From b66c72cc150428d743336a35dd0c8284919fcda0 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Tue, 8 Feb 2022 10:57:48 +0530
Subject: [PATCH] Auto clear dirtied_formats with a trigger

---
 resources/fts_sqlite.sql        |  7 +++++--
 src/calibre/db/fts/connect.py   | 15 +++++++++++++++
 src/calibre/db/tests/fts_api.py |  6 ++++++
 3 files changed, 26 insertions(+), 2 deletions(-)

diff --git a/resources/fts_sqlite.sql b/resources/fts_sqlite.sql
index 79b7b2daac..cfd503d654 100644
--- a/resources/fts_sqlite.sql
+++ b/resources/fts_sqlite.sql
@@ -6,12 +6,13 @@ CREATE TABLE fts_db.dirtied_formats ( id INTEGER PRIMARY KEY,
 
 CREATE TABLE fts_db.books_text ( id INTEGER PRIMARY KEY,
 	book INTEGER NOT NULL,
-	format TEXT NOT NULL COLLATE NOCASE,
 	timestamp REAL NOT NULL,
+	format TEXT NOT NULL COLLATE NOCASE,
     format_hash TEXT NOT NULL COLLATE NOCASE,
     format_size INTEGER NOT NULL,
-    text_hash TEXT NOT NULL COLLATE NOCASE,
     searchable_text TEXT NOT NULL DEFAULT "",
+    text_size INTEGER NOT NULL,
+    text_hash TEXT NOT NULL COLLATE NOCASE,
     UNIQUE(book, format)
 );
 
@@ -23,6 +24,7 @@ CREATE TRIGGER fts_db.books_fts_insert_trg AFTER INSERT ON fts_db.books_text
 BEGIN
     INSERT INTO books_fts(rowid, searchable_text) VALUES (NEW.id, NEW.searchable_text);
     INSERT INTO books_fts_stemmed(rowid, searchable_text) VALUES (NEW.id, NEW.searchable_text);
+    DELETE FROM dirtied_formats WHERE book=NEW.book AND format=NEW.format;
 END;
 
 CREATE TRIGGER fts_db.books_fts_delete_trg AFTER DELETE ON fts_db.books_text 
@@ -37,6 +39,7 @@ BEGIN
     INSERT INTO books_fts(rowid, searchable_text) VALUES (NEW.id, NEW.searchable_text);
     INSERT INTO books_fts_stemmed(books_fts_stemmed, rowid, searchable_text) VALUES('delete', OLD.id, OLD.searchable_text);
     INSERT INTO books_fts_stemmed(rowid, searchable_text) VALUES (NEW.id, NEW.searchable_text);
+    DELETE FROM dirtied_formats WHERE book=NEW.book AND format=NEW.format;
 END;
 
 PRAGMA fts_db.user_version=1;
diff --git a/src/calibre/db/fts/connect.py b/src/calibre/db/fts/connect.py
index 6c8271f226..0ca569c058 100644
--- a/src/calibre/db/fts/connect.py
+++ b/src/calibre/db/fts/connect.py
@@ -7,6 +7,8 @@ import builtins
 import os
 import sys
 
+from calibre.utils.date import EPOCH, utcnow
+
 from .schema_upgrade import SchemaUpgrade
 
 # TODO: db dump+restore
@@ -43,3 +45,16 @@ class FTS:
     def clear_all_dirty(self):
         conn = self.get_connection()
         conn.execute('DELETE FROM fts_db.dirtied_formats')
+
+    def add_text(self, book_id, fmt, text, text_hash='', fmt_size=0, fmt_hash=''):
+        conn = self.get_connection()
+        ts = (utcnow() - EPOCH).total_seconds()
+        fmt = fmt.upper()
+        if text:
+            conn.execute(
+                'INSERT OR REPLACE INTO fts_db.books_text '
+                '(book, timestamp, format, format_size, format_hash, searchable_text, text_size, text_hash) VALUES '
+                '(?, ?, ?, ?, ?, ?, ?, ?)', (
+                    book_id, ts, fmt, fmt_size, fmt_hash, text, len(text), text_hash))
+        else:
+            conn.execute('DELETE FROM fts_db.dirtied_formats WHERE book=? and format=?', (book_id, fmt))
diff --git a/src/calibre/db/tests/fts_api.py b/src/calibre/db/tests/fts_api.py
index d866399523..b18b42150f 100644
--- a/src/calibre/db/tests/fts_api.py
+++ b/src/calibre/db/tests/fts_api.py
@@ -43,6 +43,12 @@ class FTSAPITest(BaseTest):
         self.ae(fts.all_currently_dirty(), [])
         cache.add_format(2, 'ADDED', BytesIO(b'data2'))
         self.ae(fts.all_currently_dirty(), [(2, 'ADDED')])
+        fts.add_text(2, 'ADDED', 'data2')
+        self.ae(fts.all_currently_dirty(), [])
+        cache.add_format(2, 'ADDED', BytesIO(b'data2'))
+        self.ae(fts.all_currently_dirty(), [(2, 'ADDED')])
+        fts.add_text(2, 'ADDED', 'data2')
+        self.ae(fts.all_currently_dirty(), [])
 
 
 def find_tests():