kordoc 3.4.0 → 3.4.1
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/LICENSE +21 -21
- package/README.md +584 -584
- package/dist/{-56JCPGII.js → -YOQGQ22S.js} +5 -5
- package/dist/{chunk-MOL7MDBG.js → chunk-3TBUDJDE.js} +1 -1
- package/dist/chunk-3TBUDJDE.js.map +1 -0
- package/dist/{chunk-GE43BE46.js → chunk-AIQ3ISQU.js} +1 -1
- package/dist/chunk-AIQ3ISQU.js.map +1 -0
- package/dist/{chunk-MEPHGCPQ.js → chunk-CFQLK3LM.js} +1 -1
- package/dist/chunk-CFQLK3LM.js.map +1 -0
- package/dist/{chunk-YFNBHLLM.js → chunk-CGHPBI3X.js} +37 -24
- package/dist/chunk-CGHPBI3X.js.map +1 -0
- package/dist/{chunk-MHYDBT2Q.js → chunk-FPFB7AFX.js} +2 -2
- package/dist/chunk-FPFB7AFX.js.map +1 -0
- package/dist/chunk-GS7T56RP.cjs.map +1 -1
- package/dist/{chunk-ZPLRG2SJ.js → chunk-SE3KEL4G.js} +2 -2
- package/dist/chunk-SE3KEL4G.js.map +1 -0
- package/dist/{chunk-YYKQN7SA.cjs → chunk-W7JHJQ2O.cjs} +2 -2
- package/dist/chunk-W7JHJQ2O.cjs.map +1 -0
- package/dist/{chunk-DCZVOIEO.cjs → chunk-X34YWRL5.cjs} +1 -1
- package/dist/chunk-X34YWRL5.cjs.map +1 -0
- package/dist/cli.js +8 -8
- package/dist/cli.js.map +1 -1
- package/dist/{detect-RI2MQ33K.js → detect-NTR2FUEG.js} +2 -2
- package/dist/{formula-JCNF43NE.js → formula-KOQV353G.js} +1 -1
- package/dist/formula-KOQV353G.js.map +1 -0
- package/dist/{formula-RXVSQPXI.js → formula-MEDGYQXS.js} +1 -1
- package/dist/formula-MEDGYQXS.js.map +1 -0
- package/dist/{formula-5NKVS2LR.cjs → formula-ZYUDQHD3.cjs} +1 -1
- package/dist/formula-ZYUDQHD3.cjs.map +1 -0
- package/dist/index.cjs +206 -193
- package/dist/index.cjs.map +1 -1
- package/dist/index.js +35 -22
- package/dist/index.js.map +1 -1
- package/dist/mcp.js +6 -6
- package/dist/mcp.js.map +1 -1
- package/dist/page-range-CIRRJPXJ.js +7 -0
- package/dist/page-range-OF5I4PQY.js +8 -0
- package/dist/page-range-TPIRSA3J.cjs +8 -0
- package/dist/page-range-TPIRSA3J.cjs.map +1 -0
- package/dist/{parser-VZL3MK5X.js → parser-5HDZZWDH.js} +5 -5
- package/dist/parser-5HDZZWDH.js.map +1 -0
- package/dist/{parser-PQDICHD5.js → parser-ET64VEEX.js} +5 -5
- package/dist/parser-ET64VEEX.js.map +1 -0
- package/dist/{parser-XDW6I25C.cjs → parser-JAZCZJI6.cjs} +18 -18
- package/dist/parser-JAZCZJI6.cjs.map +1 -0
- package/dist/{provider-G4C2V2PD.cjs → provider-5K7O3Z2O.cjs} +1 -1
- package/dist/provider-5K7O3Z2O.cjs.map +1 -0
- package/dist/{provider-AKROB7WQ.js → provider-7H4CPZYS.js} +1 -1
- package/dist/provider-7H4CPZYS.js.map +1 -0
- package/dist/{provider-4ZJKV3DC.js → provider-H4WWSPOV.js} +1 -1
- package/dist/provider-H4WWSPOV.js.map +1 -0
- package/dist/{setup-57FB3LSP.js → setup-QSJ2INNS.js} +1 -1
- package/dist/setup-QSJ2INNS.js.map +1 -0
- package/dist/{watch-KERZCCNR.js → watch-56GNC2YN.js} +5 -5
- package/dist/watch-56GNC2YN.js.map +1 -0
- package/package.json +98 -98
- package/dist/chunk-DCZVOIEO.cjs.map +0 -1
- package/dist/chunk-GE43BE46.js.map +0 -1
- package/dist/chunk-MEPHGCPQ.js.map +0 -1
- package/dist/chunk-MHYDBT2Q.js.map +0 -1
- package/dist/chunk-MOL7MDBG.js.map +0 -1
- package/dist/chunk-YFNBHLLM.js.map +0 -1
- package/dist/chunk-YYKQN7SA.cjs.map +0 -1
- package/dist/chunk-ZPLRG2SJ.js.map +0 -1
- package/dist/formula-5NKVS2LR.cjs.map +0 -1
- package/dist/formula-JCNF43NE.js.map +0 -1
- package/dist/formula-RXVSQPXI.js.map +0 -1
- package/dist/page-range-737B4EZW.js +0 -8
- package/dist/page-range-LNMXJU6W.js +0 -7
- package/dist/page-range-P7SDW6LR.cjs +0 -8
- package/dist/page-range-P7SDW6LR.cjs.map +0 -1
- package/dist/parser-PQDICHD5.js.map +0 -1
- package/dist/parser-VZL3MK5X.js.map +0 -1
- package/dist/parser-XDW6I25C.cjs.map +0 -1
- package/dist/provider-4ZJKV3DC.js.map +0 -1
- package/dist/provider-AKROB7WQ.js.map +0 -1
- package/dist/provider-G4C2V2PD.cjs.map +0 -1
- package/dist/setup-57FB3LSP.js.map +0 -1
- package/dist/watch-KERZCCNR.js.map +0 -1
- /package/dist/{-56JCPGII.js.map → -YOQGQ22S.js.map} +0 -0
- /package/dist/{detect-RI2MQ33K.js.map → detect-NTR2FUEG.js.map} +0 -0
- /package/dist/{page-range-737B4EZW.js.map → page-range-CIRRJPXJ.js.map} +0 -0
- /package/dist/{page-range-LNMXJU6W.js.map → page-range-OF5I4PQY.js.map} +0 -0
package/dist/index.js
CHANGED
|
@@ -17,10 +17,10 @@ import {
|
|
|
17
17
|
sanitizeHref,
|
|
18
18
|
stripDtd,
|
|
19
19
|
toArrayBuffer
|
|
20
|
-
} from "./chunk-
|
|
20
|
+
} from "./chunk-SE3KEL4G.js";
|
|
21
21
|
import {
|
|
22
22
|
parsePageRange
|
|
23
|
-
} from "./chunk-
|
|
23
|
+
} from "./chunk-AIQ3ISQU.js";
|
|
24
24
|
|
|
25
25
|
// src/index.ts
|
|
26
26
|
import { readFile } from "fs/promises";
|
|
@@ -16907,7 +16907,7 @@ async function parseXlsxDocument(buffer, options) {
|
|
|
16907
16907
|
}
|
|
16908
16908
|
let pageFilter = null;
|
|
16909
16909
|
if (options?.pages) {
|
|
16910
|
-
const { parsePageRange: parsePageRange2 } = await import("./page-range-
|
|
16910
|
+
const { parsePageRange: parsePageRange2 } = await import("./page-range-CIRRJPXJ.js");
|
|
16911
16911
|
pageFilter = parsePageRange2(options.pages, sheets.length);
|
|
16912
16912
|
}
|
|
16913
16913
|
const blocks = [];
|
|
@@ -17545,7 +17545,7 @@ async function parseXlsDocument(buffer, options) {
|
|
|
17545
17545
|
const totalSheets = Math.min(globals.sheets.length, MAX_SHEETS2);
|
|
17546
17546
|
let pageFilter = null;
|
|
17547
17547
|
if (options?.pages) {
|
|
17548
|
-
const { parsePageRange: parsePageRange2 } = await import("./page-range-
|
|
17548
|
+
const { parsePageRange: parsePageRange2 } = await import("./page-range-CIRRJPXJ.js");
|
|
17549
17549
|
pageFilter = parsePageRange2(options.pages, totalSheets);
|
|
17550
17550
|
}
|
|
17551
17551
|
const allBlocks = [];
|
|
@@ -17968,20 +17968,32 @@ function isDisplayMath(el) {
|
|
|
17968
17968
|
|
|
17969
17969
|
// src/docx/parser.ts
|
|
17970
17970
|
var MAX_DECOMPRESS_SIZE4 = 100 * 1024 * 1024;
|
|
17971
|
-
function
|
|
17971
|
+
function matchesLocal(el, localName2) {
|
|
17972
|
+
return el.localName === localName2 || (el.tagName?.endsWith(`:${localName2}`) ?? false);
|
|
17973
|
+
}
|
|
17974
|
+
function effectiveChildElements(parent) {
|
|
17972
17975
|
const result = [];
|
|
17973
17976
|
const children = parent.childNodes;
|
|
17974
17977
|
for (let i = 0; i < children.length; i++) {
|
|
17975
17978
|
const node = children[i];
|
|
17976
|
-
if (node.nodeType
|
|
17977
|
-
|
|
17978
|
-
|
|
17979
|
-
|
|
17979
|
+
if (node.nodeType !== 1) continue;
|
|
17980
|
+
const el = node;
|
|
17981
|
+
if (matchesLocal(el, "sdt")) {
|
|
17982
|
+
for (let j = 0; j < el.childNodes.length; j++) {
|
|
17983
|
+
const c = el.childNodes[j];
|
|
17984
|
+
if (c.nodeType === 1 && matchesLocal(c, "sdtContent")) {
|
|
17985
|
+
result.push(...effectiveChildElements(c));
|
|
17986
|
+
}
|
|
17980
17987
|
}
|
|
17988
|
+
} else {
|
|
17989
|
+
result.push(el);
|
|
17981
17990
|
}
|
|
17982
17991
|
}
|
|
17983
17992
|
return result;
|
|
17984
17993
|
}
|
|
17994
|
+
function getChildElements(parent, localName2) {
|
|
17995
|
+
return effectiveChildElements(parent).filter((el) => matchesLocal(el, localName2));
|
|
17996
|
+
}
|
|
17985
17997
|
function findElements(parent, localName2) {
|
|
17986
17998
|
const result = [];
|
|
17987
17999
|
const walk = (node) => {
|
|
@@ -18372,11 +18384,8 @@ async function parseDocxDocument(buffer, options) {
|
|
|
18372
18384
|
}
|
|
18373
18385
|
const blocks = [];
|
|
18374
18386
|
const bodyEl = body[0];
|
|
18375
|
-
const
|
|
18376
|
-
for (
|
|
18377
|
-
const node = children[i];
|
|
18378
|
-
if (node.nodeType !== 1) continue;
|
|
18379
|
-
const el = node;
|
|
18387
|
+
const topLevel = effectiveChildElements(bodyEl);
|
|
18388
|
+
for (const el of topLevel) {
|
|
18380
18389
|
const localName2 = el.localName ?? el.tagName?.split(":").pop();
|
|
18381
18390
|
if (localName2 === "p") {
|
|
18382
18391
|
const block = parseParagraph2(el, styles, numbering, footnotes, rels);
|
|
@@ -18882,8 +18891,9 @@ function extractFromTable(table) {
|
|
|
18882
18891
|
if (table.cols >= 2) {
|
|
18883
18892
|
for (let r = 0; r < table.rows; r++) {
|
|
18884
18893
|
for (let c = 0; c < table.cols - 1; c++) {
|
|
18885
|
-
const labelCell = table.cells[r][c];
|
|
18886
|
-
const valueCell = table.cells[r][c + 1];
|
|
18894
|
+
const labelCell = table.cells[r]?.[c];
|
|
18895
|
+
const valueCell = table.cells[r]?.[c + 1];
|
|
18896
|
+
if (!labelCell || !valueCell) continue;
|
|
18887
18897
|
if (isLabelCell(labelCell.text)) {
|
|
18888
18898
|
fields.push({
|
|
18889
18899
|
label: labelCell.text.trim().replace(/[::]\s*$/, ""),
|
|
@@ -18904,8 +18914,8 @@ function extractFromTable(table) {
|
|
|
18904
18914
|
if (allLabels) {
|
|
18905
18915
|
for (let r = 1; r < table.rows; r++) {
|
|
18906
18916
|
for (let c = 0; c < table.cols; c++) {
|
|
18907
|
-
const label = headerRow[c]
|
|
18908
|
-
const value = table.cells[r][c]
|
|
18917
|
+
const label = headerRow[c]?.text.trim() ?? "";
|
|
18918
|
+
const value = table.cells[r]?.[c]?.text.trim() ?? "";
|
|
18909
18919
|
if (label && value) {
|
|
18910
18920
|
fields.push({ label, value, row: r, col: c });
|
|
18911
18921
|
}
|
|
@@ -19059,15 +19069,18 @@ function fillTable(table, values, filled, matchedLabels, patternFilledCells) {
|
|
|
19059
19069
|
if (!allLabels) return;
|
|
19060
19070
|
for (let r = 1; r < table.rows; r++) {
|
|
19061
19071
|
for (let c = 0; c < table.cols; c++) {
|
|
19062
|
-
const
|
|
19072
|
+
const headerCell = headerRow[c];
|
|
19073
|
+
const valueCell = table.cells[r]?.[c];
|
|
19074
|
+
if (!headerCell || !valueCell) continue;
|
|
19075
|
+
const headerLabel = normalizeLabel(headerCell.text);
|
|
19063
19076
|
const matchKey = findMatchingKey(headerLabel, values);
|
|
19064
19077
|
if (matchKey === void 0) continue;
|
|
19065
19078
|
if (matchedLabels.has(matchKey)) continue;
|
|
19066
19079
|
const newValue = values.get(matchKey);
|
|
19067
|
-
|
|
19080
|
+
valueCell.text = newValue;
|
|
19068
19081
|
matchedLabels.add(matchKey);
|
|
19069
19082
|
filled.push({
|
|
19070
|
-
label:
|
|
19083
|
+
label: headerCell.text.trim(),
|
|
19071
19084
|
value: newValue,
|
|
19072
19085
|
row: r,
|
|
19073
19086
|
col: c
|
|
@@ -23338,7 +23351,7 @@ async function parseHwp(buffer, options) {
|
|
|
23338
23351
|
async function parsePdf(buffer, options) {
|
|
23339
23352
|
let parsePdfDocument;
|
|
23340
23353
|
try {
|
|
23341
|
-
const mod = await import("./parser-
|
|
23354
|
+
const mod = await import("./parser-5HDZZWDH.js");
|
|
23342
23355
|
parsePdfDocument = mod.parsePdfDocument;
|
|
23343
23356
|
} catch {
|
|
23344
23357
|
return {
|