@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
package/dist/src/code/types.js
CHANGED
|
@@ -1,2 +1,151 @@
|
|
|
1
|
-
|
|
1
|
+
import { z } from 'zod';
|
|
2
|
+
export const CodeNodeTypeSchema = z.enum(['file', 'module', 'function', 'class', 'interface', 'type-alias', 'enum', 'variable', 'parameter']);
|
|
3
|
+
export const CodeEdgeTypeSchema = z.enum(['contains', 'imports', 'calls', 'inherits', 'implements', 'has_parameter']);
|
|
4
|
+
export const ResolutionStatusSchema = z.enum(['resolved', 'ambiguous', 'unresolved']);
|
|
5
|
+
export const ImportKindSchema = z.enum(['local', 'external', 'unresolved']);
|
|
6
|
+
export const CodeCapabilitiesSchema = z.object({
|
|
7
|
+
definitions: z.boolean(),
|
|
8
|
+
imports: z.boolean(),
|
|
9
|
+
calls: z.boolean(),
|
|
10
|
+
inheritance: z.boolean(),
|
|
11
|
+
content: z.boolean(),
|
|
12
|
+
});
|
|
13
|
+
export const ParsedImportBindingSchema = z.object({
|
|
14
|
+
localName: z.string(),
|
|
15
|
+
importedName: z.string().optional(),
|
|
16
|
+
isNamespace: z.boolean().optional(),
|
|
17
|
+
isDefault: z.boolean().optional(),
|
|
18
|
+
});
|
|
19
|
+
export const ParsedImportSchema = z.object({
|
|
20
|
+
specifier: z.string(),
|
|
21
|
+
line: z.number(),
|
|
22
|
+
kind: ImportKindSchema,
|
|
23
|
+
bindings: z.array(ParsedImportBindingSchema),
|
|
24
|
+
resolvedPath: z.string().optional(),
|
|
25
|
+
isReExport: z.boolean().optional(),
|
|
26
|
+
});
|
|
27
|
+
export const ParsedSymbolSchema = z.object({
|
|
28
|
+
type: z.enum(['function', 'class', 'interface', 'type-alias', 'enum', 'variable']),
|
|
29
|
+
name: z.string(),
|
|
30
|
+
line: z.number(),
|
|
31
|
+
containerName: z.string().optional(),
|
|
32
|
+
signature: z.string().optional(),
|
|
33
|
+
exported: z.boolean().optional(),
|
|
34
|
+
});
|
|
35
|
+
export const ParsedCallSchema = z.object({
|
|
36
|
+
callerName: z.string(),
|
|
37
|
+
callerLine: z.number(),
|
|
38
|
+
callerContainerName: z.string().optional(),
|
|
39
|
+
calleeName: z.string(),
|
|
40
|
+
qualifier: z.string().optional(),
|
|
41
|
+
line: z.number(),
|
|
42
|
+
});
|
|
43
|
+
export const ParsedInheritanceSchema = z.object({
|
|
44
|
+
className: z.string(),
|
|
45
|
+
classLine: z.number(),
|
|
46
|
+
baseName: z.string(),
|
|
47
|
+
line: z.number(),
|
|
48
|
+
kind: z.enum(['extends', 'implements']).optional(),
|
|
49
|
+
});
|
|
50
|
+
export const ParsedFileSchema = z.object({
|
|
51
|
+
path: z.string(),
|
|
52
|
+
relativePath: z.string(),
|
|
53
|
+
moduleName: z.string(),
|
|
54
|
+
language: z.string(),
|
|
55
|
+
content: z.string(),
|
|
56
|
+
lines: z.array(z.string()),
|
|
57
|
+
symbols: z.array(ParsedSymbolSchema),
|
|
58
|
+
imports: z.array(ParsedImportSchema),
|
|
59
|
+
calls: z.array(ParsedCallSchema),
|
|
60
|
+
inheritances: z.array(ParsedInheritanceSchema),
|
|
61
|
+
});
|
|
62
|
+
export const CodeNodeSchema = z.object({
|
|
63
|
+
id: z.string(),
|
|
64
|
+
type: CodeNodeTypeSchema,
|
|
65
|
+
name: z.string(),
|
|
66
|
+
path: z.string(),
|
|
67
|
+
relativePath: z.string().optional(),
|
|
68
|
+
line: z.number().optional(),
|
|
69
|
+
language: z.string(),
|
|
70
|
+
moduleName: z.string().optional(),
|
|
71
|
+
containerId: z.string().optional(),
|
|
72
|
+
containerName: z.string().optional(),
|
|
73
|
+
signature: z.string().optional(),
|
|
74
|
+
external: z.boolean().optional(),
|
|
75
|
+
});
|
|
76
|
+
export const CodeEdgeSchema = z.object({
|
|
77
|
+
id: z.string(),
|
|
78
|
+
type: CodeEdgeTypeSchema,
|
|
79
|
+
from: z.string(),
|
|
80
|
+
to: z.string().optional(),
|
|
81
|
+
fromPath: z.string(),
|
|
82
|
+
toPath: z.string().optional(),
|
|
83
|
+
line: z.number().optional(),
|
|
84
|
+
status: ResolutionStatusSchema,
|
|
85
|
+
targetName: z.string().optional(),
|
|
86
|
+
detail: z.string().optional(),
|
|
87
|
+
specifier: z.string().optional(),
|
|
88
|
+
importKind: ImportKindSchema.optional(),
|
|
89
|
+
candidates: z.array(z.object({
|
|
90
|
+
name: z.string(),
|
|
91
|
+
relativePath: z.string().optional(),
|
|
92
|
+
line: z.number().optional(),
|
|
93
|
+
})).optional(),
|
|
94
|
+
});
|
|
95
|
+
export const CodebaseIndexSchema = z.object({
|
|
96
|
+
repoRoot: z.string(),
|
|
97
|
+
files: z.array(ParsedFileSchema),
|
|
98
|
+
nodes: z.array(CodeNodeSchema),
|
|
99
|
+
edges: z.array(CodeEdgeSchema),
|
|
100
|
+
capabilities: z.record(z.string(), CodeCapabilitiesSchema),
|
|
101
|
+
});
|
|
102
|
+
export const CodeSearchResultSchema = z.object({
|
|
103
|
+
node: CodeNodeSchema,
|
|
104
|
+
resolution: ResolutionStatusSchema.optional(),
|
|
105
|
+
});
|
|
106
|
+
export const ContentMatchSchema = z.object({
|
|
107
|
+
path: z.string(),
|
|
108
|
+
relativePath: z.string(),
|
|
109
|
+
language: z.string(),
|
|
110
|
+
line: z.number(),
|
|
111
|
+
snippet: z.string(),
|
|
112
|
+
});
|
|
113
|
+
export const RelationMatchSchema = z.object({
|
|
114
|
+
from: CodeNodeSchema,
|
|
115
|
+
edge: CodeEdgeSchema,
|
|
116
|
+
to: CodeNodeSchema.optional(),
|
|
117
|
+
});
|
|
118
|
+
export const DependencyAnalysisSchema = z.object({
|
|
119
|
+
target: z.string(),
|
|
120
|
+
targetFile: CodeNodeSchema.optional(),
|
|
121
|
+
importers: z.array(RelationMatchSchema),
|
|
122
|
+
localImports: z.array(RelationMatchSchema),
|
|
123
|
+
externalImports: z.array(RelationMatchSchema),
|
|
124
|
+
unresolvedImports: z.array(RelationMatchSchema),
|
|
125
|
+
});
|
|
126
|
+
export const ClassTreeAnalysisSchema = z.object({
|
|
127
|
+
target: CodeNodeSchema,
|
|
128
|
+
parents: z.array(RelationMatchSchema),
|
|
129
|
+
children: z.array(RelationMatchSchema),
|
|
130
|
+
methods: z.array(CodeNodeSchema),
|
|
131
|
+
});
|
|
132
|
+
export const CallChainSchema = z.object({
|
|
133
|
+
nodes: z.array(CodeNodeSchema),
|
|
134
|
+
edges: z.array(CodeEdgeSchema),
|
|
135
|
+
direction: z.enum(['forward', 'reverse']).optional(),
|
|
136
|
+
status: z.enum(['resolved', 'blocked_ambiguous']).optional(),
|
|
137
|
+
blockedAt: CodeNodeSchema.optional(),
|
|
138
|
+
blockedBy: CodeEdgeSchema.optional(),
|
|
139
|
+
});
|
|
140
|
+
export const CodeCommandPayloadSchema = z.object({
|
|
141
|
+
repo: z.string(),
|
|
142
|
+
query: z.record(z.string(), z.unknown()),
|
|
143
|
+
results: z.unknown(),
|
|
144
|
+
stats: z.object({
|
|
145
|
+
filesIndexed: z.number(),
|
|
146
|
+
nodes: z.number(),
|
|
147
|
+
edges: z.number(),
|
|
148
|
+
}),
|
|
149
|
+
capabilities: z.record(z.string(), CodeCapabilitiesSchema),
|
|
150
|
+
});
|
|
2
151
|
//# sourceMappingURL=types.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"types.js","sourceRoot":"","sources":["../../../src/code/types.ts"],"names":[],"mappings":""}
|
|
1
|
+
{"version":3,"file":"types.js","sourceRoot":"","sources":["../../../src/code/types.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,QAAQ,EAAE,UAAU,EAAE,OAAO,EAAE,WAAW,EAAE,YAAY,EAAE,MAAM,EAAE,UAAU,EAAE,WAAW,CAAC,CAAC,CAAC;AAG9I,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,UAAU,EAAE,SAAS,EAAE,OAAO,EAAE,UAAU,EAAE,YAAY,EAAE,eAAe,CAAC,CAAC,CAAC;AAGtH,MAAM,CAAC,MAAM,sBAAsB,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,UAAU,EAAE,WAAW,EAAE,YAAY,CAAC,CAAC,CAAC;AAGtF,MAAM,CAAC,MAAM,gBAAgB,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,UAAU,EAAE,YAAY,CAAC,CAAC,CAAC;AAG5E,MAAM,CAAC,MAAM,sBAAsB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC7C,WAAW,EAAE,CAAC,CAAC,OAAO,EAAE;IACxB,OAAO,EAAE,CAAC,CAAC,OAAO,EAAE;IACpB,KAAK,EAAE,CAAC,CAAC,OAAO,EAAE;IAClB,WAAW,EAAE,CAAC,CAAC,OAAO,EAAE;IACxB,OAAO,EAAE,CAAC,CAAC,OAAO,EAAE;CACrB,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,yBAAyB,GAAG,CAAC,CAAC,MAAM,CAAC;IAChD,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE;IACrB,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACnC,WAAW,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE;IACnC,SAAS,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE;CAClC,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAAC,CAAC,MAAM,CAAC;IACzC,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE;IACrB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,IAAI,EAAE,gBAAgB;IACtB,QAAQ,EAAE,CAAC,CAAC,KAAK,CAAC,yBAAyB,CAAC;IAC5C,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACnC,UAAU,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE;CACnC,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAAC,CAAC,MAAM,CAAC;IACzC,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,UAAU,EAAE,OAAO,EAAE,WAAW,EAAE,YAAY,EAAE,MAAM,EAAE,UAAU,CAAC,CAAC;IAClF,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,aAAa,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACpC,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAChC,QAAQ,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE;CACjC,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,gBAAgB,GAAG,CAAC,CAAC,MAAM,CAAC;IACvC,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE;IACtB,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE;IACtB,mBAAmB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAC1C,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE;IACtB,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAChC,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;CACjB,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,uBAAuB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC9C,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE;IACrB,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE;IACrB,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,SAAS,EAAE,YAAY,CAAC,CAAC,CAAC,QAAQ,EAAE;CACnD,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,gBAAgB,GAAG,CAAC,CAAC,MAAM,CAAC;IACvC,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE;IACxB,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE;IACtB,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,OAAO,EAAE,CAAC,CAAC,MAAM,EAAE;IACnB,KAAK,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC;IAC1B,OAAO,EAAE,CAAC,CAAC,KAAK,CAAC,kBAAkB,CAAC;IACpC,OAAO,EAAE,CAAC,CAAC,KAAK,CAAC,kBAAkB,CAAC;IACpC,KAAK,EAAE,CAAC,CAAC,KAAK,CAAC,gBAAgB,CAAC;IAChC,YAAY,EAAE,CAAC,CAAC,KAAK,CAAC,uBAAuB,CAAC;CAC/C,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,cAAc,GAAG,CAAC,CAAC,MAAM,CAAC;IACrC,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE;IACd,IAAI,EAAE,kBAAkB;IACxB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACnC,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAC3B,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACjC,WAAW,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAClC,aAAa,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACpC,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAChC,QAAQ,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE;CACjC,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,cAAc,GAAG,CAAC,CAAC,MAAM,CAAC;IACrC,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE;IACd,IAAI,EAAE,kBAAkB;IACxB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACzB,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAC7B,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAC3B,MAAM,EAAE,sBAAsB;IAC9B,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACjC,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAC7B,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAChC,UAAU,EAAE,gBAAgB,CAAC,QAAQ,EAAE;IACvC,UAAU,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC;QAC3B,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;QAChB,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;QACnC,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;KAC5B,CAAC,CAAC,CAAC,QAAQ,EAAE;CACf,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,mBAAmB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC1C,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,KAAK,EAAE,CAAC,CAAC,KAAK,CAAC,gBAAgB,CAAC;IAChC,KAAK,EAAE,CAAC,CAAC,KAAK,CAAC,cAAc,CAAC;IAC9B,KAAK,EAAE,CAAC,CAAC,KAAK,CAAC,cAAc,CAAC;IAC9B,YAAY,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,sBAAsB,CAAC;CAC3D,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,sBAAsB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC7C,IAAI,EAAE,cAAc;IACpB,UAAU,EAAE,sBAAsB,CAAC,QAAQ,EAAE;CAC9C,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAAC,CAAC,MAAM,CAAC;IACzC,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE;IACxB,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,OAAO,EAAE,CAAC,CAAC,MAAM,EAAE;CACpB,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,mBAAmB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC1C,IAAI,EAAE,cAAc;IACpB,IAAI,EAAE,cAAc;IACpB,EAAE,EAAE,cAAc,CAAC,QAAQ,EAAE;CAC9B,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,wBAAwB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC/C,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE;IAClB,UAAU,EAAE,cAAc,CAAC,QAAQ,EAAE;IACrC,SAAS,EAAE,CAAC,CAAC,KAAK,CAAC,mBAAmB,CAAC;IACvC,YAAY,EAAE,CAAC,CAAC,KAAK,CAAC,mBAAmB,CAAC;IAC1C,eAAe,EAAE,CAAC,CAAC,KAAK,CAAC,mBAAmB,CAAC;IAC7C,iBAAiB,EAAE,CAAC,CAAC,KAAK,CAAC,mBAAmB,CAAC;CAChD,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,uBAAuB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC9C,MAAM,EAAE,cAAc;IACtB,OAAO,EAAE,CAAC,CAAC,KAAK,CAAC,mBAAmB,CAAC;IACrC,QAAQ,EAAE,CAAC,CAAC,KAAK,CAAC,mBAAmB,CAAC;IACtC,OAAO,EAAE,CAAC,CAAC,KAAK,CAAC,cAAc,CAAC;CACjC,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,CAAC,MAAM,CAAC;IACtC,KAAK,EAAE,CAAC,CAAC,KAAK,CAAC,cAAc,CAAC;IAC9B,KAAK,EAAE,CAAC,CAAC,KAAK,CAAC,cAAc,CAAC;IAC9B,SAAS,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,SAAS,EAAE,SAAS,CAAC,CAAC,CAAC,QAAQ,EAAE;IACpD,MAAM,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,UAAU,EAAE,mBAAmB,CAAC,CAAC,CAAC,QAAQ,EAAE;IAC5D,SAAS,EAAE,cAAc,CAAC,QAAQ,EAAE;IACpC,SAAS,EAAE,cAAc,CAAC,QAAQ,EAAE;CACrC,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,wBAAwB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC/C,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,KAAK,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC;IACxC,OAAO,EAAE,CAAC,CAAC,OAAO,EAAE;IACpB,KAAK,EAAE,CAAC,CAAC,MAAM,CAAC;QACd,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE;QACxB,KAAK,EAAE,CAAC,CAAC,MAAM,EAAE;QACjB,KAAK,EAAE,CAAC,CAAC,MAAM,EAAE;KAClB,CAAC;IACF,YAAY,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,sBAAsB,CAAC;CAC3D,CAAC,CAAC"}
|
|
@@ -0,0 +1,54 @@
|
|
|
1
|
+
import { writeFile } from 'node:fs/promises';
|
|
2
|
+
import path from 'node:path';
|
|
3
|
+
import { z } from 'zod';
|
|
4
|
+
import { inspectDocument } from './inspect.js';
|
|
5
|
+
import { formatDocInspection, formatDocPayloadJson } from './output.js';
|
|
6
|
+
import { createInspectPayload } from '../inspect/shared.js';
|
|
7
|
+
import { writeStream } from '../utils.js';
|
|
8
|
+
import { parsePositiveInt } from '../cli-validation.js';
|
|
9
|
+
const DocCommandOptionsSchema = z.object({
|
|
10
|
+
format: z.string().optional(),
|
|
11
|
+
output: z.string().optional(),
|
|
12
|
+
ci: z.boolean().optional(),
|
|
13
|
+
sampleParagraphs: z.string().optional(),
|
|
14
|
+
structure: z.boolean().optional(),
|
|
15
|
+
}).passthrough();
|
|
16
|
+
function getOptions(command) {
|
|
17
|
+
return DocCommandOptionsSchema.parse(command.optsWithGlobals());
|
|
18
|
+
}
|
|
19
|
+
async function emit(output, options) {
|
|
20
|
+
if (options.output) {
|
|
21
|
+
await writeFile(options.output, output);
|
|
22
|
+
return;
|
|
23
|
+
}
|
|
24
|
+
await writeStream(process.stdout, output);
|
|
25
|
+
}
|
|
26
|
+
export function registerDocCommands(program) {
|
|
27
|
+
const doc = program.command('doc').description('Inspect document structure and content before deep reading');
|
|
28
|
+
const inspect = doc.command('inspect <file>').description('Inspect DOCX, PDF, or ODT document metadata, structure, and content preview');
|
|
29
|
+
inspect
|
|
30
|
+
.option('--format <type>', 'output format: tabular or json', 'tabular')
|
|
31
|
+
.option('-o, --output <file>', 'write output to file')
|
|
32
|
+
.option('--ci', 'ASCII-only output, no colors')
|
|
33
|
+
.option('--sample-paragraphs <n>', 'preview paragraph count', '5')
|
|
34
|
+
.option('--no-structure', 'skip heading tree extraction');
|
|
35
|
+
inspect.action(async (file, _options, command) => {
|
|
36
|
+
const options = getOptions(command);
|
|
37
|
+
const resolved = path.resolve(file);
|
|
38
|
+
const inspectOptions = {
|
|
39
|
+
sampleParagraphs: parsePositiveInt(options.sampleParagraphs, 5, 'sample paragraph count'),
|
|
40
|
+
includeStructure: options.structure !== false,
|
|
41
|
+
};
|
|
42
|
+
const result = await inspectDocument(resolved, inspectOptions);
|
|
43
|
+
const payload = createInspectPayload(resolved, {
|
|
44
|
+
command: 'doc.inspect',
|
|
45
|
+
sampleParagraphs: inspectOptions.sampleParagraphs,
|
|
46
|
+
includeStructure: inspectOptions.includeStructure,
|
|
47
|
+
}, result);
|
|
48
|
+
const output = options.format === 'json'
|
|
49
|
+
? formatDocPayloadJson(payload)
|
|
50
|
+
: formatDocInspection(result, options.ci);
|
|
51
|
+
await emit(output.endsWith('\n') ? output : `${output}\n`, options);
|
|
52
|
+
});
|
|
53
|
+
}
|
|
54
|
+
//# sourceMappingURL=command.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"command.js","sourceRoot":"","sources":["../../../src/doc/command.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AAC7C,OAAO,IAAI,MAAM,WAAW,CAAC;AAE7B,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AACxB,OAAO,EAAE,eAAe,EAAE,MAAM,cAAc,CAAC;AAC/C,OAAO,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,aAAa,CAAC;AACxE,OAAO,EAAE,oBAAoB,EAAE,MAAM,sBAAsB,CAAC;AAC5D,OAAO,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAC1C,OAAO,EAAE,gBAAgB,EAAE,MAAM,sBAAsB,CAAC;AAGxD,MAAM,uBAAuB,GAAG,CAAC,CAAC,MAAM,CAAC;IACvC,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAC7B,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAC7B,EAAE,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE;IAC1B,gBAAgB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACvC,SAAS,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE;CAClC,CAAC,CAAC,WAAW,EAAE,CAAC;AAGjB,SAAS,UAAU,CAAC,OAAgB;IAClC,OAAO,uBAAuB,CAAC,KAAK,CAAC,OAAO,CAAC,eAAe,EAAE,CAAC,CAAC;AAClE,CAAC;AAED,KAAK,UAAU,IAAI,CAAC,MAAc,EAAE,OAA0B;IAC5D,IAAI,OAAO,CAAC,MAAM,EAAE,CAAC;QACnB,MAAM,SAAS,CAAC,OAAO,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;QACxC,OAAO;IACT,CAAC;IACD,MAAM,WAAW,CAAC,OAAO,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;AAC5C,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,OAAgB;IAClD,MAAM,GAAG,GAAG,OAAO,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,WAAW,CAAC,4DAA4D,CAAC,CAAC;IAC7G,MAAM,OAAO,GAAG,GAAG,CAAC,OAAO,CAAC,gBAAgB,CAAC,CAAC,WAAW,CAAC,6EAA6E,CAAC,CAAC;IAEzI,OAAO;SACJ,MAAM,CAAC,iBAAiB,EAAE,gCAAgC,EAAE,SAAS,CAAC;SACtE,MAAM,CAAC,qBAAqB,EAAE,sBAAsB,CAAC;SACrD,MAAM,CAAC,MAAM,EAAE,8BAA8B,CAAC;SAC9C,MAAM,CAAC,yBAAyB,EAAE,yBAAyB,EAAE,GAAG,CAAC;SACjE,MAAM,CAAC,gBAAgB,EAAE,8BAA8B,CAAC,CAAC;IAE5D,OAAO,CAAC,MAAM,CAAC,KAAK,EAAE,IAAY,EAAE,QAA2B,EAAE,OAAgB,EAAE,EAAE;QACnF,MAAM,OAAO,GAAG,UAAU,CAAC,OAAO,CAAC,CAAC;QACpC,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;QACpC,MAAM,cAAc,GAAsB;YACxC,gBAAgB,EAAE,gBAAgB,CAAC,OAAO,CAAC,gBAAgB,EAAE,CAAC,EAAE,wBAAwB,CAAC;YACzF,gBAAgB,EAAE,OAAO,CAAC,SAAS,KAAK,KAAK;SAC9C,CAAC;QACF,MAAM,MAAM,GAAG,MAAM,eAAe,CAAC,QAAQ,EAAE,cAAc,CAAC,CAAC;QAC/D,MAAM,OAAO,GAAG,oBAAoB,CAAgB,QAAQ,EAAE;YAC5D,OAAO,EAAE,aAAa;YACtB,gBAAgB,EAAE,cAAc,CAAC,gBAAgB;YACjD,gBAAgB,EAAE,cAAc,CAAC,gBAAgB;SAClD,EAAE,MAAM,CAAsB,CAAC;QAChC,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,KAAK,MAAM;YACtC,CAAC,CAAC,oBAAoB,CAAC,OAAO,CAAC;YAC/B,CAAC,CAAC,mBAAmB,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,CAAC,CAAC;QAC5C,MAAM,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,MAAM,IAAI,EAAE,OAAO,CAAC,CAAC;IACtE,CAAC,CAAC,CAAC;AACL,CAAC"}
|
|
@@ -0,0 +1,14 @@
|
|
|
1
|
+
import JSZip from 'jszip';
|
|
2
|
+
import type { DocumentProperties } from '../inspect/shared.js';
|
|
3
|
+
import type { DocRiskFlags, DocContentStats, ContentPreview } from './types.js';
|
|
4
|
+
export declare function extractDocxProperties(zip: JSZip): Promise<DocumentProperties>;
|
|
5
|
+
export declare function extractDocxRiskFlags(zip: JSZip): Promise<DocRiskFlags>;
|
|
6
|
+
export declare function extractDocxContentStats(filePath: string, zip: JSZip): Promise<DocContentStats>;
|
|
7
|
+
export declare function extractDocxPreview(filePath: string, sampleParagraphs: number): Promise<ContentPreview>;
|
|
8
|
+
export declare function buildPreviewFromMarkdown(markdown: string, sampleParagraphs: number): ContentPreview;
|
|
9
|
+
export declare function inspectDocx(filePath: string, buffer: Buffer, sampleParagraphs: number): Promise<{
|
|
10
|
+
properties: DocumentProperties;
|
|
11
|
+
riskFlags: DocRiskFlags;
|
|
12
|
+
contentStats: DocContentStats;
|
|
13
|
+
preview: ContentPreview;
|
|
14
|
+
}>;
|
|
@@ -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 +1,92 @@
|
|
|
1
|
+
import JSZip from 'jszip';
|
|
2
|
+
import officeparser from 'officeparser';
|
|
3
|
+
import { z } from 'zod';
|
|
4
|
+
import { countWords } from '../utils.js';
|
|
5
|
+
import { asOptionalString, formatDateLike } from '../inspect/shared.js';
|
|
6
|
+
function parseXmlText(xml, tag) {
|
|
7
|
+
const regex = new RegExp(`<${tag}[^>]*>([^<]*)</${tag}>`, 'i');
|
|
8
|
+
const match = xml.match(regex);
|
|
9
|
+
return match?.[1]?.trim() || undefined;
|
|
10
|
+
}
|
|
11
|
+
export async function inspectOdt(filePath, buffer, sampleParagraphs) {
|
|
12
|
+
const zip = await JSZip.loadAsync(buffer);
|
|
13
|
+
// Parse metadata from meta.xml
|
|
14
|
+
const properties = {};
|
|
15
|
+
const metaXml = await zip.file('meta.xml')?.async('text');
|
|
16
|
+
if (metaXml) {
|
|
17
|
+
properties.title = asOptionalString(parseXmlText(metaXml, 'dc:title'));
|
|
18
|
+
properties.subject = asOptionalString(parseXmlText(metaXml, 'dc:subject'));
|
|
19
|
+
properties.author = asOptionalString(parseXmlText(metaXml, 'dc:creator') ?? parseXmlText(metaXml, 'meta:initial-creator'));
|
|
20
|
+
properties.keywords = asOptionalString(parseXmlText(metaXml, 'meta:keyword'));
|
|
21
|
+
properties.createdDate = formatDateLike(parseXmlText(metaXml, 'meta:creation-date'));
|
|
22
|
+
properties.modifiedDate = formatDateLike(parseXmlText(metaXml, 'dc:date'));
|
|
23
|
+
}
|
|
24
|
+
// Parse risk flags from content.xml
|
|
25
|
+
const riskFlags = {
|
|
26
|
+
comments: false,
|
|
27
|
+
trackedChanges: false,
|
|
28
|
+
hyperlinks: false,
|
|
29
|
+
embeddedObjects: false,
|
|
30
|
+
footnotes: false,
|
|
31
|
+
endnotes: false,
|
|
32
|
+
macros: false,
|
|
33
|
+
headerFooter: false,
|
|
34
|
+
tables: false,
|
|
35
|
+
encrypted: false,
|
|
36
|
+
};
|
|
37
|
+
const contentXml = await zip.file('content.xml')?.async('text');
|
|
38
|
+
if (contentXml) {
|
|
39
|
+
if (/<office:annotation\b/.test(contentXml))
|
|
40
|
+
riskFlags.comments = true;
|
|
41
|
+
if (/<text:tracked-changes\b/.test(contentXml))
|
|
42
|
+
riskFlags.trackedChanges = true;
|
|
43
|
+
if (/<text:a\b[^>]*xlink:href=/.test(contentXml))
|
|
44
|
+
riskFlags.hyperlinks = true;
|
|
45
|
+
if (/<draw:frame\b/.test(contentXml))
|
|
46
|
+
riskFlags.embeddedObjects = true;
|
|
47
|
+
if (/<text:note\b/.test(contentXml))
|
|
48
|
+
riskFlags.footnotes = true;
|
|
49
|
+
if (/<table:table\b/.test(contentXml))
|
|
50
|
+
riskFlags.tables = true;
|
|
51
|
+
}
|
|
52
|
+
// Check for macros
|
|
53
|
+
if (zip.file('Basic/') || zip.file('Scripts/'))
|
|
54
|
+
riskFlags.macros = true;
|
|
55
|
+
// Extract text via officeparser
|
|
56
|
+
const text = z.string().parse(await officeparser.parseOffice(buffer));
|
|
57
|
+
const words = countWords(text);
|
|
58
|
+
const paragraphs = text.split(/\n\n+/).filter(p => p.trim().length > 0).length;
|
|
59
|
+
// Count images and tables from content.xml
|
|
60
|
+
let images = 0;
|
|
61
|
+
let tables = 0;
|
|
62
|
+
if (contentXml) {
|
|
63
|
+
images = (contentXml.match(/<draw:image\b/g) || []).length;
|
|
64
|
+
tables = (contentXml.match(/<table:table\b/g) || []).length;
|
|
65
|
+
}
|
|
66
|
+
const contentStats = {
|
|
67
|
+
words,
|
|
68
|
+
pages: Math.max(1, Math.ceil(words / 250)),
|
|
69
|
+
pagesEstimated: true,
|
|
70
|
+
paragraphs,
|
|
71
|
+
characters: text.length,
|
|
72
|
+
tables,
|
|
73
|
+
images,
|
|
74
|
+
};
|
|
75
|
+
// Build preview
|
|
76
|
+
const textParagraphs = text.split(/\n\n+/).filter(p => p.trim().length > 0);
|
|
77
|
+
const previewParagraphs = [];
|
|
78
|
+
const limit = Math.min(sampleParagraphs, textParagraphs.length);
|
|
79
|
+
for (let i = 0; i < limit; i++) {
|
|
80
|
+
previewParagraphs.push({
|
|
81
|
+
index: i,
|
|
82
|
+
text: textParagraphs[i].trim().replace(/\s+/g, ' '),
|
|
83
|
+
isHeading: false,
|
|
84
|
+
});
|
|
85
|
+
}
|
|
86
|
+
const preview = {
|
|
87
|
+
truncated: textParagraphs.length > sampleParagraphs,
|
|
88
|
+
paragraphs: previewParagraphs,
|
|
89
|
+
};
|
|
90
|
+
return { properties, riskFlags, contentStats, preview };
|
|
91
|
+
}
|
|
92
|
+
//# 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,CAAC,EAAE,MAAM,KAAK,CAAC;AACxB,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,CAAC,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,MAAM,YAAY,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC;IACtE,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 +1,88 @@
|
|
|
1
|
+
import pdf from 'pdf-parse';
|
|
2
|
+
import { z } from 'zod';
|
|
3
|
+
import { countWords } from '../utils.js';
|
|
4
|
+
import { asOptionalString, formatDateLike } from '../inspect/shared.js';
|
|
5
|
+
const PdfParseResultSchema = z.object({
|
|
6
|
+
text: z.string(),
|
|
7
|
+
numpages: z.number(),
|
|
8
|
+
info: z.record(z.string(), z.unknown()).default({}),
|
|
9
|
+
});
|
|
10
|
+
function parsePdfDate(value) {
|
|
11
|
+
if (!value)
|
|
12
|
+
return undefined;
|
|
13
|
+
const str = String(value);
|
|
14
|
+
// PDF dates: D:YYYYMMDDHHmmSS or plain ISO strings
|
|
15
|
+
const match = str.match(/^D:(\d{4})(\d{2})(\d{2})(\d{2})(\d{2})(\d{2})/);
|
|
16
|
+
if (match) {
|
|
17
|
+
return `${match[1]}-${match[2]}-${match[3]}T${match[4]}:${match[5]}:${match[6]}Z`;
|
|
18
|
+
}
|
|
19
|
+
return formatDateLike(str);
|
|
20
|
+
}
|
|
21
|
+
export async function inspectPdf(filePath, buffer, sampleParagraphs) {
|
|
22
|
+
// Suppress pdf.js warnings
|
|
23
|
+
const originalLog = console.log;
|
|
24
|
+
console.log = (...args) => {
|
|
25
|
+
if (typeof args[0] === 'string' && (args[0].startsWith('Warning: ') || args[0].startsWith('Info: ') || args[0].startsWith('Deprecated API usage: '))) {
|
|
26
|
+
return;
|
|
27
|
+
}
|
|
28
|
+
originalLog.apply(console, args);
|
|
29
|
+
};
|
|
30
|
+
let data;
|
|
31
|
+
try {
|
|
32
|
+
data = PdfParseResultSchema.parse(await pdf(buffer));
|
|
33
|
+
}
|
|
34
|
+
finally {
|
|
35
|
+
console.log = originalLog;
|
|
36
|
+
}
|
|
37
|
+
const info = data.info;
|
|
38
|
+
const properties = {
|
|
39
|
+
title: asOptionalString(info.Title),
|
|
40
|
+
subject: asOptionalString(info.Subject),
|
|
41
|
+
author: asOptionalString(info.Author),
|
|
42
|
+
keywords: asOptionalString(info.Keywords),
|
|
43
|
+
createdDate: parsePdfDate(info.CreationDate),
|
|
44
|
+
modifiedDate: parsePdfDate(info.ModDate),
|
|
45
|
+
};
|
|
46
|
+
// PDFs have limited risk flag detection
|
|
47
|
+
const riskFlags = {
|
|
48
|
+
comments: false,
|
|
49
|
+
trackedChanges: false,
|
|
50
|
+
hyperlinks: false,
|
|
51
|
+
embeddedObjects: false,
|
|
52
|
+
footnotes: false,
|
|
53
|
+
endnotes: false,
|
|
54
|
+
macros: false,
|
|
55
|
+
headerFooter: false,
|
|
56
|
+
tables: false,
|
|
57
|
+
encrypted: !!info.IsAcroFormPresent || info.Encrypted === true,
|
|
58
|
+
};
|
|
59
|
+
const text = data.text || '';
|
|
60
|
+
const words = countWords(text);
|
|
61
|
+
const paragraphs = text.split(/\n\n+/).filter(p => p.trim().length > 0).length;
|
|
62
|
+
const contentStats = {
|
|
63
|
+
words,
|
|
64
|
+
pages: data.numpages,
|
|
65
|
+
pagesEstimated: false,
|
|
66
|
+
paragraphs,
|
|
67
|
+
characters: text.length,
|
|
68
|
+
tables: 0,
|
|
69
|
+
images: 0,
|
|
70
|
+
};
|
|
71
|
+
// Build preview from text
|
|
72
|
+
const textParagraphs = text.split(/\n\n+/).filter(p => p.trim().length > 0);
|
|
73
|
+
const previewParagraphs = [];
|
|
74
|
+
const limit = Math.min(sampleParagraphs, textParagraphs.length);
|
|
75
|
+
for (let i = 0; i < limit; i++) {
|
|
76
|
+
previewParagraphs.push({
|
|
77
|
+
index: i,
|
|
78
|
+
text: textParagraphs[i].trim().replace(/\s+/g, ' '),
|
|
79
|
+
isHeading: false,
|
|
80
|
+
});
|
|
81
|
+
}
|
|
82
|
+
const preview = {
|
|
83
|
+
truncated: textParagraphs.length > sampleParagraphs,
|
|
84
|
+
paragraphs: previewParagraphs,
|
|
85
|
+
};
|
|
86
|
+
return { properties, riskFlags, contentStats, preview };
|
|
87
|
+
}
|
|
88
|
+
//# sourceMappingURL=inspect-pdf.js.map
|