v2 file processing db schema

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 } from '@shared/data/preference/preferenceTypes'
+import type { FileProcessorMerged, FileProcessorOverride } 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: 2026-01-23T10:31:39.487Z
+ * Generated at: 2026-03-09T06:17:20.448Z
  *
  * This file is automatically generated from classification.json
  * To update this file, modify classification.json and run:
@@ -382,6 +382,12 @@ 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
@@ -650,6 +656,9 @@ 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,
@@ -718,8 +727,8 @@ export const DefaultPreferences: PreferenceSchemas = {
 
 /**
  * 生成统计:
- * - 总配置项: 204
+ * - 总配置项: 207
  * - electronStore项: 1
- * - redux项: 203
+ * - redux项: 205
  * - localStorage项: 0
  */

packages/shared/data/preference/preferenceTypes.ts

@@ -1,3 +1,5 @@
+import type { FileProcessorOverride, PRESETS_FILE_PROCESSORS } from '@shared/data/presets/file-processing'
+
 import type { PreferenceSchemas } from './preferenceSchemas'
 
 export type PreferenceDefaultScopeType = PreferenceSchemas['default']
@@ -104,3 +106,7 @@ export type ChatMessageNavigationMode = 'none' | 'buttons' | 'anchor'
 export type MultiModelMessageStyle = 'horizontal' | 'vertical' | 'fold' | 'grid'
 
 export type MultiModelGridPopoverTrigger = 'hover' | 'click'
+
+export type FileProcessorOverrides = Record<string, FileProcessorOverride>
+
+export type FileProcessorId = (typeof PRESETS_FILE_PROCESSORS)[number]['id']

packages/shared/data/presets/file-processing.ts

@@ -0,0 +1,273 @@
+/**
+ * File Processing Presets
+ *
+ * Templates are read-only metadata about processors.
+ * User overrides are stored separately in preferences.
+ *
+ * i18n: Display names use `processor.${id}.name`
+ */
+
+// ============================================================================
+// Type Definitions
+// ============================================================================
+
+/**
+ * Processor service type
+ */
+export type FileProcessorType = 'api' | 'builtin'
+
+/**
+ * Feature type
+ */
+export type FileProcessorFeature = 'text_extraction' | 'markdown_conversion'
+
+/**
+ * Input type (category)
+ * Can be migrated to FileTypes enum in @types later
+ */
+export type FileProcessorInput = 'image' | 'document'
+
+/**
+ * Output type
+ */
+export type FileProcessorOutput = 'text' | 'markdown'
+
+/**
+ * Processor metadata (reserved for future use)
+ */
+export type FileProcessorMetadata = Record<string, never>
+
+/**
+ * Feature capability definition
+ *
+ * Each capability binds a feature with its input/output and optional API settings.
+ */
+export type CapabilityMetadata = Record<string, unknown>
+
+export type TextExtractionCapability = {
+  feature: 'text_extraction'
+  input: 'image' | 'document'
+  output: 'text'
+  apiHost?: string // API Host (template default, can be overridden)
+  modelId?: string // Model ID (template default, can be overridden)
+  metadata?: CapabilityMetadata
+  // supportedFormats?: string[] // Whitelist: only these formats supported (uncomment when needed)
+  // excludedFormats?: string[] // Blacklist: all formats except these (uncomment when needed)
+}
+
+export type MarkdownConversionCapability = {
+  feature: 'markdown_conversion'
+  input: 'document'
+  output: 'markdown'
+  apiHost?: string // API Host (template default, can be overridden)
+  modelId?: string // Model ID (template default, can be overridden)
+  metadata?: CapabilityMetadata
+  // supportedFormats?: string[] // Whitelist: only these formats supported (uncomment when needed)
+  // excludedFormats?: string[] // Blacklist: all formats except these (uncomment when needed)
+}
+
+export type FeatureCapability = TextExtractionCapability | MarkdownConversionCapability
+
+/**
+ * Processor template (read-only metadata)
+ *
+ * Note: Display name is retrieved via i18n key `processor.${id}.name`
+ */
+export type FileProcessorTemplate = {
+  id: string // Unique identifier, also used for i18n key
+  type: FileProcessorType // 'api' | 'builtin'
+  metadata?: FileProcessorMetadata // Optional processor metadata
+  capabilities: FeatureCapability[] // Feature capabilities
+}
+
+// ============================================================================
+// Override Types (for user customization)
+// ============================================================================
+
+/**
+ * Processor-specific configuration
+ *
+ * Uses a generic Record type without predefined structure.
+ * Each processor's configuration is interpreted by UI components based on processor.id.
+ *
+ * Known options fields:
+ * - Tesseract: { langs: string[] }  // Array of enabled language codes
+ *
+ * Examples:
+ * - { langs: ['chi_sim', 'eng'] }        // Tesseract language config
+ * - { quality: 'high', timeout: 30000 }  // Other processor config
+ */
+export type FileProcessorOptions = Record<string, unknown>
+
+/**
+ * Capability override (user customization for a specific feature)
+ *
+ * Stored as Record<feature, CapabilityOverride> in FileProcessorOverride.
+ */
+export type CapabilityOverride = {
+  apiHost?: string
+  modelId?: string
+  metadata?: CapabilityMetadata
+}
+
+/**
+ * User-configured processor override (stored in Preference)
+ *
+ * Design principles:
+ * - Only stores user-modified fields
+ * - apiKey is shared across all features (processor-level)
+ * - apiHost/modelId are per-feature (in capabilities Record)
+ * - Field names use camelCase (consistent with TypeScript conventions)
+ */
+export type FileProcessorOverride = {
+  apiKeys?: string[] // API Keys (shared across all features)
+  capabilities?: Partial<Record<FileProcessorFeature, CapabilityOverride>> // Per-feature overrides
+  options?: FileProcessorOptions // Processor-specific config (generic type)
+}
+
+/**
+ * Merged processor configuration (template + user override)
+ *
+ * Used by both Renderer (UI display/editing) and Main (execution).
+ * Combines the read-only template with user-configured overrides.
+ *
+ * Note: capabilities is an array (from template) with overrides merged in,
+ * NOT a Record like in FileProcessorOverride.
+ */
+export type FileProcessorMerged = {
+  id: string
+  type: FileProcessorType
+  metadata?: FileProcessorMetadata
+  capabilities: FeatureCapability[] // Merged capabilities (array)
+  apiKeys?: string[]
+  options?: FileProcessorOptions
+}
+
+// ============================================================================
+// Processor Presets
+// ============================================================================
+
+/**
+ * Built-in processor presets
+ */
+export const PRESETS_FILE_PROCESSORS = [
+  // === Image Processors (former OCR) ===
+  {
+    id: 'tesseract',
+    type: 'builtin',
+    capabilities: [
+      {
+        feature: 'text_extraction',
+        input: 'image',
+        output: 'text'
+      }
+    ]
+  },
+  {
+    id: 'system',
+    type: 'builtin',
+    capabilities: [{ feature: 'text_extraction', input: 'image', output: 'text' }]
+  },
+  {
+    id: 'paddleocr',
+    type: 'api',
+    capabilities: [
+      {
+        feature: 'text_extraction',
+        input: 'image',
+        output: 'text',
+        apiHost: 'https://paddleocr.aistudio-app.com/',
+        modelId: 'PP-OCRv5',
+        metadata: {
+          optionalPayload: {
+            useDocOrientationClassify: false,
+            useDocUnwarping: false
+          }
+        }
+      },
+      {
+        feature: 'markdown_conversion',
+        input: 'document',
+        output: 'markdown',
+        apiHost: 'https://paddleocr.aistudio-app.com/',
+        modelId: 'PaddleOCR-VL-1.5',
+        metadata: {
+          optionalPayload: {
+            useDocOrientationClassify: false,
+            useDocUnwarping: false
+          }
+        }
+      }
+    ]
+  },
+  {
+    id: 'ovocr',
+    type: 'builtin',
+    capabilities: [{ feature: 'text_extraction', input: 'image', output: 'text' }]
+  },
+
+  // === Document Processors (former Preprocess) ===
+  {
+    id: 'mineru',
+    type: 'api',
+    capabilities: [
+      {
+        feature: 'markdown_conversion',
+        input: 'document',
+        output: 'markdown',
+        apiHost: 'https://mineru.net',
+        metadata: {
+          optionalPayload: {
+            enable_formula: true,
+            enable_table: true,
+            is_ocr: true
+          }
+        }
+      }
+    ]
+  },
+  {
+    id: 'doc2x',
+    type: 'api',
+    capabilities: [
+      {
+        feature: 'markdown_conversion',
+        input: 'document',
+        output: 'markdown',
+        apiHost: 'https://v2.doc2x.noedgeai.com'
+      }
+    ]
+  },
+  {
+    id: 'mistral',
+    type: 'api',
+    capabilities: [
+      {
+        feature: 'text_extraction',
+        input: 'image',
+        output: 'text',
+        apiHost: 'https://api.mistral.ai',
+        modelId: 'mistral-ocr-latest'
+      },
+      {
+        feature: 'markdown_conversion',
+        input: 'document',
+        output: 'markdown',
+        apiHost: 'https://api.mistral.ai',
+        modelId: 'mistral-ocr-latest'
+      }
+    ]
+  },
+  {
+    id: 'open-mineru',
+    type: 'api',
+    capabilities: [
+      {
+        feature: 'markdown_conversion',
+        input: 'document',
+        output: 'markdown',
+        apiHost: 'http://127.0.0.1:8000'
+      }
+    ]
+  }
+] as const satisfies readonly FileProcessorTemplate[]