@cesarandreslopez/occ 0.3.1 → 0.4.1
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/README.md +130 -4
- package/dist/src/cli-validation.d.ts +14 -0
- package/dist/src/cli-validation.js +58 -0
- package/dist/src/cli-validation.js.map +1 -0
- package/dist/src/cli.js +26 -8
- package/dist/src/cli.js.map +1 -1
- package/dist/src/code/build.d.ts +7 -5
- package/dist/src/code/build.js +68 -11
- package/dist/src/code/build.js.map +1 -1
- package/dist/src/code/command.js +26 -19
- package/dist/src/code/command.js.map +1 -1
- package/dist/src/code/discover.d.ts +6 -4
- package/dist/src/code/discover.js +5 -0
- package/dist/src/code/discover.js.map +1 -1
- package/dist/src/code/languages.d.ts +28 -7
- package/dist/src/code/languages.js +9 -0
- package/dist/src/code/languages.js.map +1 -1
- package/dist/src/code/output.js +12 -3
- package/dist/src/code/output.js.map +1 -1
- package/dist/src/code/parsers.js +75 -10
- package/dist/src/code/parsers.js.map +1 -1
- package/dist/src/code/query.js +82 -22
- package/dist/src/code/query.js.map +1 -1
- package/dist/src/code/types.d.ts +1201 -145
- package/dist/src/code/types.js +150 -1
- package/dist/src/code/types.js.map +1 -1
- package/dist/src/doc/command.d.ts +2 -0
- package/dist/src/doc/command.js +54 -0
- package/dist/src/doc/command.js.map +1 -0
- package/dist/src/doc/inspect-docx.d.ts +14 -0
- package/dist/src/doc/inspect-docx.js +188 -0
- package/dist/src/doc/inspect-docx.js.map +1 -0
- package/dist/src/doc/inspect-odt.d.ts +8 -0
- package/dist/src/doc/inspect-odt.js +92 -0
- package/dist/src/doc/inspect-odt.js.map +1 -0
- package/dist/src/doc/inspect-pdf.d.ts +8 -0
- package/dist/src/doc/inspect-pdf.js +88 -0
- package/dist/src/doc/inspect-pdf.js.map +1 -0
- package/dist/src/doc/inspect.d.ts +2 -0
- package/dist/src/doc/inspect.js +70 -0
- package/dist/src/doc/inspect.js.map +1 -0
- package/dist/src/doc/output.d.ts +3 -0
- package/dist/src/doc/output.js +82 -0
- package/dist/src/doc/output.js.map +1 -0
- package/dist/src/doc/types.d.ts +172 -0
- package/dist/src/doc/types.js +62 -0
- package/dist/src/doc/types.js.map +1 -0
- package/dist/src/inspect/shared.d.ts +20 -0
- package/dist/src/inspect/shared.js +29 -0
- package/dist/src/inspect/shared.js.map +1 -0
- package/dist/src/inspect/xlsx-cells.d.ts +4 -0
- package/dist/src/inspect/xlsx-cells.js +31 -0
- package/dist/src/inspect/xlsx-cells.js.map +1 -0
- package/dist/src/markdown/convert.js +6 -4
- package/dist/src/markdown/convert.js.map +1 -1
- package/dist/src/output/tabular.d.ts +6 -4
- package/dist/src/output/tabular.js +5 -0
- package/dist/src/output/tabular.js.map +1 -1
- package/dist/src/output/tree.d.ts +21 -9
- package/dist/src/output/tree.js +10 -0
- package/dist/src/output/tree.js.map +1 -1
- package/dist/src/parsers/docx.js +5 -1
- package/dist/src/parsers/docx.js.map +1 -1
- package/dist/src/parsers/odf.js +4 -3
- package/dist/src/parsers/odf.js.map +1 -1
- package/dist/src/parsers/pdf.js +6 -1
- package/dist/src/parsers/pdf.js.map +1 -1
- package/dist/src/parsers/pptx.js +2 -1
- package/dist/src/parsers/pptx.js.map +1 -1
- package/dist/src/parsers/xlsx.js +11 -4
- package/dist/src/parsers/xlsx.js.map +1 -1
- package/dist/src/progress.d.ts +7 -5
- package/dist/src/progress.js +6 -0
- package/dist/src/progress.js.map +1 -1
- package/dist/src/scc.d.ts +35 -26
- package/dist/src/scc.js +26 -1
- package/dist/src/scc.js.map +1 -1
- package/dist/src/sheet/command.d.ts +2 -0
- package/dist/src/sheet/command.js +61 -0
- package/dist/src/sheet/command.js.map +1 -0
- package/dist/src/sheet/inspect.d.ts +3 -0
- package/dist/src/sheet/inspect.js +426 -0
- package/dist/src/sheet/inspect.js.map +1 -0
- package/dist/src/sheet/output.d.ts +3 -0
- package/dist/src/sheet/output.js +158 -0
- package/dist/src/sheet/output.js.map +1 -0
- package/dist/src/sheet/types.d.ts +472 -0
- package/dist/src/sheet/types.js +119 -0
- package/dist/src/sheet/types.js.map +1 -0
- package/dist/src/slide/command.d.ts +2 -0
- package/dist/src/slide/command.js +54 -0
- package/dist/src/slide/command.js.map +1 -0
- package/dist/src/slide/inspect-odp.d.ts +7 -0
- package/dist/src/slide/inspect-odp.js +105 -0
- package/dist/src/slide/inspect-odp.js.map +1 -0
- package/dist/src/slide/inspect-pptx.d.ts +11 -0
- package/dist/src/slide/inspect-pptx.js +157 -0
- package/dist/src/slide/inspect-pptx.js.map +1 -0
- package/dist/src/slide/inspect.d.ts +2 -0
- package/dist/src/slide/inspect.js +76 -0
- package/dist/src/slide/inspect.js.map +1 -0
- package/dist/src/slide/output.d.ts +3 -0
- package/dist/src/slide/output.js +92 -0
- package/dist/src/slide/output.js.map +1 -0
- package/dist/src/slide/types.d.ts +167 -0
- package/dist/src/slide/types.js +61 -0
- package/dist/src/slide/types.js.map +1 -0
- package/dist/src/stats.d.ts +6 -4
- package/dist/src/stats.js +5 -0
- package/dist/src/stats.js.map +1 -1
- package/dist/src/structure/types.d.ts +19 -11
- package/dist/src/structure/types.js +26 -21
- package/dist/src/structure/types.js.map +1 -1
- package/dist/src/table/command.d.ts +2 -0
- package/dist/src/table/command.js +58 -0
- package/dist/src/table/command.js.map +1 -0
- package/dist/src/table/inspect-docx.d.ts +2 -0
- package/dist/src/table/inspect-docx.js +108 -0
- package/dist/src/table/inspect-docx.js.map +1 -0
- package/dist/src/table/inspect-odp.d.ts +2 -0
- package/dist/src/table/inspect-odp.js +106 -0
- package/dist/src/table/inspect-odp.js.map +1 -0
- package/dist/src/table/inspect-odt.d.ts +2 -0
- package/dist/src/table/inspect-odt.js +100 -0
- package/dist/src/table/inspect-odt.js.map +1 -0
- package/dist/src/table/inspect-pptx.d.ts +2 -0
- package/dist/src/table/inspect-pptx.js +112 -0
- package/dist/src/table/inspect-pptx.js.map +1 -0
- package/dist/src/table/inspect-xlsx.d.ts +2 -0
- package/dist/src/table/inspect-xlsx.js +161 -0
- package/dist/src/table/inspect-xlsx.js.map +1 -0
- package/dist/src/table/inspect.d.ts +2 -0
- package/dist/src/table/inspect.js +65 -0
- package/dist/src/table/inspect.js.map +1 -0
- package/dist/src/table/output.d.ts +3 -0
- package/dist/src/table/output.js +93 -0
- package/dist/src/table/output.js.map +1 -0
- package/dist/src/table/types.d.ts +104 -0
- package/dist/src/table/types.js +40 -0
- package/dist/src/table/types.js.map +1 -0
- package/dist/src/types.d.ts +25 -20
- package/dist/src/types.js +21 -1
- package/dist/src/types.js.map +1 -1
- package/dist/src/utils.d.ts +11 -1
- package/dist/src/utils.js +2 -0
- package/dist/src/utils.js.map +1 -1
- package/dist/src/walker.d.ts +21 -12
- package/dist/src/walker.js +13 -0
- package/dist/src/walker.js.map +1 -1
- package/package.json +3 -2
|
@@ -0,0 +1,106 @@
|
|
|
1
|
+
import JSZip from 'jszip';
|
|
2
|
+
import { estimateTokens } from '../inspect/shared.js';
|
|
3
|
+
function stripXmlTags(xml) {
|
|
4
|
+
return xml.replace(/<[^>]*>/g, '').trim();
|
|
5
|
+
}
|
|
6
|
+
function extractIntAttribute(tag, attr) {
|
|
7
|
+
const match = tag.match(new RegExp(`${attr}\\s*=\\s*"(\\d+)"`, 'i'));
|
|
8
|
+
if (match) {
|
|
9
|
+
const val = parseInt(match[1], 10);
|
|
10
|
+
return val > 1 ? val : undefined;
|
|
11
|
+
}
|
|
12
|
+
return undefined;
|
|
13
|
+
}
|
|
14
|
+
function detectHeaders(firstRowCells, options) {
|
|
15
|
+
if (options.headerRow === 'none')
|
|
16
|
+
return null;
|
|
17
|
+
if (typeof options.headerRow === 'number' && options.headerRow === 1) {
|
|
18
|
+
return firstRowCells.map(c => c.value);
|
|
19
|
+
}
|
|
20
|
+
if (options.headerRow === 'auto') {
|
|
21
|
+
const values = firstRowCells.map(c => c.value).filter(v => v.length > 0);
|
|
22
|
+
if (values.length > 0 && values.length === firstRowCells.length && new Set(values).size === values.length) {
|
|
23
|
+
return firstRowCells.map(c => c.value);
|
|
24
|
+
}
|
|
25
|
+
}
|
|
26
|
+
return null;
|
|
27
|
+
}
|
|
28
|
+
export async function extractOdpTables(buffer, options) {
|
|
29
|
+
const zip = await JSZip.loadAsync(buffer);
|
|
30
|
+
const contentXml = await zip.file('content.xml')?.async('text');
|
|
31
|
+
if (!contentXml)
|
|
32
|
+
return [];
|
|
33
|
+
// Split by draw:page to get slide context
|
|
34
|
+
const pageMatches = [...contentXml.matchAll(/<draw:page\b([^>]*)>([\s\S]*?)<\/draw:page>/gi)];
|
|
35
|
+
const tables = [];
|
|
36
|
+
let globalTableIndex = 0;
|
|
37
|
+
for (let pi = 0; pi < pageMatches.length; pi++) {
|
|
38
|
+
const pageAttrs = pageMatches[pi][1];
|
|
39
|
+
const pageXml = pageMatches[pi][2];
|
|
40
|
+
// Try to get page name
|
|
41
|
+
const nameMatch = pageAttrs.match(/draw:name\s*=\s*"([^"]*)"/i);
|
|
42
|
+
const pageName = nameMatch?.[1];
|
|
43
|
+
const location = pageName ? `Slide: ${pageName}` : `Slide ${pi + 1}`;
|
|
44
|
+
const tableMatches = [...pageXml.matchAll(/<table:table\b[^>]*>([\s\S]*?)<\/table:table>/gi)];
|
|
45
|
+
for (const tableMatch of tableMatches) {
|
|
46
|
+
globalTableIndex++;
|
|
47
|
+
const tableXml = tableMatch[1];
|
|
48
|
+
const rowMatches = [...tableXml.matchAll(/<table:table-row\b[^>]*>([\s\S]*?)<\/table:table-row>/gi)];
|
|
49
|
+
if (rowMatches.length === 0)
|
|
50
|
+
continue;
|
|
51
|
+
const allRows = [];
|
|
52
|
+
for (const rowMatch of rowMatches) {
|
|
53
|
+
const rowXml = rowMatch[1];
|
|
54
|
+
const cellMatches = [...rowXml.matchAll(/<table:table-cell\b([^>]*?)>([\s\S]*?)<\/table:table-cell>/gi)];
|
|
55
|
+
const cells = [];
|
|
56
|
+
for (const cellMatch of cellMatches) {
|
|
57
|
+
const attrs = cellMatch[1];
|
|
58
|
+
const cellXml = cellMatch[2];
|
|
59
|
+
const value = stripXmlTags(cellXml);
|
|
60
|
+
const cell = { value };
|
|
61
|
+
const colSpan = extractIntAttribute(attrs, 'table:number-columns-spanned');
|
|
62
|
+
const rowSpan = extractIntAttribute(attrs, 'table:number-rows-spanned');
|
|
63
|
+
if (colSpan)
|
|
64
|
+
cell.colSpan = colSpan;
|
|
65
|
+
if (rowSpan)
|
|
66
|
+
cell.rowSpan = rowSpan;
|
|
67
|
+
cells.push(cell);
|
|
68
|
+
}
|
|
69
|
+
allRows.push(cells);
|
|
70
|
+
}
|
|
71
|
+
if (allRows.length === 0)
|
|
72
|
+
continue;
|
|
73
|
+
const headers = detectHeaders(allRows[0], options);
|
|
74
|
+
const dataStartIndex = headers ? 1 : 0;
|
|
75
|
+
const dataRows = allRows.slice(dataStartIndex);
|
|
76
|
+
const truncated = dataRows.length > options.sampleRows;
|
|
77
|
+
const limitedRows = dataRows.slice(0, options.sampleRows);
|
|
78
|
+
const rows = limitedRows.map((cells, idx) => ({
|
|
79
|
+
index: dataStartIndex + idx + 1,
|
|
80
|
+
cells,
|
|
81
|
+
}));
|
|
82
|
+
const columnCount = Math.max(...allRows.map(r => r.length), 0);
|
|
83
|
+
const cellCount = allRows.reduce((sum, r) => sum + r.length, 0);
|
|
84
|
+
let chars = 0;
|
|
85
|
+
for (const row of allRows) {
|
|
86
|
+
for (const cell of row) {
|
|
87
|
+
chars += cell.value.length + 1;
|
|
88
|
+
}
|
|
89
|
+
chars += 1;
|
|
90
|
+
}
|
|
91
|
+
tables.push({
|
|
92
|
+
tableIndex: globalTableIndex,
|
|
93
|
+
location,
|
|
94
|
+
rowCount: allRows.length,
|
|
95
|
+
columnCount,
|
|
96
|
+
cellCount,
|
|
97
|
+
headers,
|
|
98
|
+
rows,
|
|
99
|
+
truncated,
|
|
100
|
+
tokenEstimate: estimateTokens(chars),
|
|
101
|
+
});
|
|
102
|
+
}
|
|
103
|
+
}
|
|
104
|
+
return tables;
|
|
105
|
+
}
|
|
106
|
+
//# sourceMappingURL=inspect-odp.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"inspect-odp.js","sourceRoot":"","sources":["../../../src/table/inspect-odp.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAGtD,SAAS,YAAY,CAAC,GAAW;IAC/B,OAAO,GAAG,CAAC,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;AAC5C,CAAC;AAED,SAAS,mBAAmB,CAAC,GAAW,EAAE,IAAY;IACpD,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,IAAI,MAAM,CAAC,GAAG,IAAI,mBAAmB,EAAE,GAAG,CAAC,CAAC,CAAC;IACrE,IAAI,KAAK,EAAE,CAAC;QACV,MAAM,GAAG,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;QACnC,OAAO,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,CAAC;IACnC,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,aAAa,CAAC,aAA0B,EAAE,OAA4B;IAC7E,IAAI,OAAO,CAAC,SAAS,KAAK,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,IAAI,OAAO,OAAO,CAAC,SAAS,KAAK,QAAQ,IAAI,OAAO,CAAC,SAAS,KAAK,CAAC,EAAE,CAAC;QACrE,OAAO,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;IACzC,CAAC;IAED,IAAI,OAAO,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;QACjC,MAAM,MAAM,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACzE,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,aAAa,CAAC,MAAM,IAAI,IAAI,GAAG,CAAC,MAAM,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC,MAAM,EAAE,CAAC;YAC1G,OAAO,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;QACzC,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,gBAAgB,CAAC,MAAc,EAAE,OAA4B;IACjF,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;IAC1C,MAAM,UAAU,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IAChE,IAAI,CAAC,UAAU;QAAE,OAAO,EAAE,CAAC;IAE3B,0CAA0C;IAC1C,MAAM,WAAW,GAAG,CAAC,GAAG,UAAU,CAAC,QAAQ,CAAC,+CAA+C,CAAC,CAAC,CAAC;IAE9F,MAAM,MAAM,GAAqB,EAAE,CAAC;IACpC,IAAI,gBAAgB,GAAG,CAAC,CAAC;IAEzB,KAAK,IAAI,EAAE,GAAG,CAAC,EAAE,EAAE,GAAG,WAAW,CAAC,MAAM,EAAE,EAAE,EAAE,EAAE,CAAC;QAC/C,MAAM,SAAS,GAAG,WAAW,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QACrC,MAAM,OAAO,GAAG,WAAW,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QAEnC,uBAAuB;QACvB,MAAM,SAAS,GAAG,SAAS,CAAC,KAAK,CAAC,4BAA4B,CAAC,CAAC;QAChE,MAAM,QAAQ,GAAG,SAAS,EAAE,CAAC,CAAC,CAAC,CAAC;QAChC,MAAM,QAAQ,GAAG,QAAQ,CAAC,CAAC,CAAC,UAAU,QAAQ,EAAE,CAAC,CAAC,CAAC,SAAS,EAAE,GAAG,CAAC,EAAE,CAAC;QAErE,MAAM,YAAY,GAAG,CAAC,GAAG,OAAO,CAAC,QAAQ,CAAC,iDAAiD,CAAC,CAAC,CAAC;QAE9F,KAAK,MAAM,UAAU,IAAI,YAAY,EAAE,CAAC;YACtC,gBAAgB,EAAE,CAAC;YACnB,MAAM,QAAQ,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC;YAE/B,MAAM,UAAU,GAAG,CAAC,GAAG,QAAQ,CAAC,QAAQ,CAAC,yDAAyD,CAAC,CAAC,CAAC;YACrG,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;gBAAE,SAAS;YAEtC,MAAM,OAAO,GAAkB,EAAE,CAAC;YAElC,KAAK,MAAM,QAAQ,IAAI,UAAU,EAAE,CAAC;gBAClC,MAAM,MAAM,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;gBAC3B,MAAM,WAAW,GAAG,CAAC,GAAG,MAAM,CAAC,QAAQ,CAAC,8DAA8D,CAAC,CAAC,CAAC;gBACzG,MAAM,KAAK,GAAgB,EAAE,CAAC;gBAE9B,KAAK,MAAM,SAAS,IAAI,WAAW,EAAE,CAAC;oBACpC,MAAM,KAAK,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC;oBAC3B,MAAM,OAAO,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC;oBAC7B,MAAM,KAAK,GAAG,YAAY,CAAC,OAAO,CAAC,CAAC;oBACpC,MAAM,IAAI,GAAc,EAAE,KAAK,EAAE,CAAC;oBAElC,MAAM,OAAO,GAAG,mBAAmB,CAAC,KAAK,EAAE,8BAA8B,CAAC,CAAC;oBAC3E,MAAM,OAAO,GAAG,mBAAmB,CAAC,KAAK,EAAE,2BAA2B,CAAC,CAAC;oBACxE,IAAI,OAAO;wBAAE,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;oBACpC,IAAI,OAAO;wBAAE,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;oBAEpC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBACnB,CAAC;gBAED,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YACtB,CAAC;YAED,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;gBAAE,SAAS;YAEnC,MAAM,OAAO,GAAG,aAAa,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC;YACnD,MAAM,cAAc,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAEvC,MAAM,QAAQ,GAAG,OAAO,CAAC,KAAK,CAAC,cAAc,CAAC,CAAC;YAC/C,MAAM,SAAS,GAAG,QAAQ,CAAC,MAAM,GAAG,OAAO,CAAC,UAAU,CAAC;YACvD,MAAM,WAAW,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,OAAO,CAAC,UAAU,CAAC,CAAC;YAE1D,MAAM,IAAI,GAAG,WAAW,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,GAAG,EAAE,EAAE,CAAC,CAAC;gBAC5C,KAAK,EAAE,cAAc,GAAG,GAAG,GAAG,CAAC;gBAC/B,KAAK;aACN,CAAC,CAAC,CAAC;YAEJ,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC;YAC/D,MAAM,SAAS,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;YAEhE,IAAI,KAAK,GAAG,CAAC,CAAC;YACd,KAAK,MAAM,GAAG,IAAI,OAAO,EAAE,CAAC;gBAC1B,KAAK,MAAM,IAAI,IAAI,GAAG,EAAE,CAAC;oBACvB,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC;gBACjC,CAAC;gBACD,KAAK,IAAI,CAAC,CAAC;YACb,CAAC;YAED,MAAM,CAAC,IAAI,CAAC;gBACV,UAAU,EAAE,gBAAgB;gBAC5B,QAAQ;gBACR,QAAQ,EAAE,OAAO,CAAC,MAAM;gBACxB,WAAW;gBACX,SAAS;gBACT,OAAO;gBACP,IAAI;gBACJ,SAAS;gBACT,aAAa,EAAE,cAAc,CAAC,KAAK,CAAC;aACrC,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC"}
|
|
@@ -0,0 +1,100 @@
|
|
|
1
|
+
import JSZip from 'jszip';
|
|
2
|
+
import { estimateTokens } from '../inspect/shared.js';
|
|
3
|
+
function stripXmlTags(xml) {
|
|
4
|
+
return xml.replace(/<[^>]*>/g, '').trim();
|
|
5
|
+
}
|
|
6
|
+
function extractIntAttribute(tag, attr) {
|
|
7
|
+
const match = tag.match(new RegExp(`${attr}\\s*=\\s*"(\\d+)"`, 'i'));
|
|
8
|
+
if (match) {
|
|
9
|
+
const val = parseInt(match[1], 10);
|
|
10
|
+
return val > 1 ? val : undefined;
|
|
11
|
+
}
|
|
12
|
+
return undefined;
|
|
13
|
+
}
|
|
14
|
+
function detectHeaders(firstRowCells, options) {
|
|
15
|
+
if (options.headerRow === 'none')
|
|
16
|
+
return null;
|
|
17
|
+
if (typeof options.headerRow === 'number' && options.headerRow === 1) {
|
|
18
|
+
return firstRowCells.map(c => c.value);
|
|
19
|
+
}
|
|
20
|
+
if (options.headerRow === 'auto') {
|
|
21
|
+
const values = firstRowCells.map(c => c.value).filter(v => v.length > 0);
|
|
22
|
+
if (values.length > 0 && values.length === firstRowCells.length && new Set(values).size === values.length) {
|
|
23
|
+
return firstRowCells.map(c => c.value);
|
|
24
|
+
}
|
|
25
|
+
}
|
|
26
|
+
return null;
|
|
27
|
+
}
|
|
28
|
+
function extractTablesFromXml(contentXml, options, locationPrefix) {
|
|
29
|
+
const tableMatches = [...contentXml.matchAll(/<table:table\b[^>]*>([\s\S]*?)<\/table:table>/gi)];
|
|
30
|
+
if (tableMatches.length === 0)
|
|
31
|
+
return [];
|
|
32
|
+
const tables = [];
|
|
33
|
+
for (let ti = 0; ti < tableMatches.length; ti++) {
|
|
34
|
+
const tableXml = tableMatches[ti][1];
|
|
35
|
+
const rowMatches = [...tableXml.matchAll(/<table:table-row\b[^>]*>([\s\S]*?)<\/table:table-row>/gi)];
|
|
36
|
+
if (rowMatches.length === 0)
|
|
37
|
+
continue;
|
|
38
|
+
const allRows = [];
|
|
39
|
+
for (const rowMatch of rowMatches) {
|
|
40
|
+
const rowXml = rowMatch[1];
|
|
41
|
+
const cellMatches = [...rowXml.matchAll(/<table:table-cell\b([^>]*?)>([\s\S]*?)<\/table:table-cell>/gi)];
|
|
42
|
+
const cells = [];
|
|
43
|
+
for (const cellMatch of cellMatches) {
|
|
44
|
+
const attrs = cellMatch[1];
|
|
45
|
+
const cellXml = cellMatch[2];
|
|
46
|
+
const value = stripXmlTags(cellXml);
|
|
47
|
+
const cell = { value };
|
|
48
|
+
const colSpan = extractIntAttribute(attrs, 'table:number-columns-spanned');
|
|
49
|
+
const rowSpan = extractIntAttribute(attrs, 'table:number-rows-spanned');
|
|
50
|
+
if (colSpan)
|
|
51
|
+
cell.colSpan = colSpan;
|
|
52
|
+
if (rowSpan)
|
|
53
|
+
cell.rowSpan = rowSpan;
|
|
54
|
+
cells.push(cell);
|
|
55
|
+
}
|
|
56
|
+
allRows.push(cells);
|
|
57
|
+
}
|
|
58
|
+
if (allRows.length === 0)
|
|
59
|
+
continue;
|
|
60
|
+
// Detect headers
|
|
61
|
+
const headers = detectHeaders(allRows[0], options);
|
|
62
|
+
const dataStartIndex = headers ? 1 : 0;
|
|
63
|
+
const dataRows = allRows.slice(dataStartIndex);
|
|
64
|
+
const truncated = dataRows.length > options.sampleRows;
|
|
65
|
+
const limitedRows = dataRows.slice(0, options.sampleRows);
|
|
66
|
+
const rows = limitedRows.map((cells, idx) => ({
|
|
67
|
+
index: dataStartIndex + idx + 1,
|
|
68
|
+
cells,
|
|
69
|
+
}));
|
|
70
|
+
const columnCount = Math.max(...allRows.map(r => r.length), 0);
|
|
71
|
+
const cellCount = allRows.reduce((sum, r) => sum + r.length, 0);
|
|
72
|
+
let chars = 0;
|
|
73
|
+
for (const row of allRows) {
|
|
74
|
+
for (const cell of row) {
|
|
75
|
+
chars += cell.value.length + 1;
|
|
76
|
+
}
|
|
77
|
+
chars += 1;
|
|
78
|
+
}
|
|
79
|
+
tables.push({
|
|
80
|
+
tableIndex: ti + 1,
|
|
81
|
+
location: locationPrefix,
|
|
82
|
+
rowCount: allRows.length,
|
|
83
|
+
columnCount,
|
|
84
|
+
cellCount,
|
|
85
|
+
headers,
|
|
86
|
+
rows,
|
|
87
|
+
truncated,
|
|
88
|
+
tokenEstimate: estimateTokens(chars),
|
|
89
|
+
});
|
|
90
|
+
}
|
|
91
|
+
return tables;
|
|
92
|
+
}
|
|
93
|
+
export async function extractOdtTables(buffer, options) {
|
|
94
|
+
const zip = await JSZip.loadAsync(buffer);
|
|
95
|
+
const contentXml = await zip.file('content.xml')?.async('text');
|
|
96
|
+
if (!contentXml)
|
|
97
|
+
return [];
|
|
98
|
+
return extractTablesFromXml(contentXml, options, null);
|
|
99
|
+
}
|
|
100
|
+
//# sourceMappingURL=inspect-odt.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"inspect-odt.js","sourceRoot":"","sources":["../../../src/table/inspect-odt.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAGtD,SAAS,YAAY,CAAC,GAAW;IAC/B,OAAO,GAAG,CAAC,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;AAC5C,CAAC;AAED,SAAS,mBAAmB,CAAC,GAAW,EAAE,IAAY;IACpD,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,IAAI,MAAM,CAAC,GAAG,IAAI,mBAAmB,EAAE,GAAG,CAAC,CAAC,CAAC;IACrE,IAAI,KAAK,EAAE,CAAC;QACV,MAAM,GAAG,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;QACnC,OAAO,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,CAAC;IACnC,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,aAAa,CAAC,aAA0B,EAAE,OAA4B;IAC7E,IAAI,OAAO,CAAC,SAAS,KAAK,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,IAAI,OAAO,OAAO,CAAC,SAAS,KAAK,QAAQ,IAAI,OAAO,CAAC,SAAS,KAAK,CAAC,EAAE,CAAC;QACrE,OAAO,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;IACzC,CAAC;IAED,IAAI,OAAO,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;QACjC,MAAM,MAAM,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACzE,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,aAAa,CAAC,MAAM,IAAI,IAAI,GAAG,CAAC,MAAM,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC,MAAM,EAAE,CAAC;YAC1G,OAAO,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;QACzC,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC;AAED,SAAS,oBAAoB,CAAC,UAAkB,EAAE,OAA4B,EAAE,cAA6B;IAC3G,MAAM,YAAY,GAAG,CAAC,GAAG,UAAU,CAAC,QAAQ,CAAC,iDAAiD,CAAC,CAAC,CAAC;IACjG,IAAI,YAAY,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IAEzC,MAAM,MAAM,GAAqB,EAAE,CAAC;IAEpC,KAAK,IAAI,EAAE,GAAG,CAAC,EAAE,EAAE,GAAG,YAAY,CAAC,MAAM,EAAE,EAAE,EAAE,EAAE,CAAC;QAChD,MAAM,QAAQ,GAAG,YAAY,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QACrC,MAAM,UAAU,GAAG,CAAC,GAAG,QAAQ,CAAC,QAAQ,CAAC,yDAAyD,CAAC,CAAC,CAAC;QACrG,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QAEtC,MAAM,OAAO,GAAkB,EAAE,CAAC;QAElC,KAAK,MAAM,QAAQ,IAAI,UAAU,EAAE,CAAC;YAClC,MAAM,MAAM,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;YAC3B,MAAM,WAAW,GAAG,CAAC,GAAG,MAAM,CAAC,QAAQ,CAAC,8DAA8D,CAAC,CAAC,CAAC;YACzG,MAAM,KAAK,GAAgB,EAAE,CAAC;YAE9B,KAAK,MAAM,SAAS,IAAI,WAAW,EAAE,CAAC;gBACpC,MAAM,KAAK,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC;gBAC3B,MAAM,OAAO,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC;gBAC7B,MAAM,KAAK,GAAG,YAAY,CAAC,OAAO,CAAC,CAAC;gBACpC,MAAM,IAAI,GAAc,EAAE,KAAK,EAAE,CAAC;gBAElC,MAAM,OAAO,GAAG,mBAAmB,CAAC,KAAK,EAAE,8BAA8B,CAAC,CAAC;gBAC3E,MAAM,OAAO,GAAG,mBAAmB,CAAC,KAAK,EAAE,2BAA2B,CAAC,CAAC;gBACxE,IAAI,OAAO;oBAAE,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;gBACpC,IAAI,OAAO;oBAAE,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;gBAEpC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACnB,CAAC;YAED,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QACtB,CAAC;QAED,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QAEnC,iBAAiB;QACjB,MAAM,OAAO,GAAG,aAAa,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC;QACnD,MAAM,cAAc,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAEvC,MAAM,QAAQ,GAAG,OAAO,CAAC,KAAK,CAAC,cAAc,CAAC,CAAC;QAC/C,MAAM,SAAS,GAAG,QAAQ,CAAC,MAAM,GAAG,OAAO,CAAC,UAAU,CAAC;QACvD,MAAM,WAAW,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,OAAO,CAAC,UAAU,CAAC,CAAC;QAE1D,MAAM,IAAI,GAAG,WAAW,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,GAAG,EAAE,EAAE,CAAC,CAAC;YAC5C,KAAK,EAAE,cAAc,GAAG,GAAG,GAAG,CAAC;YAC/B,KAAK;SACN,CAAC,CAAC,CAAC;QAEJ,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC;QAC/D,MAAM,SAAS,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;QAEhE,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,KAAK,MAAM,GAAG,IAAI,OAAO,EAAE,CAAC;YAC1B,KAAK,MAAM,IAAI,IAAI,GAAG,EAAE,CAAC;gBACvB,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC;YACjC,CAAC;YACD,KAAK,IAAI,CAAC,CAAC;QACb,CAAC;QAED,MAAM,CAAC,IAAI,CAAC;YACV,UAAU,EAAE,EAAE,GAAG,CAAC;YAClB,QAAQ,EAAE,cAAc;YACxB,QAAQ,EAAE,OAAO,CAAC,MAAM;YACxB,WAAW;YACX,SAAS;YACT,OAAO;YACP,IAAI;YACJ,SAAS;YACT,aAAa,EAAE,cAAc,CAAC,KAAK,CAAC;SACrC,CAAC,CAAC;IACL,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,gBAAgB,CAAC,MAAc,EAAE,OAA4B;IACjF,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;IAC1C,MAAM,UAAU,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IAChE,IAAI,CAAC,UAAU;QAAE,OAAO,EAAE,CAAC;IAE3B,OAAO,oBAAoB,CAAC,UAAU,EAAE,OAAO,EAAE,IAAI,CAAC,CAAC;AACzD,CAAC"}
|
|
@@ -0,0 +1,112 @@
|
|
|
1
|
+
import JSZip from 'jszip';
|
|
2
|
+
import { estimateTokens } from '../inspect/shared.js';
|
|
3
|
+
function extractCellText(tcXml) {
|
|
4
|
+
const matches = tcXml.match(/<a:t>([^<]*)<\/a:t>/g) || [];
|
|
5
|
+
return matches.map(m => {
|
|
6
|
+
const text = m.match(/<a:t>([^<]*)<\/a:t>/);
|
|
7
|
+
return text?.[1] || '';
|
|
8
|
+
}).join(' ').trim();
|
|
9
|
+
}
|
|
10
|
+
function extractIntAttribute(xml, attr) {
|
|
11
|
+
const match = xml.match(new RegExp(`${attr}\\s*=\\s*"(\\d+)"`, 'i'));
|
|
12
|
+
if (match) {
|
|
13
|
+
const val = parseInt(match[1], 10);
|
|
14
|
+
return val > 1 ? val : undefined;
|
|
15
|
+
}
|
|
16
|
+
return undefined;
|
|
17
|
+
}
|
|
18
|
+
function detectHeaders(firstRowCells, options) {
|
|
19
|
+
if (options.headerRow === 'none')
|
|
20
|
+
return null;
|
|
21
|
+
if (typeof options.headerRow === 'number' && options.headerRow === 1) {
|
|
22
|
+
return firstRowCells.map(c => c.value);
|
|
23
|
+
}
|
|
24
|
+
if (options.headerRow === 'auto') {
|
|
25
|
+
const values = firstRowCells.map(c => c.value).filter(v => v.length > 0);
|
|
26
|
+
if (values.length > 0 && values.length === firstRowCells.length && new Set(values).size === values.length) {
|
|
27
|
+
return firstRowCells.map(c => c.value);
|
|
28
|
+
}
|
|
29
|
+
}
|
|
30
|
+
return null;
|
|
31
|
+
}
|
|
32
|
+
export async function extractPptxTables(buffer, options) {
|
|
33
|
+
const zip = await JSZip.loadAsync(buffer);
|
|
34
|
+
const slideFiles = Object.keys(zip.files)
|
|
35
|
+
.filter(name => /^ppt\/slides\/slide\d+\.xml$/.test(name))
|
|
36
|
+
.sort((a, b) => {
|
|
37
|
+
const numA = parseInt(a.match(/slide(\d+)/)?.[1] || '0', 10);
|
|
38
|
+
const numB = parseInt(b.match(/slide(\d+)/)?.[1] || '0', 10);
|
|
39
|
+
return numA - numB;
|
|
40
|
+
});
|
|
41
|
+
const tables = [];
|
|
42
|
+
let tableIndex = 0;
|
|
43
|
+
for (let si = 0; si < slideFiles.length; si++) {
|
|
44
|
+
const slideXml = await zip.file(slideFiles[si])?.async('text');
|
|
45
|
+
if (!slideXml)
|
|
46
|
+
continue;
|
|
47
|
+
// Find all <a:tbl> elements
|
|
48
|
+
const tableMatches = [...slideXml.matchAll(/<a:tbl\b[^>]*>([\s\S]*?)<\/a:tbl>/gi)];
|
|
49
|
+
for (const tableMatch of tableMatches) {
|
|
50
|
+
tableIndex++;
|
|
51
|
+
const tableXml = tableMatch[1];
|
|
52
|
+
// Find rows
|
|
53
|
+
const rowMatches = [...tableXml.matchAll(/<a:tr\b[^>]*>([\s\S]*?)<\/a:tr>/gi)];
|
|
54
|
+
if (rowMatches.length === 0)
|
|
55
|
+
continue;
|
|
56
|
+
const allRows = [];
|
|
57
|
+
for (const rowMatch of rowMatches) {
|
|
58
|
+
const rowXml = rowMatch[1];
|
|
59
|
+
const cellMatches = [...rowXml.matchAll(/<a:tc\b([^>]*?)>([\s\S]*?)<\/a:tc>/gi)];
|
|
60
|
+
const cells = [];
|
|
61
|
+
for (const cellMatch of cellMatches) {
|
|
62
|
+
const attrs = cellMatch[1];
|
|
63
|
+
const cellXml = cellMatch[2];
|
|
64
|
+
const value = extractCellText(cellXml);
|
|
65
|
+
const cell = { value };
|
|
66
|
+
const gridSpan = extractIntAttribute(attrs, 'gridSpan');
|
|
67
|
+
const rowSpan = extractIntAttribute(attrs, 'rowSpan');
|
|
68
|
+
if (gridSpan)
|
|
69
|
+
cell.colSpan = gridSpan;
|
|
70
|
+
if (rowSpan)
|
|
71
|
+
cell.rowSpan = rowSpan;
|
|
72
|
+
cells.push(cell);
|
|
73
|
+
}
|
|
74
|
+
allRows.push(cells);
|
|
75
|
+
}
|
|
76
|
+
if (allRows.length === 0)
|
|
77
|
+
continue;
|
|
78
|
+
// Detect headers
|
|
79
|
+
const headers = detectHeaders(allRows[0], options);
|
|
80
|
+
const dataStartIndex = headers ? 1 : 0;
|
|
81
|
+
const dataRows = allRows.slice(dataStartIndex);
|
|
82
|
+
const truncated = dataRows.length > options.sampleRows;
|
|
83
|
+
const limitedRows = dataRows.slice(0, options.sampleRows);
|
|
84
|
+
const rows = limitedRows.map((cells, idx) => ({
|
|
85
|
+
index: dataStartIndex + idx + 1,
|
|
86
|
+
cells,
|
|
87
|
+
}));
|
|
88
|
+
const columnCount = Math.max(...allRows.map(r => r.length), 0);
|
|
89
|
+
const cellCount = allRows.reduce((sum, r) => sum + r.length, 0);
|
|
90
|
+
let chars = 0;
|
|
91
|
+
for (const row of allRows) {
|
|
92
|
+
for (const cell of row) {
|
|
93
|
+
chars += cell.value.length + 1;
|
|
94
|
+
}
|
|
95
|
+
chars += 1;
|
|
96
|
+
}
|
|
97
|
+
tables.push({
|
|
98
|
+
tableIndex,
|
|
99
|
+
location: `Slide ${si + 1}`,
|
|
100
|
+
rowCount: allRows.length,
|
|
101
|
+
columnCount,
|
|
102
|
+
cellCount,
|
|
103
|
+
headers,
|
|
104
|
+
rows,
|
|
105
|
+
truncated,
|
|
106
|
+
tokenEstimate: estimateTokens(chars),
|
|
107
|
+
});
|
|
108
|
+
}
|
|
109
|
+
}
|
|
110
|
+
return tables;
|
|
111
|
+
}
|
|
112
|
+
//# sourceMappingURL=inspect-pptx.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"inspect-pptx.js","sourceRoot":"","sources":["../../../src/table/inspect-pptx.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAGtD,SAAS,eAAe,CAAC,KAAa;IACpC,MAAM,OAAO,GAAG,KAAK,CAAC,KAAK,CAAC,sBAAsB,CAAC,IAAI,EAAE,CAAC;IAC1D,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE;QACrB,MAAM,IAAI,GAAG,CAAC,CAAC,KAAK,CAAC,qBAAqB,CAAC,CAAC;QAC5C,OAAO,IAAI,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;IACzB,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;AACtB,CAAC;AAED,SAAS,mBAAmB,CAAC,GAAW,EAAE,IAAY;IACpD,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,IAAI,MAAM,CAAC,GAAG,IAAI,mBAAmB,EAAE,GAAG,CAAC,CAAC,CAAC;IACrE,IAAI,KAAK,EAAE,CAAC;QACV,MAAM,GAAG,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;QACnC,OAAO,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,CAAC;IACnC,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,aAAa,CAAC,aAA0B,EAAE,OAA4B;IAC7E,IAAI,OAAO,CAAC,SAAS,KAAK,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,IAAI,OAAO,OAAO,CAAC,SAAS,KAAK,QAAQ,IAAI,OAAO,CAAC,SAAS,KAAK,CAAC,EAAE,CAAC;QACrE,OAAO,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;IACzC,CAAC;IAED,IAAI,OAAO,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;QACjC,MAAM,MAAM,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACzE,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,aAAa,CAAC,MAAM,IAAI,IAAI,GAAG,CAAC,MAAM,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC,MAAM,EAAE,CAAC;YAC1G,OAAO,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;QACzC,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,MAAc,EAAE,OAA4B;IAClF,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;IAE1C,MAAM,UAAU,GAAG,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC;SACtC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,8BAA8B,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;SACzD,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;QACb,MAAM,IAAI,GAAG,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,GAAG,EAAE,EAAE,CAAC,CAAC;QAC7D,MAAM,IAAI,GAAG,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,GAAG,EAAE,EAAE,CAAC,CAAC;QAC7D,OAAO,IAAI,GAAG,IAAI,CAAC;IACrB,CAAC,CAAC,CAAC;IAEL,MAAM,MAAM,GAAqB,EAAE,CAAC;IACpC,IAAI,UAAU,GAAG,CAAC,CAAC;IAEnB,KAAK,IAAI,EAAE,GAAG,CAAC,EAAE,EAAE,GAAG,UAAU,CAAC,MAAM,EAAE,EAAE,EAAE,EAAE,CAAC;QAC9C,MAAM,QAAQ,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,EAAE,CAAC,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;QAC/D,IAAI,CAAC,QAAQ;YAAE,SAAS;QAExB,4BAA4B;QAC5B,MAAM,YAAY,GAAG,CAAC,GAAG,QAAQ,CAAC,QAAQ,CAAC,qCAAqC,CAAC,CAAC,CAAC;QAEnF,KAAK,MAAM,UAAU,IAAI,YAAY,EAAE,CAAC;YACtC,UAAU,EAAE,CAAC;YACb,MAAM,QAAQ,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC;YAE/B,YAAY;YACZ,MAAM,UAAU,GAAG,CAAC,GAAG,QAAQ,CAAC,QAAQ,CAAC,mCAAmC,CAAC,CAAC,CAAC;YAC/E,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;gBAAE,SAAS;YAEtC,MAAM,OAAO,GAAkB,EAAE,CAAC;YAElC,KAAK,MAAM,QAAQ,IAAI,UAAU,EAAE,CAAC;gBAClC,MAAM,MAAM,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;gBAC3B,MAAM,WAAW,GAAG,CAAC,GAAG,MAAM,CAAC,QAAQ,CAAC,sCAAsC,CAAC,CAAC,CAAC;gBACjF,MAAM,KAAK,GAAgB,EAAE,CAAC;gBAE9B,KAAK,MAAM,SAAS,IAAI,WAAW,EAAE,CAAC;oBACpC,MAAM,KAAK,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC;oBAC3B,MAAM,OAAO,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC;oBAC7B,MAAM,KAAK,GAAG,eAAe,CAAC,OAAO,CAAC,CAAC;oBACvC,MAAM,IAAI,GAAc,EAAE,KAAK,EAAE,CAAC;oBAElC,MAAM,QAAQ,GAAG,mBAAmB,CAAC,KAAK,EAAE,UAAU,CAAC,CAAC;oBACxD,MAAM,OAAO,GAAG,mBAAmB,CAAC,KAAK,EAAE,SAAS,CAAC,CAAC;oBACtD,IAAI,QAAQ;wBAAE,IAAI,CAAC,OAAO,GAAG,QAAQ,CAAC;oBACtC,IAAI,OAAO;wBAAE,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;oBAEpC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBACnB,CAAC;gBAED,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YACtB,CAAC;YAED,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;gBAAE,SAAS;YAEnC,iBAAiB;YACjB,MAAM,OAAO,GAAG,aAAa,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC;YACnD,MAAM,cAAc,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAEvC,MAAM,QAAQ,GAAG,OAAO,CAAC,KAAK,CAAC,cAAc,CAAC,CAAC;YAC/C,MAAM,SAAS,GAAG,QAAQ,CAAC,MAAM,GAAG,OAAO,CAAC,UAAU,CAAC;YACvD,MAAM,WAAW,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,OAAO,CAAC,UAAU,CAAC,CAAC;YAE1D,MAAM,IAAI,GAAG,WAAW,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,GAAG,EAAE,EAAE,CAAC,CAAC;gBAC5C,KAAK,EAAE,cAAc,GAAG,GAAG,GAAG,CAAC;gBAC/B,KAAK;aACN,CAAC,CAAC,CAAC;YAEJ,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC;YAC/D,MAAM,SAAS,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;YAEhE,IAAI,KAAK,GAAG,CAAC,CAAC;YACd,KAAK,MAAM,GAAG,IAAI,OAAO,EAAE,CAAC;gBAC1B,KAAK,MAAM,IAAI,IAAI,GAAG,EAAE,CAAC;oBACvB,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC;gBACjC,CAAC;gBACD,KAAK,IAAI,CAAC,CAAC;YACb,CAAC;YAED,MAAM,CAAC,IAAI,CAAC;gBACV,UAAU;gBACV,QAAQ,EAAE,SAAS,EAAE,GAAG,CAAC,EAAE;gBAC3B,QAAQ,EAAE,OAAO,CAAC,MAAM;gBACxB,WAAW;gBACX,SAAS;gBACT,OAAO;gBACP,IAAI;gBACJ,SAAS;gBACT,aAAa,EAAE,cAAc,CAAC,KAAK,CAAC;aACrC,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC"}
|
|
@@ -0,0 +1,161 @@
|
|
|
1
|
+
import XLSX from 'xlsx';
|
|
2
|
+
import { getCell, renderCell, isNonEmptyCell } from '../inspect/xlsx-cells.js';
|
|
3
|
+
import { estimateTokens } from '../inspect/shared.js';
|
|
4
|
+
function buildMergeMap(sheet) {
|
|
5
|
+
const map = new Map();
|
|
6
|
+
const merges = sheet['!merges'] ?? [];
|
|
7
|
+
for (const merge of merges) {
|
|
8
|
+
const key = XLSX.utils.encode_cell({ r: merge.s.r, c: merge.s.c });
|
|
9
|
+
map.set(key, {
|
|
10
|
+
colSpan: merge.e.c - merge.s.c + 1,
|
|
11
|
+
rowSpan: merge.e.r - merge.s.r + 1,
|
|
12
|
+
});
|
|
13
|
+
}
|
|
14
|
+
return map;
|
|
15
|
+
}
|
|
16
|
+
function buildMergedCellSet(sheet) {
|
|
17
|
+
const set = new Set();
|
|
18
|
+
const merges = sheet['!merges'] ?? [];
|
|
19
|
+
for (const merge of merges) {
|
|
20
|
+
for (let r = merge.s.r; r <= merge.e.r; r++) {
|
|
21
|
+
for (let c = merge.s.c; c <= merge.e.c; c++) {
|
|
22
|
+
if (r === merge.s.r && c === merge.s.c)
|
|
23
|
+
continue;
|
|
24
|
+
set.add(XLSX.utils.encode_cell({ r, c }));
|
|
25
|
+
}
|
|
26
|
+
}
|
|
27
|
+
}
|
|
28
|
+
return set;
|
|
29
|
+
}
|
|
30
|
+
function detectHeaders(sheet, startRow, startCol, endCol, headerRow) {
|
|
31
|
+
if (headerRow === 'none')
|
|
32
|
+
return null;
|
|
33
|
+
const rowIndex = typeof headerRow === 'number' ? headerRow - 1 : startRow;
|
|
34
|
+
const values = [];
|
|
35
|
+
let nonEmpty = 0;
|
|
36
|
+
for (let c = startCol; c <= endCol; c++) {
|
|
37
|
+
const cell = getCell(sheet, rowIndex, c);
|
|
38
|
+
const val = renderCell(cell);
|
|
39
|
+
values.push(val);
|
|
40
|
+
if (val)
|
|
41
|
+
nonEmpty++;
|
|
42
|
+
}
|
|
43
|
+
if (nonEmpty === 0)
|
|
44
|
+
return null;
|
|
45
|
+
if (headerRow === 'auto') {
|
|
46
|
+
// Check if all non-empty values are unique strings
|
|
47
|
+
const nonEmptyValues = values.filter(v => v.length > 0);
|
|
48
|
+
if (nonEmptyValues.length > 0 && new Set(nonEmptyValues).size === nonEmptyValues.length) {
|
|
49
|
+
// Check that values look like headers (all strings in the row)
|
|
50
|
+
let allString = true;
|
|
51
|
+
for (let c = startCol; c <= endCol; c++) {
|
|
52
|
+
const cell = getCell(sheet, rowIndex, c);
|
|
53
|
+
if (cell && isNonEmptyCell(cell) && cell.t !== 's' && cell.t !== 'z') {
|
|
54
|
+
allString = false;
|
|
55
|
+
break;
|
|
56
|
+
}
|
|
57
|
+
}
|
|
58
|
+
if (allString)
|
|
59
|
+
return values;
|
|
60
|
+
}
|
|
61
|
+
return null;
|
|
62
|
+
}
|
|
63
|
+
return values;
|
|
64
|
+
}
|
|
65
|
+
export async function extractXlsxTables(buffer, options) {
|
|
66
|
+
const workbook = XLSX.read(buffer, {
|
|
67
|
+
type: 'buffer',
|
|
68
|
+
cellText: true,
|
|
69
|
+
cellDates: true,
|
|
70
|
+
sheetStubs: true,
|
|
71
|
+
});
|
|
72
|
+
const tables = [];
|
|
73
|
+
let tableIndex = 0;
|
|
74
|
+
for (let si = 0; si < workbook.SheetNames.length; si++) {
|
|
75
|
+
const sheetName = workbook.SheetNames[si];
|
|
76
|
+
const sheet = workbook.Sheets[sheetName];
|
|
77
|
+
if (!sheet || !sheet['!ref'])
|
|
78
|
+
continue;
|
|
79
|
+
tableIndex++;
|
|
80
|
+
const range = XLSX.utils.decode_range(sheet['!ref']);
|
|
81
|
+
const mergeMap = buildMergeMap(sheet);
|
|
82
|
+
const mergedCellSet = buildMergedCellSet(sheet);
|
|
83
|
+
const startRow = range.s.r;
|
|
84
|
+
const startCol = range.s.c;
|
|
85
|
+
const endCol = range.e.c;
|
|
86
|
+
const endRow = range.e.r;
|
|
87
|
+
// Detect headers
|
|
88
|
+
const headers = detectHeaders(sheet, startRow, startCol, endCol, options.headerRow);
|
|
89
|
+
const dataStartRow = headers ? startRow + 1 : startRow;
|
|
90
|
+
const totalRows = endRow - startRow + 1;
|
|
91
|
+
const totalCols = endCol - startCol + 1;
|
|
92
|
+
// Build rows
|
|
93
|
+
const maxRows = options.sampleRows;
|
|
94
|
+
const dataRowCount = endRow - dataStartRow + 1;
|
|
95
|
+
const truncated = dataRowCount > maxRows;
|
|
96
|
+
const rowLimit = Math.min(dataRowCount, maxRows);
|
|
97
|
+
const rows = [];
|
|
98
|
+
let chars = 0;
|
|
99
|
+
// Count total chars for token estimate (including header row)
|
|
100
|
+
if (headers) {
|
|
101
|
+
for (const h of headers)
|
|
102
|
+
chars += h.length + 1;
|
|
103
|
+
chars += 1;
|
|
104
|
+
}
|
|
105
|
+
let nonEmptyRowCount = 0;
|
|
106
|
+
for (let r = dataStartRow; r <= endRow; r++) {
|
|
107
|
+
const cells = [];
|
|
108
|
+
let rowHasContent = false;
|
|
109
|
+
for (let c = startCol; c <= endCol; c++) {
|
|
110
|
+
const cellAddr = XLSX.utils.encode_cell({ r, c });
|
|
111
|
+
if (mergedCellSet.has(cellAddr)) {
|
|
112
|
+
cells.push({ value: '' });
|
|
113
|
+
continue;
|
|
114
|
+
}
|
|
115
|
+
const cell = getCell(sheet, r, c);
|
|
116
|
+
const value = renderCell(cell);
|
|
117
|
+
const tableCell = { value };
|
|
118
|
+
const merge = mergeMap.get(cellAddr);
|
|
119
|
+
if (merge) {
|
|
120
|
+
if (merge.colSpan > 1)
|
|
121
|
+
tableCell.colSpan = merge.colSpan;
|
|
122
|
+
if (merge.rowSpan > 1)
|
|
123
|
+
tableCell.rowSpan = merge.rowSpan;
|
|
124
|
+
}
|
|
125
|
+
if (value)
|
|
126
|
+
rowHasContent = true;
|
|
127
|
+
cells.push(tableCell);
|
|
128
|
+
chars += value.length + 1;
|
|
129
|
+
}
|
|
130
|
+
chars += 1;
|
|
131
|
+
if (!rowHasContent)
|
|
132
|
+
continue;
|
|
133
|
+
nonEmptyRowCount++;
|
|
134
|
+
if (rows.length < rowLimit) {
|
|
135
|
+
rows.push({ index: r - startRow + 1, cells });
|
|
136
|
+
}
|
|
137
|
+
}
|
|
138
|
+
// Cell count
|
|
139
|
+
let cellCount = 0;
|
|
140
|
+
for (let r = startRow; r <= endRow; r++) {
|
|
141
|
+
for (let c = startCol; c <= endCol; c++) {
|
|
142
|
+
const cell = getCell(sheet, r, c);
|
|
143
|
+
if (cell && isNonEmptyCell(cell))
|
|
144
|
+
cellCount++;
|
|
145
|
+
}
|
|
146
|
+
}
|
|
147
|
+
tables.push({
|
|
148
|
+
tableIndex,
|
|
149
|
+
location: `Sheet: ${sheetName}`,
|
|
150
|
+
rowCount: totalRows,
|
|
151
|
+
columnCount: totalCols,
|
|
152
|
+
cellCount,
|
|
153
|
+
headers,
|
|
154
|
+
rows,
|
|
155
|
+
truncated: truncated && rows.length >= rowLimit,
|
|
156
|
+
tokenEstimate: estimateTokens(chars),
|
|
157
|
+
});
|
|
158
|
+
}
|
|
159
|
+
return tables;
|
|
160
|
+
}
|
|
161
|
+
//# sourceMappingURL=inspect-xlsx.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"inspect-xlsx.js","sourceRoot":"","sources":["../../../src/table/inspect-xlsx.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,MAAM,CAAC;AAExB,OAAO,EAAE,OAAO,EAAE,UAAU,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAC;AAC/E,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAQtD,SAAS,aAAa,CAAC,KAAgB;IACrC,MAAM,GAAG,GAAG,IAAI,GAAG,EAAqB,CAAC;IACzC,MAAM,MAAM,GAAG,KAAK,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;IACtC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,EAAE,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;QACnE,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE;YACX,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;YAClC,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;SACnC,CAAC,CAAC;IACL,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,SAAS,kBAAkB,CAAC,KAAgB;IAC1C,MAAM,GAAG,GAAG,IAAI,GAAG,EAAU,CAAC;IAC9B,MAAM,MAAM,GAAG,KAAK,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;IACtC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;YAC5C,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;gBAC5C,IAAI,CAAC,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC;oBAAE,SAAS;gBACjD,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAC5C,CAAC;QACH,CAAC;IACH,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,SAAS,aAAa,CACpB,KAAgB,EAChB,QAAgB,EAChB,QAAgB,EAChB,MAAc,EACd,SAAmC;IAEnC,IAAI,SAAS,KAAK,MAAM;QAAE,OAAO,IAAI,CAAC;IAEtC,MAAM,QAAQ,GAAG,OAAO,SAAS,KAAK,QAAQ,CAAC,CAAC,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC;IAE1E,MAAM,MAAM,GAAa,EAAE,CAAC;IAC5B,IAAI,QAAQ,GAAG,CAAC,CAAC;IACjB,KAAK,IAAI,CAAC,GAAG,QAAQ,EAAE,CAAC,IAAI,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACxC,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC;QACzC,MAAM,GAAG,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACjB,IAAI,GAAG;YAAE,QAAQ,EAAE,CAAC;IACtB,CAAC;IAED,IAAI,QAAQ,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IAEhC,IAAI,SAAS,KAAK,MAAM,EAAE,CAAC;QACzB,mDAAmD;QACnD,MAAM,cAAc,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACxD,IAAI,cAAc,CAAC,MAAM,GAAG,CAAC,IAAI,IAAI,GAAG,CAAC,cAAc,CAAC,CAAC,IAAI,KAAK,cAAc,CAAC,MAAM,EAAE,CAAC;YACxF,+DAA+D;YAC/D,IAAI,SAAS,GAAG,IAAI,CAAC;YACrB,KAAK,IAAI,CAAC,GAAG,QAAQ,EAAE,CAAC,IAAI,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBACxC,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC;gBACzC,IAAI,IAAI,IAAI,cAAc,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,CAAC,KAAK,GAAG,IAAI,IAAI,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;oBACrE,SAAS,GAAG,KAAK,CAAC;oBAClB,MAAM;gBACR,CAAC;YACH,CAAC;YACD,IAAI,SAAS;gBAAE,OAAO,MAAM,CAAC;QAC/B,CAAC;QACD,OAAO,IAAI,CAAC;IACd,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,MAAc,EAAE,OAA4B;IAClF,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;QACjC,IAAI,EAAE,QAAQ;QACd,QAAQ,EAAE,IAAI;QACd,SAAS,EAAE,IAAI;QACf,UAAU,EAAE,IAAI;KACjB,CAAC,CAAC;IAEH,MAAM,MAAM,GAAqB,EAAE,CAAC;IACpC,IAAI,UAAU,GAAG,CAAC,CAAC;IAEnB,KAAK,IAAI,EAAE,GAAG,CAAC,EAAE,EAAE,GAAG,QAAQ,CAAC,UAAU,CAAC,MAAM,EAAE,EAAE,EAAE,EAAE,CAAC;QACvD,MAAM,SAAS,GAAG,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC,CAAC;QAC1C,MAAM,KAAK,GAAG,QAAQ,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC;QACzC,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC;YAAE,SAAS;QAEvC,UAAU,EAAE,CAAC;QACb,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;QACrD,MAAM,QAAQ,GAAG,aAAa,CAAC,KAAK,CAAC,CAAC;QACtC,MAAM,aAAa,GAAG,kBAAkB,CAAC,KAAK,CAAC,CAAC;QAEhD,MAAM,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QAC3B,MAAM,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QAC3B,MAAM,MAAM,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QACzB,MAAM,MAAM,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QAEzB,iBAAiB;QACjB,MAAM,OAAO,GAAG,aAAa,CAAC,KAAK,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,OAAO,CAAC,SAAS,CAAC,CAAC;QACpF,MAAM,YAAY,GAAG,OAAO,CAAC,CAAC,CAAC,QAAQ,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC;QAEvD,MAAM,SAAS,GAAG,MAAM,GAAG,QAAQ,GAAG,CAAC,CAAC;QACxC,MAAM,SAAS,GAAG,MAAM,GAAG,QAAQ,GAAG,CAAC,CAAC;QAExC,aAAa;QACb,MAAM,OAAO,GAAG,OAAO,CAAC,UAAU,CAAC;QACnC,MAAM,YAAY,GAAG,MAAM,GAAG,YAAY,GAAG,CAAC,CAAC;QAC/C,MAAM,SAAS,GAAG,YAAY,GAAG,OAAO,CAAC;QACzC,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,YAAY,EAAE,OAAO,CAAC,CAAC;QAEjD,MAAM,IAAI,GAA4C,EAAE,CAAC;QACzD,IAAI,KAAK,GAAG,CAAC,CAAC;QAEd,8DAA8D;QAC9D,IAAI,OAAO,EAAE,CAAC;YACZ,KAAK,MAAM,CAAC,IAAI,OAAO;gBAAE,KAAK,IAAI,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;YAC/C,KAAK,IAAI,CAAC,CAAC;QACb,CAAC;QAED,IAAI,gBAAgB,GAAG,CAAC,CAAC;QACzB,KAAK,IAAI,CAAC,GAAG,YAAY,EAAE,CAAC,IAAI,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YAC5C,MAAM,KAAK,GAAgB,EAAE,CAAC;YAC9B,IAAI,aAAa,GAAG,KAAK,CAAC;YAE1B,KAAK,IAAI,CAAC,GAAG,QAAQ,EAAE,CAAC,IAAI,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBACxC,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC;gBAClD,IAAI,aAAa,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE,CAAC;oBAChC,KAAK,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC,CAAC;oBAC1B,SAAS;gBACX,CAAC;gBAED,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;gBAClC,MAAM,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC;gBAC/B,MAAM,SAAS,GAAc,EAAE,KAAK,EAAE,CAAC;gBAEvC,MAAM,KAAK,GAAG,QAAQ,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;gBACrC,IAAI,KAAK,EAAE,CAAC;oBACV,IAAI,KAAK,CAAC,OAAO,GAAG,CAAC;wBAAE,SAAS,CAAC,OAAO,GAAG,KAAK,CAAC,OAAO,CAAC;oBACzD,IAAI,KAAK,CAAC,OAAO,GAAG,CAAC;wBAAE,SAAS,CAAC,OAAO,GAAG,KAAK,CAAC,OAAO,CAAC;gBAC3D,CAAC;gBAED,IAAI,KAAK;oBAAE,aAAa,GAAG,IAAI,CAAC;gBAChC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;gBACtB,KAAK,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC;YAC5B,CAAC;YACD,KAAK,IAAI,CAAC,CAAC;YAEX,IAAI,CAAC,aAAa;gBAAE,SAAS;YAC7B,gBAAgB,EAAE,CAAC;YAEnB,IAAI,IAAI,CAAC,MAAM,GAAG,QAAQ,EAAE,CAAC;gBAC3B,IAAI,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,CAAC,GAAG,QAAQ,GAAG,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC;YAChD,CAAC;QACH,CAAC;QAED,aAAa;QACb,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,KAAK,IAAI,CAAC,GAAG,QAAQ,EAAE,CAAC,IAAI,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YACxC,KAAK,IAAI,CAAC,GAAG,QAAQ,EAAE,CAAC,IAAI,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBACxC,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;gBAClC,IAAI,IAAI,IAAI,cAAc,CAAC,IAAI,CAAC;oBAAE,SAAS,EAAE,CAAC;YAChD,CAAC;QACH,CAAC;QAED,MAAM,CAAC,IAAI,CAAC;YACV,UAAU;YACV,QAAQ,EAAE,UAAU,SAAS,EAAE;YAC/B,QAAQ,EAAE,SAAS;YACnB,WAAW,EAAE,SAAS;YACtB,SAAS;YACT,OAAO;YACP,IAAI;YACJ,SAAS,EAAE,SAAS,IAAI,IAAI,CAAC,MAAM,IAAI,QAAQ;YAC/C,aAAa,EAAE,cAAc,CAAC,KAAK,CAAC;SACrC,CAAC,CAAC;IACL,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC"}
|
|
@@ -0,0 +1,65 @@
|
|
|
1
|
+
import { readFile, stat } from 'node:fs/promises';
|
|
2
|
+
import path from 'node:path';
|
|
3
|
+
import { z } from 'zod';
|
|
4
|
+
import { getExtension } from '../utils.js';
|
|
5
|
+
import { extractDocxTables } from './inspect-docx.js';
|
|
6
|
+
import { extractXlsxTables } from './inspect-xlsx.js';
|
|
7
|
+
import { extractPptxTables } from './inspect-pptx.js';
|
|
8
|
+
import { extractOdtTables } from './inspect-odt.js';
|
|
9
|
+
import { extractOdpTables } from './inspect-odp.js';
|
|
10
|
+
const TableFormatSchema = z.enum(['docx', 'xlsx', 'pptx', 'odt', 'odp', 'pdf']);
|
|
11
|
+
const SUPPORTED_FORMATS = new Set(TableFormatSchema.options);
|
|
12
|
+
export async function inspectTables(filePath, options) {
|
|
13
|
+
const resolvedPath = path.resolve(filePath);
|
|
14
|
+
const ext = getExtension(resolvedPath);
|
|
15
|
+
if (!SUPPORTED_FORMATS.has(ext)) {
|
|
16
|
+
throw new Error(`Unsupported format: .${ext || '(none)'} (supported: docx, xlsx, pptx, odt, odp, pdf)`);
|
|
17
|
+
}
|
|
18
|
+
const fileStats = await stat(resolvedPath);
|
|
19
|
+
const format = TableFormatSchema.parse(ext);
|
|
20
|
+
if (format === 'pdf') {
|
|
21
|
+
return {
|
|
22
|
+
file: resolvedPath,
|
|
23
|
+
format,
|
|
24
|
+
size: fileStats.size,
|
|
25
|
+
tableCount: 0,
|
|
26
|
+
tables: [],
|
|
27
|
+
notes: ['PDF format does not support structural table extraction. Tables in PDFs are rendered as positioned text without semantic structure. Consider converting to DOCX first for table extraction.'],
|
|
28
|
+
totalTokenEstimate: 0,
|
|
29
|
+
};
|
|
30
|
+
}
|
|
31
|
+
const buffer = await readFile(resolvedPath);
|
|
32
|
+
let tables;
|
|
33
|
+
switch (format) {
|
|
34
|
+
case 'docx':
|
|
35
|
+
tables = await extractDocxTables(resolvedPath, options);
|
|
36
|
+
break;
|
|
37
|
+
case 'xlsx':
|
|
38
|
+
tables = await extractXlsxTables(buffer, options);
|
|
39
|
+
break;
|
|
40
|
+
case 'pptx':
|
|
41
|
+
tables = await extractPptxTables(buffer, options);
|
|
42
|
+
break;
|
|
43
|
+
case 'odt':
|
|
44
|
+
tables = await extractOdtTables(buffer, options);
|
|
45
|
+
break;
|
|
46
|
+
case 'odp':
|
|
47
|
+
tables = await extractOdpTables(buffer, options);
|
|
48
|
+
break;
|
|
49
|
+
}
|
|
50
|
+
// Filter to specific table if requested
|
|
51
|
+
if (options.table && tables) {
|
|
52
|
+
tables = tables.filter(t => t.tableIndex === options.table);
|
|
53
|
+
}
|
|
54
|
+
const totalTokenEstimate = tables.reduce((sum, t) => sum + t.tokenEstimate, 0);
|
|
55
|
+
return {
|
|
56
|
+
file: resolvedPath,
|
|
57
|
+
format,
|
|
58
|
+
size: fileStats.size,
|
|
59
|
+
tableCount: tables.length,
|
|
60
|
+
tables,
|
|
61
|
+
notes: [],
|
|
62
|
+
totalTokenEstimate,
|
|
63
|
+
};
|
|
64
|
+
}
|
|
65
|
+
//# sourceMappingURL=inspect.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"inspect.js","sourceRoot":"","sources":["../../../src/table/inspect.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,kBAAkB,CAAC;AAClD,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AACxB,OAAO,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AAC3C,OAAO,EAAE,iBAAiB,EAAE,MAAM,mBAAmB,CAAC;AACtD,OAAO,EAAE,iBAAiB,EAAE,MAAM,mBAAmB,CAAC;AACtD,OAAO,EAAE,iBAAiB,EAAE,MAAM,mBAAmB,CAAC;AACtD,OAAO,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAC;AACpD,OAAO,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAC;AAGpD,MAAM,iBAAiB,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC;AAChF,MAAM,iBAAiB,GAAwB,IAAI,GAAG,CAAC,iBAAiB,CAAC,OAAO,CAAC,CAAC;AAElF,MAAM,CAAC,KAAK,UAAU,aAAa,CAAC,QAAgB,EAAE,OAA4B;IAChF,MAAM,YAAY,GAAG,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC;IAC5C,MAAM,GAAG,GAAG,YAAY,CAAC,YAAY,CAAC,CAAC;IAEvC,IAAI,CAAC,iBAAiB,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;QAChC,MAAM,IAAI,KAAK,CAAC,wBAAwB,GAAG,IAAI,QAAQ,+CAA+C,CAAC,CAAC;IAC1G,CAAC;IAED,MAAM,SAAS,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,CAAC;IAC3C,MAAM,MAAM,GAAG,iBAAiB,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IAE5C,IAAI,MAAM,KAAK,KAAK,EAAE,CAAC;QACrB,OAAO;YACL,IAAI,EAAE,YAAY;YAClB,MAAM;YACN,IAAI,EAAE,SAAS,CAAC,IAAI;YACpB,UAAU,EAAE,CAAC;YACb,MAAM,EAAE,EAAE;YACV,KAAK,EAAE,CAAC,6LAA6L,CAAC;YACtM,kBAAkB,EAAE,CAAC;SACtB,CAAC;IACJ,CAAC;IAED,MAAM,MAAM,GAAG,MAAM,QAAQ,CAAC,YAAY,CAAC,CAAC;IAC5C,IAAI,MAAM,CAAC;IAEX,QAAQ,MAAM,EAAE,CAAC;QACf,KAAK,MAAM;YACT,MAAM,GAAG,MAAM,iBAAiB,CAAC,YAAY,EAAE,OAAO,CAAC,CAAC;YACxD,MAAM;QACR,KAAK,MAAM;YACT,MAAM,GAAG,MAAM,iBAAiB,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;YAClD,MAAM;QACR,KAAK,MAAM;YACT,MAAM,GAAG,MAAM,iBAAiB,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;YAClD,MAAM;QACR,KAAK,KAAK;YACR,MAAM,GAAG,MAAM,gBAAgB,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;YACjD,MAAM;QACR,KAAK,KAAK;YACR,MAAM,GAAG,MAAM,gBAAgB,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;YACjD,MAAM;IACV,CAAC;IAED,wCAAwC;IACxC,IAAI,OAAO,CAAC,KAAK,IAAI,MAAM,EAAE,CAAC;QAC5B,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,UAAU,KAAK,OAAO,CAAC,KAAK,CAAC,CAAC;IAC9D,CAAC;IAED,MAAM,kBAAkB,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,aAAa,EAAE,CAAC,CAAC,CAAC;IAE/E,OAAO;QACL,IAAI,EAAE,YAAY;QAClB,MAAM;QACN,IAAI,EAAE,SAAS,CAAC,IAAI;QACpB,UAAU,EAAE,MAAM,CAAC,MAAM;QACzB,MAAM;QACN,KAAK,EAAE,EAAE;QACT,kBAAkB;KACnB,CAAC;AACJ,CAAC"}
|