@cesarandreslopez/occ 0.3.1 → 0.4.0
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.js +8 -0
- package/dist/src/cli.js.map +1 -1
- package/dist/src/code/build.js +22 -9
- package/dist/src/code/build.js.map +1 -1
- package/dist/src/code/command.js +2 -2
- package/dist/src/code/command.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 +52 -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 +5 -3
- 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 +91 -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 +82 -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 +59 -0
- package/dist/src/doc/types.js +2 -0
- package/dist/src/doc/types.js.map +1 -0
- package/dist/src/inspect/shared.d.ts +18 -0
- package/dist/src/inspect/shared.js +18 -0
- package/dist/src/inspect/shared.js.map +1 -0
- package/dist/src/parsers/xlsx.js +4 -4
- package/dist/src/parsers/xlsx.js.map +1 -1
- package/dist/src/sheet/command.d.ts +2 -0
- package/dist/src/sheet/command.js +70 -0
- package/dist/src/sheet/command.js.map +1 -0
- package/dist/src/sheet/inspect.d.ts +7 -0
- package/dist/src/sheet/inspect.js +454 -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 +117 -0
- package/dist/src/sheet/types.js +2 -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 +59 -0
- package/dist/src/slide/types.js +2 -0
- package/dist/src/slide/types.js.map +1 -0
- package/dist/src/table/command.d.ts +2 -0
- package/dist/src/table/command.js +68 -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 +63 -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 +38 -0
- package/dist/src/table/types.js +2 -0
- package/dist/src/table/types.js.map +1 -0
- package/package.json +1 -1
|
@@ -0,0 +1,188 @@
|
|
|
1
|
+
import JSZip from 'jszip';
|
|
2
|
+
import mammoth from 'mammoth';
|
|
3
|
+
import { countWords } from '../utils.js';
|
|
4
|
+
import { asOptionalString, formatDateLike } from '../inspect/shared.js';
|
|
5
|
+
function parseXmlText(xml, tag) {
|
|
6
|
+
const regex = new RegExp(`<${tag}[^>]*>([^<]*)</${tag}>`, 'i');
|
|
7
|
+
const match = xml.match(regex);
|
|
8
|
+
return match?.[1]?.trim() || undefined;
|
|
9
|
+
}
|
|
10
|
+
export async function extractDocxProperties(zip) {
|
|
11
|
+
const props = {};
|
|
12
|
+
const coreXml = await zip.file('docProps/core.xml')?.async('text');
|
|
13
|
+
if (coreXml) {
|
|
14
|
+
props.title = asOptionalString(parseXmlText(coreXml, 'dc:title'));
|
|
15
|
+
props.subject = asOptionalString(parseXmlText(coreXml, 'dc:subject'));
|
|
16
|
+
props.author = asOptionalString(parseXmlText(coreXml, 'dc:creator'));
|
|
17
|
+
props.keywords = asOptionalString(parseXmlText(coreXml, 'cp:keywords'));
|
|
18
|
+
props.lastModifiedBy = asOptionalString(parseXmlText(coreXml, 'cp:lastModifiedBy'));
|
|
19
|
+
props.createdDate = formatDateLike(parseXmlText(coreXml, 'dcterms:created'));
|
|
20
|
+
props.modifiedDate = formatDateLike(parseXmlText(coreXml, 'dcterms:modified'));
|
|
21
|
+
}
|
|
22
|
+
const appXml = await zip.file('docProps/app.xml')?.async('text');
|
|
23
|
+
if (appXml) {
|
|
24
|
+
props.company = asOptionalString(parseXmlText(appXml, 'Company'));
|
|
25
|
+
}
|
|
26
|
+
return props;
|
|
27
|
+
}
|
|
28
|
+
export async function extractDocxRiskFlags(zip) {
|
|
29
|
+
const flags = {
|
|
30
|
+
comments: false,
|
|
31
|
+
trackedChanges: false,
|
|
32
|
+
hyperlinks: false,
|
|
33
|
+
embeddedObjects: false,
|
|
34
|
+
footnotes: false,
|
|
35
|
+
endnotes: false,
|
|
36
|
+
macros: false,
|
|
37
|
+
headerFooter: false,
|
|
38
|
+
tables: false,
|
|
39
|
+
encrypted: false,
|
|
40
|
+
};
|
|
41
|
+
// Check comments
|
|
42
|
+
const commentsXml = await zip.file('word/comments.xml')?.async('text');
|
|
43
|
+
if (commentsXml && /<w:comment\b/.test(commentsXml)) {
|
|
44
|
+
flags.comments = true;
|
|
45
|
+
}
|
|
46
|
+
// Check document.xml for tracked changes, tables, hyperlinks
|
|
47
|
+
const docXml = await zip.file('word/document.xml')?.async('text');
|
|
48
|
+
if (docXml) {
|
|
49
|
+
if (/<w:ins\b/.test(docXml) || /<w:del\b/.test(docXml))
|
|
50
|
+
flags.trackedChanges = true;
|
|
51
|
+
if (/<w:tbl\b/.test(docXml))
|
|
52
|
+
flags.tables = true;
|
|
53
|
+
if (/<w:hyperlink\b/.test(docXml))
|
|
54
|
+
flags.hyperlinks = true;
|
|
55
|
+
}
|
|
56
|
+
// Check for macros
|
|
57
|
+
if (zip.file('word/vbaProject.bin'))
|
|
58
|
+
flags.macros = true;
|
|
59
|
+
// Check for images
|
|
60
|
+
const mediaFiles = Object.keys(zip.files).filter(name => name.startsWith('word/media/'));
|
|
61
|
+
if (mediaFiles.length > 0)
|
|
62
|
+
flags.embeddedObjects = true;
|
|
63
|
+
// Check footnotes/endnotes
|
|
64
|
+
const footnotesXml = await zip.file('word/footnotes.xml')?.async('text');
|
|
65
|
+
if (footnotesXml && /<w:footnote\b/.test(footnotesXml)) {
|
|
66
|
+
// Word always has footnotes.xml with separator/continuationSeparator; check for actual footnotes
|
|
67
|
+
const realFootnotes = (footnotesXml.match(/<w:footnote\b/g) || []).length;
|
|
68
|
+
if (realFootnotes > 2)
|
|
69
|
+
flags.footnotes = true; // skip the 2 built-in
|
|
70
|
+
}
|
|
71
|
+
const endnotesXml = await zip.file('word/endnotes.xml')?.async('text');
|
|
72
|
+
if (endnotesXml && /<w:endnote\b/.test(endnotesXml)) {
|
|
73
|
+
const realEndnotes = (endnotesXml.match(/<w:endnote\b/g) || []).length;
|
|
74
|
+
if (realEndnotes > 2)
|
|
75
|
+
flags.endnotes = true;
|
|
76
|
+
}
|
|
77
|
+
// Check headers/footers
|
|
78
|
+
const headerFiles = Object.keys(zip.files).filter(name => /^word\/(header|footer)\d*\.xml$/.test(name));
|
|
79
|
+
if (headerFiles.length > 0)
|
|
80
|
+
flags.headerFooter = true;
|
|
81
|
+
return flags;
|
|
82
|
+
}
|
|
83
|
+
export async function extractDocxContentStats(filePath, zip) {
|
|
84
|
+
const result = await mammoth.extractRawText({ path: filePath });
|
|
85
|
+
const text = result.value || '';
|
|
86
|
+
const words = countWords(text);
|
|
87
|
+
const paragraphs = text.split(/\n\n+/).filter(p => p.trim().length > 0).length;
|
|
88
|
+
const characters = text.length;
|
|
89
|
+
// Pages from app.xml if available
|
|
90
|
+
const appXml = await zip.file('docProps/app.xml')?.async('text');
|
|
91
|
+
let pages = 0;
|
|
92
|
+
let pagesEstimated = true;
|
|
93
|
+
if (appXml) {
|
|
94
|
+
const pagesStr = parseXmlText(appXml, 'Pages');
|
|
95
|
+
if (pagesStr) {
|
|
96
|
+
const parsed = parseInt(pagesStr, 10);
|
|
97
|
+
if (parsed > 0) {
|
|
98
|
+
pages = parsed;
|
|
99
|
+
pagesEstimated = false;
|
|
100
|
+
}
|
|
101
|
+
}
|
|
102
|
+
}
|
|
103
|
+
if (pages === 0) {
|
|
104
|
+
pages = Math.max(1, Math.ceil(words / 250));
|
|
105
|
+
pagesEstimated = true;
|
|
106
|
+
}
|
|
107
|
+
// Count tables from HTML conversion
|
|
108
|
+
let tables = 0;
|
|
109
|
+
try {
|
|
110
|
+
const htmlResult = await mammoth.convertToHtml({ path: filePath });
|
|
111
|
+
const html = htmlResult.value || '';
|
|
112
|
+
tables = (html.match(/<table\b/gi) || []).length;
|
|
113
|
+
}
|
|
114
|
+
catch { /* ignore */ }
|
|
115
|
+
// Count images
|
|
116
|
+
const mediaFiles = Object.keys(zip.files).filter(name => name.startsWith('word/media/'));
|
|
117
|
+
const images = mediaFiles.length;
|
|
118
|
+
return { words, pages, pagesEstimated, paragraphs, characters, tables, images };
|
|
119
|
+
}
|
|
120
|
+
export async function extractDocxPreview(filePath, sampleParagraphs) {
|
|
121
|
+
const result = await mammoth.convertToHtml({ path: filePath });
|
|
122
|
+
const html = result.value || '';
|
|
123
|
+
if (!html.trim()) {
|
|
124
|
+
return { truncated: false, paragraphs: [] };
|
|
125
|
+
}
|
|
126
|
+
// Use turndown to convert to markdown for heading detection
|
|
127
|
+
const TurndownService = (await import('turndown')).default;
|
|
128
|
+
const turndown = new TurndownService({ headingStyle: 'atx' });
|
|
129
|
+
const markdown = turndown.turndown(html);
|
|
130
|
+
return buildPreviewFromMarkdown(markdown, sampleParagraphs);
|
|
131
|
+
}
|
|
132
|
+
export function buildPreviewFromMarkdown(markdown, sampleParagraphs) {
|
|
133
|
+
const lines = markdown.split('\n');
|
|
134
|
+
const previews = [];
|
|
135
|
+
let paragraphIndex = 0;
|
|
136
|
+
let i = 0;
|
|
137
|
+
while (i < lines.length && previews.length < sampleParagraphs) {
|
|
138
|
+
const line = lines[i].trim();
|
|
139
|
+
if (!line) {
|
|
140
|
+
i++;
|
|
141
|
+
continue;
|
|
142
|
+
}
|
|
143
|
+
const headingMatch = line.match(/^(#{1,6})\s+(.+)$/);
|
|
144
|
+
if (headingMatch) {
|
|
145
|
+
previews.push({
|
|
146
|
+
index: paragraphIndex,
|
|
147
|
+
text: headingMatch[2],
|
|
148
|
+
isHeading: true,
|
|
149
|
+
headingLevel: headingMatch[1].length,
|
|
150
|
+
});
|
|
151
|
+
}
|
|
152
|
+
else {
|
|
153
|
+
// Collect contiguous non-empty lines as one paragraph
|
|
154
|
+
let text = line;
|
|
155
|
+
while (i + 1 < lines.length && lines[i + 1].trim() && !lines[i + 1].trim().match(/^#{1,6}\s/)) {
|
|
156
|
+
i++;
|
|
157
|
+
text += ' ' + lines[i].trim();
|
|
158
|
+
}
|
|
159
|
+
previews.push({
|
|
160
|
+
index: paragraphIndex,
|
|
161
|
+
text,
|
|
162
|
+
isHeading: false,
|
|
163
|
+
});
|
|
164
|
+
}
|
|
165
|
+
paragraphIndex++;
|
|
166
|
+
i++;
|
|
167
|
+
}
|
|
168
|
+
// Check for remaining content
|
|
169
|
+
let hasMore = false;
|
|
170
|
+
for (let j = i; j < lines.length; j++) {
|
|
171
|
+
if (lines[j].trim()) {
|
|
172
|
+
hasMore = true;
|
|
173
|
+
break;
|
|
174
|
+
}
|
|
175
|
+
}
|
|
176
|
+
return { truncated: hasMore, paragraphs: previews };
|
|
177
|
+
}
|
|
178
|
+
export async function inspectDocx(filePath, buffer, sampleParagraphs) {
|
|
179
|
+
const zip = await JSZip.loadAsync(buffer);
|
|
180
|
+
const [properties, riskFlags, contentStats, preview] = await Promise.all([
|
|
181
|
+
extractDocxProperties(zip),
|
|
182
|
+
extractDocxRiskFlags(zip),
|
|
183
|
+
extractDocxContentStats(filePath, zip),
|
|
184
|
+
extractDocxPreview(filePath, sampleParagraphs),
|
|
185
|
+
]);
|
|
186
|
+
return { properties, riskFlags, contentStats, preview };
|
|
187
|
+
}
|
|
188
|
+
//# sourceMappingURL=inspect-docx.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"inspect-docx.js","sourceRoot":"","sources":["../../../src/doc/inspect-docx.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,OAAO,MAAM,SAAS,CAAC;AAC9B,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,gBAAgB,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAIxE,SAAS,YAAY,CAAC,GAAW,EAAE,GAAW;IAC5C,MAAM,KAAK,GAAG,IAAI,MAAM,CAAC,IAAI,GAAG,kBAAkB,GAAG,GAAG,EAAE,GAAG,CAAC,CAAC;IAC/D,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;IAC/B,OAAO,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,IAAI,SAAS,CAAC;AACzC,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,qBAAqB,CAAC,GAAU;IACpD,MAAM,KAAK,GAAuB,EAAE,CAAC;IAErC,MAAM,OAAO,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,mBAAmB,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IACnE,IAAI,OAAO,EAAE,CAAC;QACZ,KAAK,CAAC,KAAK,GAAG,gBAAgB,CAAC,YAAY,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC,CAAC;QAClE,KAAK,CAAC,OAAO,GAAG,gBAAgB,CAAC,YAAY,CAAC,OAAO,EAAE,YAAY,CAAC,CAAC,CAAC;QACtE,KAAK,CAAC,MAAM,GAAG,gBAAgB,CAAC,YAAY,CAAC,OAAO,EAAE,YAAY,CAAC,CAAC,CAAC;QACrE,KAAK,CAAC,QAAQ,GAAG,gBAAgB,CAAC,YAAY,CAAC,OAAO,EAAE,aAAa,CAAC,CAAC,CAAC;QACxE,KAAK,CAAC,cAAc,GAAG,gBAAgB,CAAC,YAAY,CAAC,OAAO,EAAE,mBAAmB,CAAC,CAAC,CAAC;QACpF,KAAK,CAAC,WAAW,GAAG,cAAc,CAAC,YAAY,CAAC,OAAO,EAAE,iBAAiB,CAAC,CAAC,CAAC;QAC7E,KAAK,CAAC,YAAY,GAAG,cAAc,CAAC,YAAY,CAAC,OAAO,EAAE,kBAAkB,CAAC,CAAC,CAAC;IACjF,CAAC;IAED,MAAM,MAAM,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,kBAAkB,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IACjE,IAAI,MAAM,EAAE,CAAC;QACX,KAAK,CAAC,OAAO,GAAG,gBAAgB,CAAC,YAAY,CAAC,MAAM,EAAE,SAAS,CAAC,CAAC,CAAC;IACpE,CAAC;IAED,OAAO,KAAK,CAAC;AACf,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,oBAAoB,CAAC,GAAU;IACnD,MAAM,KAAK,GAAiB;QAC1B,QAAQ,EAAE,KAAK;QACf,cAAc,EAAE,KAAK;QACrB,UAAU,EAAE,KAAK;QACjB,eAAe,EAAE,KAAK;QACtB,SAAS,EAAE,KAAK;QAChB,QAAQ,EAAE,KAAK;QACf,MAAM,EAAE,KAAK;QACb,YAAY,EAAE,KAAK;QACnB,MAAM,EAAE,KAAK;QACb,SAAS,EAAE,KAAK;KACjB,CAAC;IAEF,iBAAiB;IACjB,MAAM,WAAW,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,mBAAmB,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IACvE,IAAI,WAAW,IAAI,cAAc,CAAC,IAAI,CAAC,WAAW,CAAC,EAAE,CAAC;QACpD,KAAK,CAAC,QAAQ,GAAG,IAAI,CAAC;IACxB,CAAC;IAED,6DAA6D;IAC7D,MAAM,MAAM,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,mBAAmB,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IAClE,IAAI,MAAM,EAAE,CAAC;QACX,IAAI,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC;YAAE,KAAK,CAAC,cAAc,GAAG,IAAI,CAAC;QACpF,IAAI,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC;YAAE,KAAK,CAAC,MAAM,GAAG,IAAI,CAAC;QACjD,IAAI,gBAAgB,CAAC,IAAI,CAAC,MAAM,CAAC;YAAE,KAAK,CAAC,UAAU,GAAG,IAAI,CAAC;IAC7D,CAAC;IAED,mBAAmB;IACnB,IAAI,GAAG,CAAC,IAAI,CAAC,qBAAqB,CAAC;QAAE,KAAK,CAAC,MAAM,GAAG,IAAI,CAAC;IAEzD,mBAAmB;IACnB,MAAM,UAAU,GAAG,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,UAAU,CAAC,aAAa,CAAC,CAAC,CAAC;IACzF,IAAI,UAAU,CAAC,MAAM,GAAG,CAAC;QAAE,KAAK,CAAC,eAAe,GAAG,IAAI,CAAC;IAExD,2BAA2B;IAC3B,MAAM,YAAY,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,oBAAoB,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IACzE,IAAI,YAAY,IAAI,eAAe,CAAC,IAAI,CAAC,YAAY,CAAC,EAAE,CAAC;QACvD,iGAAiG;QACjG,MAAM,aAAa,GAAG,CAAC,YAAY,CAAC,KAAK,CAAC,gBAAgB,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC;QAC1E,IAAI,aAAa,GAAG,CAAC;YAAE,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,CAAC,sBAAsB;IACvE,CAAC;IAED,MAAM,WAAW,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,mBAAmB,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IACvE,IAAI,WAAW,IAAI,cAAc,CAAC,IAAI,CAAC,WAAW,CAAC,EAAE,CAAC;QACpD,MAAM,YAAY,GAAG,CAAC,WAAW,CAAC,KAAK,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC;QACvE,IAAI,YAAY,GAAG,CAAC;YAAE,KAAK,CAAC,QAAQ,GAAG,IAAI,CAAC;IAC9C,CAAC;IAED,wBAAwB;IACxB,MAAM,WAAW,GAAG,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,iCAAiC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IACxG,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC;QAAE,KAAK,CAAC,YAAY,GAAG,IAAI,CAAC;IAEtD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,uBAAuB,CAAC,QAAgB,EAAE,GAAU;IACxE,MAAM,MAAM,GAAG,MAAM,OAAO,CAAC,cAAc,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,CAAC;IAChE,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC;IAChC,MAAM,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC;IAC/B,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC;IAC/E,MAAM,UAAU,GAAG,IAAI,CAAC,MAAM,CAAC;IAE/B,kCAAkC;IAClC,MAAM,MAAM,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,kBAAkB,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IACjE,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,IAAI,cAAc,GAAG,IAAI,CAAC;IAC1B,IAAI,MAAM,EAAE,CAAC;QACX,MAAM,QAAQ,GAAG,YAAY,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;QAC/C,IAAI,QAAQ,EAAE,CAAC;YACb,MAAM,MAAM,GAAG,QAAQ,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;YACtC,IAAI,MAAM,GAAG,CAAC,EAAE,CAAC;gBACf,KAAK,GAAG,MAAM,CAAC;gBACf,cAAc,GAAG,KAAK,CAAC;YACzB,CAAC;QACH,CAAC;IACH,CAAC;IACD,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;QAChB,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,CAAC,CAAC;QAC5C,cAAc,GAAG,IAAI,CAAC;IACxB,CAAC;IAED,oCAAoC;IACpC,IAAI,MAAM,GAAG,CAAC,CAAC;IACf,IAAI,CAAC;QACH,MAAM,UAAU,GAAG,MAAM,OAAO,CAAC,aAAa,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,CAAC;QACnE,MAAM,IAAI,GAAG,UAAU,CAAC,KAAK,IAAI,EAAE,CAAC;QACpC,MAAM,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC;IACnD,CAAC;IAAC,MAAM,CAAC,CAAC,YAAY,CAAC,CAAC;IAExB,eAAe;IACf,MAAM,UAAU,GAAG,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,UAAU,CAAC,aAAa,CAAC,CAAC,CAAC;IACzF,MAAM,MAAM,GAAG,UAAU,CAAC,MAAM,CAAC;IAEjC,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,cAAc,EAAE,UAAU,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC;AAClF,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,kBAAkB,CAAC,QAAgB,EAAE,gBAAwB;IACjF,MAAM,MAAM,GAAG,MAAM,OAAO,CAAC,aAAa,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,CAAC;IAC/D,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC;IAEhC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC;QACjB,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,UAAU,EAAE,EAAE,EAAE,CAAC;IAC9C,CAAC;IAED,4DAA4D;IAC5D,MAAM,eAAe,GAAG,CAAC,MAAM,MAAM,CAAC,UAAU,CAAC,CAAC,CAAC,OAAO,CAAC;IAC3D,MAAM,QAAQ,GAAG,IAAI,eAAe,CAAC,EAAE,YAAY,EAAE,KAAK,EAAE,CAAC,CAAC;IAC9D,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;IAEzC,OAAO,wBAAwB,CAAC,QAAQ,EAAE,gBAAgB,CAAC,CAAC;AAC9D,CAAC;AAED,MAAM,UAAU,wBAAwB,CAAC,QAAgB,EAAE,gBAAwB;IACjF,MAAM,KAAK,GAAG,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACnC,MAAM,QAAQ,GAAiC,EAAE,CAAC;IAClD,IAAI,cAAc,GAAG,CAAC,CAAC;IAEvB,IAAI,CAAC,GAAG,CAAC,CAAC;IACV,OAAO,CAAC,GAAG,KAAK,CAAC,MAAM,IAAI,QAAQ,CAAC,MAAM,GAAG,gBAAgB,EAAE,CAAC;QAC9D,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QAC7B,IAAI,CAAC,IAAI,EAAE,CAAC;YACV,CAAC,EAAE,CAAC;YACJ,SAAS;QACX,CAAC;QAED,MAAM,YAAY,GAAG,IAAI,CAAC,KAAK,CAAC,mBAAmB,CAAC,CAAC;QACrD,IAAI,YAAY,EAAE,CAAC;YACjB,QAAQ,CAAC,IAAI,CAAC;gBACZ,KAAK,EAAE,cAAc;gBACrB,IAAI,EAAE,YAAY,CAAC,CAAC,CAAC;gBACrB,SAAS,EAAE,IAAI;gBACf,YAAY,EAAE,YAAY,CAAC,CAAC,CAAC,CAAC,MAAM;aACrC,CAAC,CAAC;QACL,CAAC;aAAM,CAAC;YACN,sDAAsD;YACtD,IAAI,IAAI,GAAG,IAAI,CAAC;YAChB,OAAO,CAAC,GAAG,CAAC,GAAG,KAAK,CAAC,MAAM,IAAI,KAAK,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,WAAW,CAAC,EAAE,CAAC;gBAC9F,CAAC,EAAE,CAAC;gBACJ,IAAI,IAAI,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YAChC,CAAC;YACD,QAAQ,CAAC,IAAI,CAAC;gBACZ,KAAK,EAAE,cAAc;gBACrB,IAAI;gBACJ,SAAS,EAAE,KAAK;aACjB,CAAC,CAAC;QACL,CAAC;QACD,cAAc,EAAE,CAAC;QACjB,CAAC,EAAE,CAAC;IACN,CAAC;IAED,8BAA8B;IAC9B,IAAI,OAAO,GAAG,KAAK,CAAC;IACpB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC;YACpB,OAAO,GAAG,IAAI,CAAC;YACf,MAAM;QACR,CAAC;IACH,CAAC;IAED,OAAO,EAAE,SAAS,EAAE,OAAO,EAAE,UAAU,EAAE,QAAQ,EAAE,CAAC;AACtD,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,WAAW,CAC/B,QAAgB,EAChB,MAAc,EACd,gBAAwB;IAOxB,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;IAC1C,MAAM,CAAC,UAAU,EAAE,SAAS,EAAE,YAAY,EAAE,OAAO,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;QACvE,qBAAqB,CAAC,GAAG,CAAC;QAC1B,oBAAoB,CAAC,GAAG,CAAC;QACzB,uBAAuB,CAAC,QAAQ,EAAE,GAAG,CAAC;QACtC,kBAAkB,CAAC,QAAQ,EAAE,gBAAgB,CAAC;KAC/C,CAAC,CAAC;IAEH,OAAO,EAAE,UAAU,EAAE,SAAS,EAAE,YAAY,EAAE,OAAO,EAAE,CAAC;AAC1D,CAAC"}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
import type { DocumentProperties } from '../inspect/shared.js';
|
|
2
|
+
import type { DocRiskFlags, DocContentStats, ContentPreview } from './types.js';
|
|
3
|
+
export declare function inspectOdt(filePath: string, buffer: Buffer, sampleParagraphs: number): Promise<{
|
|
4
|
+
properties: DocumentProperties;
|
|
5
|
+
riskFlags: DocRiskFlags;
|
|
6
|
+
contentStats: DocContentStats;
|
|
7
|
+
preview: ContentPreview;
|
|
8
|
+
}>;
|
|
@@ -0,0 +1,91 @@
|
|
|
1
|
+
import JSZip from 'jszip';
|
|
2
|
+
import officeparser from 'officeparser';
|
|
3
|
+
import { countWords } from '../utils.js';
|
|
4
|
+
import { asOptionalString, formatDateLike } from '../inspect/shared.js';
|
|
5
|
+
function parseXmlText(xml, tag) {
|
|
6
|
+
const regex = new RegExp(`<${tag}[^>]*>([^<]*)</${tag}>`, 'i');
|
|
7
|
+
const match = xml.match(regex);
|
|
8
|
+
return match?.[1]?.trim() || undefined;
|
|
9
|
+
}
|
|
10
|
+
export async function inspectOdt(filePath, buffer, sampleParagraphs) {
|
|
11
|
+
const zip = await JSZip.loadAsync(buffer);
|
|
12
|
+
// Parse metadata from meta.xml
|
|
13
|
+
const properties = {};
|
|
14
|
+
const metaXml = await zip.file('meta.xml')?.async('text');
|
|
15
|
+
if (metaXml) {
|
|
16
|
+
properties.title = asOptionalString(parseXmlText(metaXml, 'dc:title'));
|
|
17
|
+
properties.subject = asOptionalString(parseXmlText(metaXml, 'dc:subject'));
|
|
18
|
+
properties.author = asOptionalString(parseXmlText(metaXml, 'dc:creator') ?? parseXmlText(metaXml, 'meta:initial-creator'));
|
|
19
|
+
properties.keywords = asOptionalString(parseXmlText(metaXml, 'meta:keyword'));
|
|
20
|
+
properties.createdDate = formatDateLike(parseXmlText(metaXml, 'meta:creation-date'));
|
|
21
|
+
properties.modifiedDate = formatDateLike(parseXmlText(metaXml, 'dc:date'));
|
|
22
|
+
}
|
|
23
|
+
// Parse risk flags from content.xml
|
|
24
|
+
const riskFlags = {
|
|
25
|
+
comments: false,
|
|
26
|
+
trackedChanges: false,
|
|
27
|
+
hyperlinks: false,
|
|
28
|
+
embeddedObjects: false,
|
|
29
|
+
footnotes: false,
|
|
30
|
+
endnotes: false,
|
|
31
|
+
macros: false,
|
|
32
|
+
headerFooter: false,
|
|
33
|
+
tables: false,
|
|
34
|
+
encrypted: false,
|
|
35
|
+
};
|
|
36
|
+
const contentXml = await zip.file('content.xml')?.async('text');
|
|
37
|
+
if (contentXml) {
|
|
38
|
+
if (/<office:annotation\b/.test(contentXml))
|
|
39
|
+
riskFlags.comments = true;
|
|
40
|
+
if (/<text:tracked-changes\b/.test(contentXml))
|
|
41
|
+
riskFlags.trackedChanges = true;
|
|
42
|
+
if (/<text:a\b[^>]*xlink:href=/.test(contentXml))
|
|
43
|
+
riskFlags.hyperlinks = true;
|
|
44
|
+
if (/<draw:frame\b/.test(contentXml))
|
|
45
|
+
riskFlags.embeddedObjects = true;
|
|
46
|
+
if (/<text:note\b/.test(contentXml))
|
|
47
|
+
riskFlags.footnotes = true;
|
|
48
|
+
if (/<table:table\b/.test(contentXml))
|
|
49
|
+
riskFlags.tables = true;
|
|
50
|
+
}
|
|
51
|
+
// Check for macros
|
|
52
|
+
if (zip.file('Basic/') || zip.file('Scripts/'))
|
|
53
|
+
riskFlags.macros = true;
|
|
54
|
+
// Extract text via officeparser
|
|
55
|
+
const text = await officeparser.parseOffice(buffer);
|
|
56
|
+
const words = countWords(text);
|
|
57
|
+
const paragraphs = text.split(/\n\n+/).filter(p => p.trim().length > 0).length;
|
|
58
|
+
// Count images and tables from content.xml
|
|
59
|
+
let images = 0;
|
|
60
|
+
let tables = 0;
|
|
61
|
+
if (contentXml) {
|
|
62
|
+
images = (contentXml.match(/<draw:image\b/g) || []).length;
|
|
63
|
+
tables = (contentXml.match(/<table:table\b/g) || []).length;
|
|
64
|
+
}
|
|
65
|
+
const contentStats = {
|
|
66
|
+
words,
|
|
67
|
+
pages: Math.max(1, Math.ceil(words / 250)),
|
|
68
|
+
pagesEstimated: true,
|
|
69
|
+
paragraphs,
|
|
70
|
+
characters: text.length,
|
|
71
|
+
tables,
|
|
72
|
+
images,
|
|
73
|
+
};
|
|
74
|
+
// Build preview
|
|
75
|
+
const textParagraphs = text.split(/\n\n+/).filter(p => p.trim().length > 0);
|
|
76
|
+
const previewParagraphs = [];
|
|
77
|
+
const limit = Math.min(sampleParagraphs, textParagraphs.length);
|
|
78
|
+
for (let i = 0; i < limit; i++) {
|
|
79
|
+
previewParagraphs.push({
|
|
80
|
+
index: i,
|
|
81
|
+
text: textParagraphs[i].trim().replace(/\s+/g, ' '),
|
|
82
|
+
isHeading: false,
|
|
83
|
+
});
|
|
84
|
+
}
|
|
85
|
+
const preview = {
|
|
86
|
+
truncated: textParagraphs.length > sampleParagraphs,
|
|
87
|
+
paragraphs: previewParagraphs,
|
|
88
|
+
};
|
|
89
|
+
return { properties, riskFlags, contentStats, preview };
|
|
90
|
+
}
|
|
91
|
+
//# sourceMappingURL=inspect-odt.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"inspect-odt.js","sourceRoot":"","sources":["../../../src/doc/inspect-odt.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,YAAY,MAAM,cAAc,CAAC;AACxC,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,gBAAgB,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAIxE,SAAS,YAAY,CAAC,GAAW,EAAE,GAAW;IAC5C,MAAM,KAAK,GAAG,IAAI,MAAM,CAAC,IAAI,GAAG,kBAAkB,GAAG,GAAG,EAAE,GAAG,CAAC,CAAC;IAC/D,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;IAC/B,OAAO,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,IAAI,SAAS,CAAC;AACzC,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,UAAU,CAC9B,QAAgB,EAChB,MAAc,EACd,gBAAwB;IAOxB,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;IAE1C,+BAA+B;IAC/B,MAAM,UAAU,GAAuB,EAAE,CAAC;IAC1C,MAAM,OAAO,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IAC1D,IAAI,OAAO,EAAE,CAAC;QACZ,UAAU,CAAC,KAAK,GAAG,gBAAgB,CAAC,YAAY,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC,CAAC;QACvE,UAAU,CAAC,OAAO,GAAG,gBAAgB,CAAC,YAAY,CAAC,OAAO,EAAE,YAAY,CAAC,CAAC,CAAC;QAC3E,UAAU,CAAC,MAAM,GAAG,gBAAgB,CAAC,YAAY,CAAC,OAAO,EAAE,YAAY,CAAC,IAAI,YAAY,CAAC,OAAO,EAAE,sBAAsB,CAAC,CAAC,CAAC;QAC3H,UAAU,CAAC,QAAQ,GAAG,gBAAgB,CAAC,YAAY,CAAC,OAAO,EAAE,cAAc,CAAC,CAAC,CAAC;QAC9E,UAAU,CAAC,WAAW,GAAG,cAAc,CAAC,YAAY,CAAC,OAAO,EAAE,oBAAoB,CAAC,CAAC,CAAC;QACrF,UAAU,CAAC,YAAY,GAAG,cAAc,CAAC,YAAY,CAAC,OAAO,EAAE,SAAS,CAAC,CAAC,CAAC;IAC7E,CAAC;IAED,oCAAoC;IACpC,MAAM,SAAS,GAAiB;QAC9B,QAAQ,EAAE,KAAK;QACf,cAAc,EAAE,KAAK;QACrB,UAAU,EAAE,KAAK;QACjB,eAAe,EAAE,KAAK;QACtB,SAAS,EAAE,KAAK;QAChB,QAAQ,EAAE,KAAK;QACf,MAAM,EAAE,KAAK;QACb,YAAY,EAAE,KAAK;QACnB,MAAM,EAAE,KAAK;QACb,SAAS,EAAE,KAAK;KACjB,CAAC;IAEF,MAAM,UAAU,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IAChE,IAAI,UAAU,EAAE,CAAC;QACf,IAAI,sBAAsB,CAAC,IAAI,CAAC,UAAU,CAAC;YAAE,SAAS,CAAC,QAAQ,GAAG,IAAI,CAAC;QACvE,IAAI,yBAAyB,CAAC,IAAI,CAAC,UAAU,CAAC;YAAE,SAAS,CAAC,cAAc,GAAG,IAAI,CAAC;QAChF,IAAI,2BAA2B,CAAC,IAAI,CAAC,UAAU,CAAC;YAAE,SAAS,CAAC,UAAU,GAAG,IAAI,CAAC;QAC9E,IAAI,eAAe,CAAC,IAAI,CAAC,UAAU,CAAC;YAAE,SAAS,CAAC,eAAe,GAAG,IAAI,CAAC;QACvE,IAAI,cAAc,CAAC,IAAI,CAAC,UAAU,CAAC;YAAE,SAAS,CAAC,SAAS,GAAG,IAAI,CAAC;QAChE,IAAI,gBAAgB,CAAC,IAAI,CAAC,UAAU,CAAC;YAAE,SAAS,CAAC,MAAM,GAAG,IAAI,CAAC;IACjE,CAAC;IAED,mBAAmB;IACnB,IAAI,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC;QAAE,SAAS,CAAC,MAAM,GAAG,IAAI,CAAC;IAExE,gCAAgC;IAChC,MAAM,IAAI,GAAG,MAAM,YAAY,CAAC,WAAW,CAAC,MAAM,CAAsB,CAAC;IACzE,MAAM,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC;IAC/B,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC;IAE/E,2CAA2C;IAC3C,IAAI,MAAM,GAAG,CAAC,CAAC;IACf,IAAI,MAAM,GAAG,CAAC,CAAC;IACf,IAAI,UAAU,EAAE,CAAC;QACf,MAAM,GAAG,CAAC,UAAU,CAAC,KAAK,CAAC,gBAAgB,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC;QAC3D,MAAM,GAAG,CAAC,UAAU,CAAC,KAAK,CAAC,iBAAiB,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC;IAC9D,CAAC;IAED,MAAM,YAAY,GAAoB;QACpC,KAAK;QACL,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,CAAC;QAC1C,cAAc,EAAE,IAAI;QACpB,UAAU;QACV,UAAU,EAAE,IAAI,CAAC,MAAM;QACvB,MAAM;QACN,MAAM;KACP,CAAC;IAEF,gBAAgB;IAChB,MAAM,cAAc,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IAC5E,MAAM,iBAAiB,GAAiC,EAAE,CAAC;IAC3D,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,gBAAgB,EAAE,cAAc,CAAC,MAAM,CAAC,CAAC;IAChE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE,CAAC;QAC/B,iBAAiB,CAAC,IAAI,CAAC;YACrB,KAAK,EAAE,CAAC;YACR,IAAI,EAAE,cAAc,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC;YACnD,SAAS,EAAE,KAAK;SACjB,CAAC,CAAC;IACL,CAAC;IAED,MAAM,OAAO,GAAmB;QAC9B,SAAS,EAAE,cAAc,CAAC,MAAM,GAAG,gBAAgB;QACnD,UAAU,EAAE,iBAAiB;KAC9B,CAAC;IAEF,OAAO,EAAE,UAAU,EAAE,SAAS,EAAE,YAAY,EAAE,OAAO,EAAE,CAAC;AAC1D,CAAC"}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
import type { DocumentProperties } from '../inspect/shared.js';
|
|
2
|
+
import type { DocRiskFlags, DocContentStats, ContentPreview } from './types.js';
|
|
3
|
+
export declare function inspectPdf(filePath: string, buffer: Buffer, sampleParagraphs: number): Promise<{
|
|
4
|
+
properties: DocumentProperties;
|
|
5
|
+
riskFlags: DocRiskFlags;
|
|
6
|
+
contentStats: DocContentStats;
|
|
7
|
+
preview: ContentPreview;
|
|
8
|
+
}>;
|
|
@@ -0,0 +1,82 @@
|
|
|
1
|
+
import pdf from 'pdf-parse';
|
|
2
|
+
import { countWords } from '../utils.js';
|
|
3
|
+
import { asOptionalString, formatDateLike } from '../inspect/shared.js';
|
|
4
|
+
function parsePdfDate(value) {
|
|
5
|
+
if (!value)
|
|
6
|
+
return undefined;
|
|
7
|
+
const str = String(value);
|
|
8
|
+
// PDF dates: D:YYYYMMDDHHmmSS or plain ISO strings
|
|
9
|
+
const match = str.match(/^D:(\d{4})(\d{2})(\d{2})(\d{2})(\d{2})(\d{2})/);
|
|
10
|
+
if (match) {
|
|
11
|
+
return `${match[1]}-${match[2]}-${match[3]}T${match[4]}:${match[5]}:${match[6]}Z`;
|
|
12
|
+
}
|
|
13
|
+
return formatDateLike(str);
|
|
14
|
+
}
|
|
15
|
+
export async function inspectPdf(filePath, buffer, sampleParagraphs) {
|
|
16
|
+
// Suppress pdf.js warnings
|
|
17
|
+
const originalLog = console.log;
|
|
18
|
+
console.log = (...args) => {
|
|
19
|
+
if (typeof args[0] === 'string' && (args[0].startsWith('Warning: ') || args[0].startsWith('Info: ') || args[0].startsWith('Deprecated API usage: '))) {
|
|
20
|
+
return;
|
|
21
|
+
}
|
|
22
|
+
originalLog.apply(console, args);
|
|
23
|
+
};
|
|
24
|
+
let data;
|
|
25
|
+
try {
|
|
26
|
+
data = await pdf(buffer);
|
|
27
|
+
}
|
|
28
|
+
finally {
|
|
29
|
+
console.log = originalLog;
|
|
30
|
+
}
|
|
31
|
+
const info = data.info || {};
|
|
32
|
+
const properties = {
|
|
33
|
+
title: asOptionalString(info.Title),
|
|
34
|
+
subject: asOptionalString(info.Subject),
|
|
35
|
+
author: asOptionalString(info.Author),
|
|
36
|
+
keywords: asOptionalString(info.Keywords),
|
|
37
|
+
createdDate: parsePdfDate(info.CreationDate),
|
|
38
|
+
modifiedDate: parsePdfDate(info.ModDate),
|
|
39
|
+
};
|
|
40
|
+
// PDFs have limited risk flag detection
|
|
41
|
+
const riskFlags = {
|
|
42
|
+
comments: false,
|
|
43
|
+
trackedChanges: false,
|
|
44
|
+
hyperlinks: false,
|
|
45
|
+
embeddedObjects: false,
|
|
46
|
+
footnotes: false,
|
|
47
|
+
endnotes: false,
|
|
48
|
+
macros: false,
|
|
49
|
+
headerFooter: false,
|
|
50
|
+
tables: false,
|
|
51
|
+
encrypted: !!info.IsAcroFormPresent || info.Encrypted === true,
|
|
52
|
+
};
|
|
53
|
+
const text = data.text || '';
|
|
54
|
+
const words = countWords(text);
|
|
55
|
+
const paragraphs = text.split(/\n\n+/).filter(p => p.trim().length > 0).length;
|
|
56
|
+
const contentStats = {
|
|
57
|
+
words,
|
|
58
|
+
pages: data.numpages,
|
|
59
|
+
pagesEstimated: false,
|
|
60
|
+
paragraphs,
|
|
61
|
+
characters: text.length,
|
|
62
|
+
tables: 0,
|
|
63
|
+
images: 0,
|
|
64
|
+
};
|
|
65
|
+
// Build preview from text
|
|
66
|
+
const textParagraphs = text.split(/\n\n+/).filter(p => p.trim().length > 0);
|
|
67
|
+
const previewParagraphs = [];
|
|
68
|
+
const limit = Math.min(sampleParagraphs, textParagraphs.length);
|
|
69
|
+
for (let i = 0; i < limit; i++) {
|
|
70
|
+
previewParagraphs.push({
|
|
71
|
+
index: i,
|
|
72
|
+
text: textParagraphs[i].trim().replace(/\s+/g, ' '),
|
|
73
|
+
isHeading: false,
|
|
74
|
+
});
|
|
75
|
+
}
|
|
76
|
+
const preview = {
|
|
77
|
+
truncated: textParagraphs.length > sampleParagraphs,
|
|
78
|
+
paragraphs: previewParagraphs,
|
|
79
|
+
};
|
|
80
|
+
return { properties, riskFlags, contentStats, preview };
|
|
81
|
+
}
|
|
82
|
+
//# sourceMappingURL=inspect-pdf.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"inspect-pdf.js","sourceRoot":"","sources":["../../../src/doc/inspect-pdf.ts"],"names":[],"mappings":"AACA,OAAO,GAAG,MAAM,WAAW,CAAC;AAC5B,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,gBAAgB,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAIxE,SAAS,YAAY,CAAC,KAAc;IAClC,IAAI,CAAC,KAAK;QAAE,OAAO,SAAS,CAAC;IAC7B,MAAM,GAAG,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;IAC1B,mDAAmD;IACnD,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,+CAA+C,CAAC,CAAC;IACzE,IAAI,KAAK,EAAE,CAAC;QACV,OAAO,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC;IACpF,CAAC;IACD,OAAO,cAAc,CAAC,GAAG,CAAC,CAAC;AAC7B,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,UAAU,CAC9B,QAAgB,EAChB,MAAc,EACd,gBAAwB;IAOxB,2BAA2B;IAC3B,MAAM,WAAW,GAAG,OAAO,CAAC,GAAG,CAAC;IAChC,OAAO,CAAC,GAAG,GAAG,CAAC,GAAG,IAAe,EAAE,EAAE;QACnC,IAAI,OAAO,IAAI,CAAC,CAAC,CAAC,KAAK,QAAQ,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,wBAAwB,CAAC,CAAC,EAAE,CAAC;YACrJ,OAAO;QACT,CAAC;QACD,WAAW,CAAC,KAAK,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;IACnC,CAAC,CAAC;IAEF,IAAI,IAAuE,CAAC;IAC5E,IAAI,CAAC;QACH,IAAI,GAAG,MAAM,GAAG,CAAC,MAAM,CAAC,CAAC;IAC3B,CAAC;YAAS,CAAC;QACT,OAAO,CAAC,GAAG,GAAG,WAAW,CAAC;IAC5B,CAAC;IAED,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,IAAI,EAAE,CAAC;IAE7B,MAAM,UAAU,GAAuB;QACrC,KAAK,EAAE,gBAAgB,CAAC,IAAI,CAAC,KAAK,CAAC;QACnC,OAAO,EAAE,gBAAgB,CAAC,IAAI,CAAC,OAAO,CAAC;QACvC,MAAM,EAAE,gBAAgB,CAAC,IAAI,CAAC,MAAM,CAAC;QACrC,QAAQ,EAAE,gBAAgB,CAAC,IAAI,CAAC,QAAQ,CAAC;QACzC,WAAW,EAAE,YAAY,CAAC,IAAI,CAAC,YAAY,CAAC;QAC5C,YAAY,EAAE,YAAY,CAAC,IAAI,CAAC,OAAO,CAAC;KACzC,CAAC;IAEF,wCAAwC;IACxC,MAAM,SAAS,GAAiB;QAC9B,QAAQ,EAAE,KAAK;QACf,cAAc,EAAE,KAAK;QACrB,UAAU,EAAE,KAAK;QACjB,eAAe,EAAE,KAAK;QACtB,SAAS,EAAE,KAAK;QAChB,QAAQ,EAAE,KAAK;QACf,MAAM,EAAE,KAAK;QACb,YAAY,EAAE,KAAK;QACnB,MAAM,EAAE,KAAK;QACb,SAAS,EAAE,CAAC,CAAC,IAAI,CAAC,iBAAiB,IAAK,IAAgC,CAAC,SAAS,KAAK,IAAI;KAC5F,CAAC;IAEF,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,IAAI,EAAE,CAAC;IAC7B,MAAM,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC;IAC/B,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC;IAE/E,MAAM,YAAY,GAAoB;QACpC,KAAK;QACL,KAAK,EAAE,IAAI,CAAC,QAAQ;QACpB,cAAc,EAAE,KAAK;QACrB,UAAU;QACV,UAAU,EAAE,IAAI,CAAC,MAAM;QACvB,MAAM,EAAE,CAAC;QACT,MAAM,EAAE,CAAC;KACV,CAAC;IAEF,0BAA0B;IAC1B,MAAM,cAAc,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IAC5E,MAAM,iBAAiB,GAAiC,EAAE,CAAC;IAC3D,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,gBAAgB,EAAE,cAAc,CAAC,MAAM,CAAC,CAAC;IAChE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE,CAAC;QAC/B,iBAAiB,CAAC,IAAI,CAAC;YACrB,KAAK,EAAE,CAAC;YACR,IAAI,EAAE,cAAc,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC;YACnD,SAAS,EAAE,KAAK;SACjB,CAAC,CAAC;IACL,CAAC;IAED,MAAM,OAAO,GAAmB;QAC9B,SAAS,EAAE,cAAc,CAAC,MAAM,GAAG,gBAAgB;QACnD,UAAU,EAAE,iBAAiB;KAC9B,CAAC;IAEF,OAAO,EAAE,UAAU,EAAE,SAAS,EAAE,YAAY,EAAE,OAAO,EAAE,CAAC;AAC1D,CAAC"}
|
|
@@ -0,0 +1,70 @@
|
|
|
1
|
+
import path from 'node:path';
|
|
2
|
+
import { readFile } from 'node:fs/promises';
|
|
3
|
+
import { getExtension } from '../utils.js';
|
|
4
|
+
import { estimateTokens } from '../inspect/shared.js';
|
|
5
|
+
import { documentToMarkdown } from '../markdown/convert.js';
|
|
6
|
+
import { extractFromMarkdown } from '../structure/index.js';
|
|
7
|
+
import { inspectDocx } from './inspect-docx.js';
|
|
8
|
+
import { inspectPdf } from './inspect-pdf.js';
|
|
9
|
+
import { inspectOdt } from './inspect-odt.js';
|
|
10
|
+
const SUPPORTED_EXTENSIONS = new Set(['docx', 'pdf', 'odt']);
|
|
11
|
+
export async function inspectDocument(filePath, options) {
|
|
12
|
+
const resolvedPath = path.resolve(filePath);
|
|
13
|
+
const ext = getExtension(resolvedPath);
|
|
14
|
+
if (!SUPPORTED_EXTENSIONS.has(ext)) {
|
|
15
|
+
throw new Error(`Unsupported document format: .${ext || '(none)'} (expected .docx, .pdf, or .odt)`);
|
|
16
|
+
}
|
|
17
|
+
const buffer = await readFile(resolvedPath);
|
|
18
|
+
const format = ext;
|
|
19
|
+
let result;
|
|
20
|
+
switch (format) {
|
|
21
|
+
case 'docx':
|
|
22
|
+
result = await inspectDocx(resolvedPath, buffer, options.sampleParagraphs);
|
|
23
|
+
break;
|
|
24
|
+
case 'pdf':
|
|
25
|
+
result = await inspectPdf(resolvedPath, buffer, options.sampleParagraphs);
|
|
26
|
+
break;
|
|
27
|
+
case 'odt':
|
|
28
|
+
result = await inspectOdt(resolvedPath, buffer, options.sampleParagraphs);
|
|
29
|
+
break;
|
|
30
|
+
}
|
|
31
|
+
// Structure extraction
|
|
32
|
+
let structure = null;
|
|
33
|
+
if (options.includeStructure) {
|
|
34
|
+
try {
|
|
35
|
+
const markdown = await documentToMarkdown(resolvedPath);
|
|
36
|
+
if (markdown) {
|
|
37
|
+
const docStructure = extractFromMarkdown(markdown);
|
|
38
|
+
if (docStructure.totalNodes > 0) {
|
|
39
|
+
const topLevelSections = docStructure.rootNodes.map(n => n.title);
|
|
40
|
+
structure = {
|
|
41
|
+
headingCount: docStructure.totalNodes,
|
|
42
|
+
maxDepth: docStructure.maxDepth,
|
|
43
|
+
topLevelSections,
|
|
44
|
+
tree: docStructure.rootNodes,
|
|
45
|
+
};
|
|
46
|
+
}
|
|
47
|
+
}
|
|
48
|
+
}
|
|
49
|
+
catch { /* structure extraction is best-effort */ }
|
|
50
|
+
}
|
|
51
|
+
// Token estimates
|
|
52
|
+
const fullText = [
|
|
53
|
+
result.preview.paragraphs.map(p => p.text).join(' '),
|
|
54
|
+
].join(' ');
|
|
55
|
+
const fullTokenEstimate = estimateTokens(result.contentStats.characters);
|
|
56
|
+
const previewTokenEstimate = estimateTokens(result.preview.paragraphs.reduce((sum, p) => sum + p.text.length, 0));
|
|
57
|
+
return {
|
|
58
|
+
file: resolvedPath,
|
|
59
|
+
format,
|
|
60
|
+
size: buffer.length,
|
|
61
|
+
properties: result.properties,
|
|
62
|
+
riskFlags: result.riskFlags,
|
|
63
|
+
contentStats: result.contentStats,
|
|
64
|
+
structure,
|
|
65
|
+
contentPreview: result.preview,
|
|
66
|
+
fullTokenEstimate,
|
|
67
|
+
previewTokenEstimate,
|
|
68
|
+
};
|
|
69
|
+
}
|
|
70
|
+
//# sourceMappingURL=inspect.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"inspect.js","sourceRoot":"","sources":["../../../src/doc/inspect.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,EAAE,QAAQ,EAAE,MAAM,kBAAkB,CAAC;AAC5C,OAAO,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AAC3C,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,kBAAkB,EAAE,MAAM,wBAAwB,CAAC;AAC5D,OAAO,EAAE,mBAAmB,EAAE,MAAM,uBAAuB,CAAC;AAC5D,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAC9C,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAG9C,MAAM,oBAAoB,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC;AAE7D,MAAM,CAAC,KAAK,UAAU,eAAe,CAAC,QAAgB,EAAE,OAA0B;IAChF,MAAM,YAAY,GAAG,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC;IAC5C,MAAM,GAAG,GAAG,YAAY,CAAC,YAAY,CAAC,CAAC;IACvC,IAAI,CAAC,oBAAoB,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;QACnC,MAAM,IAAI,KAAK,CAAC,iCAAiC,GAAG,IAAI,QAAQ,kCAAkC,CAAC,CAAC;IACtG,CAAC;IAED,MAAM,MAAM,GAAG,MAAM,QAAQ,CAAC,YAAY,CAAC,CAAC;IAC5C,MAAM,MAAM,GAAG,GAA6B,CAAC;IAE7C,IAAI,MAA+C,CAAC;IACpD,QAAQ,MAAM,EAAE,CAAC;QACf,KAAK,MAAM;YACT,MAAM,GAAG,MAAM,WAAW,CAAC,YAAY,EAAE,MAAM,EAAE,OAAO,CAAC,gBAAgB,CAAC,CAAC;YAC3E,MAAM;QACR,KAAK,KAAK;YACR,MAAM,GAAG,MAAM,UAAU,CAAC,YAAY,EAAE,MAAM,EAAE,OAAO,CAAC,gBAAgB,CAAC,CAAC;YAC1E,MAAM;QACR,KAAK,KAAK;YACR,MAAM,GAAG,MAAM,UAAU,CAAC,YAAY,EAAE,MAAM,EAAE,OAAO,CAAC,gBAAgB,CAAC,CAAC;YAC1E,MAAM;IACV,CAAC;IAED,uBAAuB;IACvB,IAAI,SAAS,GAA4B,IAAI,CAAC;IAC9C,IAAI,OAAO,CAAC,gBAAgB,EAAE,CAAC;QAC7B,IAAI,CAAC;YACH,MAAM,QAAQ,GAAG,MAAM,kBAAkB,CAAC,YAAY,CAAC,CAAC;YACxD,IAAI,QAAQ,EAAE,CAAC;gBACb,MAAM,YAAY,GAAG,mBAAmB,CAAC,QAAQ,CAAC,CAAC;gBACnD,IAAI,YAAY,CAAC,UAAU,GAAG,CAAC,EAAE,CAAC;oBAChC,MAAM,gBAAgB,GAAG,YAAY,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;oBAClE,SAAS,GAAG;wBACV,YAAY,EAAE,YAAY,CAAC,UAAU;wBACrC,QAAQ,EAAE,YAAY,CAAC,QAAQ;wBAC/B,gBAAgB;wBAChB,IAAI,EAAE,YAAY,CAAC,SAAS;qBAC7B,CAAC;gBACJ,CAAC;YACH,CAAC;QACH,CAAC;QAAC,MAAM,CAAC,CAAC,yCAAyC,CAAC,CAAC;IACvD,CAAC;IAED,kBAAkB;IAClB,MAAM,QAAQ,GAAG;QACf,MAAM,CAAC,OAAO,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC;KACrD,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IACZ,MAAM,iBAAiB,GAAG,cAAc,CAAC,MAAM,CAAC,YAAY,CAAC,UAAU,CAAC,CAAC;IACzE,MAAM,oBAAoB,GAAG,cAAc,CACzC,MAAM,CAAC,OAAO,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC,CACrE,CAAC;IAEF,OAAO;QACL,IAAI,EAAE,YAAY;QAClB,MAAM;QACN,IAAI,EAAE,MAAM,CAAC,MAAM;QACnB,UAAU,EAAE,MAAM,CAAC,UAAU;QAC7B,SAAS,EAAE,MAAM,CAAC,SAAS;QAC3B,YAAY,EAAE,MAAM,CAAC,YAAY;QACjC,SAAS;QACT,cAAc,EAAE,MAAM,CAAC,OAAO;QAC9B,iBAAiB;QACjB,oBAAoB;KACrB,CAAC;AACJ,CAAC"}
|
|
@@ -0,0 +1,82 @@
|
|
|
1
|
+
import chalk from 'chalk';
|
|
2
|
+
import { formatBytes, formatNumber } from '../utils.js';
|
|
3
|
+
import { sectionHeader, stripAnsi } from '../output/tabular.js';
|
|
4
|
+
const colors = {
|
|
5
|
+
header: (value) => chalk.bold(value),
|
|
6
|
+
key: (value) => chalk.cyan(value),
|
|
7
|
+
value: (value) => value,
|
|
8
|
+
dim: (value) => chalk.dim(value),
|
|
9
|
+
ok: (value) => chalk.green(value),
|
|
10
|
+
warn: (value) => chalk.yellow(value),
|
|
11
|
+
};
|
|
12
|
+
const noColor = Object.fromEntries(Object.keys(colors).map(key => [key, (value) => value]));
|
|
13
|
+
function palette(ci = false) {
|
|
14
|
+
return ci ? noColor : colors;
|
|
15
|
+
}
|
|
16
|
+
export function formatDocPayloadJson(payload) {
|
|
17
|
+
return JSON.stringify(payload, null, 2);
|
|
18
|
+
}
|
|
19
|
+
export function formatDocInspection(result, ci = false) {
|
|
20
|
+
const c = palette(ci);
|
|
21
|
+
const sections = [];
|
|
22
|
+
// Overview
|
|
23
|
+
const overviewLines = [
|
|
24
|
+
`${c.key('File')}: ${c.value(result.file)}`,
|
|
25
|
+
`${c.key('Format')}: ${c.value(result.format.toUpperCase())}`,
|
|
26
|
+
`${c.key('Size')}: ${c.value(formatBytes(result.size))}`,
|
|
27
|
+
];
|
|
28
|
+
const propEntries = Object.entries(result.properties).filter(([, value]) => value);
|
|
29
|
+
if (propEntries.length > 0) {
|
|
30
|
+
overviewLines.push(`${c.key('Properties')}: ${propEntries.map(([key, value]) => `${key}=${value}`).join(' | ')}`);
|
|
31
|
+
}
|
|
32
|
+
const riskFlags = Object.entries(result.riskFlags).filter(([, enabled]) => enabled).map(([key]) => key);
|
|
33
|
+
overviewLines.push(`${c.key('Risk Flags')}: ${riskFlags.length > 0 ? c.warn(riskFlags.join(', ')) : c.ok('none')}`);
|
|
34
|
+
sections.push(overviewLines.join('\n'));
|
|
35
|
+
// Content Stats
|
|
36
|
+
const pagesLabel = result.contentStats.pagesEstimated ? ' (estimated)' : '';
|
|
37
|
+
const statsLines = [
|
|
38
|
+
` ${c.key('Words')}: ${formatNumber(result.contentStats.words)} ${c.key('Pages')}: ${formatNumber(result.contentStats.pages)}${pagesLabel}`,
|
|
39
|
+
` ${c.key('Paragraphs')}: ${formatNumber(result.contentStats.paragraphs)} ${c.key('Characters')}: ${formatNumber(result.contentStats.characters)}`,
|
|
40
|
+
` ${c.key('Tables')}: ${formatNumber(result.contentStats.tables)} ${c.key('Images')}: ${formatNumber(result.contentStats.images)}`,
|
|
41
|
+
` ${c.key('Token Estimate')}: preview=${formatNumber(result.previewTokenEstimate)} | full=${formatNumber(result.fullTokenEstimate)}`,
|
|
42
|
+
];
|
|
43
|
+
const statsWidth = Math.max(...statsLines.map(line => stripAnsi(line).length), 56);
|
|
44
|
+
sections.push(`${c.header(sectionHeader('Content Stats', statsWidth, ci))}\n${statsLines.join('\n')}`);
|
|
45
|
+
// Structure
|
|
46
|
+
if (result.structure) {
|
|
47
|
+
const structLines = [];
|
|
48
|
+
for (const node of result.structure.tree) {
|
|
49
|
+
formatTreeNode(node, structLines, '');
|
|
50
|
+
}
|
|
51
|
+
const structTitle = `Structure (${result.structure.headingCount} headings, depth ${result.structure.maxDepth})`;
|
|
52
|
+
const structWidth = Math.max(...structLines.map(line => line.length), structTitle.length + 6, 56);
|
|
53
|
+
sections.push(`${c.header(sectionHeader(structTitle, structWidth, ci))}\n${structLines.join('\n')}`);
|
|
54
|
+
}
|
|
55
|
+
// Content Preview
|
|
56
|
+
if (result.contentPreview.paragraphs.length > 0) {
|
|
57
|
+
const previewLines = [];
|
|
58
|
+
for (const p of result.contentPreview.paragraphs) {
|
|
59
|
+
if (p.isHeading) {
|
|
60
|
+
previewLines.push(` ${c.key(`[H${p.headingLevel}]`)} ${p.text}`);
|
|
61
|
+
}
|
|
62
|
+
else {
|
|
63
|
+
previewLines.push(` ${p.text.length > 120 ? p.text.slice(0, 120) + '...' : p.text}`);
|
|
64
|
+
}
|
|
65
|
+
}
|
|
66
|
+
if (result.contentPreview.truncated) {
|
|
67
|
+
const remainingWords = result.contentStats.words - result.contentPreview.paragraphs.reduce((sum, p) => sum + p.text.split(/\s+/).length, 0);
|
|
68
|
+
const remainingParas = result.contentStats.paragraphs - result.contentPreview.paragraphs.length;
|
|
69
|
+
previewLines.push(c.dim(` (${remainingParas} more paragraphs, ~${formatNumber(remainingWords)} remaining words)`));
|
|
70
|
+
}
|
|
71
|
+
const previewWidth = Math.max(...previewLines.map(line => stripAnsi(line).length), 56);
|
|
72
|
+
sections.push(`${c.header(sectionHeader('Content Preview', previewWidth, ci))}\n${previewLines.join('\n')}`);
|
|
73
|
+
}
|
|
74
|
+
return sections.join('\n\n') + '\n';
|
|
75
|
+
}
|
|
76
|
+
function formatTreeNode(node, lines, indent) {
|
|
77
|
+
lines.push(` ${indent}${node.structureCode ?? ''} ${node.title}`);
|
|
78
|
+
for (const child of node.children) {
|
|
79
|
+
formatTreeNode(child, lines, indent + ' ');
|
|
80
|
+
}
|
|
81
|
+
}
|
|
82
|
+
//# sourceMappingURL=output.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"output.js","sourceRoot":"","sources":["../../../src/doc/output.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,EAAE,WAAW,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AACxD,OAAO,EAAE,aAAa,EAAE,SAAS,EAAE,MAAM,sBAAsB,CAAC;AAchE,MAAM,MAAM,GAAgB;IAC1B,MAAM,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC;IACpC,GAAG,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC;IACjC,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK;IACvB,GAAG,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC;IAChC,EAAE,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC;IACjC,IAAI,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC;CACrC,CAAC;AAEF,MAAM,OAAO,GAAgB,MAAM,CAAC,WAAW,CAC7C,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC,KAAa,EAAE,EAAE,CAAC,KAAK,CAAC,CAAC,CACtC,CAAC;AAE5B,SAAS,OAAO,CAAC,EAAE,GAAG,KAAK;IACzB,OAAO,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC;AAC/B,CAAC;AAED,MAAM,UAAU,oBAAoB,CAAC,OAA0B;IAC7D,OAAO,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;AAC1C,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,MAA0B,EAAE,EAAE,GAAG,KAAK;IACxE,MAAM,CAAC,GAAG,OAAO,CAAC,EAAE,CAAC,CAAC;IACtB,MAAM,QAAQ,GAAa,EAAE,CAAC;IAE9B,WAAW;IACX,MAAM,aAAa,GAAG;QACpB,GAAG,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE;QAC3C,GAAG,CAAC,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,MAAM,CAAC,WAAW,EAAE,CAAC,EAAE;QAC7D,GAAG,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,EAAE;KACzD,CAAC;IAEF,MAAM,WAAW,GAAG,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,KAAK,CAAC,CAAC;IACnF,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC3B,aAAa,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,YAAY,CAAC,KAAK,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,IAAI,KAAK,EAAE,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;IACpH,CAAC;IAED,MAAM,SAAS,GAAG,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,OAAO,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC;IACxG,aAAa,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,YAAY,CAAC,KAAK,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;IAEpH,QAAQ,CAAC,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IAExC,gBAAgB;IAChB,MAAM,UAAU,GAAG,MAAM,CAAC,YAAY,CAAC,cAAc,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC,CAAC,EAAE,CAAC;IAC5E,MAAM,UAAU,GAAG;QACjB,KAAK,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,KAAK,YAAY,CAAC,MAAM,CAAC,YAAY,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,KAAK,YAAY,CAAC,MAAM,CAAC,YAAY,CAAC,KAAK,CAAC,GAAG,UAAU,EAAE;QACnJ,KAAK,CAAC,CAAC,GAAG,CAAC,YAAY,CAAC,KAAK,YAAY,CAAC,MAAM,CAAC,YAAY,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC,GAAG,CAAC,YAAY,CAAC,KAAK,YAAY,CAAC,MAAM,CAAC,YAAY,CAAC,UAAU,CAAC,EAAE;QACvJ,KAAK,CAAC,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,YAAY,CAAC,MAAM,CAAC,YAAY,CAAC,MAAM,CAAC,cAAc,CAAC,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,YAAY,CAAC,MAAM,CAAC,YAAY,CAAC,MAAM,CAAC,EAAE;QAC7I,KAAK,CAAC,CAAC,GAAG,CAAC,gBAAgB,CAAC,aAAa,YAAY,CAAC,MAAM,CAAC,oBAAoB,CAAC,WAAW,YAAY,CAAC,MAAM,CAAC,iBAAiB,CAAC,EAAE;KACtI,CAAC;IAEF,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC,CAAC;IACnF,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,aAAa,CAAC,eAAe,EAAE,UAAU,EAAE,EAAE,CAAC,CAAC,KAAK,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAEvG,YAAY;IACZ,IAAI,MAAM,CAAC,SAAS,EAAE,CAAC;QACrB,MAAM,WAAW,GAAa,EAAE,CAAC;QACjC,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;YACzC,cAAc,CAAC,IAAI,EAAE,WAAW,EAAE,EAAE,CAAC,CAAC;QACxC,CAAC;QACD,MAAM,WAAW,GAAG,cAAc,MAAM,CAAC,SAAS,CAAC,YAAY,oBAAoB,MAAM,CAAC,SAAS,CAAC,QAAQ,GAAG,CAAC;QAChH,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC;QAClG,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,aAAa,CAAC,WAAW,EAAE,WAAW,EAAE,EAAE,CAAC,CAAC,KAAK,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACvG,CAAC;IAED,kBAAkB;IAClB,IAAI,MAAM,CAAC,cAAc,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAChD,MAAM,YAAY,GAAa,EAAE,CAAC;QAClC,KAAK,MAAM,CAAC,IAAI,MAAM,CAAC,cAAc,CAAC,UAAU,EAAE,CAAC;YACjD,IAAI,CAAC,CAAC,SAAS,EAAE,CAAC;gBAChB,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,YAAY,GAAG,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;YACpE,CAAC;iBAAM,CAAC;gBACN,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;YACxF,CAAC;QACH,CAAC;QACD,IAAI,MAAM,CAAC,cAAc,CAAC,SAAS,EAAE,CAAC;YACpC,MAAM,cAAc,GAAG,MAAM,CAAC,YAAY,CAAC,KAAK,GAAG,MAAM,CAAC,cAAc,CAAC,UAAU,CAAC,MAAM,CACxF,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,MAAM,EAAE,CAAC,CAChD,CAAC;YACF,MAAM,cAAc,GAAG,MAAM,CAAC,YAAY,CAAC,UAAU,GAAG,MAAM,CAAC,cAAc,CAAC,UAAU,CAAC,MAAM,CAAC;YAChG,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,cAAc,sBAAsB,YAAY,CAAC,cAAc,CAAC,mBAAmB,CAAC,CAAC,CAAC;QACtH,CAAC;QACD,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC,CAAC;QACvF,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,aAAa,CAAC,iBAAiB,EAAE,YAAY,EAAE,EAAE,CAAC,CAAC,KAAK,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAC/G,CAAC;IAED,OAAO,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,GAAG,IAAI,CAAC;AACtC,CAAC;AAED,SAAS,cAAc,CAAC,IAAwE,EAAE,KAAe,EAAE,MAAc;IAC/H,KAAK,CAAC,IAAI,CAAC,KAAK,MAAM,GAAG,IAAI,CAAC,aAAa,IAAI,EAAE,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC,CAAC;IACrE,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;QAClC,cAAc,CAAC,KAAK,EAAE,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC,CAAC;IAC9C,CAAC;AACH,CAAC"}
|
|
@@ -0,0 +1,59 @@
|
|
|
1
|
+
import type { DocumentProperties } from '../inspect/shared.js';
|
|
2
|
+
import type { StructureNode } from '../structure/types.js';
|
|
3
|
+
export interface DocRiskFlags {
|
|
4
|
+
comments: boolean;
|
|
5
|
+
trackedChanges: boolean;
|
|
6
|
+
hyperlinks: boolean;
|
|
7
|
+
embeddedObjects: boolean;
|
|
8
|
+
footnotes: boolean;
|
|
9
|
+
endnotes: boolean;
|
|
10
|
+
macros: boolean;
|
|
11
|
+
headerFooter: boolean;
|
|
12
|
+
tables: boolean;
|
|
13
|
+
encrypted: boolean;
|
|
14
|
+
}
|
|
15
|
+
export interface DocContentStats {
|
|
16
|
+
words: number;
|
|
17
|
+
pages: number;
|
|
18
|
+
pagesEstimated: boolean;
|
|
19
|
+
paragraphs: number;
|
|
20
|
+
characters: number;
|
|
21
|
+
tables: number;
|
|
22
|
+
images: number;
|
|
23
|
+
}
|
|
24
|
+
export interface StructureSummary {
|
|
25
|
+
headingCount: number;
|
|
26
|
+
maxDepth: number;
|
|
27
|
+
topLevelSections: string[];
|
|
28
|
+
tree: StructureNode[];
|
|
29
|
+
}
|
|
30
|
+
export interface ContentPreview {
|
|
31
|
+
truncated: boolean;
|
|
32
|
+
paragraphs: Array<{
|
|
33
|
+
index: number;
|
|
34
|
+
text: string;
|
|
35
|
+
isHeading: boolean;
|
|
36
|
+
headingLevel?: number;
|
|
37
|
+
}>;
|
|
38
|
+
}
|
|
39
|
+
export interface DocumentInspection {
|
|
40
|
+
file: string;
|
|
41
|
+
format: 'docx' | 'pdf' | 'odt';
|
|
42
|
+
size: number;
|
|
43
|
+
properties: DocumentProperties;
|
|
44
|
+
riskFlags: DocRiskFlags;
|
|
45
|
+
contentStats: DocContentStats;
|
|
46
|
+
structure: StructureSummary | null;
|
|
47
|
+
contentPreview: ContentPreview;
|
|
48
|
+
fullTokenEstimate: number;
|
|
49
|
+
previewTokenEstimate: number;
|
|
50
|
+
}
|
|
51
|
+
export interface DocInspectPayload {
|
|
52
|
+
file: string;
|
|
53
|
+
query: Record<string, unknown>;
|
|
54
|
+
results: DocumentInspection;
|
|
55
|
+
}
|
|
56
|
+
export interface InspectDocOptions {
|
|
57
|
+
sampleParagraphs: number;
|
|
58
|
+
includeStructure: boolean;
|
|
59
|
+
}
|