@tangle-network/agent-app 0.45.21 → 0.45.23
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/dist/DesignCanvas-ZG7DJS7C.js +10 -0
- package/dist/DesignCanvasEditor-SLFMHOII.js +13 -0
- package/dist/app-auth/index.js +2 -2
- package/dist/assistant/index.d.ts +1 -1
- package/dist/assistant/index.js +5 -5
- package/dist/{attachment-validation-CsmUzuI3.d.ts → attachment-validation-BfJDBU1g.d.ts} +41 -5
- package/dist/{auth-DJjJBtYw.d.ts → auth-BYtiazA4.d.ts} +1 -1
- package/dist/chat-react/index.js +2 -2
- package/dist/chat-routes/index.d.ts +8 -4
- package/dist/chat-routes/index.js +21 -4
- package/dist/chat-routes/index.js.map +1 -1
- package/dist/{chunk-KPRBFNGT.js → chunk-3IHYJIDJ.js} +3 -3
- package/dist/{chunk-MYS7DHMM.js → chunk-4QVG4WJQ.js} +2 -2
- package/dist/{chunk-RMHZKMQ4.js → chunk-73CQJOIU.js} +5 -5
- package/dist/{chunk-AH3IOSWF.js → chunk-7PFFFRVU.js} +2077 -341
- package/dist/chunk-7PFFFRVU.js.map +1 -0
- package/dist/chunk-CCVG2TL6.js +327 -0
- package/dist/chunk-CCVG2TL6.js.map +1 -0
- package/dist/{chunk-U4BXAZCN.js → chunk-DRB77FYL.js} +2 -2
- package/dist/{chunk-6MUJROBT.js → chunk-E4DHYENH.js} +2 -2
- package/dist/chunk-F7VH4G5C.js +283 -0
- package/dist/chunk-F7VH4G5C.js.map +1 -0
- package/dist/chunk-I24YIKHO.js +183 -0
- package/dist/chunk-I24YIKHO.js.map +1 -0
- package/dist/{chunk-OO4DROKM.js → chunk-OASVAPRL.js} +3 -3
- package/dist/{chunk-ZSGTG6PK.js → chunk-P5PIAQVS.js} +2 -2
- package/dist/chunk-PC2WYTK7.js +405 -0
- package/dist/chunk-PC2WYTK7.js.map +1 -0
- package/dist/{chunk-YEFFHORB.js → chunk-TX6S7XXU.js} +8 -2
- package/dist/chunk-TX6S7XXU.js.map +1 -0
- package/dist/chunk-UGWQLQDS.js +35 -0
- package/dist/chunk-UGWQLQDS.js.map +1 -0
- package/dist/{chunk-O22STZ3P.js → chunk-X5DQ5QOP.js} +2 -2
- package/dist/chunk-XEA5EG6P.js +1478 -0
- package/dist/chunk-XEA5EG6P.js.map +1 -0
- package/dist/{chunk-PCJHEZA7.js → chunk-YABHE2QY.js} +2 -2
- package/dist/design-canvas/index.d.ts +3 -3
- package/dist/design-canvas/index.js +2 -2
- package/dist/design-canvas-react/engine.js +4 -4
- package/dist/design-canvas-react/index.js +7 -7
- package/dist/design-canvas-react/lazy.js +1 -1
- package/dist/documents/index.d.ts +233 -0
- package/dist/documents/index.js +495 -0
- package/dist/documents/index.js.map +1 -0
- package/dist/documents/pdf-inspector.d.ts +60 -0
- package/dist/documents/pdf-inspector.js +100 -0
- package/dist/documents/pdf-inspector.js.map +1 -0
- package/dist/eval/index.d.ts +1 -1
- package/dist/fold-DACU7gtf.d.ts +320 -0
- package/dist/legibility/cli.d.ts +1 -0
- package/dist/legibility/cli.js +143 -0
- package/dist/legibility/cli.js.map +1 -0
- package/dist/legibility/index.d.ts +572 -0
- package/dist/legibility/index.js +33 -0
- package/dist/legibility/index.js.map +1 -0
- package/dist/{mcp-BiKqnsJV.d.ts → mcp-xxoZZfor.d.ts} +2 -2
- package/dist/openui/index.d.ts +208 -0
- package/dist/openui/index.js +350 -0
- package/dist/openui/index.js.map +1 -0
- package/dist/openui-react/index.d.ts +119 -0
- package/dist/openui-react/index.js +149 -0
- package/dist/openui-react/index.js.map +1 -0
- package/dist/platform/index.js +2 -2
- package/dist/preset-cloudflare/index.d.ts +2 -2
- package/dist/rate-limit-BOVmJRXZ.d.ts +35 -0
- package/dist/record/drizzle.d.ts +682 -0
- package/dist/record/drizzle.js +579 -0
- package/dist/record/drizzle.js.map +1 -0
- package/dist/record/index.d.ts +16 -0
- package/dist/record/index.js +35 -0
- package/dist/record/index.js.map +1 -0
- package/dist/runtime/index.d.ts +3 -3
- package/dist/sandbox/index.d.ts +2 -2
- package/dist/sandbox/index.js +3 -2
- package/dist/sequences/index.d.ts +3 -3
- package/dist/sequences/index.js +1 -1
- package/dist/session-shell/index.d.ts +173 -1
- package/dist/session-shell/index.js +13 -1
- package/dist/teams/index.js +9 -9
- package/dist/teams/invitations-api.js +3 -3
- package/dist/tools/index.d.ts +5 -5
- package/dist/tools/index.js +3 -2
- package/dist/{types-DbU-oO5h.d.ts → types-B25Mib1r.d.ts} +11 -0
- package/dist/types-DrBidf3O.d.ts +123 -0
- package/dist/values-C7RidJuI.d.ts +156 -0
- package/dist/web/index.d.ts +197 -31
- package/dist/web/index.js +15 -3
- package/dist/web-react/async/index.d.ts +309 -0
- package/dist/web-react/async/index.js +327 -0
- package/dist/web-react/async/index.js.map +1 -0
- package/dist/web-react/index.d.ts +586 -2
- package/dist/web-react/index.js +84 -6
- package/dist/work-product/index.d.ts +1 -1
- package/dist/work-product/index.js +2 -1
- package/dist/work-product/index.js.map +1 -1
- package/dist/workspace-react/index.js +1 -1
- package/package.json +49 -3
- package/dist/DesignCanvas-NEIW3AER.js +0 -10
- package/dist/DesignCanvasEditor-QD33OI2W.js +0 -13
- package/dist/chunk-AH3IOSWF.js.map +0 -1
- package/dist/chunk-GINKOZFD.js +0 -214
- package/dist/chunk-GINKOZFD.js.map +0 -1
- package/dist/chunk-J25RST7G.js +0 -126
- package/dist/chunk-J25RST7G.js.map +0 -1
- package/dist/chunk-WBHPN5DY.js +0 -158
- package/dist/chunk-WBHPN5DY.js.map +0 -1
- package/dist/chunk-YEFFHORB.js.map +0 -1
- /package/dist/{DesignCanvas-NEIW3AER.js.map → DesignCanvas-ZG7DJS7C.js.map} +0 -0
- /package/dist/{DesignCanvasEditor-QD33OI2W.js.map → DesignCanvasEditor-SLFMHOII.js.map} +0 -0
- /package/dist/{chunk-KPRBFNGT.js.map → chunk-3IHYJIDJ.js.map} +0 -0
- /package/dist/{chunk-MYS7DHMM.js.map → chunk-4QVG4WJQ.js.map} +0 -0
- /package/dist/{chunk-RMHZKMQ4.js.map → chunk-73CQJOIU.js.map} +0 -0
- /package/dist/{chunk-U4BXAZCN.js.map → chunk-DRB77FYL.js.map} +0 -0
- /package/dist/{chunk-6MUJROBT.js.map → chunk-E4DHYENH.js.map} +0 -0
- /package/dist/{chunk-OO4DROKM.js.map → chunk-OASVAPRL.js.map} +0 -0
- /package/dist/{chunk-ZSGTG6PK.js.map → chunk-P5PIAQVS.js.map} +0 -0
- /package/dist/{chunk-O22STZ3P.js.map → chunk-X5DQ5QOP.js.map} +0 -0
- /package/dist/{chunk-PCJHEZA7.js.map → chunk-YABHE2QY.js.map} +0 -0
|
@@ -0,0 +1,233 @@
|
|
|
1
|
+
import { P as PdfEngine, a as PdfTextFormat, D as DocumentOutcome, E as ExtractedDocument, b as PdfClassification, c as DocxExtractionDetail, d as DocumentFormat } from '../types-DrBidf3O.js';
|
|
2
|
+
export { e as DocumentExtractionError, f as DocumentExtractionErrorCode, g as DocumentExtractionStage, h as PdfContentKind, i as PdfExtractionDetail, j as PdfPageClassification } from '../types-DrBidf3O.js';
|
|
3
|
+
|
|
4
|
+
/**
|
|
5
|
+
* The extraction pipeline: size gate → media type → (PDF: classify, then
|
|
6
|
+
* extract) / (DOCX: OPC read) / (text: strict decode).
|
|
7
|
+
*
|
|
8
|
+
* Two rules make this safe to build a product on.
|
|
9
|
+
*
|
|
10
|
+
* 1. **Classify before extracting a PDF.** The wasm engine's `extractText`
|
|
11
|
+
* panics (`RuntimeError: unreachable`) on any page with no text layer, and
|
|
12
|
+
* that is measured, not theoretical — including on `mixed` documents where
|
|
13
|
+
* only ONE page is an image. Classification runs first and decides whether
|
|
14
|
+
* extraction is even reachable, which is also what produces the OCR signal.
|
|
15
|
+
* 2. **A scanned PDF is `pdf-needs-ocr`, never an empty string.** Both hand-
|
|
16
|
+
* rolled predecessors of this module got this right and it is the single
|
|
17
|
+
* behaviour a product cannot afford to lose: silent empty text lets a
|
|
18
|
+
* downstream model answer confidently about a document nobody read.
|
|
19
|
+
*/
|
|
20
|
+
|
|
21
|
+
/** 25 MB — the cap both predecessor implementations converged on, and roughly
|
|
22
|
+
* the point where a Worker's memory budget stops tolerating a full decode. */
|
|
23
|
+
declare const DEFAULT_MAX_DOCUMENT_BYTES: number;
|
|
24
|
+
interface ExtractDocumentOptions {
|
|
25
|
+
/** The type the upload declared. Falls back to `filename`'s extension when
|
|
26
|
+
* missing or `application/octet-stream`. */
|
|
27
|
+
readonly mediaType?: string;
|
|
28
|
+
/** Used to resolve the media type, and named in error messages. */
|
|
29
|
+
readonly filename?: string;
|
|
30
|
+
/** Required to extract PDFs. Supplied by the caller because the wasm asset
|
|
31
|
+
* must come from the consumer's bundler — see
|
|
32
|
+
* `@tangle-network/agent-app/documents/pdf-inspector`. */
|
|
33
|
+
readonly pdf?: PdfEngine;
|
|
34
|
+
/** Byte cap on the uploaded ARCHIVE, default
|
|
35
|
+
* {@link DEFAULT_MAX_DOCUMENT_BYTES}. */
|
|
36
|
+
readonly maxBytes?: number;
|
|
37
|
+
/**
|
|
38
|
+
* Byte cap on what any ONE part of an Office package may expand to, default
|
|
39
|
+
* {@link DEFAULT_MAX_ZIP_ENTRY_BYTES}.
|
|
40
|
+
*
|
|
41
|
+
* A separate dial from {@link maxBytes} because they bound different things:
|
|
42
|
+
* deflate reaches ~1000:1 on repetitive input, so capping the upload does
|
|
43
|
+
* nothing to cap the decompression it asks for.
|
|
44
|
+
*/
|
|
45
|
+
readonly maxUncompressedPartBytes?: number;
|
|
46
|
+
/**
|
|
47
|
+
* Plain text (default) preserves source line breaks, which line-anchored
|
|
48
|
+
* parsers depend on; markdown carries headings and page markers but reflows
|
|
49
|
+
* lines into paragraphs.
|
|
50
|
+
*
|
|
51
|
+
* A *preference*, because of exactly one measured case: on a partially
|
|
52
|
+
* scanned PDF the plain-text engine panics, and only the markdown engine can
|
|
53
|
+
* recover the pages that do have text. The result reports what you actually
|
|
54
|
+
* got in {@link PdfExtractionDetail.textFormat} — check it when the
|
|
55
|
+
* distinction matters.
|
|
56
|
+
*/
|
|
57
|
+
readonly preferredPdfFormat?: PdfTextFormat;
|
|
58
|
+
}
|
|
59
|
+
/**
|
|
60
|
+
* Classify a PDF: whether it has a text layer, and which pages do not.
|
|
61
|
+
*
|
|
62
|
+
* Safe on every input — classification is what makes extraction safe, so it can
|
|
63
|
+
* never be the thing that panics. Call this directly when a product wants to
|
|
64
|
+
* show "3 of 40 pages need OCR" before deciding to extract.
|
|
65
|
+
*/
|
|
66
|
+
declare function classifyPdfDocument(bytes: Uint8Array, pdf: PdfEngine): DocumentOutcome<PdfClassification>;
|
|
67
|
+
/**
|
|
68
|
+
* Extract a document's text.
|
|
69
|
+
*
|
|
70
|
+
* Every failure is a typed {@link DocumentExtractionError} naming the stage it
|
|
71
|
+
* stopped at; nothing throws, and success never carries empty text.
|
|
72
|
+
*/
|
|
73
|
+
declare function extractDocument(input: ArrayBuffer | Uint8Array, options?: ExtractDocumentOptions): Promise<DocumentOutcome<ExtractedDocument>>;
|
|
74
|
+
interface DocumentExtractorConfig {
|
|
75
|
+
readonly pdf?: PdfEngine;
|
|
76
|
+
readonly maxBytes?: number;
|
|
77
|
+
readonly maxUncompressedPartBytes?: number;
|
|
78
|
+
readonly preferredPdfFormat?: PdfTextFormat;
|
|
79
|
+
}
|
|
80
|
+
interface DocumentExtractor {
|
|
81
|
+
readonly extract: (input: ArrayBuffer | Uint8Array, options?: Omit<ExtractDocumentOptions, 'pdf' | 'maxBytes' | 'maxUncompressedPartBytes' | 'preferredPdfFormat'>) => Promise<DocumentOutcome<ExtractedDocument>>;
|
|
82
|
+
/** Classify a PDF without extracting. Fails with `pdf-engine-unavailable`
|
|
83
|
+
* when the extractor was built without a PDF engine. */
|
|
84
|
+
readonly classifyPdf: (input: ArrayBuffer | Uint8Array) => DocumentOutcome<PdfClassification>;
|
|
85
|
+
}
|
|
86
|
+
/**
|
|
87
|
+
* Bind an engine and limits once, at worker start, so call sites read
|
|
88
|
+
* `documents.extract(bytes, { mediaType })` and never re-wire the wasm.
|
|
89
|
+
*/
|
|
90
|
+
declare function createDocumentExtractor(config?: DocumentExtractorConfig): DocumentExtractor;
|
|
91
|
+
|
|
92
|
+
/**
|
|
93
|
+
* The slice of PKZIP an OOXML package needs: read the central directory, then
|
|
94
|
+
* inflate one named entry.
|
|
95
|
+
*
|
|
96
|
+
* Why not a zip library: DOCX text extraction is the only reason this module
|
|
97
|
+
* needs a zip, and the runtime already supplies the hard part —
|
|
98
|
+
* `DecompressionStream('deflate-raw')` is a standard API present in workerd
|
|
99
|
+
* and in Node, so the whole reader is header arithmetic plus a stream. Pulling
|
|
100
|
+
* a zip package in would put ten transitive Node-shaped dependencies into every
|
|
101
|
+
* consumer's Worker bundle to avoid ~120 lines.
|
|
102
|
+
*
|
|
103
|
+
* The CENTRAL DIRECTORY is the source of truth, not the local headers: an
|
|
104
|
+
* archive written with data descriptors (streamed output) carries zeros for
|
|
105
|
+
* the sizes in its local headers, and reading those is the classic way a
|
|
106
|
+
* hand-rolled zip reader silently returns nothing.
|
|
107
|
+
*/
|
|
108
|
+
type ZipOutcome<T> = {
|
|
109
|
+
succeeded: true;
|
|
110
|
+
value: T;
|
|
111
|
+
} | {
|
|
112
|
+
succeeded: false;
|
|
113
|
+
error: string;
|
|
114
|
+
};
|
|
115
|
+
interface ZipEntry {
|
|
116
|
+
readonly name: string;
|
|
117
|
+
readonly compressionMethod: number;
|
|
118
|
+
readonly compressedSize: number;
|
|
119
|
+
readonly uncompressedSize: number;
|
|
120
|
+
readonly localHeaderOffset: number;
|
|
121
|
+
readonly encrypted: boolean;
|
|
122
|
+
}
|
|
123
|
+
/** Read every central-directory entry. Names are decoded as UTF-8, which is
|
|
124
|
+
* what OOXML producers write (general-purpose bit 11). */
|
|
125
|
+
declare function readZipDirectory(bytes: Uint8Array): ZipOutcome<ZipEntry[]>;
|
|
126
|
+
/**
|
|
127
|
+
* Ceiling on what ONE entry may expand to. A deflate stream reaches ~1000:1 on
|
|
128
|
+
* repetitive input, so an archive small enough to pass any upload gate can ask
|
|
129
|
+
* for gigabytes: measured here, a 305,893-byte archive declared a single entry
|
|
130
|
+
* expanding to 314,572,800 bytes. A Cloudflare Worker isolate has 128 MB, so
|
|
131
|
+
* that is an out-of-memory crash, not a slow path — and a crash is not a typed
|
|
132
|
+
* error a product can report. 32 MB is far past any real Office part (the main
|
|
133
|
+
* document of a 400-page contract is single-digit MB) and far under the
|
|
134
|
+
* isolate's budget.
|
|
135
|
+
*/
|
|
136
|
+
declare const DEFAULT_MAX_ZIP_ENTRY_BYTES: number;
|
|
137
|
+
/** Options for {@link readZipEntry}. */
|
|
138
|
+
interface ReadZipEntryOptions {
|
|
139
|
+
/** Ceiling on the entry's UNCOMPRESSED size. Defaults to
|
|
140
|
+
* {@link DEFAULT_MAX_ZIP_ENTRY_BYTES}. Checked against the directory's
|
|
141
|
+
* declared size before any inflate runs, and again against the real output
|
|
142
|
+
* as it streams, because the declared size is the archive's claim. */
|
|
143
|
+
readonly maxUncompressedBytes?: number;
|
|
144
|
+
}
|
|
145
|
+
/** Inflate one entry's bytes. The local header supplies only the variable-length
|
|
146
|
+
* field sizes needed to find the payload; sizes come from the directory. */
|
|
147
|
+
declare function readZipEntry(bytes: Uint8Array, entry: ZipEntry, options?: ReadZipEntryOptions): Promise<ZipOutcome<Uint8Array>>;
|
|
148
|
+
|
|
149
|
+
/**
|
|
150
|
+
* DOCX → plain text, one paragraph per line.
|
|
151
|
+
*
|
|
152
|
+
* The main document part is resolved through the OPC relationship graph
|
|
153
|
+
* (`_rels/.rels` → the `officeDocument` relationship's target) rather than
|
|
154
|
+
* assuming `word/document.xml`, because the part name is a producer's choice,
|
|
155
|
+
* not a constant.
|
|
156
|
+
*
|
|
157
|
+
* WordprocessingML text lives in `<w:t>` runs. Everything else is markup, with
|
|
158
|
+
* four exceptions that carry meaning a reader would otherwise lose: `<w:tab/>`,
|
|
159
|
+
* `<w:br/>`, `<w:cr/>`, and the paragraph boundary `</w:p>`. Two element bodies
|
|
160
|
+
* are deliberately DROPPED — `<w:instrText>` (field instruction codes such as
|
|
161
|
+
* `HYPERLINK "…"`, which are machinery, not prose) and `<w:delText>` (text a
|
|
162
|
+
* tracked-changes revision deleted; extracting it would show a reader words the
|
|
163
|
+
* author removed).
|
|
164
|
+
*
|
|
165
|
+
* One line per paragraph, not blank-line-separated paragraphs: extractors
|
|
166
|
+
* downstream match line-anchored patterns against form and contract text, so a
|
|
167
|
+
* predictable line structure is the useful shape.
|
|
168
|
+
*/
|
|
169
|
+
|
|
170
|
+
/** Resolve XML character/entity references. Unknown entities are left verbatim
|
|
171
|
+
* rather than dropped, so nothing silently disappears from a document. */
|
|
172
|
+
declare function decodeXmlEntities(input: string): string;
|
|
173
|
+
interface DocxText {
|
|
174
|
+
readonly text: string;
|
|
175
|
+
readonly paragraphCount: number;
|
|
176
|
+
}
|
|
177
|
+
/**
|
|
178
|
+
* Walk WordprocessingML and emit its readable text.
|
|
179
|
+
*
|
|
180
|
+
* Hand-scanned rather than regex-split because comments and CDATA sections may
|
|
181
|
+
* contain `>`; a `/<[^>]*>/` scan ends the tag early on the first one and
|
|
182
|
+
* corrupts everything after it.
|
|
183
|
+
*/
|
|
184
|
+
declare function wordprocessingXmlToText(xml: string): DocxText;
|
|
185
|
+
/**
|
|
186
|
+
* Extract the readable text of a `.docx` / `.docm` package. Every failure names
|
|
187
|
+
* the archive-level cause; a package this cannot read is an error, never an
|
|
188
|
+
* empty document.
|
|
189
|
+
*
|
|
190
|
+
* `zip.maxUncompressedBytes` bounds what each part may expand to. It defaults
|
|
191
|
+
* to the zip reader's own ceiling, which is what keeps a deflate bomb inside a
|
|
192
|
+
* small `.docx` a typed error instead of an out-of-memory crash.
|
|
193
|
+
*/
|
|
194
|
+
declare function extractDocxText(bytes: Uint8Array, mediaType: string, zip?: ReadZipEntryOptions): Promise<DocumentOutcome<{
|
|
195
|
+
text: string;
|
|
196
|
+
detail: DocxExtractionDetail;
|
|
197
|
+
}>>;
|
|
198
|
+
|
|
199
|
+
/**
|
|
200
|
+
* Media type → extraction format. Resolution order is declared type first,
|
|
201
|
+
* filename extension second (products routinely receive `application/octet-stream`
|
|
202
|
+
* from a browser upload), and nothing else — byte sniffing belongs to the
|
|
203
|
+
* upload gate that reads untrusted bytes, not to the extractor that has already
|
|
204
|
+
* been handed a decision.
|
|
205
|
+
*/
|
|
206
|
+
|
|
207
|
+
/** OOXML WordprocessingML — the `.docx` a word processor writes. */
|
|
208
|
+
declare const DOCX_MEDIA_TYPE = "application/vnd.openxmlformats-officedocument.wordprocessingml.document";
|
|
209
|
+
/** Macro-enabled WordprocessingML (`.docm`). Same part structure, so text
|
|
210
|
+
* extraction is identical; the VBA project is never read or executed. */
|
|
211
|
+
declare const DOCM_MEDIA_TYPE = "application/vnd.ms-word.document.macroEnabled.12";
|
|
212
|
+
declare const PDF_MEDIA_TYPE = "application/pdf";
|
|
213
|
+
interface ResolvedMediaType {
|
|
214
|
+
readonly format: DocumentFormat;
|
|
215
|
+
/** Canonical media type for the resolved format. */
|
|
216
|
+
readonly mediaType: string;
|
|
217
|
+
}
|
|
218
|
+
/** Strip parameters and case from a media type: `Text/Plain; charset=utf-8` →
|
|
219
|
+
* `text/plain`. */
|
|
220
|
+
declare function normalizeMediaType(mediaType: string): string;
|
|
221
|
+
interface ResolveMediaTypeInput {
|
|
222
|
+
/** The type the upload declared. May be empty or `application/octet-stream`. */
|
|
223
|
+
readonly mediaType?: string;
|
|
224
|
+
/** Consulted only when `mediaType` is missing or generic. */
|
|
225
|
+
readonly filename?: string;
|
|
226
|
+
}
|
|
227
|
+
/**
|
|
228
|
+
* Decide which extractor handles these bytes. Fails loud rather than guessing:
|
|
229
|
+
* an unrecognized type names itself and lists what is supported.
|
|
230
|
+
*/
|
|
231
|
+
declare function resolveMediaType(input: ResolveMediaTypeInput): DocumentOutcome<ResolvedMediaType>;
|
|
232
|
+
|
|
233
|
+
export { DEFAULT_MAX_DOCUMENT_BYTES, DEFAULT_MAX_ZIP_ENTRY_BYTES, DOCM_MEDIA_TYPE, DOCX_MEDIA_TYPE, type DocumentExtractor, type DocumentExtractorConfig, DocumentFormat, DocumentOutcome, DocxExtractionDetail, type DocxText, type ExtractDocumentOptions, ExtractedDocument, PDF_MEDIA_TYPE, PdfClassification, PdfEngine, PdfTextFormat, type ReadZipEntryOptions, type ResolveMediaTypeInput, type ResolvedMediaType, type ZipEntry, type ZipOutcome, classifyPdfDocument, createDocumentExtractor, decodeXmlEntities, extractDocument, extractDocxText, normalizeMediaType, readZipDirectory, readZipEntry, resolveMediaType, wordprocessingXmlToText };
|