books: schema, migration and ingest script (WIP)

src/lib/server/db/schema/books.ts

@@ -0,0 +1,162 @@
+import { sql } from 'drizzle-orm';
+import {
+	bigint,
+	check,
+	index,
+	integer,
+	jsonb,
+	primaryKey,
+	real,
+	text,
+	uniqueIndex
+} from 'drizzle-orm/pg-core';
+import { booksSchema } from './schemas';
+import { id, timestamps, timestamptz } from './helpers';
+
+// Book tracker + reader. A work is one book (any edition we care about); a work can
+// exist without files (placeholder for a book Stan wants but doesn't have yet).
+// Files live on stan-vm under $BOOKS_ROOT/library (target /srv/books/library); covers under $BOOKS_ROOT/covers.
+
+export const bookMetadataStatus = booksSchema.enum('metadata_status', [
+	'pending', // not looked up yet
+	'matched', // confident match (ISBN or exact title+author)
+	'low_confidence', // fuzzy match, needs a human look
+	'not_found',
+	'manual' // edited by hand; enrichment never overwrites it
+]);
+
+export const bookWorks = booksSchema.table(
+	'works',
+	{
+		id: id(),
+		title: text('title').notNull(),
+		subtitle: text('subtitle'),
+		authors: text('authors').array().notNull().default([]),
+		isbn10: text('isbn10'),
+		isbn13: text('isbn13'),
+		publisher: text('publisher'),
+		year: integer('year'),
+		language: text('language'), // ISO 639-1 where known, e.g. 'en', 'nl'
+		pageCount: integer('page_count'),
+		description: text('description'),
+		tags: text('tags').array().notNull().default([]),
+		coverUrl: text('cover_url'), // remote source of the cover
+		coverPath: text('cover_path'), // local file, relative to /srv/books/covers
+		// External ids
+		openlibraryWorkId: text('openlibrary_work_id'), // e.g. OL45883W
+		openlibraryEditionId: text('openlibrary_edition_id'), // e.g. OL7353617M
+		googleBooksId: text('google_books_id'),
+		annasMd5: text('annas_md5').array().notNull().default([]), // Anna's Archive / libgen file md5s
+		sourceRefs: jsonb('source_refs').$type<Record<string, unknown>>().notNull().default({}),
+		metadataStatus: bookMetadataStatus('metadata_status').notNull().default('pending'),
+		metadataSource: text('metadata_source'), // 'openlibrary' | 'googlebooks' | 'libgen' | 'manual'
+		metadataFetchedAt: timestamptz('metadata_fetched_at'),
+		notes: text('notes'),
+		...timestamps
+	},
+	(t) => [
+		uniqueIndex('works_isbn13_key')
+			.on(t.isbn13)
+			.where(sql`${t.isbn13} is not null`),
+		index('works_title_idx').on(sql`lower(${t.title})`),
+		check('works_year_check', sql`${t.year} is null or ${t.year} between 0 and 2100`)
+	]
+);
+
+export const bookFiles = booksSchema.table(
+	'files',
+	{
+		id: id(),
+		workId: integer('work_id')
+			.notNull()
+			.references(() => bookWorks.id, { onDelete: 'cascade' }),
+		path: text('path').notNull(), // relative to /srv/books/library
+		format: text('format').notNull(), // epub | pdf | mobi | azw3 | djvu
+		sizeBytes: bigint('size_bytes', { mode: 'number' }).notNull(),
+		sha256: text('sha256').notNull(),
+		md5: text('md5'),
+		source: text('source'), // 'gdrive' | 'textbooks-repo' | 'annas' | 'upload'
+		...timestamps
+	},
+	(t) => [
+		uniqueIndex('files_path_key').on(t.path),
+		uniqueIndex('files_sha256_key').on(t.sha256),
+		index('files_work_id_idx').on(t.workId)
+	]
+);
+
+export const bookShelves = booksSchema.table(
+	'shelves',
+	{
+		id: id(),
+		slug: text('slug').notNull(),
+		name: text('name').notNull(),
+		sortOrder: integer('sort_order').notNull().default(0),
+		...timestamps
+	},
+	(t) => [uniqueIndex('shelves_slug_key').on(t.slug)]
+);
+
+export const bookShelfWorks = booksSchema.table(
+	'shelf_works',
+	{
+		shelfId: integer('shelf_id')
+			.notNull()
+			.references(() => bookShelves.id, { onDelete: 'cascade' }),
+		workId: integer('work_id')
+			.notNull()
+			.references(() => bookWorks.id, { onDelete: 'cascade' }),
+		sortOrder: integer('sort_order').notNull().default(0),
+		addedAt: timestamptz('added_at').notNull().defaultNow()
+	},
+	(t) => [
+		primaryKey({ columns: [t.shelfId, t.workId] }),
+		index('shelf_works_work_id_idx').on(t.workId)
+	]
+);
+
+export const bookProgress = booksSchema.table(
+	'progress',
+	{
+		id: id(),
+		workId: integer('work_id')
+			.notNull()
+			.references(() => bookWorks.id, { onDelete: 'cascade' }),
+		fileId: integer('file_id').references(() => bookFiles.id, { onDelete: 'set null' }),
+		locator: text('locator'), // EPUB CFI or 'page:123' for PDFs
+		page: integer('page'),
+		percent: real('percent').notNull().default(0),
+		startedAt: timestamptz('started_at'),
+		finishedAt: timestamptz('finished_at'),
+		...timestamps
+	},
+	(t) => [
+		uniqueIndex('progress_work_id_key').on(t.workId),
+		check('progress_percent_check', sql`${t.percent} between 0 and 100`)
+	]
+);
+
+export const bookNotes = booksSchema.table(
+	'notes',
+	{
+		id: id(),
+		workId: integer('work_id')
+			.notNull()
+			.references(() => bookWorks.id, { onDelete: 'cascade' }),
+		fileId: integer('file_id').references(() => bookFiles.id, { onDelete: 'set null' }),
+		locator: text('locator'),
+		page: integer('page'),
+		highlight: text('highlight'), // the selected text, if any
+		note: text('note'),
+		color: text('color'),
+		...timestamps
+	},
+	(t) => [index('notes_work_id_idx').on(t.workId)]
+);
+
+export type BookWork = typeof bookWorks.$inferSelect;
+export type NewBookWork = typeof bookWorks.$inferInsert;
+export type BookFile = typeof bookFiles.$inferSelect;
+export type BookShelf = typeof bookShelves.$inferSelect;
+export type BookProgress = typeof bookProgress.$inferSelect;
+export type BookNote = typeof bookNotes.$inferSelect;

src/lib/server/db/schema/index.ts

@@ -12,3 +12,4 @@ export * from './schemas';
 export * from './money-tracker';
 export * from './cs-prices';
 export * from './house-tracker';
+export * from './books';

src/lib/server/db/schema/schemas.ts

@@ -12,3 +12,4 @@ export const csSchema = pgSchema('cs');
 export const animaldexSchema = pgSchema('animaldex');
 export const siteSchema = pgSchema('site');
 export const metaSchema = pgSchema('meta');
+export const booksSchema = pgSchema('books');