@pdfvector/instance-contract 0.0.27 → 0.0.29

package/.tsc/lib/supported-mimes.js

@@ -0,0 +1,122 @@
+/**
+ * All MIME types accepted for file uploads across all API endpoints.
+ * Single source of truth — imported by all contract schemas.
+ */
+export const supportedFileMimes = [
+    // PDF
+    "application/pdf",
+    // Office documents
+    "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+    "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+    "application/vnd.openxmlformats-officedocument.presentationml.presentation",
+    // CSV
+    "text/csv",
+    "application/csv",
+    // Images
+    "image/png",
+    "image/jpeg",
+    "image/tiff",
+    "image/bmp",
+    "image/heif",
+    "image/heic",
+    // Plain text & structured text
+    "text/plain",
+    "text/markdown",
+    "text/tab-separated-values",
+    "text/xml",
+    "application/xml",
+    // RTF
+    "application/rtf",
+    "text/rtf",
+    // HTML
+    "text/html",
+    // OpenDocument & EPUB
+    "application/epub+zip",
+    "application/vnd.oasis.opendocument.text",
+    "application/vnd.oasis.opendocument.spreadsheet",
+    "application/vnd.oasis.opendocument.presentation",
+    // Bibliography / Academic
+    "application/x-bibtex",
+];
+export const supportedFileFormatsDescription = "PDF, DOCX, XLSX, PPTX, CSV, PNG, JPG, TIFF, BMP, HEIF, TXT, MD, TSV, XML, RTF, HTML, ODT, ODS, ODP, EPUB, BIB, RIS, NBIB, ENW";
+/**
+ * Human-readable description of supported file types with extensions.
+ * Used in route-level API descriptions.
+ */
+export const supportedFileTypesLong = "PDF, Word (.docx), Excel (.xlsx), PowerPoint (.pptx), CSV, " +
+    "Image (.png, .jpg, .tiff, .bmp, .heif), " +
+    "Plain Text (.txt, .md, .tsv, .xml), RTF, HTML, " +
+    "OpenDocument (.odt, .ods, .odp), EPUB, " +
+    "and Bibliography (.bib, .ris, .nbib, .enw) files";
+/**
+ * Per-tier file type support descriptions for model input fields.
+ * Format lists are stated once at the bottom to avoid wall-of-text in Scalar.
+ */
+const formatNote = "\n\n" +
+    "Supported formats by tier:\n" +
+    "- All tiers: PDF, Word, Excel, PowerPoint, CSV, Text, HTML, RTF, OpenDocument, EPUB, Bibliography.\n" +
+    "- Pro adds: Image (PNG, JPG).\n" +
+    "- Max/Auto adds: Image (PNG, JPG, TIFF, BMP, HEIF).";
+/** Model tier descriptions for document parse endpoints. */
+export const documentParseModelDescription = "Model tier for parsing.\n\n" +
+    "- auto (default): Intelligent fallback. Up to 1000 pages, 500MB.\n" +
+    "- nano: 1 credit/page. Simple plain text documents. Up to 30 pages, 10MB.\n" +
+    "- mini: 2 credits/page. Documents with tables and structured content. Up to 30 pages, 10MB.\n" +
+    "- pro: 4 credits/page. Tables, handwritten text, figures, math, Arabic. Up to 30 pages, 40MB.\n" +
+    "- max: 8 credits/page. Full Pro capabilities + enhanced multilingual. Up to 1000 pages, 500MB." +
+    formatNote;
+/** Model tier descriptions for document extract endpoints. */
+export const documentExtractModelDescription = "Model tier for extracting structured data.\n\n" +
+    "- auto (default): Automatically selects the best tier based on document complexity.\n" +
+    "- nano: 2 credits/page. Fastest. Best for simple documents with straightforward schemas.\n" +
+    "- mini: 4 credits/page. Balanced speed and accuracy. Moderately complex schemas.\n" +
+    "- pro: 8 credits/page. High accuracy for complex documents with large or nested schemas.\n" +
+    "- max: 16 credits/page. Maximum accuracy. Best for difficult extractions requiring deep reasoning." +
+    formatNote;
+/** Model tier descriptions for document ask endpoints. */
+export const documentAskModelDescription = "Model tier for answering the question.\n\n" +
+    "- auto (default): Automatically selects the best tier based on document complexity.\n" +
+    "- nano: 2 credits/page. Fastest. Best for simple questions about straightforward documents.\n" +
+    "- mini: 4 credits/page. Balanced speed and accuracy. Moderately complex questions.\n" +
+    "- pro: 8 credits/page. High accuracy for nuanced questions about complex documents.\n" +
+    "- max: 16 credits/page. Maximum accuracy. Best for difficult questions requiring deep reasoning." +
+    formatNote;
+/** Model tier descriptions for invoice/identity/bankStatement parse endpoints (pro/max/auto only). */
+export const specializedParseModelDescription = (type) => "Model tier for parsing.\n\n" +
+    "- auto (default): Intelligent fallback.\n" +
+    `- pro: 6 credits/page. Extracts structured ${type} fields with standard accuracy.\n` +
+    `- max: 10 credits/page. Extracts structured ${type} fields with highest accuracy and fallback.`;
+/** Model tier descriptions for invoice/identity/bankStatement extract endpoints. */
+export const specializedExtractModelDescription = "Model tier for extracting structured data.\n\n" +
+    "- auto (default): Automatically selects the best tier based on document complexity.\n" +
+    "- nano: 6 credits/page. Fastest. Best for simple documents with straightforward schemas.\n" +
+    "- mini: 10 credits/page. Balanced speed and accuracy. Moderately complex schemas.\n" +
+    "- pro: 14 credits/page. High accuracy for complex documents with large or nested schemas.\n" +
+    "- max: 18 credits/page. Maximum accuracy. Best for difficult extractions requiring deep reasoning." +
+    formatNote;
+/** Model tier descriptions for invoice/identity/bankStatement ask endpoints. */
+export const specializedAskModelDescription = "Model tier for answering the question.\n\n" +
+    "- auto (default): Automatically selects the best tier based on document complexity.\n" +
+    "- nano: 6 credits/page. Fastest. Best for simple questions about straightforward documents.\n" +
+    "- mini: 10 credits/page. Balanced speed and accuracy. Moderately complex questions.\n" +
+    "- pro: 14 credits/page. High accuracy for nuanced questions about complex documents.\n" +
+    "- max: 18 credits/page. Maximum accuracy. Best for difficult questions requiring deep reasoning." +
+    formatNote;
+/** Output model description for parse results. */
+export const outputModelDescription = "Model tier used to parse the document.\n\n" +
+    "- nano: Fastest, best for simple documents.\n" +
+    "- mini: Balanced speed and accuracy.\n" +
+    "- pro: High accuracy for complex documents.\n" +
+    "- max: Maximum accuracy with deep reasoning.";
+/** Output model description for extract results. */
+export const outputExtractModelDescription = "Model tier used to extract the data.\n\n" +
+    "- nano: Fastest, best for simple documents.\n" +
+    "- mini: Balanced speed and accuracy.\n" +
+    "- pro: High accuracy for complex documents.\n" +
+    "- max: Maximum accuracy with deep reasoning.";
+/** Output model description for ask results. */
+export const outputAskModelDescription = "Model tier used to answer the question.\n\n" +
+    "- nano: Fastest, best for simple questions.\n" +
+    "- mini: Balanced speed and accuracy.\n" +
+    "- pro: High accuracy for complex questions.\n" +
+    "- max: Maximum accuracy with deep reasoning.";

package/CHANGELOG.md

@@ -1,5 +1,19 @@
 # @pdfvector/instance-contract
 
+## 0.0.29
+### Patch Changes
+
+
+
+- [#149](https://github.com/phuctm97/pdfvector/pull/149) [`bcffa28`](https://github.com/phuctm97/pdfvector/commit/bcffa2840c98af3e8ab0c971e61d580beedd3b85) Thanks [@khanhduyvt0101](https://github.com/khanhduyvt0101)! - Add academic paper graph and similar papers APIs with integration tests
+
+## 0.0.28
+### Patch Changes
+
+
+
+- [#148](https://github.com/phuctm97/pdfvector/pull/148) [`9a8d292`](https://github.com/phuctm97/pdfvector/commit/9a8d2920d022837dd09eb297949c2a40acd04b68) Thanks [@khanhduyvt0101](https://github.com/khanhduyvt0101)! - Add support for academic and additional file types
+
 ## 0.0.27
 ### Patch Changes
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@pdfvector/instance-contract",
-  "version": "0.0.27",
+  "version": "0.0.29",
   "type": "module",
   "description": "API contract definitions for PDFVector instance server",
   "license": "MIT",