kordoc 3.11.0 → 3.13.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +13 -0
- package/dist/{-QMW2GDWP.js → -MRPDPE2X.js} +5 -5
- package/dist/{chunk-GETZFABK.js → chunk-33CIKNL4.js} +2 -2
- package/dist/{chunk-RCSDSN5Y.cjs → chunk-MEXUFQVC.cjs} +2 -2
- package/dist/{chunk-RCSDSN5Y.cjs.map → chunk-MEXUFQVC.cjs.map} +1 -1
- package/dist/{chunk-D4JKKSSI.js → chunk-O4WFPOFN.js} +2 -2
- package/dist/{chunk-SNSAZ7FA.js → chunk-Q6UXVJXI.js} +2 -2
- package/dist/{chunk-TUBD3EEF.js → chunk-TWACKS3K.js} +21 -5
- package/dist/{chunk-TUBD3EEF.js.map → chunk-TWACKS3K.js.map} +1 -1
- package/dist/{chunk-O27B5IRL.js → chunk-YXOT7P5B.js} +2 -2
- package/dist/cli.js +7 -7
- package/dist/index.cjs +134 -118
- package/dist/index.cjs.map +1 -1
- package/dist/index.js +18 -2
- package/dist/index.js.map +1 -1
- package/dist/mcp.js +5 -5
- package/dist/{parser-X34HD2KQ.cjs → parser-DZBI3EA2.cjs} +101 -15
- package/dist/parser-DZBI3EA2.cjs.map +1 -0
- package/dist/{parser-3EARWV2H.js → parser-XUNGOH76.js} +90 -4
- package/dist/parser-XUNGOH76.js.map +1 -0
- package/dist/{parser-STGI5I5K.js → parser-Z3IL2ESY.js} +89 -3
- package/dist/parser-Z3IL2ESY.js.map +1 -0
- package/dist/render-VZ4OGHZU.js +9 -0
- package/dist/{watch-KVYJ46S6.js → watch-G5NFIT5L.js} +5 -5
- package/package.json +1 -1
- package/dist/parser-3EARWV2H.js.map +0 -1
- package/dist/parser-STGI5I5K.js.map +0 -1
- package/dist/parser-X34HD2KQ.cjs.map +0 -1
- package/dist/render-FPWZ6YNZ.js +0 -9
- /package/dist/{-QMW2GDWP.js.map → -MRPDPE2X.js.map} +0 -0
- /package/dist/{chunk-GETZFABK.js.map → chunk-33CIKNL4.js.map} +0 -0
- /package/dist/{chunk-D4JKKSSI.js.map → chunk-O4WFPOFN.js.map} +0 -0
- /package/dist/{chunk-SNSAZ7FA.js.map → chunk-Q6UXVJXI.js.map} +0 -0
- /package/dist/{chunk-O27B5IRL.js.map → chunk-YXOT7P5B.js.map} +0 -0
- /package/dist/{render-FPWZ6YNZ.js.map → render-VZ4OGHZU.js.map} +0 -0
- /package/dist/{watch-KVYJ46S6.js.map → watch-G5NFIT5L.js.map} +0 -0
package/dist/mcp.js
CHANGED
|
@@ -10,22 +10,22 @@ import {
|
|
|
10
10
|
parse,
|
|
11
11
|
patchHwp,
|
|
12
12
|
patchHwpx
|
|
13
|
-
} from "./chunk-
|
|
13
|
+
} from "./chunk-TWACKS3K.js";
|
|
14
14
|
import {
|
|
15
15
|
detectFormat,
|
|
16
16
|
detectZipFormat
|
|
17
17
|
} from "./chunk-MEPHGCPQ.js";
|
|
18
|
-
import "./chunk-
|
|
18
|
+
import "./chunk-33CIKNL4.js";
|
|
19
19
|
import {
|
|
20
20
|
blocksToMarkdown
|
|
21
|
-
} from "./chunk-
|
|
21
|
+
} from "./chunk-YXOT7P5B.js";
|
|
22
22
|
import "./chunk-MOL7MDBG.js";
|
|
23
23
|
import {
|
|
24
24
|
KordocError,
|
|
25
25
|
VERSION,
|
|
26
26
|
sanitizeError,
|
|
27
27
|
toArrayBuffer
|
|
28
|
-
} from "./chunk-
|
|
28
|
+
} from "./chunk-Q6UXVJXI.js";
|
|
29
29
|
|
|
30
30
|
// src/mcp.ts
|
|
31
31
|
import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
|
|
@@ -196,7 +196,7 @@ server.tool(
|
|
|
196
196
|
break;
|
|
197
197
|
case "pdf":
|
|
198
198
|
try {
|
|
199
|
-
const { extractPdfMetadataOnly } = await import("./parser-
|
|
199
|
+
const { extractPdfMetadataOnly } = await import("./parser-XUNGOH76.js");
|
|
200
200
|
metadata = await extractPdfMetadataOnly(buffer);
|
|
201
201
|
} catch {
|
|
202
202
|
metadata = void 0;
|
|
@@ -6,7 +6,7 @@
|
|
|
6
6
|
|
|
7
7
|
|
|
8
8
|
|
|
9
|
-
var
|
|
9
|
+
var _chunkMEXUFQVCcjs = require('./chunk-MEXUFQVC.cjs');
|
|
10
10
|
|
|
11
11
|
|
|
12
12
|
var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
|
|
@@ -240,10 +240,40 @@ var EDGE_MIN_SPAN = 12;
|
|
|
240
240
|
var EDGE_INSET = 15;
|
|
241
241
|
var EDGE_NEAR = 10;
|
|
242
242
|
var EDGE_CONNECT_TOL = 5;
|
|
243
|
+
var CHAIN_Y_TOL = 1.5;
|
|
244
|
+
var CHAIN_GAP = 3;
|
|
245
|
+
function chainCollinearRules(horizontals) {
|
|
246
|
+
if (horizontals.length <= 1) return horizontals;
|
|
247
|
+
const sorted = [...horizontals].sort((a, b) => a.y1 - b.y1 || a.x1 - b.x1);
|
|
248
|
+
const rules = [];
|
|
249
|
+
let bandStart = 0;
|
|
250
|
+
const flushBand = (end) => {
|
|
251
|
+
const band = sorted.slice(bandStart, end).sort((a, b) => a.x1 - b.x1);
|
|
252
|
+
let cur = { ...band[0] };
|
|
253
|
+
for (let i = 1; i < band.length; i++) {
|
|
254
|
+
const seg = band[i];
|
|
255
|
+
if (seg.x1 - cur.x2 <= CHAIN_GAP) {
|
|
256
|
+
if (seg.x2 > cur.x2) cur.x2 = seg.x2;
|
|
257
|
+
if (seg.lineWidth > cur.lineWidth) cur.lineWidth = seg.lineWidth;
|
|
258
|
+
} else {
|
|
259
|
+
rules.push(cur);
|
|
260
|
+
cur = { ...seg };
|
|
261
|
+
}
|
|
262
|
+
}
|
|
263
|
+
rules.push(cur);
|
|
264
|
+
};
|
|
265
|
+
for (let i = 1; i <= sorted.length; i++) {
|
|
266
|
+
if (i === sorted.length || sorted[i].y1 - sorted[bandStart].y1 > CHAIN_Y_TOL) {
|
|
267
|
+
flushBand(i);
|
|
268
|
+
bandStart = i;
|
|
269
|
+
}
|
|
270
|
+
}
|
|
271
|
+
return rules;
|
|
272
|
+
}
|
|
243
273
|
function closeOpenTableEdges(horizontals, verticals) {
|
|
244
274
|
if (horizontals.length < EDGE_MIN_RULES) return verticals;
|
|
245
275
|
const groups = [];
|
|
246
|
-
for (const hl of horizontals) {
|
|
276
|
+
for (const hl of chainCollinearRules(horizontals)) {
|
|
247
277
|
let placed = false;
|
|
248
278
|
for (const g2 of groups) {
|
|
249
279
|
if (Math.abs(g2[0].x1 - hl.x1) <= EDGE_ALIGN_TOL && Math.abs(g2[0].x2 - hl.x2) <= EDGE_ALIGN_TOL) {
|
|
@@ -2056,14 +2086,14 @@ function isProseSpread(items) {
|
|
|
2056
2086
|
for (let i = 1; i < sorted.length; i++) {
|
|
2057
2087
|
gaps.push(sorted[i].x - (sorted[i - 1].x + sorted[i - 1].w));
|
|
2058
2088
|
}
|
|
2059
|
-
const maxGap =
|
|
2089
|
+
const maxGap = _chunkMEXUFQVCcjs.safeMax.call(void 0, gaps);
|
|
2060
2090
|
const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
|
|
2061
2091
|
return maxGap < 40 && avgLen < 5;
|
|
2062
2092
|
}
|
|
2063
2093
|
function detectColumns(yLines) {
|
|
2064
2094
|
const allItems = yLines.flat();
|
|
2065
2095
|
if (allItems.length === 0) return null;
|
|
2066
|
-
const pageWidth =
|
|
2096
|
+
const pageWidth = _chunkMEXUFQVCcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkMEXUFQVCcjs.safeMin.call(void 0, allItems.map((i) => i.x));
|
|
2067
2097
|
if (pageWidth < 100) return null;
|
|
2068
2098
|
let bigoLineIdx = -1;
|
|
2069
2099
|
for (let i = 0; i < yLines.length; i++) {
|
|
@@ -2285,9 +2315,9 @@ function detectHeadings(blocks, medianFontSize) {
|
|
|
2285
2315
|
if (/^\d+$/.test(text)) continue;
|
|
2286
2316
|
const ratio = block.style.fontSize / medianFontSize;
|
|
2287
2317
|
let level = 0;
|
|
2288
|
-
if (ratio >=
|
|
2289
|
-
else if (ratio >=
|
|
2290
|
-
else if (ratio >=
|
|
2318
|
+
if (ratio >= _chunkMEXUFQVCcjs.HEADING_RATIO_H1) level = 1;
|
|
2319
|
+
else if (ratio >= _chunkMEXUFQVCcjs.HEADING_RATIO_H2) level = 2;
|
|
2320
|
+
else if (ratio >= _chunkMEXUFQVCcjs.HEADING_RATIO_H3) level = 3;
|
|
2291
2321
|
if (level > 0) {
|
|
2292
2322
|
block.type = "heading";
|
|
2293
2323
|
block.level = level;
|
|
@@ -2298,6 +2328,10 @@ function detectHeadings(blocks, medianFontSize) {
|
|
|
2298
2328
|
function shouldDemoteTable(table) {
|
|
2299
2329
|
const allCells = table.cells.flatMap((row) => row.map((c) => c.text.trim())).filter(Boolean);
|
|
2300
2330
|
const allText = allCells.join(" ");
|
|
2331
|
+
if (table.rows >= 2 && table.cols >= 2 && table.cells[0].every((c) => {
|
|
2332
|
+
const t = c.text.trim();
|
|
2333
|
+
return t.length > 0 && t.length <= 12 && !/[□■◆○●▶ㅇ<>]/.test(t);
|
|
2334
|
+
}) && table.cells.slice(1).some((row) => row.some((c) => c.text.trim() !== ""))) return false;
|
|
2301
2335
|
if (table.rows <= 3 && table.cols <= 3) {
|
|
2302
2336
|
const totalCells2 = table.rows * table.cols;
|
|
2303
2337
|
const emptyCells2 = totalCells2 - allCells.length;
|
|
@@ -2687,6 +2721,54 @@ function wrapStrikethroughRuns(items) {
|
|
|
2687
2721
|
arr[arr.length - 1].text = arr[arr.length - 1].text + "~~";
|
|
2688
2722
|
}
|
|
2689
2723
|
}
|
|
2724
|
+
var PROSEBOX_FULLWIDTH_MIN = 0.6;
|
|
2725
|
+
var PROSEBOX_LONG_CELL_CHARS = 80;
|
|
2726
|
+
var PROSEBOX_LONG_CELL_MIN = 3;
|
|
2727
|
+
var PROSEBOX_LONG_CELL_RATIO = 0.4;
|
|
2728
|
+
var PROSEBOX_X_TOL = 8;
|
|
2729
|
+
function verticalCoverageAt(verticals, x, yMin, yMax) {
|
|
2730
|
+
const tol = PROSEBOX_X_TOL;
|
|
2731
|
+
const spans = [];
|
|
2732
|
+
for (const v of verticals) {
|
|
2733
|
+
if (Math.abs(v.x1 - x) > tol) continue;
|
|
2734
|
+
const lo = Math.max(v.y1, yMin), hi = Math.min(v.y2, yMax);
|
|
2735
|
+
if (hi > lo) spans.push([lo, hi]);
|
|
2736
|
+
}
|
|
2737
|
+
if (spans.length === 0) return 0;
|
|
2738
|
+
spans.sort((a, b) => a[0] - b[0]);
|
|
2739
|
+
let total = 0, s = spans[0][0], e = spans[0][1];
|
|
2740
|
+
for (let i = 1; i < spans.length; i++) {
|
|
2741
|
+
if (spans[i][0] <= e) {
|
|
2742
|
+
if (spans[i][1] > e) e = spans[i][1];
|
|
2743
|
+
} else {
|
|
2744
|
+
total += e - s;
|
|
2745
|
+
s = spans[i][0];
|
|
2746
|
+
e = spans[i][1];
|
|
2747
|
+
}
|
|
2748
|
+
}
|
|
2749
|
+
return total + (e - s);
|
|
2750
|
+
}
|
|
2751
|
+
function isProseBoxGrid(grid, verticals, table) {
|
|
2752
|
+
const numCols = grid.colXs.length - 1;
|
|
2753
|
+
if (numCols < 2 || grid.rowYs.length < 3) return false;
|
|
2754
|
+
const gyMax = grid.rowYs[0], gyMin = grid.rowYs[grid.rowYs.length - 1];
|
|
2755
|
+
const span = gyMax - gyMin;
|
|
2756
|
+
if (span <= 0) return false;
|
|
2757
|
+
const interior = grid.colXs.slice(1, -1);
|
|
2758
|
+
let fullWidthHeight = 0;
|
|
2759
|
+
for (let r = 0; r < grid.rowYs.length - 1; r++) {
|
|
2760
|
+
const top = grid.rowYs[r], bot = grid.rowYs[r + 1];
|
|
2761
|
+
const h = top - bot;
|
|
2762
|
+
if (h <= 0) continue;
|
|
2763
|
+
const hasDivider = interior.some((cx) => verticalCoverageAt(verticals, cx, bot, top) >= h * 0.5);
|
|
2764
|
+
if (!hasDivider) fullWidthHeight += h;
|
|
2765
|
+
}
|
|
2766
|
+
if (fullWidthHeight < span * PROSEBOX_FULLWIDTH_MIN) return false;
|
|
2767
|
+
const texts = table.cells.flat().map((c) => c.text.trim()).filter(Boolean);
|
|
2768
|
+
const longCells = texts.filter((s) => s.length > PROSEBOX_LONG_CELL_CHARS).length;
|
|
2769
|
+
if (longCells < PROSEBOX_LONG_CELL_MIN || longCells < texts.length * PROSEBOX_LONG_CELL_RATIO) return false;
|
|
2770
|
+
return true;
|
|
2771
|
+
}
|
|
2690
2772
|
function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
|
|
2691
2773
|
const blocks = [];
|
|
2692
2774
|
const usedItems = /* @__PURE__ */ new Set();
|
|
@@ -2761,6 +2843,10 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
|
|
|
2761
2843
|
};
|
|
2762
2844
|
const hasContent = finalGrid.some((row) => row.some((cell) => cell.text.trim() !== ""));
|
|
2763
2845
|
if (!hasContent) continue;
|
|
2846
|
+
if (isProseBoxGrid(grid, verticals, irTable)) {
|
|
2847
|
+
for (const it of tableItems) usedItems.delete(it);
|
|
2848
|
+
continue;
|
|
2849
|
+
}
|
|
2764
2850
|
const tableBbox = {
|
|
2765
2851
|
page: pageNum,
|
|
2766
2852
|
x: grid.bbox.x1,
|
|
@@ -2807,7 +2893,7 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
|
|
|
2807
2893
|
}
|
|
2808
2894
|
if (remaining.length > 0) {
|
|
2809
2895
|
const allY = remaining.map((i) => i.y);
|
|
2810
|
-
const pageH =
|
|
2896
|
+
const pageH = _chunkMEXUFQVCcjs.safeMax.call(void 0, allY) - _chunkMEXUFQVCcjs.safeMin.call(void 0, allY);
|
|
2811
2897
|
const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
|
|
2812
2898
|
const textBlocks = [];
|
|
2813
2899
|
for (const group of groups) {
|
|
@@ -2983,7 +3069,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false) {
|
|
|
2983
3069
|
blocks.push({ type: "paragraph", text: tableText, pageNumber: pageNum, bbox, style: dominantStyle(items) });
|
|
2984
3070
|
} else {
|
|
2985
3071
|
const allY = items.map((i) => i.y);
|
|
2986
|
-
const pageHeight =
|
|
3072
|
+
const pageHeight = _chunkMEXUFQVCcjs.safeMax.call(void 0, allY) - _chunkMEXUFQVCcjs.safeMin.call(void 0, allY);
|
|
2987
3073
|
const gapThreshold = Math.max(15, pageHeight * 0.03);
|
|
2988
3074
|
const orderedGroups = proseCutX !== null ? splitTwoColumnProse(items, proseCutX) : xyCutOrder(items, gapThreshold);
|
|
2989
3075
|
for (const group of orderedGroups) {
|
|
@@ -3236,7 +3322,7 @@ async function loadPdfWithTimeout(buffer) {
|
|
|
3236
3322
|
new Promise((_, reject) => {
|
|
3237
3323
|
timer = setTimeout(() => {
|
|
3238
3324
|
loadingTask.destroy();
|
|
3239
|
-
reject(new (0,
|
|
3325
|
+
reject(new (0, _chunkMEXUFQVCcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
|
|
3240
3326
|
}, PDF_LOAD_TIMEOUT_MS);
|
|
3241
3327
|
})
|
|
3242
3328
|
]);
|
|
@@ -3249,7 +3335,7 @@ async function parsePdfDocument(buffer, options) {
|
|
|
3249
3335
|
const doc = await loadPdfWithTimeout(buffer);
|
|
3250
3336
|
try {
|
|
3251
3337
|
const pageCount = doc.numPages;
|
|
3252
|
-
if (pageCount === 0) throw new (0,
|
|
3338
|
+
if (pageCount === 0) throw new (0, _chunkMEXUFQVCcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
|
|
3253
3339
|
const metadata = { pageCount };
|
|
3254
3340
|
await extractPdfMetadata(doc, metadata);
|
|
3255
3341
|
const blocks = [];
|
|
@@ -3309,11 +3395,11 @@ async function parsePdfDocument(buffer, options) {
|
|
|
3309
3395
|
pageText += pageText ? "\n" + t : t;
|
|
3310
3396
|
}
|
|
3311
3397
|
pageQuality.push(computePageQuality(i, pageText));
|
|
3312
|
-
if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0,
|
|
3398
|
+
if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkMEXUFQVCcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
|
|
3313
3399
|
parsedPages++;
|
|
3314
3400
|
_optionalChain([options, 'optionalAccess', _36 => _36.onProgress, 'optionalCall', _37 => _37(parsedPages, totalTarget)]);
|
|
3315
3401
|
} catch (pageErr) {
|
|
3316
|
-
if (pageErr instanceof
|
|
3402
|
+
if (pageErr instanceof _chunkMEXUFQVCcjs.KordocError) throw pageErr;
|
|
3317
3403
|
warnings.push({ page: i, message: `\uD398\uC774\uC9C0 ${i} \uD30C\uC2F1 \uC2E4\uD328: ${pageErr instanceof Error ? pageErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
|
|
3318
3404
|
}
|
|
3319
3405
|
}
|
|
@@ -3381,7 +3467,7 @@ async function parsePdfDocument(buffer, options) {
|
|
|
3381
3467
|
detectKoreanListBlocks(blocks);
|
|
3382
3468
|
const outline = blocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
|
|
3383
3469
|
sanitizeBlockControlChars(blocks);
|
|
3384
|
-
let markdown = cleanPdfText(
|
|
3470
|
+
let markdown = cleanPdfText(_chunkMEXUFQVCcjs.blocksToMarkdown.call(void 0, blocks));
|
|
3385
3471
|
return {
|
|
3386
3472
|
markdown,
|
|
3387
3473
|
blocks,
|
|
@@ -3440,4 +3526,4 @@ async function extractPdfMetadataOnly(buffer) {
|
|
|
3440
3526
|
|
|
3441
3527
|
|
|
3442
3528
|
exports.cleanPdfText = cleanPdfText; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.detectTableCaptions = detectTableCaptions; exports.extractPdfMetadataOnly = extractPdfMetadataOnly; exports.mergeCrossPageTables = mergeCrossPageTables; exports.parsePdfDocument = parsePdfDocument; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks;
|
|
3443
|
-
//# sourceMappingURL=parser-
|
|
3529
|
+
//# sourceMappingURL=parser-DZBI3EA2.cjs.map
|