kordoc 4.12.0 → 4.12.2
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 +26 -5
- package/dist/{-B3EUUBPV.js → -EZH6NKTE.js} +8 -8
- package/dist/{chunk-HGUNWAXK.js → chunk-2B7OIEJJ.js} +9 -5
- package/dist/chunk-2B7OIEJJ.js.map +1 -0
- package/dist/{chunk-QKFCGJJY.js → chunk-2HBYWA5X.js} +2 -2
- package/dist/{chunk-HFJWCAJ2.cjs → chunk-2LKYJ4G3.cjs} +6 -6
- package/dist/chunk-2LKYJ4G3.cjs.map +1 -0
- package/dist/{chunk-COGFPMEB.js → chunk-2LO525TL.js} +2 -2
- package/dist/{chunk-TROD36DK.cjs → chunk-2LPBIW2B.cjs} +3 -3
- package/dist/chunk-2LPBIW2B.cjs.map +1 -0
- package/dist/{chunk-BQSWSMRG.cjs → chunk-62IU4NV3.cjs} +2 -2
- package/dist/chunk-62IU4NV3.cjs.map +1 -0
- package/dist/{chunk-SMGSCP57.js → chunk-7XUXSDH5.js} +308 -53
- package/dist/chunk-7XUXSDH5.js.map +1 -0
- package/dist/{chunk-YASFVKDT.js → chunk-DBGEUDBY.js} +2 -2
- package/dist/chunk-DCZVOIEO.cjs.map +1 -1
- package/dist/{chunk-4PKNWBXJ.js → chunk-FBLMVHDK.js} +2 -2
- package/dist/{chunk-I2OETEEL.js → chunk-FVBR7SC3.js} +2 -2
- package/dist/chunk-GS7T56RP.cjs.map +1 -1
- package/dist/{chunk-46UDI56N.js → chunk-IOJX4Q63.js} +3 -3
- package/dist/{chunk-BQE7AMFS.js → chunk-JCMBQC5N.js} +2 -2
- package/dist/{chunk-TF5DS74F.js → chunk-KX7IYCCZ.js} +3 -3
- package/dist/{chunk-PFUW4I4I.js → chunk-LD2K4K5V.js} +170 -18
- package/dist/chunk-LD2K4K5V.js.map +1 -0
- package/dist/{chunk-MJPRWCI5.js → chunk-MUYU4JI5.js} +2 -2
- package/dist/{chunk-SKGYPM6L.js → chunk-OEYECZGV.js} +308 -53
- package/dist/chunk-OEYECZGV.js.map +1 -0
- package/dist/{chunk-AT23GFSK.cjs → chunk-ONNAHSPV.cjs} +323 -68
- package/dist/chunk-ONNAHSPV.cjs.map +1 -0
- package/dist/{chunk-GHVXLV52.js → chunk-Q6ETEJFK.js} +9 -5
- package/dist/chunk-Q6ETEJFK.js.map +1 -0
- package/dist/{chunk-2IR3PEZU.js → chunk-QBCMFNHS.js} +2 -2
- package/dist/{chunk-DRLKMFDV.js → chunk-QM2RSJ2S.js} +3 -3
- package/dist/chunk-T7GRHAIW.cjs.map +1 -1
- package/dist/{chunk-TJYZY2O6.js → chunk-UL2MDA5C.js} +3 -3
- package/dist/{chunk-TYYJSOZS.cjs → chunk-V5BUPZ6I.cjs} +23 -19
- package/dist/chunk-V5BUPZ6I.cjs.map +1 -0
- package/dist/{chunk-7JQ2ITT2.js → chunk-XXDZ2KTW.js} +2 -2
- package/dist/chunks-TT7ZE5QQ.js +10 -0
- package/dist/cli.js +21 -21
- package/dist/cli.js.map +1 -1
- package/dist/{formula-WGKKR5OQ.js → formula-6RQX4NMJ.js} +4 -4
- package/dist/{formula-YMGXXRBQ.cjs → formula-NBEO6CFC.cjs} +9 -9
- package/dist/formula-NBEO6CFC.cjs.map +1 -0
- package/dist/{formula-G4CD5KQ6.js → formula-NW33S4TA.js} +3 -3
- package/dist/{image-ocr-MBWZK3ZT.js → image-ocr-CAVU2TNZ.js} +5 -5
- package/dist/{image-ocr-M4RP4INO.js → image-ocr-SO56VETS.js} +7 -7
- package/dist/{image-ocr-BE2HHMVN.cjs → image-ocr-V4KY373B.cjs} +11 -11
- package/dist/image-ocr-V4KY373B.cjs.map +1 -0
- package/dist/index.cjs +547 -395
- package/dist/index.cjs.map +1 -1
- package/dist/index.d.cts +5 -2
- package/dist/index.d.ts +5 -2
- package/dist/index.js +166 -14
- package/dist/index.js.map +1 -1
- package/dist/mcp.js +15 -15
- package/dist/{model-bundle-BIMP5DIH.js → model-bundle-EIJF34DR.js} +5 -5
- package/dist/{models-QPOEGHVA.js → models-6V4CPQXC.js} +5 -5
- package/dist/page-range-P7SDW6LR.cjs.map +1 -1
- package/dist/{parser-CMNHBVTL.js → parser-GIREMEPD.js} +307 -6
- package/dist/parser-GIREMEPD.js.map +1 -0
- package/dist/{parser-NR3XO4SM.js → parser-OBHNXEQM.js} +307 -6
- package/dist/parser-OBHNXEQM.js.map +1 -0
- package/dist/{parser-QJY5W47V.cjs → parser-OMT25WH2.cjs} +334 -33
- package/dist/parser-OMT25WH2.cjs.map +1 -0
- package/dist/pdf-ocr-IRXQ5GBM.js +13 -0
- package/dist/pdf-ocr-SL4GCOGS.js +16 -0
- package/dist/pdf-ocr-TFYVK4FI.cjs +14 -0
- package/dist/pdf-ocr-TFYVK4FI.cjs.map +1 -0
- package/dist/{profile-io-UZ43CLWX.js → profile-io-PZVMM66V.js} +3 -3
- package/dist/{rasterize-JC2JVVOF.js → rasterize-NRWP2C6L.js} +2 -2
- package/dist/render-V5ZPTZQ7.js +10 -0
- package/dist/seal-LVG2DGAB.js +10 -0
- package/dist/{setup-WDUVPFX3.js → setup-V55AJP6H.js} +3 -3
- package/dist/{watch-UI7OXIES.js → watch-7CQSMDDP.js} +9 -9
- package/package.json +1 -1
- package/dist/chunk-AT23GFSK.cjs.map +0 -1
- package/dist/chunk-BQSWSMRG.cjs.map +0 -1
- package/dist/chunk-GHVXLV52.js.map +0 -1
- package/dist/chunk-HFJWCAJ2.cjs.map +0 -1
- package/dist/chunk-HGUNWAXK.js.map +0 -1
- package/dist/chunk-PFUW4I4I.js.map +0 -1
- package/dist/chunk-SKGYPM6L.js.map +0 -1
- package/dist/chunk-SMGSCP57.js.map +0 -1
- package/dist/chunk-TROD36DK.cjs.map +0 -1
- package/dist/chunk-TYYJSOZS.cjs.map +0 -1
- package/dist/chunks-Q4GCPRRU.js +0 -10
- package/dist/formula-YMGXXRBQ.cjs.map +0 -1
- package/dist/image-ocr-BE2HHMVN.cjs.map +0 -1
- package/dist/parser-CMNHBVTL.js.map +0 -1
- package/dist/parser-NR3XO4SM.js.map +0 -1
- package/dist/parser-QJY5W47V.cjs.map +0 -1
- package/dist/pdf-ocr-DF3OTBKH.js +0 -13
- package/dist/pdf-ocr-R4DIKF4Q.js +0 -16
- package/dist/pdf-ocr-T3RHIZKY.cjs +0 -14
- package/dist/pdf-ocr-T3RHIZKY.cjs.map +0 -1
- package/dist/render-KSQ6UGCP.js +0 -10
- package/dist/seal-EOFFQ4JR.js +0 -10
- /package/dist/{-B3EUUBPV.js.map → -EZH6NKTE.js.map} +0 -0
- /package/dist/{chunk-QKFCGJJY.js.map → chunk-2HBYWA5X.js.map} +0 -0
- /package/dist/{chunk-COGFPMEB.js.map → chunk-2LO525TL.js.map} +0 -0
- /package/dist/{chunk-YASFVKDT.js.map → chunk-DBGEUDBY.js.map} +0 -0
- /package/dist/{chunk-4PKNWBXJ.js.map → chunk-FBLMVHDK.js.map} +0 -0
- /package/dist/{chunk-I2OETEEL.js.map → chunk-FVBR7SC3.js.map} +0 -0
- /package/dist/{chunk-46UDI56N.js.map → chunk-IOJX4Q63.js.map} +0 -0
- /package/dist/{chunk-BQE7AMFS.js.map → chunk-JCMBQC5N.js.map} +0 -0
- /package/dist/{chunk-TF5DS74F.js.map → chunk-KX7IYCCZ.js.map} +0 -0
- /package/dist/{chunk-MJPRWCI5.js.map → chunk-MUYU4JI5.js.map} +0 -0
- /package/dist/{chunk-2IR3PEZU.js.map → chunk-QBCMFNHS.js.map} +0 -0
- /package/dist/{chunk-DRLKMFDV.js.map → chunk-QM2RSJ2S.js.map} +0 -0
- /package/dist/{chunk-TJYZY2O6.js.map → chunk-UL2MDA5C.js.map} +0 -0
- /package/dist/{chunk-7JQ2ITT2.js.map → chunk-XXDZ2KTW.js.map} +0 -0
- /package/dist/{chunks-Q4GCPRRU.js.map → chunks-TT7ZE5QQ.js.map} +0 -0
- /package/dist/{formula-WGKKR5OQ.js.map → formula-6RQX4NMJ.js.map} +0 -0
- /package/dist/{formula-G4CD5KQ6.js.map → formula-NW33S4TA.js.map} +0 -0
- /package/dist/{image-ocr-MBWZK3ZT.js.map → image-ocr-CAVU2TNZ.js.map} +0 -0
- /package/dist/{image-ocr-M4RP4INO.js.map → image-ocr-SO56VETS.js.map} +0 -0
- /package/dist/{model-bundle-BIMP5DIH.js.map → model-bundle-EIJF34DR.js.map} +0 -0
- /package/dist/{models-QPOEGHVA.js.map → models-6V4CPQXC.js.map} +0 -0
- /package/dist/{pdf-ocr-DF3OTBKH.js.map → pdf-ocr-IRXQ5GBM.js.map} +0 -0
- /package/dist/{pdf-ocr-R4DIKF4Q.js.map → pdf-ocr-SL4GCOGS.js.map} +0 -0
- /package/dist/{profile-io-UZ43CLWX.js.map → profile-io-PZVMM66V.js.map} +0 -0
- /package/dist/{rasterize-JC2JVVOF.js.map → rasterize-NRWP2C6L.js.map} +0 -0
- /package/dist/{render-KSQ6UGCP.js.map → render-V5ZPTZQ7.js.map} +0 -0
- /package/dist/{seal-EOFFQ4JR.js.map → seal-LVG2DGAB.js.map} +0 -0
- /package/dist/{setup-WDUVPFX3.js.map → setup-V55AJP6H.js.map} +0 -0
- /package/dist/{watch-UI7OXIES.js.map → watch-7CQSMDDP.js.map} +0 -0
package/dist/mcp.js
CHANGED
|
@@ -39,30 +39,30 @@ import {
|
|
|
39
39
|
resolveBuiltinTemplate,
|
|
40
40
|
unknownFontWarnings,
|
|
41
41
|
usesGaejosikMunche
|
|
42
|
-
} from "./chunk-
|
|
42
|
+
} from "./chunk-LD2K4K5V.js";
|
|
43
43
|
import "./chunk-7FCNOKOR.js";
|
|
44
44
|
import "./chunk-N3DLQ2YH.js";
|
|
45
45
|
import "./chunk-MOL7MDBG.js";
|
|
46
|
-
import "./chunk-
|
|
46
|
+
import "./chunk-KX7IYCCZ.js";
|
|
47
47
|
import "./chunk-DZIXKL7E.js";
|
|
48
|
-
import "./chunk-
|
|
48
|
+
import "./chunk-DBGEUDBY.js";
|
|
49
49
|
import {
|
|
50
50
|
blocksToMarkdown
|
|
51
|
-
} from "./chunk-
|
|
52
|
-
import "./chunk-
|
|
53
|
-
import "./chunk-
|
|
51
|
+
} from "./chunk-Q6ETEJFK.js";
|
|
52
|
+
import "./chunk-IOJX4Q63.js";
|
|
53
|
+
import "./chunk-FVBR7SC3.js";
|
|
54
54
|
import {
|
|
55
55
|
assertWithinRoot,
|
|
56
56
|
getAccessRoot,
|
|
57
57
|
isOfflineMode
|
|
58
|
-
} from "./chunk-
|
|
58
|
+
} from "./chunk-QBCMFNHS.js";
|
|
59
59
|
import {
|
|
60
60
|
KordocError,
|
|
61
61
|
VERSION,
|
|
62
62
|
classifyError,
|
|
63
63
|
sanitizeError,
|
|
64
64
|
toArrayBuffer
|
|
65
|
-
} from "./chunk-
|
|
65
|
+
} from "./chunk-FBLMVHDK.js";
|
|
66
66
|
|
|
67
67
|
// src/mcp.ts
|
|
68
68
|
import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
|
|
@@ -312,7 +312,7 @@ server.tool(
|
|
|
312
312
|
break;
|
|
313
313
|
case "pdf":
|
|
314
314
|
try {
|
|
315
|
-
const { extractPdfMetadataOnly } = await import("./parser-
|
|
315
|
+
const { extractPdfMetadataOnly } = await import("./parser-OBHNXEQM.js");
|
|
316
316
|
metadata = await extractPdfMetadataOnly(buffer);
|
|
317
317
|
} catch {
|
|
318
318
|
metadata = void 0;
|
|
@@ -694,7 +694,7 @@ server.tool(
|
|
|
694
694
|
}
|
|
695
695
|
const image = new Uint8Array(await readFile(imgResolved));
|
|
696
696
|
const ext = extname(imgResolved).slice(1).toLowerCase();
|
|
697
|
-
const { placeSealHwpx } = await import("./seal-
|
|
697
|
+
const { placeSealHwpx } = await import("./seal-LVG2DGAB.js");
|
|
698
698
|
const result = await placeSealHwpx(buffer, [{
|
|
699
699
|
anchor,
|
|
700
700
|
occurrence,
|
|
@@ -796,7 +796,7 @@ server.tool(
|
|
|
796
796
|
}
|
|
797
797
|
const outPath = output_path ? safeOutputPath(output_path, /* @__PURE__ */ new Set([format === "png" ? ".png" : ".svg"])) : void 0;
|
|
798
798
|
const { buffer } = await readValidatedFile(file_path, MAX_FILE_SIZE, /* @__PURE__ */ new Set([".hwpx"]));
|
|
799
|
-
const { renderHwpxToSvg } = await import("./render-
|
|
799
|
+
const { renderHwpxToSvg } = await import("./render-V5ZPTZQ7.js");
|
|
800
800
|
const result = await renderHwpxToSvg(buffer, { highlights, reflow: true, reflowMode: reflow_mode });
|
|
801
801
|
const summary = [
|
|
802
802
|
`\uB80C\uB354 \uC644\uB8CC: ${result.pageCount}\uD398\uC774\uC9C0, ${Math.round(result.width)}x${Math.round(result.height)}pt (\uD14D\uC2A4\uD2B8 ${result.stats.texts}\xB7\uC774\uBBF8\uC9C0 ${result.stats.images}\xB7\uD45C ${result.stats.tables})`,
|
|
@@ -808,7 +808,7 @@ server.tool(
|
|
|
808
808
|
summary.push(`\uC800\uC7A5: ${outPath}`);
|
|
809
809
|
return { content: [{ type: "text", text: summary.join("\n") }] };
|
|
810
810
|
}
|
|
811
|
-
const { rasterizeSvg } = await import("./rasterize-
|
|
811
|
+
const { rasterizeSvg } = await import("./rasterize-NRWP2C6L.js");
|
|
812
812
|
const raster = await rasterizeSvg(result.svg, result.width, result.height);
|
|
813
813
|
if (outPath) {
|
|
814
814
|
await mkdir(dirname(outPath), { recursive: true });
|
|
@@ -906,7 +906,7 @@ server.tool(
|
|
|
906
906
|
if (!parsed.success) {
|
|
907
907
|
return { content: [{ type: "text", text: `\uD30C\uC2F1 \uC2E4\uD328: ${parsed.error}` }], isError: true };
|
|
908
908
|
}
|
|
909
|
-
const { blocksToChunks } = await import("./chunks-
|
|
909
|
+
const { blocksToChunks } = await import("./chunks-TT7ZE5QQ.js");
|
|
910
910
|
const chunks = blocksToChunks(parsed.blocks, { granularity, includeTableCells: include_table_cells });
|
|
911
911
|
return { content: [{ type: "text", text: capResponseText(JSON.stringify(chunks, null, 2)) }] };
|
|
912
912
|
} catch (err) {
|
|
@@ -925,7 +925,7 @@ server.tool(
|
|
|
925
925
|
try {
|
|
926
926
|
const out = safeOutputPath(output_path, PROFILE_EXTENSIONS);
|
|
927
927
|
const { buffer } = await readValidatedFile(hwpx_path);
|
|
928
|
-
const { hwpxToProfile } = await import("./-
|
|
928
|
+
const { hwpxToProfile } = await import("./-EZH6NKTE.js");
|
|
929
929
|
const profile = await hwpxToProfile(buffer);
|
|
930
930
|
await mkdir(dirname(out), { recursive: true });
|
|
931
931
|
await writeFile(out, JSON.stringify(profile, null, 2));
|
|
@@ -1014,7 +1014,7 @@ server.tool(
|
|
|
1014
1014
|
}
|
|
1015
1015
|
let profile;
|
|
1016
1016
|
if (profile_path) {
|
|
1017
|
-
const { parseFormatProfileJson } = await import("./profile-io-
|
|
1017
|
+
const { parseFormatProfileJson } = await import("./profile-io-PZVMM66V.js");
|
|
1018
1018
|
profile = parseFormatProfileJson(await readFile(safePath(profile_path, PROFILE_EXTENSIONS), "utf-8"));
|
|
1019
1019
|
}
|
|
1020
1020
|
const out = safeOutputPath(output_path, /* @__PURE__ */ new Set([".hwpx"]));
|
|
@@ -1,14 +1,14 @@
|
|
|
1
1
|
#!/usr/bin/env node
|
|
2
2
|
import {
|
|
3
3
|
ALL_OCR_MODELS
|
|
4
|
-
} from "./chunk-
|
|
4
|
+
} from "./chunk-2HBYWA5X.js";
|
|
5
5
|
import {
|
|
6
6
|
ALL_FORMULA_MODELS,
|
|
7
7
|
getModelStatusIn,
|
|
8
8
|
getModelsDir
|
|
9
|
-
} from "./chunk-
|
|
10
|
-
import "./chunk-
|
|
11
|
-
import "./chunk-
|
|
9
|
+
} from "./chunk-JCMBQC5N.js";
|
|
10
|
+
import "./chunk-QBCMFNHS.js";
|
|
11
|
+
import "./chunk-FBLMVHDK.js";
|
|
12
12
|
|
|
13
13
|
// src/shared/model-bundle.ts
|
|
14
14
|
import { copyFile, mkdir } from "fs/promises";
|
|
@@ -101,4 +101,4 @@ export {
|
|
|
101
101
|
importModels,
|
|
102
102
|
modelCacheStatus
|
|
103
103
|
};
|
|
104
|
-
//# sourceMappingURL=model-bundle-
|
|
104
|
+
//# sourceMappingURL=model-bundle-EIJF34DR.js.map
|
|
@@ -9,10 +9,10 @@ import {
|
|
|
9
9
|
getOcrModelsDir,
|
|
10
10
|
ocrModelPath,
|
|
11
11
|
parseCharacterDict
|
|
12
|
-
} from "./chunk-
|
|
13
|
-
import "./chunk-
|
|
14
|
-
import "./chunk-
|
|
15
|
-
import "./chunk-
|
|
12
|
+
} from "./chunk-2HBYWA5X.js";
|
|
13
|
+
import "./chunk-JCMBQC5N.js";
|
|
14
|
+
import "./chunk-QBCMFNHS.js";
|
|
15
|
+
import "./chunk-FBLMVHDK.js";
|
|
16
16
|
export {
|
|
17
17
|
ALL_OCR_MODELS,
|
|
18
18
|
OCR_DET_MODEL,
|
|
@@ -24,4 +24,4 @@ export {
|
|
|
24
24
|
ocrModelPath,
|
|
25
25
|
parseCharacterDict
|
|
26
26
|
};
|
|
27
|
-
//# sourceMappingURL=models-
|
|
27
|
+
//# sourceMappingURL=models-6V4CPQXC.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/
|
|
1
|
+
{"version":3,"sources":["/Users/chris_gomdori/workspace/kordoc/dist/page-range-P7SDW6LR.cjs"],"names":[],"mappings":"AAAA;AACE;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACF,0DAAC","file":"/Users/chris_gomdori/workspace/kordoc/dist/page-range-P7SDW6LR.cjs"}
|
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import {
|
|
2
2
|
blocksToMarkdown,
|
|
3
3
|
encodePng
|
|
4
|
-
} from "./chunk-
|
|
4
|
+
} from "./chunk-2B7OIEJJ.js";
|
|
5
5
|
import {
|
|
6
6
|
parsePageRange
|
|
7
7
|
} from "./chunk-GE43BE46.js";
|
|
@@ -18,12 +18,12 @@ import {
|
|
|
18
18
|
mergeCrossPageTables,
|
|
19
19
|
normalizeItems,
|
|
20
20
|
removeHeaderFooterBlocks
|
|
21
|
-
} from "./chunk-
|
|
21
|
+
} from "./chunk-7XUXSDH5.js";
|
|
22
22
|
import "./chunk-ONQHGZTS.js";
|
|
23
23
|
import {
|
|
24
24
|
KordocError,
|
|
25
25
|
sanitizeHref
|
|
26
|
-
} from "./chunk-
|
|
26
|
+
} from "./chunk-MUYU4JI5.js";
|
|
27
27
|
|
|
28
28
|
// src/pdf/image-extract.ts
|
|
29
29
|
import { OPS, ImageKind } from "pdfjs-dist/legacy/build/pdf.mjs";
|
|
@@ -322,6 +322,300 @@ function summarizeDocumentQuality(pages) {
|
|
|
322
322
|
};
|
|
323
323
|
}
|
|
324
324
|
|
|
325
|
+
// src/pdf/symbol-fonts.ts
|
|
326
|
+
var WINGDINGS = [
|
|
327
|
+
"\u{1F589}",
|
|
328
|
+
"\u2702",
|
|
329
|
+
"\u2701",
|
|
330
|
+
"\u{1F453}",
|
|
331
|
+
"\u{1F56D}",
|
|
332
|
+
"\u{1F56E}",
|
|
333
|
+
"\u{1F56F}",
|
|
334
|
+
"\u{1F57F}",
|
|
335
|
+
"\u2706",
|
|
336
|
+
"\u{1F582}",
|
|
337
|
+
"\u{1F583}",
|
|
338
|
+
"\u{1F4EA}",
|
|
339
|
+
"\u{1F4EB}",
|
|
340
|
+
"\u{1F4EC}",
|
|
341
|
+
"\u{1F4ED}",
|
|
342
|
+
"\u{1F4C1}",
|
|
343
|
+
"\u{1F4C2}",
|
|
344
|
+
"\u{1F4C4}",
|
|
345
|
+
"\u{1F5CF}",
|
|
346
|
+
"\u{1F5D0}",
|
|
347
|
+
"\u{1F5C4}",
|
|
348
|
+
"\u231B",
|
|
349
|
+
"\u{1F5AE}",
|
|
350
|
+
"\u{1F5B0}",
|
|
351
|
+
"\u{1F5B2}",
|
|
352
|
+
"\u{1F5B3}",
|
|
353
|
+
"\u{1F5B4}",
|
|
354
|
+
"\u{1F5AB}",
|
|
355
|
+
"\u{1F5AC}",
|
|
356
|
+
"\u2707",
|
|
357
|
+
"\u270D",
|
|
358
|
+
"\u{1F58E}",
|
|
359
|
+
"\u270C",
|
|
360
|
+
"\u{1F44C}",
|
|
361
|
+
"\u{1F44D}",
|
|
362
|
+
"\u{1F44E}",
|
|
363
|
+
"\u261C",
|
|
364
|
+
"\u261E",
|
|
365
|
+
"\u261D",
|
|
366
|
+
"\u261F",
|
|
367
|
+
"\u{1F590}",
|
|
368
|
+
"\u263A",
|
|
369
|
+
"\u{1F610}",
|
|
370
|
+
"\u2639",
|
|
371
|
+
"\u{1F4A3}",
|
|
372
|
+
"\u2620",
|
|
373
|
+
"\u{1F3F3}",
|
|
374
|
+
"\u{1F3F1}",
|
|
375
|
+
"\u2708",
|
|
376
|
+
"\u263C",
|
|
377
|
+
"\u{1F4A7}",
|
|
378
|
+
"\u2744",
|
|
379
|
+
"\u{1F546}",
|
|
380
|
+
"\u271E",
|
|
381
|
+
"\u{1F548}",
|
|
382
|
+
"\u2720",
|
|
383
|
+
"\u2721",
|
|
384
|
+
"\u262A",
|
|
385
|
+
"\u262F",
|
|
386
|
+
"\u0950",
|
|
387
|
+
"\u2638",
|
|
388
|
+
"\u2648",
|
|
389
|
+
"\u2649",
|
|
390
|
+
"\u264A",
|
|
391
|
+
"\u264B",
|
|
392
|
+
"\u264C",
|
|
393
|
+
"\u264D",
|
|
394
|
+
"\u264E",
|
|
395
|
+
"\u264F",
|
|
396
|
+
"\u2650",
|
|
397
|
+
"\u2651",
|
|
398
|
+
"\u2652",
|
|
399
|
+
"\u2653",
|
|
400
|
+
"\u{1F670}",
|
|
401
|
+
"\u{1F675}",
|
|
402
|
+
"\u25CF",
|
|
403
|
+
"\u25CB",
|
|
404
|
+
"\u25A0",
|
|
405
|
+
"\u25A1",
|
|
406
|
+
"\u25A1",
|
|
407
|
+
"\u2751",
|
|
408
|
+
"\u2752",
|
|
409
|
+
"\u2B27",
|
|
410
|
+
"\u29EB",
|
|
411
|
+
"\u25C6",
|
|
412
|
+
"\u2756",
|
|
413
|
+
"\u2B25",
|
|
414
|
+
"\u2327",
|
|
415
|
+
"\u2BB9",
|
|
416
|
+
"\u2318",
|
|
417
|
+
"\u{1F3F5}",
|
|
418
|
+
"\u{1F3F6}",
|
|
419
|
+
"\u{1F676}",
|
|
420
|
+
"\u{1F677}",
|
|
421
|
+
"",
|
|
422
|
+
"\u24EA",
|
|
423
|
+
"\u2460",
|
|
424
|
+
"\u2461",
|
|
425
|
+
"\u2462",
|
|
426
|
+
"\u2463",
|
|
427
|
+
"\u2464",
|
|
428
|
+
"\u2465",
|
|
429
|
+
"\u2466",
|
|
430
|
+
"\u2467",
|
|
431
|
+
"\u2468",
|
|
432
|
+
"\u2469",
|
|
433
|
+
"\u24FF",
|
|
434
|
+
"\u2776",
|
|
435
|
+
"\u2777",
|
|
436
|
+
"\u2778",
|
|
437
|
+
"\u2779",
|
|
438
|
+
"\u277A",
|
|
439
|
+
"\u277B",
|
|
440
|
+
"\u277C",
|
|
441
|
+
"\u277D",
|
|
442
|
+
"\u277E",
|
|
443
|
+
"\u277F",
|
|
444
|
+
"\u{1F662}",
|
|
445
|
+
"\u{1F660}",
|
|
446
|
+
"\u{1F661}",
|
|
447
|
+
"\u{1F663}",
|
|
448
|
+
"\u{1F65E}",
|
|
449
|
+
"\u{1F65C}",
|
|
450
|
+
"\u{1F65D}",
|
|
451
|
+
"\u{1F65F}",
|
|
452
|
+
"\xB7",
|
|
453
|
+
"\u2022",
|
|
454
|
+
"\u25AA",
|
|
455
|
+
"\u26AA",
|
|
456
|
+
"\u25CB",
|
|
457
|
+
"\u25EF",
|
|
458
|
+
"\u25C9",
|
|
459
|
+
"\u25CE",
|
|
460
|
+
"\u{1F53F}",
|
|
461
|
+
"\u25AA",
|
|
462
|
+
"\u25FB",
|
|
463
|
+
"\u{1F7C2}",
|
|
464
|
+
"\u2726",
|
|
465
|
+
"\u2605",
|
|
466
|
+
"\u2736",
|
|
467
|
+
"\u2734",
|
|
468
|
+
"\u2739",
|
|
469
|
+
"\u2735",
|
|
470
|
+
"\u2BD0",
|
|
471
|
+
"\u2316",
|
|
472
|
+
"\u27E1",
|
|
473
|
+
"\u2311",
|
|
474
|
+
"\u2BD1",
|
|
475
|
+
"\u272A",
|
|
476
|
+
"\u2730",
|
|
477
|
+
"\u{1F550}",
|
|
478
|
+
"\u{1F551}",
|
|
479
|
+
"\u{1F552}",
|
|
480
|
+
"\u{1F553}",
|
|
481
|
+
"\u{1F554}",
|
|
482
|
+
"\u{1F555}",
|
|
483
|
+
"\u{1F556}",
|
|
484
|
+
"\u{1F557}",
|
|
485
|
+
"\u{1F558}",
|
|
486
|
+
"\u{1F559}",
|
|
487
|
+
"\u{1F55A}",
|
|
488
|
+
"\u{1F55B}",
|
|
489
|
+
"\u2BB0",
|
|
490
|
+
"\u2BB1",
|
|
491
|
+
"\u2BB2",
|
|
492
|
+
"\u2BB3",
|
|
493
|
+
"\u2BB4",
|
|
494
|
+
"\u2BB5",
|
|
495
|
+
"\u2BB6",
|
|
496
|
+
"\u2BB7",
|
|
497
|
+
"\u{1F66A}",
|
|
498
|
+
"\u{1F66B}",
|
|
499
|
+
"\u{1F655}",
|
|
500
|
+
"\u{1F654}",
|
|
501
|
+
"\u{1F657}",
|
|
502
|
+
"\u{1F656}",
|
|
503
|
+
"\u{1F650}",
|
|
504
|
+
"\u{1F651}",
|
|
505
|
+
"\u{1F652}",
|
|
506
|
+
"\u{1F653}",
|
|
507
|
+
"\u232B",
|
|
508
|
+
"\u2326",
|
|
509
|
+
"\u2B98",
|
|
510
|
+
"\u2B9A",
|
|
511
|
+
"\u2B99",
|
|
512
|
+
"\u2B9B",
|
|
513
|
+
"\u2B88",
|
|
514
|
+
"\u2B8A",
|
|
515
|
+
"\u2B89",
|
|
516
|
+
"\u2B8B",
|
|
517
|
+
"\u2190",
|
|
518
|
+
"\u2192",
|
|
519
|
+
"\u2191",
|
|
520
|
+
"\u2193",
|
|
521
|
+
"\u2196",
|
|
522
|
+
"\u2197",
|
|
523
|
+
"\u2199",
|
|
524
|
+
"\u2198",
|
|
525
|
+
"\u2B05",
|
|
526
|
+
"\u2794",
|
|
527
|
+
"\u2B06",
|
|
528
|
+
"\u2B07",
|
|
529
|
+
"\u2B09",
|
|
530
|
+
"\u2B08",
|
|
531
|
+
"\u2B0B",
|
|
532
|
+
"\u2B0A",
|
|
533
|
+
"\u21E6",
|
|
534
|
+
"\u21E8",
|
|
535
|
+
"\u21E7",
|
|
536
|
+
"\u21E9",
|
|
537
|
+
"\u2B04",
|
|
538
|
+
"\u21F3",
|
|
539
|
+
"\u2B00",
|
|
540
|
+
"\u2B01",
|
|
541
|
+
"\u2B03",
|
|
542
|
+
"\u2B02",
|
|
543
|
+
"\u{1F8AC}",
|
|
544
|
+
"\u{1F8AD}",
|
|
545
|
+
"\u2717",
|
|
546
|
+
"\u2714",
|
|
547
|
+
"\u2612",
|
|
548
|
+
"\u2611",
|
|
549
|
+
""
|
|
550
|
+
];
|
|
551
|
+
var WINANSI_REVERSE = {
|
|
552
|
+
8364: 128,
|
|
553
|
+
8218: 130,
|
|
554
|
+
402: 131,
|
|
555
|
+
8222: 132,
|
|
556
|
+
8230: 133,
|
|
557
|
+
8224: 134,
|
|
558
|
+
8225: 135,
|
|
559
|
+
710: 136,
|
|
560
|
+
8240: 137,
|
|
561
|
+
352: 138,
|
|
562
|
+
8249: 139,
|
|
563
|
+
338: 140,
|
|
564
|
+
381: 142,
|
|
565
|
+
8216: 145,
|
|
566
|
+
8217: 146,
|
|
567
|
+
8220: 147,
|
|
568
|
+
8221: 148,
|
|
569
|
+
8226: 149,
|
|
570
|
+
8211: 150,
|
|
571
|
+
8212: 151,
|
|
572
|
+
732: 152,
|
|
573
|
+
8482: 153,
|
|
574
|
+
353: 154,
|
|
575
|
+
8250: 155,
|
|
576
|
+
339: 156,
|
|
577
|
+
382: 158,
|
|
578
|
+
376: 159
|
|
579
|
+
};
|
|
580
|
+
function symbolFontTable(fontName) {
|
|
581
|
+
if (!fontName) return void 0;
|
|
582
|
+
if (/wingdings(?![\s-]*[23])/i.test(fontName)) return WINGDINGS;
|
|
583
|
+
return void 0;
|
|
584
|
+
}
|
|
585
|
+
function remapSymbolText(text, table) {
|
|
586
|
+
let out = "";
|
|
587
|
+
for (const ch of text) {
|
|
588
|
+
let code = ch.codePointAt(0);
|
|
589
|
+
if (code >= 128 && WINANSI_REVERSE[code] !== void 0) code = WINANSI_REVERSE[code];
|
|
590
|
+
if (code >= 33 && code <= 255) {
|
|
591
|
+
const mapped = table[code - 33];
|
|
592
|
+
out += mapped !== void 0 && mapped !== "" ? mapped : ch;
|
|
593
|
+
} else {
|
|
594
|
+
out += ch;
|
|
595
|
+
}
|
|
596
|
+
}
|
|
597
|
+
return out;
|
|
598
|
+
}
|
|
599
|
+
function remapSymbolFontItems(items, resolveFontName) {
|
|
600
|
+
const cache = /* @__PURE__ */ new Map();
|
|
601
|
+
let changed = 0;
|
|
602
|
+
for (const it of items) {
|
|
603
|
+
if (!it.fontName || !it.text) continue;
|
|
604
|
+
let table = cache.get(it.fontName);
|
|
605
|
+
if (!cache.has(it.fontName)) {
|
|
606
|
+
table = symbolFontTable(resolveFontName(it.fontName));
|
|
607
|
+
cache.set(it.fontName, table);
|
|
608
|
+
}
|
|
609
|
+
if (!table) continue;
|
|
610
|
+
const mapped = remapSymbolText(it.text, table);
|
|
611
|
+
if (mapped !== it.text) {
|
|
612
|
+
it.text = mapped;
|
|
613
|
+
changed++;
|
|
614
|
+
}
|
|
615
|
+
}
|
|
616
|
+
return changed;
|
|
617
|
+
}
|
|
618
|
+
|
|
325
619
|
// src/pdf/text-clean.ts
|
|
326
620
|
function sanitizeBlockControlChars(blocks) {
|
|
327
621
|
for (const b of blocks) {
|
|
@@ -420,7 +714,7 @@ function applyLinkAnnotations(items, annots) {
|
|
|
420
714
|
|
|
421
715
|
// src/pdf/formula-ocr.ts
|
|
422
716
|
async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
|
|
423
|
-
const formulaMod = await import("./formula-
|
|
717
|
+
const formulaMod = await import("./formula-NW33S4TA.js");
|
|
424
718
|
const { FormulaPipeline, ensureFormulaModels } = formulaMod;
|
|
425
719
|
await ensureFormulaModels((p) => {
|
|
426
720
|
if (p.phase === "download" && p.total) {
|
|
@@ -643,6 +937,13 @@ async function parsePdfDocument(buffer, options) {
|
|
|
643
937
|
} catch {
|
|
644
938
|
}
|
|
645
939
|
const opList = await page.getOperatorList();
|
|
940
|
+
remapSymbolFontItems(visible, (loadedName) => {
|
|
941
|
+
try {
|
|
942
|
+
return page.commonObjs.has(loadedName) ? page.commonObjs.get(loadedName)?.name : void 0;
|
|
943
|
+
} catch {
|
|
944
|
+
return void 0;
|
|
945
|
+
}
|
|
946
|
+
});
|
|
646
947
|
const pageArea = viewport.width * viewport.height;
|
|
647
948
|
if (pageArea > 0) {
|
|
648
949
|
const imageRegions = extractImageRegions(opList.fnArray, opList.argsArray);
|
|
@@ -705,7 +1006,7 @@ async function parsePdfDocument(buffer, options) {
|
|
|
705
1006
|
}
|
|
706
1007
|
if (targets.size > 0) {
|
|
707
1008
|
try {
|
|
708
|
-
const { runPdfOcr } = await import("./pdf-ocr-
|
|
1009
|
+
const { runPdfOcr } = await import("./pdf-ocr-IRXQ5GBM.js");
|
|
709
1010
|
const mode = typeof options.ocr === "function" ? options.ocr : "builtin";
|
|
710
1011
|
const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, options.onProgress, options.tables !== false);
|
|
711
1012
|
if (ocrPageBlocks.size > 0) {
|
|
@@ -841,4 +1142,4 @@ export {
|
|
|
841
1142
|
parsePdfDocument,
|
|
842
1143
|
removeHeaderFooterBlocks
|
|
843
1144
|
};
|
|
844
|
-
//# sourceMappingURL=parser-
|
|
1145
|
+
//# sourceMappingURL=parser-GIREMEPD.js.map
|