kordoc 3.8.1 → 3.8.2

package/dist/mcp.js

@@ -10,7 +10,7 @@ import {
   parse,
   patchHwp,
   patchHwpx
-} from "./chunk-KT2BCHXI.js";
+} from "./chunk-DP37KF2X.js";
 import {
   detectFormat,
   detectZipFormat
@@ -21,7 +21,7 @@ import {
   blocksToMarkdown,
   sanitizeError,
   toArrayBuffer
-} from "./chunk-IFYJFWD2.js";
+} from "./chunk-JHZUFBUV.js";
 import "./chunk-MOL7MDBG.js";
 
 // src/mcp.ts
@@ -193,7 +193,7 @@ server.tool(
           break;
         case "pdf":
           try {
-            const { extractPdfMetadataOnly } = await import("./parser-FFEBMLSH.js");
+            const { extractPdfMetadataOnly } = await import("./parser-KNQDRLZQ.js");
             metadata = await extractPdfMetadataOnly(buffer);
           } catch {
             metadata = void 0;

package/dist/{parser-FFEBMLSH.js → parser-KNQDRLZQ.js}

@@ -7,7 +7,7 @@ import {
   blocksToMarkdown,
   safeMax,
   safeMin
-} from "./chunk-IFYJFWD2.js";
+} from "./chunk-JHZUFBUV.js";
 import {
   parsePageRange
 } from "./chunk-MOL7MDBG.js";
@@ -21,21 +21,30 @@ function extractLines(fnArray, argsArray) {
   const horizontals = [];
   const verticals = [];
   let lineWidth = 1;
+  let ctm = [1, 0, 0, 1, 0, 0];
+  const ctmStack = [];
+  const applyCtm = (x, y) => [ctm[0] * x + ctm[2] * y + ctm[4], ctm[1] * x + ctm[3] * y + ctm[5]];
+  const ctmScale = () => (Math.hypot(ctm[0], ctm[1]) + Math.hypot(ctm[2], ctm[3])) / 2;
   let currentPath = [];
   let pathStartX = 0, pathStartY = 0;
   let curX = 0, curY = 0;
-  function pushRectangle(path, rx, ry, rw, rh) {
-    if (Math.abs(rh) < ORIENTATION_TOL * 2) {
-      path.push({ x1: rx, y1: ry + rh / 2, x2: rx + rw, y2: ry + rh / 2 });
-    } else if (Math.abs(rw) < ORIENTATION_TOL * 2) {
-      path.push({ x1: rx + rw / 2, y1: ry, x2: rx + rw / 2, y2: ry + rh });
+  function pushSeg(x1, y1, x2, y2) {
+    const [tx1, ty1] = applyCtm(x1, y1);
+    const [tx2, ty2] = applyCtm(x2, y2);
+    currentPath.push({ x1: tx1, y1: ty1, x2: tx2, y2: ty2 });
+  }
+  function pushRectangle(rx, ry, rw, rh) {
+    const effH = Math.abs(rh) * Math.hypot(ctm[2], ctm[3]);
+    const effW = Math.abs(rw) * Math.hypot(ctm[0], ctm[1]);
+    if (effH < ORIENTATION_TOL * 2) {
+      pushSeg(rx, ry + rh / 2, rx + rw, ry + rh / 2);
+    } else if (effW < ORIENTATION_TOL * 2) {
+      pushSeg(rx + rw / 2, ry, rx + rw / 2, ry + rh);
     } else {
-      path.push(
-        { x1: rx, y1: ry, x2: rx + rw, y2: ry },
-        { x1: rx + rw, y1: ry, x2: rx + rw, y2: ry + rh },
-        { x1: rx + rw, y1: ry + rh, x2: rx, y2: ry + rh },
-        { x1: rx, y1: ry + rh, x2: rx, y2: ry }
-      );
+      pushSeg(rx, ry, rx + rw, ry);
+      pushSeg(rx + rw, ry, rx + rw, ry + rh);
+      pushSeg(rx + rw, ry + rh, rx, ry + rh);
+      pushSeg(rx, ry + rh, rx, ry);
     }
   }
   function flushPath(isStroke) {
@@ -43,8 +52,9 @@ function extractLines(fnArray, argsArray) {
       currentPath = [];
       return;
     }
+    const effWidth = lineWidth * ctmScale();
     for (const seg of currentPath) {
-      classifyAndAdd(seg, lineWidth, horizontals, verticals);
+      classifyAndAdd(seg, effWidth, horizontals, verticals);
     }
     currentPath = [];
   }
@@ -55,6 +65,24 @@ function extractLines(fnArray, argsArray) {
       case OPS.setLineWidth:
         lineWidth = args[0] || 1;
         break;
+      case OPS.save:
+        ctmStack.push(ctm.slice());
+        break;
+      case OPS.restore:
+        ctm = ctmStack.pop() ?? [1, 0, 0, 1, 0, 0];
+        break;
+      case OPS.transform: {
+        const t = args;
+        ctm = [
+          ctm[0] * t[0] + ctm[2] * t[1],
+          ctm[1] * t[0] + ctm[3] * t[1],
+          ctm[0] * t[2] + ctm[2] * t[3],
+          ctm[1] * t[2] + ctm[3] * t[3],
+          ctm[0] * t[4] + ctm[2] * t[5] + ctm[4],
+          ctm[1] * t[4] + ctm[3] * t[5] + ctm[5]
+        ];
+        break;
+      }
       case OPS.constructPath: {
         const arg0 = args[0];
         if (Array.isArray(arg0)) {
@@ -69,16 +97,16 @@ function extractLines(fnArray, argsArray) {
               pathStartY = curY;
             } else if (subOp === OPS.lineTo) {
               const x2 = coords[ci++], y2 = coords[ci++];
-              currentPath.push({ x1: curX, y1: curY, x2, y2 });
+              pushSeg(curX, curY, x2, y2);
               curX = x2;
               curY = y2;
             } else if (subOp === OPS.rectangle) {
               const rx = coords[ci++], ry = coords[ci++];
               const rw = coords[ci++], rh = coords[ci++];
-              pushRectangle(currentPath, rx, ry, rw, rh);
+              pushRectangle(rx, ry, rw, rh);
             } else if (subOp === OPS.closePath) {
               if (curX !== pathStartX || curY !== pathStartY) {
-                currentPath.push({ x1: curX, y1: curY, x2: pathStartX, y2: pathStartY });
+                pushSeg(curX, curY, pathStartX, pathStartY);
               }
               curX = pathStartX;
               curY = pathStartY;
@@ -104,7 +132,7 @@ function extractLines(fnArray, argsArray) {
                 pathStartY = curY;
               } else if (drawOp === 1 /* lineTo */) {
                 const x2 = pathData[di++], y2 = pathData[di++];
-                currentPath.push({ x1: curX, y1: curY, x2, y2 });
+                pushSeg(curX, curY, x2, y2);
                 curX = x2;
                 curY = y2;
               } else if (drawOp === 2 /* curveTo */) {
@@ -113,7 +141,7 @@ function extractLines(fnArray, argsArray) {
                 di += 4;
               } else if (drawOp === 4 /* closePath */) {
                 if (curX !== pathStartX || curY !== pathStartY) {
-                  currentPath.push({ x1: curX, y1: curY, x2: pathStartX, y2: pathStartY });
+                  pushSeg(curX, curY, pathStartX, pathStartY);
                 }
                 curX = pathStartX;
                 curY = pathStartY;
@@ -2949,16 +2977,28 @@ g.pdfjsWorker = pdfjsWorker;
 
 // src/pdf/parser.ts
 import { getDocument, GlobalWorkerOptions } from "pdfjs-dist/legacy/build/pdf.mjs";
+import { createRequire } from "module";
+import { dirname, join } from "path";
 GlobalWorkerOptions.workerSrc = "";
 var MAX_PAGES = 5e3;
 var MAX_TOTAL_TEXT = 100 * 1024 * 1024;
 var PDF_LOAD_TIMEOUT_MS = 3e4;
+var pdfjsAssets = {};
+try {
+  const _require = createRequire(import.meta.url);
+  const pkgDir = dirname(_require.resolve("pdfjs-dist/package.json"));
+  pdfjsAssets.cMapUrl = join(pkgDir, "cmaps") + "/";
+  pdfjsAssets.cMapPacked = true;
+  pdfjsAssets.standardFontDataUrl = join(pkgDir, "standard_fonts") + "/";
+} catch {
+}
 async function loadPdfWithTimeout(buffer) {
   const loadingTask = getDocument({
     data: new Uint8Array(buffer),
     useSystemFonts: true,
     disableFontFace: true,
-    isEvalSupported: false
+    isEvalSupported: false,
+    ...pdfjsAssets
   });
   let timer;
   try {
@@ -3171,4 +3211,4 @@ export {
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-FFEBMLSH.js.map
+//# sourceMappingURL=parser-KNQDRLZQ.js.map