From 399f8cbd41ba3f8a9dc0f6511675c31db050cdb8 Mon Sep 17 00:00:00 2001
From: icarus <eurfelux@gmail.com>
Date: Mon, 20 Oct 2025 00:27:52 +0800
Subject: [PATCH] feat(db): add ocr provider schema with capabilities and
 config

Add new schema for OCR providers including fields for id, name, capabilities, and config. Capabilities and config are stored as JSON to accommodate various provider types and configurations.
---
 src/main/data/db/schemas/ocr/provider.ts | 46 ++++++++++++++++++++++++
 1 file changed, 46 insertions(+)
 create mode 100644 src/main/data/db/schemas/ocr/provider.ts

diff --git a/src/main/data/db/schemas/ocr/provider.ts b/src/main/data/db/schemas/ocr/provider.ts
new file mode 100644
index 0000000000..8a181db40d
--- /dev/null
+++ b/src/main/data/db/schemas/ocr/provider.ts
@@ -0,0 +1,46 @@
+import type { OcrProviderCapabilityRecord, OcrProviderConfig } from '@types'
+import { index, sqliteTable, text } from 'drizzle-orm/sqlite-core'
+
+import { createUpdateTimestamps } from '../columnHelpers'
+
+export const ocrProviderTable = sqliteTable(
+  'ocr_provider',
+  {
+    /**
+     * Unique identifier for the provider.
+     * For built-in providers, it's 'tesseract', 'system', etc.
+     * For custom providers, it can be any unique string.
+     * As the primary key, it ensures the uniqueness of each provider.
+     */
+    id: text('id').primaryKey(),
+
+    /**
+     * Display name of the provider, e.g., "Tesseract OCR".
+     * Cannot be null.
+     */
+    name: text('name').notNull(),
+
+    /**
+     * Object describing the provider's capabilities, e.g., { image: true }.
+     * Stored as JSON in a text column. Drizzle's `mode: 'json'` handles
+     * serialization and deserialization automatically. `$type` provides strong typing.
+     * Cannot be null; should store an empty object `{}` even if no specific capabilities.
+     */
+    capabilities: text('capabilities', { mode: 'json' }).$type<OcrProviderCapabilityRecord>().notNull(),
+
+    /**
+     * Provider-specific configuration. This is a polymorphic field, its structure varies by provider type.
+     * For example, Tesseract's configuration is entirely different from PaddleOCR's.
+     * Storing it as JSON is the most flexible approach to accommodate any configuration structure.
+     * This field is nullable because `config` in the `OcrProvider` type is optional.
+     */
+    config: text('config', { mode: 'json' }).$type<OcrProviderConfig>(),
+
+    /** Timestamps. */
+    ...createUpdateTimestamps
+  },
+  (t) => [index('name').on(t.name)]
+)
+
+export type OcrProviderInsert = typeof ocrProviderTable.$inferInsert
+export type OcrProviderSelect = typeof ocrProviderTable.$inferSelect