parser-C636QIOB.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {
@@ -27163,4 +27236,4 @@ export {
   parseHwpml,
   fillForm
 };
-//# sourceMappingURL=chunk-MLNHZN32.js.map
+//# sourceMappingURL=chunk-HC2SEH6I.js.map