v2 file processing db schema

packages/shared/__tests__/file-processing.test.ts

@@ -0,0 +1,92 @@
+import { describe, expect, it } from 'vitest'
+
+import {
+  FeatureCapabilitySchema,
+  FileProcessorOverrideSchema,
+  FileProcessorTemplateSchema,
+  FileProcessorTemplatesSchema,
+  PRESETS_FILE_PROCESSORS
+} from '../data/presets/file-processing'
+import { FILE_TYPE } from '../data/types/file'
+
+describe('FeatureCapabilitySchema', () => {
+  it('supports multiple input types for a single capability', () => {
+    const result = FeatureCapabilitySchema.safeParse({
+      feature: 'text_extraction',
+      inputs: [FILE_TYPE.IMAGE, FILE_TYPE.DOCUMENT],
+      output: FILE_TYPE.TEXT
+    })
+
+    expect(result.success).toBe(true)
+  })
+})
+
+describe('FileProcessorTemplatesSchema', () => {
+  it('validates built-in presets', () => {
+    expect(() => FileProcessorTemplatesSchema.parse(PRESETS_FILE_PROCESSORS)).not.toThrow()
+  })
+
+  it('rejects duplicate features in a single processor template', () => {
+    const result = FileProcessorTemplateSchema.safeParse({
+      id: 'paddleocr',
+      type: 'api',
+      capabilities: [
+        {
+          feature: 'text_extraction',
+          inputs: [FILE_TYPE.IMAGE],
+          output: FILE_TYPE.TEXT
+        },
+        {
+          feature: 'text_extraction',
+          inputs: [FILE_TYPE.DOCUMENT],
+          output: FILE_TYPE.TEXT
+        }
+      ]
+    })
+
+    expect(result.success).toBe(false)
+  })
+})
+
+describe('FileProcessorOverrideSchema', () => {
+  it('accepts valid overrides', () => {
+    const result = FileProcessorOverrideSchema.safeParse({
+      apiKeys: ['test-key'],
+      capabilities: {
+        text_extraction: {
+          apiHost: 'https://example.com',
+          modelId: 'model-1'
+        }
+      },
+      options: {
+        langs: ['eng', 'chi_sim']
+      }
+    })
+
+    expect(result.success).toBe(true)
+  })
+
+  it('rejects invalid urls', () => {
+    const result = FileProcessorOverrideSchema.safeParse({
+      capabilities: {
+        markdown_conversion: {
+          apiHost: 'not-a-url'
+        }
+      }
+    })
+
+    expect(result.success).toBe(false)
+  })
+
+  it('rejects unknown feature overrides', () => {
+    const result = FileProcessorOverrideSchema.safeParse({
+      capabilities: {
+        vision: {
+          apiHost: 'https://example.com'
+        }
+      }
+    })
+
+    expect(result.success).toBe(false)
+  })
+})

packages/shared/__tests__/file-type.test.ts

@@ -0,0 +1,15 @@
+import { describe, expect, it } from 'vitest'
+
+import { FILE_TYPE, FileTypeSchema } from '../data/types/file'
+
+describe('FileTypeSchema', () => {
+  it('accepts canonical file types', () => {
+    expect(FileTypeSchema.safeParse(FILE_TYPE.IMAGE).success).toBe(true)
+    expect(FileTypeSchema.safeParse(FILE_TYPE.DOCUMENT).success).toBe(true)
+    expect(FileTypeSchema.safeParse(FILE_TYPE.TEXT).success).toBe(true)
+  })
+
+  it('rejects unknown file types', () => {
+    expect(FileTypeSchema.safeParse('markdown').success).toBe(false)
+  })
+})

packages/shared/data/api/schemas/fileProcessing.ts

@@ -0,0 +1,49 @@
+/**
+ * File Processing API Schema definitions
+ *
+ * Contains file processing endpoints for:
+ * - Listing available processors
+ * - Reading and updating processor configuration
+ */
+
+import type { FileProcessorId, FileProcessorOverride } from '@shared/data/preference/preferenceTypes'
+import type { FileProcessorMerged } from '@shared/data/presets/file-processing'
+
+// ============================================================================
+// API Schema Definitions
+// ============================================================================
+
+/**
+ * File Processing API Schema definitions
+ */
+export interface FileProcessingSchemas {
+  /**
+   * List available processors
+   * @example GET /file-processing/processors
+   */
+  '/file-processing/processors': {
+    /** Get list of available processors */
+    GET: {
+      response: FileProcessorMerged[]
+    }
+  }
+
+  /**
+   * Get or update processor configuration
+   * @example GET /file-processing/processors/tesseract
+   * @example PATCH /file-processing/processors/tesseract { "apiKeys": ["xxx"] }
+   */
+  '/file-processing/processors/:id': {
+    /** Get processor configuration */
+    GET: {
+      params: { id: FileProcessorId }
+      response: FileProcessorMerged
+    }
+    /** Update processor configuration */
+    PATCH: {
+      params: { id: FileProcessorId }
+      body: FileProcessorOverride
+      response: FileProcessorMerged
+    }
+  }
+}

packages/shared/data/api/schemas/index.ts

@@ -20,6 +20,7 @@
  */
 
 import type { AssertValidSchemas } from '../apiTypes'
+import type { FileProcessingSchemas } from './fileProcessing'
 import type { MessageSchemas } from './messages'
 import type { TestSchemas } from './test'
 import type { TopicSchemas } from './topics'
@@ -36,4 +37,4 @@ import type { TopicSchemas } from './topics'
  * 1. Create the schema file (e.g., topic.ts)
  * 2. Import and add to intersection below
  */
-export type ApiSchemas = AssertValidSchemas<TestSchemas & TopicSchemas & MessageSchemas>
+export type ApiSchemas = AssertValidSchemas<TestSchemas & TopicSchemas & MessageSchemas & FileProcessingSchemas>

packages/shared/data/preference/preferenceSchemas.ts

@@ -1,6 +1,6 @@
 /**
  * Auto-generated preferences configuration
- * Generated at: 2026-03-08T07:54:09.830Z
+ * Generated at: 2026-03-12T12:32:08.496Z
  *
  * This file is automatically generated from classification.json
  * To update this file, modify classification.json and run:
@@ -404,6 +404,12 @@ export interface PreferenceSchemas {
     'feature.translate.model_prompt': string
     // redux/settings/targetLanguage
     'feature.translate.target_language': string
+    // redux/preprocess/defaultProvider
+    'file_processing.default.markdown_conversion': PreferenceTypes.FileProcessorId
+    // redux/ocr/imageProviderId
+    'file_processing.default.text_extraction': PreferenceTypes.FileProcessorId | null
+    // target-key-definitions/complex/complex
+    'file_processing.overrides': PreferenceTypes.FileProcessorOverrides
     // redux/shortcuts/shortcuts.exit_fullscreen
     'shortcut.app.exit_fullscreen': Record<string, unknown>
     // redux/shortcuts/shortcuts.search_message
@@ -683,6 +689,9 @@ export const DefaultPreferences: PreferenceSchemas = {
     'feature.selection.trigger_mode': PreferenceTypes.SelectionTriggerMode.Selected,
     'feature.translate.model_prompt': TRANSLATE_PROMPT,
     'feature.translate.target_language': 'en-us',
+    'file_processing.default.markdown_conversion': 'mineru',
+    'file_processing.default.text_extraction': null,
+    'file_processing.overrides': {} as PreferenceTypes.FileProcessorOverrides,
     'shortcut.app.exit_fullscreen': { editable: false, enabled: true, key: ['Escape'], system: true },
     'shortcut.app.search_message': {
       editable: true,
@@ -751,8 +760,8 @@ export const DefaultPreferences: PreferenceSchemas = {
 
 /**
  * 生成统计:
- * - 总配置项: 215
+ * - 总配置项: 218
  * - electronStore项: 1
- * - redux项: 203
+ * - redux项: 205
  * - localStorage项: 0
  */

packages/shared/data/preference/preferenceTypes.ts

@@ -105,6 +105,44 @@ export type MultiModelMessageStyle = 'horizontal' | 'vertical' | 'fold' | 'grid'
 
 export type MultiModelGridPopoverTrigger = 'hover' | 'click'
 
+export const FILE_PROCESSOR_TYPES = ['api', 'builtin'] as const
+
+export type FileProcessorType = (typeof FILE_PROCESSOR_TYPES)[number]
+
+export const FILE_PROCESSOR_FEATURES = ['text_extraction', 'markdown_conversion'] as const
+
+export type FileProcessorFeature = (typeof FILE_PROCESSOR_FEATURES)[number]
+
+export const FILE_PROCESSOR_IDS = [
+  'tesseract',
+  'system',
+  'paddleocr',
+  'ovocr',
+  'mineru',
+  'doc2x',
+  'mistral',
+  'open-mineru'
+] as const
+
+export type FileProcessorId = (typeof FILE_PROCESSOR_IDS)[number]
+
+export type FileProcessorOptions = Record<string, unknown>
+
+export type CapabilityOverride = {
+  apiHost?: string
+  modelId?: string
+  metadata?: Record<string, unknown>
+}
+
+export type FileProcessorCapabilityOverrides = Partial<Record<FileProcessorFeature, CapabilityOverride>>
+
+export type FileProcessorOverride = {
+  apiKeys?: string[]
+  capabilities?: FileProcessorCapabilityOverrides
+  options?: FileProcessorOptions
+}
+
+export type FileProcessorOverrides = Partial<Record<FileProcessorId, FileProcessorOverride>>
 // ============================================================================
 // WebSearch Types
 // ============================================================================