Add support for FTS index to export/import

src/calibre/db/backend.py

@@ -16,7 +16,7 @@ import shutil
 import sys
 import time
 import uuid
-from contextlib import suppress
+from contextlib import suppress, closing
 from functools import partial
 
 from calibre import as_unicode, force_unicode, isbytestring, prints
@@ -1255,7 +1255,6 @@ class DB:
     def dump_and_restore(self, callback=None, sql=None):
         import codecs
         from apsw import Shell
-        from contextlib import closing
         if callback is None:
             callback = lambda x: x
         uv = int(self.user_version)
@@ -2184,14 +2183,18 @@ class DB:
         self.executemany('INSERT INTO data (book,format,uncompressed_size,name) VALUES (?,?,?,?)', vals)
 
     def backup_database(self, path):
-        dest_db = apsw.Connection(path)
+        with closing(apsw.Connection(path)) as dest_db:
             with dest_db.backup('main', self.conn, 'main') as b:
                 while not b.done:
-                try:
-                    b.step(100)
-                except apsw.BusyError:
-                    pass
+                    with suppress(apsw.BusyError):
+                        b.step(128)
             dest_db.cursor().execute('DELETE FROM metadata_dirtied; VACUUM;')
-        dest_db.close()
 
+    def backup_fts_database(self, path):
+        with closing(apsw.Connection(path)) as dest_db:
+            with dest_db.backup('main', self.conn, 'fts_db') as b:
+                while not b.done:
+                    with suppress(apsw.BusyError):
+                        b.step(128)
+            dest_db.cursor().execute('VACUUM;')
     # }}}

src/calibre/db/cache.py

@@ -473,6 +473,8 @@ class Cache:
             self.fts_job_queue.put(None)
             self.fts_queue_thread = None
             self.fts_job_queue = Queue()
+        if fts:
+            self._update_fts_indexing_numbers()
         return fts
 
     @write_api
@@ -497,6 +499,7 @@ class Cache:
             if not path or not is_fmt_ok(fmt):
                 with self.write_lock:
                     self.backend.remove_dirty_fts(book_id, fmt)
+                    self._update_fts_indexing_numbers()
                 return True
 
             with self.read_lock, open(path, 'rb') as src, PersistentTemporaryFile(suffix=f'.{fmt.lower()}') as pt:
@@ -2617,7 +2620,9 @@ class Cache:
         key_prefix = as_hex_unicode(library_key)
         book_ids = self._all_book_ids()
         total = len(book_ids) + 1
-        format_metadata = {}
+        has_fts = self.is_fts_enabled()
+        if has_fts:
+            total += 1
         if progress is not None:
             progress('metadata.db', 0, total)
         pt = PersistentTemporaryFile('-export.db')
@@ -2627,12 +2632,28 @@ class Cache:
         with lopen(pt.name, 'rb') as f:
             exporter.add_file(f, dbkey)
         os.remove(pt.name)
+        poff = 1
+        if has_fts:
+            poff += 1
+            if progress is not None:
+                progress('full-text-search.db', 1, total)
+            pt = PersistentTemporaryFile('-export.db')
+            pt.close()
+            self.backend.backup_fts_database(pt.name)
+            ftsdbkey = key_prefix + ':::' + 'full-text-search.db'
+            with lopen(pt.name, 'rb') as f:
+                exporter.add_file(f, ftsdbkey)
+            os.remove(pt.name)
+
+        format_metadata = {}
         metadata = {'format_data':format_metadata, 'metadata.db':dbkey, 'total':total}
+        if has_fts:
+            metadata['full-text-search.db'] = ftsdbkey
         for i, book_id in enumerate(book_ids):
             if abort is not None and abort.is_set():
                 return
             if progress is not None:
-                progress(self._field_for('title', book_id), i + 1, total)
+                progress(self._field_for('title', book_id), i + poff, total)
             format_metadata[book_id] = {}
             for fmt in self._formats(book_id):
                 mdata = self.format_metadata(book_id, fmt)
@@ -2743,6 +2764,7 @@ def import_library(library_key, importer, library_path, progress=None, abort=Non
     from calibre.db.backend import DB
     metadata = importer.metadata[library_key]
     total = metadata['total']
+    poff = 1
     if progress is not None:
         progress('metadata.db', 0, total)
     if abort is not None and abort.is_set():
@@ -2751,6 +2773,16 @@ def import_library(library_key, importer, library_path, progress=None, abort=Non
         src = importer.start_file(metadata['metadata.db'], 'metadata.db for ' + library_path)
         shutil.copyfileobj(src, f)
         src.close()
+    if 'full-text-search.db' in metadata:
+        if progress is not None:
+            progress('full-text-search.db', 1, total)
+        if abort is not None and abort.is_set():
+            return
+        poff += 1
+        with open(os.path.join(library_path, 'full-text-search.db'), 'wb') as f:
+            src = importer.start_file(metadata['full-text-search.db'], 'full-text-search.db for ' + library_path)
+            shutil.copyfileobj(src, f)
+            src.close()
     cache = Cache(DB(library_path, load_user_formatter_functions=False))
     cache.init()
     format_data = {int(book_id):data for book_id, data in iteritems(metadata['format_data'])}
@@ -2759,7 +2791,7 @@ def import_library(library_key, importer, library_path, progress=None, abort=Non
             return
         title = cache._field_for('title', book_id)
         if progress is not None:
-            progress(title, i + 1, total)
+            progress(title, i + poff, total)
         cache._update_path((book_id,), mark_as_dirtied=False)
         for fmt, fmtkey in iteritems(fmt_key_map):
             if fmt == '.cover':

src/calibre/db/fts/connect.py

@@ -19,8 +19,6 @@ from calibre.utils.date import EPOCH, utcnow
 from .pool import Pool
 from .schema_upgrade import SchemaUpgrade
 
-# TODO: calibre export/import should preserve indexed data
-
 
 def print(*args, **kwargs):
     kwargs['file'] = sys.__stdout__

src/calibre/db/tests/filesystem.py

@@ -5,7 +5,7 @@ __license__   = 'GPL v3'
 __copyright__ = '2013, Kovid Goyal <kovid at kovidgoyal.net>'
 __docformat__ = 'restructuredtext en'
 
-import unittest, os
+import unittest, os, time
 from io import BytesIO
 
 from calibre.constants import iswindows
@@ -166,6 +166,23 @@ class FilesystemTest(BaseTest):
                     for fmt in cache.formats(book_id):
                         self.assertEqual(cache.format(book_id, fmt), ic.format(book_id, fmt))
                         self.assertEqual(cache.format_metadata(book_id, fmt)['mtime'], cache.format_metadata(book_id, fmt)['mtime'])
+        cache.add_format(1, 'TXT', BytesIO(b'testing exim'))
+        cache.fts_indexing_sleep_time = 0.001
+        cache.enable_fts()
+        cache.set_fts_num_of_workers(4)
+        st = time.monotonic()
+        while cache.fts_indexing_left > 0 and time.monotonic() - st < 15:
+            time.sleep(0.05)
+        if cache.fts_indexing_left > 0:
+            raise ValueError('FTS indexing did not complete')
+        self.assertEqual(cache.fts_search('exim')[0]['id'], 1)
+        with TemporaryDirectory('export_lib') as tdir, TemporaryDirectory('import_lib') as idir:
+            exporter = Exporter(tdir)
+            cache.export_library('l', exporter)
+            exporter.commit()
+            importer = Importer(tdir)
+            ic = import_library('l', importer, idir)
+            self.assertEqual(ic.fts_search('exim')[0]['id'], 1)
 
     def test_find_books_in_directory(self):
         from calibre.db.adding import find_books_in_directory, compile_rule