kordoc 4.2.1 → 4.2.2
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 +5 -0
- package/dist/{-BSO3CDQQ.js → -NFHVGTVM.js} +9 -9
- package/dist/{chunk-Z6LPEWIX.js → chunk-2L7MIKOM.js} +3 -3
- package/dist/chunk-2L7MIKOM.js.map +1 -0
- package/dist/{chunk-35U3MVA2.cjs → chunk-DLFZAU4X.cjs} +3 -3
- package/dist/{chunk-35U3MVA2.cjs.map → chunk-DLFZAU4X.cjs.map} +1 -1
- package/dist/{chunk-QSB2RTX5.js → chunk-DSL6EVFJ.js} +6 -4
- package/dist/chunk-DSL6EVFJ.js.map +1 -0
- package/dist/{chunk-R22KDUKW.cjs → chunk-EFF2JPPA.cjs} +9 -9
- package/dist/{chunk-R22KDUKW.cjs.map → chunk-EFF2JPPA.cjs.map} +1 -1
- package/dist/{chunk-BHILWIE6.js → chunk-FHYLLKIP.js} +3 -3
- package/dist/chunk-FHYLLKIP.js.map +1 -0
- package/dist/{chunk-PLTG6ZRF.js → chunk-GHMVNMUN.js} +3 -3
- package/dist/{chunk-HTCKGSSW.js → chunk-GUYIXVV7.js} +3 -3
- package/dist/chunk-GUYIXVV7.js.map +1 -0
- package/dist/{chunk-7IMNZ2BL.js → chunk-IXHBRMDB.js} +6 -4
- package/dist/chunk-IXHBRMDB.js.map +1 -0
- package/dist/{chunk-RBUBZ2TA.cjs → chunk-KKZATKQW.cjs} +3 -3
- package/dist/chunk-KKZATKQW.cjs.map +1 -0
- package/dist/{chunk-QOC7S5VE.js → chunk-L5B7PQOP.js} +2 -2
- package/dist/{chunk-WJSGYVFD.js → chunk-LUJN3VFB.js} +3 -3
- package/dist/{chunk-KRAYUNFM.js → chunk-MOOVF6CT.js} +2 -2
- package/dist/{chunk-BKBYTPSL.js → chunk-MYD3QLBL.js} +2 -2
- package/dist/{chunk-L5JWKIOT.js → chunk-OLDIDBJ2.js} +3 -3
- package/dist/{chunk-F5ECB53I.js → chunk-T4AWY45Q.js} +2 -2
- package/dist/{chunk-WWYAYZ5U.js → chunk-TXYQJXOV.js} +48 -14
- package/dist/chunk-TXYQJXOV.js.map +1 -0
- package/dist/{chunk-LSXXGUWA.cjs → chunk-WMU2KJHF.cjs} +8 -6
- package/dist/chunk-WMU2KJHF.cjs.map +1 -0
- package/dist/{chunk-EZHWPKCH.js → chunk-X77U36TV.js} +2 -2
- package/dist/{chunk-54WVWWT4.js → chunk-ZF7QL4IY.js} +1 -1
- package/dist/chunk-ZF7QL4IY.js.map +1 -0
- package/dist/chunks-N4CWHTJI.js +10 -0
- package/dist/cli.js +17 -17
- package/dist/{image-ocr-N5Z32R4U.cjs → image-ocr-3AFBDDJI.cjs} +9 -9
- package/dist/{image-ocr-N5Z32R4U.cjs.map → image-ocr-3AFBDDJI.cjs.map} +1 -1
- package/dist/{image-ocr-SLUEM3FA.js → image-ocr-S4REURLR.js} +5 -5
- package/dist/{image-ocr-DALT7PFQ.js → image-ocr-TGHAMIQL.js} +4 -4
- package/dist/index.cjs +446 -412
- package/dist/index.cjs.map +1 -1
- package/dist/index.d.cts +3 -0
- package/dist/index.d.ts +3 -0
- package/dist/index.js +44 -10
- package/dist/index.js.map +1 -1
- package/dist/mcp.js +15 -15
- package/dist/{parser-F5KY3YSD.js → parser-C3APVZGT.js} +6 -6
- package/dist/{parser-33TK5SMH.js → parser-KMC4YOZH.js} +5 -5
- package/dist/{parser-R4CXIXRN.cjs → parser-MRM6WICO.cjs} +24 -24
- package/dist/{parser-R4CXIXRN.cjs.map → parser-MRM6WICO.cjs.map} +1 -1
- package/dist/pdf-ocr-5BRMIJAT.cjs +13 -0
- package/dist/{pdf-ocr-WGWZDHMV.cjs.map → pdf-ocr-5BRMIJAT.cjs.map} +1 -1
- package/dist/pdf-ocr-UDTEQEWA.js +12 -0
- package/dist/{pdf-ocr-BXQZKU3N.js → pdf-ocr-UURMV6FD.js} +5 -5
- package/dist/{profile-io-CBYHKLFU.js → profile-io-RLQ62KS5.js} +3 -3
- package/dist/{rasterize-AWMM2CSW.js → rasterize-LBABF2WI.js} +2 -2
- package/dist/render-CYXQEHS7.js +10 -0
- package/dist/seal-SV6PFFYU.js +10 -0
- package/dist/{watch-HKXCISVO.js → watch-XEUEGOXW.js} +9 -9
- package/package.json +1 -1
- package/dist/chunk-54WVWWT4.js.map +0 -1
- package/dist/chunk-7IMNZ2BL.js.map +0 -1
- package/dist/chunk-BHILWIE6.js.map +0 -1
- package/dist/chunk-HTCKGSSW.js.map +0 -1
- package/dist/chunk-LSXXGUWA.cjs.map +0 -1
- package/dist/chunk-QSB2RTX5.js.map +0 -1
- package/dist/chunk-RBUBZ2TA.cjs.map +0 -1
- package/dist/chunk-WWYAYZ5U.js.map +0 -1
- package/dist/chunk-Z6LPEWIX.js.map +0 -1
- package/dist/chunks-LRFS5NJE.js +0 -10
- package/dist/pdf-ocr-34F5AQKK.js +0 -12
- package/dist/pdf-ocr-WGWZDHMV.cjs +0 -13
- package/dist/render-Y3IE2FIK.js +0 -10
- package/dist/seal-7RXFCRMT.js +0 -10
- /package/dist/{-BSO3CDQQ.js.map → -NFHVGTVM.js.map} +0 -0
- /package/dist/{chunk-PLTG6ZRF.js.map → chunk-GHMVNMUN.js.map} +0 -0
- /package/dist/{chunk-QOC7S5VE.js.map → chunk-L5B7PQOP.js.map} +0 -0
- /package/dist/{chunk-WJSGYVFD.js.map → chunk-LUJN3VFB.js.map} +0 -0
- /package/dist/{chunk-KRAYUNFM.js.map → chunk-MOOVF6CT.js.map} +0 -0
- /package/dist/{chunk-BKBYTPSL.js.map → chunk-MYD3QLBL.js.map} +0 -0
- /package/dist/{chunk-L5JWKIOT.js.map → chunk-OLDIDBJ2.js.map} +0 -0
- /package/dist/{chunk-F5ECB53I.js.map → chunk-T4AWY45Q.js.map} +0 -0
- /package/dist/{chunk-EZHWPKCH.js.map → chunk-X77U36TV.js.map} +0 -0
- /package/dist/{chunks-LRFS5NJE.js.map → chunks-N4CWHTJI.js.map} +0 -0
- /package/dist/{image-ocr-SLUEM3FA.js.map → image-ocr-S4REURLR.js.map} +0 -0
- /package/dist/{image-ocr-DALT7PFQ.js.map → image-ocr-TGHAMIQL.js.map} +0 -0
- /package/dist/{parser-F5KY3YSD.js.map → parser-C3APVZGT.js.map} +0 -0
- /package/dist/{parser-33TK5SMH.js.map → parser-KMC4YOZH.js.map} +0 -0
- /package/dist/{pdf-ocr-34F5AQKK.js.map → pdf-ocr-UDTEQEWA.js.map} +0 -0
- /package/dist/{pdf-ocr-BXQZKU3N.js.map → pdf-ocr-UURMV6FD.js.map} +0 -0
- /package/dist/{profile-io-CBYHKLFU.js.map → profile-io-RLQ62KS5.js.map} +0 -0
- /package/dist/{rasterize-AWMM2CSW.js.map → rasterize-LBABF2WI.js.map} +0 -0
- /package/dist/{render-Y3IE2FIK.js.map → render-CYXQEHS7.js.map} +0 -0
- /package/dist/{seal-7RXFCRMT.js.map → seal-SV6PFFYU.js.map} +0 -0
- /package/dist/{watch-HKXCISVO.js.map → watch-XEUEGOXW.js.map} +0 -0
|
@@ -6,7 +6,7 @@ import {
|
|
|
6
6
|
HEADING_RATIO_H1,
|
|
7
7
|
HEADING_RATIO_H2,
|
|
8
8
|
HEADING_RATIO_H3
|
|
9
|
-
} from "./chunk-
|
|
9
|
+
} from "./chunk-ZF7QL4IY.js";
|
|
10
10
|
import {
|
|
11
11
|
detectFormat,
|
|
12
12
|
detectOle2Format,
|
|
@@ -27,7 +27,7 @@ import {
|
|
|
27
27
|
paraTextPureT,
|
|
28
28
|
patchZipEntries,
|
|
29
29
|
scanSectionXml
|
|
30
|
-
} from "./chunk-
|
|
30
|
+
} from "./chunk-LUJN3VFB.js";
|
|
31
31
|
import {
|
|
32
32
|
MAX_COLS,
|
|
33
33
|
MAX_ROWS,
|
|
@@ -37,14 +37,14 @@ import {
|
|
|
37
37
|
dedupeRunningHeaders,
|
|
38
38
|
flattenLayoutTables,
|
|
39
39
|
mapPuaText
|
|
40
|
-
} from "./chunk-
|
|
40
|
+
} from "./chunk-DSL6EVFJ.js";
|
|
41
41
|
import {
|
|
42
42
|
SPACE_EM_FIXED,
|
|
43
43
|
charWidthEm1000,
|
|
44
44
|
fitRatioForFewerLines,
|
|
45
45
|
measureTextWidth,
|
|
46
46
|
simulateWrap
|
|
47
|
-
} from "./chunk-
|
|
47
|
+
} from "./chunk-GHMVNMUN.js";
|
|
48
48
|
import {
|
|
49
49
|
MAX_DECOMPRESS_SIZE,
|
|
50
50
|
MAX_XML_DEPTH,
|
|
@@ -56,7 +56,7 @@ import {
|
|
|
56
56
|
createXmlParser,
|
|
57
57
|
extractTextFromNode,
|
|
58
58
|
findChildByLocalName
|
|
59
|
-
} from "./chunk-
|
|
59
|
+
} from "./chunk-FHYLLKIP.js";
|
|
60
60
|
import {
|
|
61
61
|
KordocError,
|
|
62
62
|
classifyError,
|
|
@@ -68,7 +68,7 @@ import {
|
|
|
68
68
|
sanitizeHref,
|
|
69
69
|
stripDtd,
|
|
70
70
|
toArrayBuffer
|
|
71
|
-
} from "./chunk-
|
|
71
|
+
} from "./chunk-2L7MIKOM.js";
|
|
72
72
|
|
|
73
73
|
// src/index.ts
|
|
74
74
|
import { readFile } from "fs/promises";
|
|
@@ -184,6 +184,26 @@ function comResultToParseResult(pages, pageCount, warnings) {
|
|
|
184
184
|
}
|
|
185
185
|
|
|
186
186
|
// src/hwpx/styles.ts
|
|
187
|
+
function isRealStrikeShape(shape) {
|
|
188
|
+
switch (shape) {
|
|
189
|
+
case "SOLID":
|
|
190
|
+
case "DASH":
|
|
191
|
+
case "DOT":
|
|
192
|
+
case "DASH_DOT":
|
|
193
|
+
case "DASH_DOT_DOT":
|
|
194
|
+
case "LONG_DASH":
|
|
195
|
+
case "CIRCLE":
|
|
196
|
+
case "DOUBLE_SLIM":
|
|
197
|
+
case "SLIM_THICK":
|
|
198
|
+
case "THICK_SLIM":
|
|
199
|
+
case "SLIM_THICK_SLIM":
|
|
200
|
+
case "WAVE":
|
|
201
|
+
case "DOUBLE_WAVE":
|
|
202
|
+
return true;
|
|
203
|
+
default:
|
|
204
|
+
return false;
|
|
205
|
+
}
|
|
206
|
+
}
|
|
187
207
|
async function extractHwpxStyles(zip, decompressed) {
|
|
188
208
|
const result = {
|
|
189
209
|
charProperties: /* @__PURE__ */ new Map(),
|
|
@@ -248,6 +268,10 @@ function parseCharProperties(doc, map) {
|
|
|
248
268
|
const localTag = (k.tagName || "").replace(/^[^:]+:/, "");
|
|
249
269
|
if (localTag === "bold") prop.bold = true;
|
|
250
270
|
else if (localTag === "italic") prop.italic = true;
|
|
271
|
+
else if (localTag === "strikeout") {
|
|
272
|
+
const shape = k.getAttribute("shape") || "";
|
|
273
|
+
if (isRealStrikeShape(shape)) prop.strike = true;
|
|
274
|
+
}
|
|
251
275
|
}
|
|
252
276
|
const fontFaces = el.getElementsByTagName("*");
|
|
253
277
|
for (let j = 0; j < fontFaces.length; j++) {
|
|
@@ -2214,21 +2238,22 @@ function extractRunSpans(para2, ctx, mode, requireMixed) {
|
|
|
2214
2238
|
else {
|
|
2215
2239
|
if (cp?.bold) span.bold = true;
|
|
2216
2240
|
if (cp?.italic) span.italic = true;
|
|
2241
|
+
if (cp?.strike) span.strike = true;
|
|
2217
2242
|
}
|
|
2218
|
-
if (span.bold || span.italic || span.code) styled = true;
|
|
2243
|
+
if (span.bold || span.italic || span.strike || span.code) styled = true;
|
|
2219
2244
|
spans.push(span);
|
|
2220
2245
|
}
|
|
2221
2246
|
if (!styled || spans.length === 0) return null;
|
|
2222
2247
|
const merged = [];
|
|
2223
2248
|
for (const s of spans) {
|
|
2224
2249
|
const last = merged[merged.length - 1];
|
|
2225
|
-
if (last && !!last.bold === !!s.bold && !!last.italic === !!s.italic && !!last.code === !!s.code) {
|
|
2250
|
+
if (last && !!last.bold === !!s.bold && !!last.italic === !!s.italic && !!last.strike === !!s.strike && !!last.code === !!s.code) {
|
|
2226
2251
|
last.text += s.text;
|
|
2227
2252
|
} else {
|
|
2228
2253
|
merged.push(s);
|
|
2229
2254
|
}
|
|
2230
2255
|
}
|
|
2231
|
-
if (requireMixed && !merged.some((s) => !(s.bold || s.italic || s.code))) return null;
|
|
2256
|
+
if (requireMixed && !merged.some((s) => s.strike) && !merged.some((s) => !(s.bold || s.italic || s.code))) return null;
|
|
2232
2257
|
return merged;
|
|
2233
2258
|
}
|
|
2234
2259
|
function gongmunDepthFromIndent(para2, left, ctx) {
|
|
@@ -4647,7 +4672,10 @@ function parseParagraph(records, start, end, ctx) {
|
|
|
4647
4672
|
if (headingLevel > 0) block.level = headingLevel;
|
|
4648
4673
|
if (ctx.docInfo && charShapeIds.length > 0) {
|
|
4649
4674
|
const style = resolveCharStyle(charShapeIds, ctx.docInfo);
|
|
4650
|
-
if (style)
|
|
4675
|
+
if (style) {
|
|
4676
|
+
block.style = style;
|
|
4677
|
+
if (style.strike) block.text = headMarker ? `${headMarker} ~~${trimmed}~~` : `~~${trimmed}~~`;
|
|
4678
|
+
}
|
|
4651
4679
|
}
|
|
4652
4680
|
if (footnotes.length > 0) block.footnoteText = footnotes.join("; ");
|
|
4653
4681
|
blocks.push(block);
|
|
@@ -5038,7 +5066,13 @@ function resolveCharStyle(charShapeIds, docInfo) {
|
|
|
5038
5066
|
if (cs.fontSize > 0) style.fontSize = cs.fontSize / 10;
|
|
5039
5067
|
if (cs.attrFlags & 1) style.italic = true;
|
|
5040
5068
|
if (cs.attrFlags & 2) style.bold = true;
|
|
5041
|
-
|
|
5069
|
+
if (hasRealStrike(cs.attrFlags)) style.strike = true;
|
|
5070
|
+
return style.fontSize || style.bold || style.italic || style.strike ? style : void 0;
|
|
5071
|
+
}
|
|
5072
|
+
function hasRealStrike(attrFlags) {
|
|
5073
|
+
const strikeBits = attrFlags >> 18 & 7;
|
|
5074
|
+
const shapeId = attrFlags >> 26 & 15;
|
|
5075
|
+
return strikeBits !== 0 && shapeId <= 12;
|
|
5042
5076
|
}
|
|
5043
5077
|
|
|
5044
5078
|
// src/hwp3/parser.ts
|
|
@@ -27513,7 +27547,7 @@ async function parse(input, options) {
|
|
|
27513
27547
|
}
|
|
27514
27548
|
async function parseImage(buffer, options) {
|
|
27515
27549
|
try {
|
|
27516
|
-
const { parseImageDocument } = await import("./image-ocr-
|
|
27550
|
+
const { parseImageDocument } = await import("./image-ocr-S4REURLR.js");
|
|
27517
27551
|
const { blocks, warnings } = await parseImageDocument(buffer, options);
|
|
27518
27552
|
return {
|
|
27519
27553
|
success: true,
|
|
@@ -27571,7 +27605,7 @@ async function parseHwp(buffer, options) {
|
|
|
27571
27605
|
async function parsePdf(buffer, options) {
|
|
27572
27606
|
let parsePdfDocument;
|
|
27573
27607
|
try {
|
|
27574
|
-
const mod = await import("./parser-
|
|
27608
|
+
const mod = await import("./parser-C3APVZGT.js");
|
|
27575
27609
|
parsePdfDocument = mod.parsePdfDocument;
|
|
27576
27610
|
} catch {
|
|
27577
27611
|
return {
|
|
@@ -27705,4 +27739,4 @@ export {
|
|
|
27705
27739
|
parseHwpml,
|
|
27706
27740
|
fillForm
|
|
27707
27741
|
};
|
|
27708
|
-
//# sourceMappingURL=chunk-
|
|
27742
|
+
//# sourceMappingURL=chunk-TXYQJXOV.js.map
|