From c28160058103f6da4ed5a01d847b4f7fabb2de7f Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Tue, 27 Aug 2013 09:18:36 +0530
Subject: [PATCH] newdb: Optimize setting the same value as before for author
 data

Do not mark books as dirty if the same value as before is set for
author_sort or author_link.
---
 src/calibre/db/cache.py         | 14 ++++++++------
 src/calibre/db/tables.py        |  4 ++++
 src/calibre/db/tests/writing.py | 18 ++++++++++++++++++
 3 files changed, 30 insertions(+), 6 deletions(-)

diff --git a/src/calibre/db/cache.py b/src/calibre/db/cache.py
index 74455b046f..96c4f65e99 100644
--- a/src/calibre/db/cache.py
+++ b/src/calibre/db/cache.py
@@ -1531,11 +1531,11 @@ class Cache(object):
 
     @write_api
     def set_sort_for_authors(self, author_id_to_sort_map, update_books=True):
-        self.fields['authors'].table.set_sort_names(author_id_to_sort_map, self.backend)
+        sort_map = self.fields['authors'].table.set_sort_names(author_id_to_sort_map, self.backend)
         changed_books = set()
         if update_books:
             val_map = {}
-            for author_id in author_id_to_sort_map:
+            for author_id in sort_map:
                 books = self._books_for_field('authors', author_id)
                 changed_books |= books
                 for book_id in books:
@@ -1545,16 +1545,18 @@ class Cache(object):
                     val_map[book_id] = ' & '.join(sorts)
             if val_map:
                 self._set_field('author_sort', val_map)
-        self._mark_as_dirty(changed_books)
+        if changed_books:
+            self._mark_as_dirty(changed_books)
         return changed_books
 
     @write_api
     def set_link_for_authors(self, author_id_to_link_map):
-        self.fields['authors'].table.set_links(author_id_to_link_map, self.backend)
+        link_map = self.fields['authors'].table.set_links(author_id_to_link_map, self.backend)
         changed_books = set()
-        for author_id in author_id_to_link_map:
+        for author_id in link_map:
             changed_books |= self._books_for_field('authors', author_id)
-        self._mark_as_dirty(changed_books)
+        if changed_books:
+            self._mark_as_dirty(changed_books)
         return changed_books
 
     @read_api
diff --git a/src/calibre/db/tables.py b/src/calibre/db/tables.py
index 57ca395f4c..edb16d8e3a 100644
--- a/src/calibre/db/tables.py
+++ b/src/calibre/db/tables.py
@@ -366,15 +366,19 @@ class AuthorsTable(ManyToManyTable):
 
     def set_sort_names(self, aus_map, db):
         aus_map = {aid:(a or '').strip() for aid, a in aus_map.iteritems()}
+        aus_map = {aid:a for aid, a in aus_map.iteritems() if a != self.asort_map.get(aid, None)}
         self.asort_map.update(aus_map)
         db.conn.executemany('UPDATE authors SET sort=? WHERE id=?',
             [(v, k) for k, v in aus_map.iteritems()])
+        return aus_map
 
     def set_links(self, link_map, db):
         link_map = {aid:(l or '').strip() for aid, l in link_map.iteritems()}
+        link_map = {aid:l for aid, l in link_map.iteritems() if l != self.alink_map.get(aid, None)}
         self.alink_map.update(link_map)
         db.conn.executemany('UPDATE authors SET link=? WHERE id=?',
             [(v, k) for k, v in link_map.iteritems()])
+        return link_map
 
     def remove_books(self, book_ids, db):
         clean = ManyToManyTable.remove_books(self, book_ids, db)
diff --git a/src/calibre/db/tests/writing.py b/src/calibre/db/tests/writing.py
index 68f740743d..a199da6b28 100644
--- a/src/calibre/db/tests/writing.py
+++ b/src/calibre/db/tests/writing.py
@@ -607,3 +607,21 @@ class WritingTest(BaseTest):
         cache = self.init_cache()
         self.assertEqual(cache.all_book_ids(), all_ids, 'dump and restore broke database')
     # }}}
+
+    def test_set_author_data(self):  # {{{
+        cache = self.init_cache()
+        adata = cache.author_data()
+        ldata = {aid:str(aid) for aid in adata}
+        self.assertEqual({1,2,3}, cache.set_link_for_authors(ldata))
+        for c in (cache, self.init_cache()):
+            self.assertEqual(ldata, {aid:d['link'] for aid, d in c.author_data().iteritems()})
+        self.assertEqual({3}, cache.set_link_for_authors({aid:'xxx' if aid == max(adata) else str(aid) for aid in adata}),
+                         'Setting the author link to the same value as before, incorrectly marked some books as dirty')
+        sdata = {aid:'%s, changed' % aid for aid in adata}
+        self.assertEqual({1,2,3}, cache.set_sort_for_authors(sdata))
+        for bid in (1, 2, 3):
+            self.assertIn(', changed', cache.field_for('author_sort', bid))
+        sdata = {aid:'%s, changed' % (aid*2 if aid == max(adata) else aid) for aid in adata}
+        self.assertEqual({3}, cache.set_sort_for_authors(sdata),
+                         'Setting the author sort to the same value as before, incorrectly marked some books as dirty')
+    # }}}