kordoc 4.16.1 → 4.16.2
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 +6 -7
- package/dist/{-3DURLXSO.js → -4G6MA4X6.js} +11 -11
- package/dist/chunk-3IB7LBHC.js +57 -0
- package/dist/chunk-3IB7LBHC.js.map +1 -0
- package/dist/{chunk-EEVAPRPD.js → chunk-4VODJKFL.js} +3 -3
- package/dist/{chunk-NCXRSQNE.js → chunk-5KS5PWJR.js} +2 -2
- package/dist/chunk-62W7QCEX.cjs +57 -0
- package/dist/chunk-62W7QCEX.cjs.map +1 -0
- package/dist/{chunk-ANZTLTKU.js → chunk-7KNGSQU7.js} +4 -4
- package/dist/chunk-7KNGSQU7.js.map +1 -0
- package/dist/{chunk-63YGGAQW.js → chunk-AUWGDP4G.js} +2 -2
- package/dist/{chunk-2BZFKIRB.cjs → chunk-CEIATRDW.cjs} +3 -3
- package/dist/{chunk-2BZFKIRB.cjs.map → chunk-CEIATRDW.cjs.map} +1 -1
- package/dist/{chunk-BIHS4VWR.js → chunk-DES2LGFZ.js} +2 -2
- package/dist/{chunk-N65ZXWNQ.js → chunk-F4NRLLY4.js} +2 -2
- package/dist/{chunk-NV37CQFN.cjs → chunk-FKUDAIJI.cjs} +12 -12
- package/dist/{chunk-NV37CQFN.cjs.map → chunk-FKUDAIJI.cjs.map} +1 -1
- package/dist/{chunk-VCRBR72A.js → chunk-GO5IFOUF.js} +2 -2
- package/dist/{chunk-BTO7ZEE4.js → chunk-IQGIT2VX.js} +37 -7
- package/dist/chunk-IQGIT2VX.js.map +1 -0
- package/dist/chunk-J3IF6KGW.js +59 -0
- package/dist/chunk-J3IF6KGW.js.map +1 -0
- package/dist/{chunk-VNOQTDTE.js → chunk-JELRWAZC.js} +2 -2
- package/dist/{chunk-CXWWX6BQ.js → chunk-JFLAM3MM.js} +2 -2
- package/dist/{chunk-7XAGIPOX.js → chunk-KIOQ4N37.js} +2 -2
- package/dist/{chunk-UZHT3LBF.js → chunk-LEIAEWZ7.js} +4 -4
- package/dist/chunk-LEIAEWZ7.js.map +1 -0
- package/dist/{chunk-NUPXPKIQ.js → chunk-MLIS253L.js} +2 -2
- package/dist/{chunk-DJ2BBQ5P.js → chunk-OXLPVQYK.js} +3 -3
- package/dist/{chunk-3VLVNKJV.js → chunk-P6ZDXTV6.js} +2 -2
- package/dist/{chunk-OBRPYCIS.js → chunk-QOCS7W7R.js} +58 -23
- package/dist/chunk-QOCS7W7R.js.map +1 -0
- package/dist/{chunk-24N4IJ5I.js → chunk-S2PL227Q.js} +2 -2
- package/dist/{chunk-PVMROO3A.js → chunk-SFEQYKYJ.js} +86 -94
- package/dist/chunk-SFEQYKYJ.js.map +1 -0
- package/dist/{chunk-YRRBA4HW.cjs → chunk-STM3ORY7.cjs} +2 -2
- package/dist/{chunk-YRRBA4HW.cjs.map → chunk-STM3ORY7.cjs.map} +1 -1
- package/dist/{chunk-7FKPPIDZ.js → chunk-TXUKTOG2.js} +166 -14
- package/dist/chunk-TXUKTOG2.js.map +1 -0
- package/dist/{chunk-3WIQSM3Z.js → chunk-UTP2WCHS.js} +4 -4
- package/dist/{chunk-7IRHSRPU.cjs → chunk-WDQTFAXY.cjs} +97 -25
- package/dist/chunk-WDQTFAXY.cjs.map +1 -0
- package/dist/{chunk-D2QCXK2X.cjs → chunk-XTFN5WKS.cjs} +7 -7
- package/dist/chunk-XTFN5WKS.cjs.map +1 -0
- package/dist/chunks-NUNIKS2J.js +10 -0
- package/dist/cli.js +28 -27
- package/dist/cli.js.map +1 -1
- package/dist/{formula-IPLFHEBD.js → formula-EPKLPDU7.js} +3 -3
- package/dist/{formula-U4BTEZ6I.js → formula-NARUOXRX.js} +4 -4
- package/dist/{formula-TRTONZMA.cjs → formula-WIMV5TUX.cjs} +9 -9
- package/dist/{formula-TRTONZMA.cjs.map → formula-WIMV5TUX.cjs.map} +1 -1
- package/dist/{image-ocr-3IJ25PS4.js → image-ocr-3WJ4QH3T.js} +8 -8
- package/dist/{image-ocr-GTDG4SAJ.js → image-ocr-5BUJQ4TK.js} +6 -6
- package/dist/{image-ocr-THFLQ3DC.cjs → image-ocr-AXWXR6JV.cjs} +15 -15
- package/dist/{image-ocr-THFLQ3DC.cjs.map → image-ocr-AXWXR6JV.cjs.map} +1 -1
- package/dist/index.cjs +723 -659
- package/dist/index.cjs.map +1 -1
- package/dist/index.d.cts +7 -1
- package/dist/index.d.ts +7 -1
- package/dist/index.js +82 -18
- package/dist/index.js.map +1 -1
- package/dist/mcp.js +25 -23
- package/dist/mcp.js.map +1 -1
- package/dist/{model-bundle-PVWIAVEX.js → model-bundle-IMBW7D2Z.js} +5 -5
- package/dist/{models-VCRF3JMT.js → models-SY75IBOE.js} +5 -5
- package/dist/{parser-2CN2DIPV.cjs → parser-2LU4EVU3.cjs} +86 -83
- package/dist/parser-2LU4EVU3.cjs.map +1 -0
- package/dist/{parser-T4RITZDP.js → parser-N6476XAC.js} +15 -12
- package/dist/{parser-UEROAPYZ.js.map → parser-N6476XAC.js.map} +1 -1
- package/dist/{parser-UEROAPYZ.js → parser-ZQM3664Z.js} +12 -9
- package/dist/{parser-T4RITZDP.js.map → parser-ZQM3664Z.js.map} +1 -1
- package/dist/pdf-ocr-6DEE23IP.js +15 -0
- package/dist/pdf-ocr-GHEJVPX2.cjs +16 -0
- package/dist/{pdf-ocr-NARV2AQI.cjs.map → pdf-ocr-GHEJVPX2.cjs.map} +1 -1
- package/dist/pdf-ocr-MQTWODVU.js +18 -0
- package/dist/{profile-io-YD54CDXY.js → profile-io-GJPKCYPT.js} +3 -3
- package/dist/{rasterize-3THYOOUN.js → rasterize-UOALJQDS.js} +3 -3
- package/dist/{redact-doc-65L2YJN6.js → redact-doc-22SYW3KH.js} +11 -11
- package/dist/{render-YU7QJVZ7.js → render-MJ4H3R23.js} +7 -7
- package/dist/seal-XWMRUTI4.js +10 -0
- package/dist/{setup-CFHKXDY4.js → setup-2URYVUEO.js} +3 -3
- package/dist/visual-USPS4QRC.js +20 -0
- package/dist/{watch-D6FCAYKJ.js → watch-ALXZGTOB.js} +12 -12
- package/package.json +1 -1
- package/dist/chunk-7FKPPIDZ.js.map +0 -1
- package/dist/chunk-7IRHSRPU.cjs.map +0 -1
- package/dist/chunk-ANZTLTKU.js.map +0 -1
- package/dist/chunk-BTO7ZEE4.js.map +0 -1
- package/dist/chunk-D2QCXK2X.cjs.map +0 -1
- package/dist/chunk-DO74GYSM.cjs +0 -11
- package/dist/chunk-DO74GYSM.cjs.map +0 -1
- package/dist/chunk-KBAQLAEL.js +0 -11
- package/dist/chunk-KBAQLAEL.js.map +0 -1
- package/dist/chunk-OBRPYCIS.js.map +0 -1
- package/dist/chunk-PVMROO3A.js.map +0 -1
- package/dist/chunk-T77M2QP7.js +0 -13
- package/dist/chunk-T77M2QP7.js.map +0 -1
- package/dist/chunk-UZHT3LBF.js.map +0 -1
- package/dist/chunks-PYYQURNO.js +0 -10
- package/dist/parser-2CN2DIPV.cjs.map +0 -1
- package/dist/pdf-ocr-NARV2AQI.cjs +0 -16
- package/dist/pdf-ocr-SYLMM727.js +0 -18
- package/dist/pdf-ocr-YMQZDJBO.js +0 -15
- package/dist/seal-6CQGFTML.js +0 -10
- package/dist/visual-ED2H75Q5.js +0 -20
- /package/dist/{-3DURLXSO.js.map → -4G6MA4X6.js.map} +0 -0
- /package/dist/{chunk-EEVAPRPD.js.map → chunk-4VODJKFL.js.map} +0 -0
- /package/dist/{chunk-NCXRSQNE.js.map → chunk-5KS5PWJR.js.map} +0 -0
- /package/dist/{chunk-63YGGAQW.js.map → chunk-AUWGDP4G.js.map} +0 -0
- /package/dist/{chunk-BIHS4VWR.js.map → chunk-DES2LGFZ.js.map} +0 -0
- /package/dist/{chunk-N65ZXWNQ.js.map → chunk-F4NRLLY4.js.map} +0 -0
- /package/dist/{chunk-VCRBR72A.js.map → chunk-GO5IFOUF.js.map} +0 -0
- /package/dist/{chunk-VNOQTDTE.js.map → chunk-JELRWAZC.js.map} +0 -0
- /package/dist/{chunk-CXWWX6BQ.js.map → chunk-JFLAM3MM.js.map} +0 -0
- /package/dist/{chunk-7XAGIPOX.js.map → chunk-KIOQ4N37.js.map} +0 -0
- /package/dist/{chunk-NUPXPKIQ.js.map → chunk-MLIS253L.js.map} +0 -0
- /package/dist/{chunk-DJ2BBQ5P.js.map → chunk-OXLPVQYK.js.map} +0 -0
- /package/dist/{chunk-3VLVNKJV.js.map → chunk-P6ZDXTV6.js.map} +0 -0
- /package/dist/{chunk-24N4IJ5I.js.map → chunk-S2PL227Q.js.map} +0 -0
- /package/dist/{chunk-3WIQSM3Z.js.map → chunk-UTP2WCHS.js.map} +0 -0
- /package/dist/{chunks-PYYQURNO.js.map → chunks-NUNIKS2J.js.map} +0 -0
- /package/dist/{formula-IPLFHEBD.js.map → formula-EPKLPDU7.js.map} +0 -0
- /package/dist/{formula-U4BTEZ6I.js.map → formula-NARUOXRX.js.map} +0 -0
- /package/dist/{image-ocr-3IJ25PS4.js.map → image-ocr-3WJ4QH3T.js.map} +0 -0
- /package/dist/{image-ocr-GTDG4SAJ.js.map → image-ocr-5BUJQ4TK.js.map} +0 -0
- /package/dist/{model-bundle-PVWIAVEX.js.map → model-bundle-IMBW7D2Z.js.map} +0 -0
- /package/dist/{models-VCRF3JMT.js.map → models-SY75IBOE.js.map} +0 -0
- /package/dist/{pdf-ocr-SYLMM727.js.map → pdf-ocr-6DEE23IP.js.map} +0 -0
- /package/dist/{pdf-ocr-YMQZDJBO.js.map → pdf-ocr-MQTWODVU.js.map} +0 -0
- /package/dist/{profile-io-YD54CDXY.js.map → profile-io-GJPKCYPT.js.map} +0 -0
- /package/dist/{rasterize-3THYOOUN.js.map → rasterize-UOALJQDS.js.map} +0 -0
- /package/dist/{redact-doc-65L2YJN6.js.map → redact-doc-22SYW3KH.js.map} +0 -0
- /package/dist/{render-YU7QJVZ7.js.map → render-MJ4H3R23.js.map} +0 -0
- /package/dist/{seal-6CQGFTML.js.map → seal-XWMRUTI4.js.map} +0 -0
- /package/dist/{setup-CFHKXDY4.js.map → setup-2URYVUEO.js.map} +0 -0
- /package/dist/{visual-ED2H75Q5.js.map → visual-USPS4QRC.js.map} +0 -0
- /package/dist/{watch-D6FCAYKJ.js.map → watch-ALXZGTOB.js.map} +0 -0
|
@@ -4,7 +4,7 @@ import {
|
|
|
4
4
|
blocksToPages,
|
|
5
5
|
encodePng,
|
|
6
6
|
escapeLiteralDollar
|
|
7
|
-
} from "./chunk-
|
|
7
|
+
} from "./chunk-7KNGSQU7.js";
|
|
8
8
|
import {
|
|
9
9
|
hasRequestedPagesAfter,
|
|
10
10
|
parsePageRange
|
|
@@ -44,13 +44,15 @@ import {
|
|
|
44
44
|
startsNewItem,
|
|
45
45
|
trimTrailingEmptyTableCols,
|
|
46
46
|
wrapJoiner
|
|
47
|
-
} from "./chunk-
|
|
48
|
-
import
|
|
49
|
-
|
|
47
|
+
} from "./chunk-TXUKTOG2.js";
|
|
48
|
+
import {
|
|
49
|
+
stripScriptTags
|
|
50
|
+
} from "./chunk-3IB7LBHC.js";
|
|
51
|
+
import "./chunk-F4NRLLY4.js";
|
|
50
52
|
import {
|
|
51
53
|
KordocError,
|
|
52
54
|
sanitizeHref
|
|
53
|
-
} from "./chunk-
|
|
55
|
+
} from "./chunk-MLIS253L.js";
|
|
54
56
|
|
|
55
57
|
// src/pdf/ocr-region-merge.ts
|
|
56
58
|
function mergeOcrImageRegions(blocks, page, regions, ocrBlocks) {
|
|
@@ -2417,7 +2419,7 @@ function cleanPdfText(text, opts) {
|
|
|
2417
2419
|
}
|
|
2418
2420
|
const mark = /^(?:#{1,6}|-) /.exec(line)?.[0] ?? "";
|
|
2419
2421
|
return mark + collapseEvenSpacing(line.slice(mark.length), false);
|
|
2420
|
-
}).replace(/\\~\\~/g, "~~").replace(/~~~~/g, "").replace(/<u>\s*<\/u>/g, "").replace(/\n{3,}/g, "\n\n").trim();
|
|
2422
|
+
}).replace(/\\~\\~/g, "~~").replace(/~~~~/g, "").replace(/<u>\s*<\/u>/g, "").replace(/<(sup|sub)>\s*<\/\1>/g, "").replace(/\n{3,}/g, "\n\n").trim();
|
|
2421
2423
|
}
|
|
2422
2424
|
function startsWithMarker(line) {
|
|
2423
2425
|
const t = line.trimStart();
|
|
@@ -2505,7 +2507,7 @@ function mergeLinkRuns(markdown) {
|
|
|
2505
2507
|
|
|
2506
2508
|
// src/pdf/formula-ocr.ts
|
|
2507
2509
|
async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
|
|
2508
|
-
const formulaMod = await import("./formula-
|
|
2510
|
+
const formulaMod = await import("./formula-EPKLPDU7.js");
|
|
2509
2511
|
const { FormulaPipeline, ensureFormulaModels } = formulaMod;
|
|
2510
2512
|
await ensureFormulaModels((p) => {
|
|
2511
2513
|
if (p.phase === "download" && p.total) {
|
|
@@ -2888,10 +2890,11 @@ async function parsePdfDocument(buffer, options) {
|
|
|
2888
2890
|
}
|
|
2889
2891
|
if (targets.size > 0) {
|
|
2890
2892
|
try {
|
|
2891
|
-
const { runPdfOcr } = await import("./pdf-ocr-
|
|
2893
|
+
const { runPdfOcr } = await import("./pdf-ocr-6DEE23IP.js");
|
|
2892
2894
|
const mode = typeof options?.ocr === "function" ? options.ocr : "builtin";
|
|
2893
2895
|
const regionPages = new Map([...ocrRegions].filter(([p]) => options?.ocr !== "force" && !isImageBased && !pageQuality.find((q) => q.page === p)?.needsOcr));
|
|
2894
2896
|
const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, options?.onProgress, options?.tables !== false, vectorPageOps, regionPages);
|
|
2897
|
+
for (const obs of ocrPageBlocks.values()) stripScriptTags({ blocks: obs });
|
|
2895
2898
|
if (ocrPageBlocks.size > 0) {
|
|
2896
2899
|
const replacePages = /* @__PURE__ */ new Set();
|
|
2897
2900
|
for (const [p, obs] of ocrPageBlocks) {
|
|
@@ -3087,4 +3090,4 @@ export {
|
|
|
3087
3090
|
parsePdfDocument,
|
|
3088
3091
|
removeHeaderFooterBlocks
|
|
3089
3092
|
};
|
|
3090
|
-
//# sourceMappingURL=parser-
|
|
3093
|
+
//# sourceMappingURL=parser-ZQM3664Z.js.map
|