Move updating covers in children to the end of publishing process,

update checkcover report

diff --git a/apps/catalogue/__init__.py b/apps/catalogue/__init__.py
index 45acadc..1fbdc20 100644 (file)
--- a/apps/catalogue/__init__.py
+++ b/apps/catalogue/__init__.py
@@ -19,6 +19,9 @@ class Settings(AppSettings):
             'fb2': 'wolnelektury_pl_fb2',
         }
 
+    REDAKCJA_URL = "http://redakcja.wolnelektury.pl"
+    GOOD_LICENSES = set([r'CC BY \d\.\d', r'CC BY-SA \d\.\d'])
+
     def _more_DONT_BUILD(self, value):
         for format_ in ['cover', 'pdf', 'epub', 'mobi', 'fb2', 'txt']:
             attname = 'NO_BUILD_%s' % format_.upper()

diff --git a/apps/catalogue/management/commands/checkcovers.py b/apps/catalogue/management/commands/checkcovers.py
index ef61702..1af39e1 100644 (file)
--- a/apps/catalogue/management/commands/checkcovers.py
+++ b/apps/catalogue/management/commands/checkcovers.py
@@ -5,6 +5,7 @@
 from optparse import make_option
 from django.contrib.sites.models import Site
 from django.core.management.base import BaseCommand
+from catalogue import app_settings
 
 
 def ancestor_has_cover(book):
@@ -39,14 +40,16 @@ class Command(BaseCommand):
 
         without_cover = []
         with_ancestral_cover = []
-        by_flickr_author = defaultdict(list)
-        not_flickr = []
-        by_license = defaultdict(list)
+        not_redakcja = []
+        bad_license = defaultdict(list)
         no_license = []
 
-        re_flickr = re.compile(ur'https?://(?:www\.|secure\.)?flickr.com/photos/([^/]*)/.*')
         re_license = re.compile(ur'.*,\s*(CC.*)')
 
+        redakcja_url = app_settings.REDAKCJA_URL
+        good_license = re.compile("(%s)" % ")|(".join(
+                            app_settings.GOOD_LICENSES))
+
         with transaction.commit_on_success():
             for book in Book.objects.all().order_by('slug').iterator():
                 extra_info = book.extra_info
@@ -56,76 +59,71 @@ class Command(BaseCommand):
                     else:
                         without_cover.append(book)
                 else:
-                    match = re_flickr.match(extra_info.get('cover_source', ''))
-                    if match:
-                        by_flickr_author[match.group(1)].append(book)
-                    else:
-                        not_flickr.append(book)
+                    if not extra_info.get('cover_source', ''
+                                ).startswith(redakcja_url):
+                        not_redakcja.append(book)
                     match = re_license.match(extra_info.get('cover_by', ''))
                     if match:
-                        by_license[match.group(1)].append(book)
+                        if not good_license.match(match.group(1)):
+                            bad_license[match.group(1)].append(book)
                     else:
                         no_license.append(book)
 
-        print """%d books with no covers, %d with ancestral covers.
-Licenses used: %s (%d covers without license).
-Flickr authors: %s (%d covers not from flickr).
+        print """%d books with no covers, %d with inherited covers.
+Bad licenses used: %s (%d covers without license).
+%d covers not from %s.
 """ % (
             len(without_cover),
             len(with_ancestral_cover),
-            ", ".join(sorted(by_license.keys())),
+            ", ".join(sorted(bad_license.keys())) or "none",
             len(no_license),
-            ", ".join(sorted(by_flickr_author.keys())),
-            len(not_flickr),
+            len(not_redakcja),
+            redakcja_url,
             )
 
         if verbose:
-            print
-            print "By license:"
-            print "==========="
-            for lic, books in by_license.items():
+            if bad_license:
+                print
+                print "Bad license:"
+                print "============"
+                for lic, books in bad_license.items():
+                    print
+                    print lic
+                    for book in books:
+                        print full_url(book)
+
+            if no_license:
                 print
-                print lic
-                for book in books:
+                print "No license:"
+                print "==========="
+                for book in no_license:
+                    print
                     print full_url(book)
+                    print book.extra_info.get('cover_by')
+                    print book.extra_info.get('cover_source')
+                    print book.extra_info.get('cover_url')
 
-            print
-            print "No license:"
-            print "==========="
-            for book in no_license:
+            if not_redakcja:
                 print
-                print full_url(book)
-                print book.extra_info.get('cover_by')
-                print book.extra_info.get('cover_source')
-                print book.extra_info.get('cover_url')
-
-            print
-            print "By Flickr author:"
-            print "================="
-            for author, books in by_flickr_author.items():
+                print "Not from Redakcja or source missing:"
+                print "===================================="
+                for book in not_redakcja:
+                    print
+                    print full_url(book)
+                    print book.extra_info.get('cover_by')
+                    print book.extra_info.get('cover_source')
+                    print book.extra_info.get('cover_url')
+
+            if without_cover:
                 print
-                print "author: http://flickr.com/photos/%s/" % author
-                for book in books:
+                print "No cover:"
+                print "========="
+                for book in without_cover:
                     print full_url(book)
 
-            print
-            print "Not from Flickr or source missing:"
-            print "=================================="
-            for book in not_flickr:
+            if with_ancestral_cover:
                 print
-                print full_url(book)
-                print book.extra_info.get('cover_by')
-                print book.extra_info.get('cover_source')
-                print book.extra_info.get('cover_url')
-
-            print
-            print "No cover:"
-            print "========="
-            for book in without_cover:
-                print full_url(book)
-
-            print
-            print "With ancestral cover:"
-            print "====================="
-            for book in with_ancestral_cover:
-                print full_url(book)
+                print "With ancestral cover:"
+                print "====================="
+                for book in with_ancestral_cover:
+                    print full_url(book)

diff --git a/apps/catalogue/models/book.py b/apps/catalogue/models/book.py
index c2b54e4..6a31f4b 100644 (file)
--- a/apps/catalogue/models/book.py
+++ b/apps/catalogue/models/book.py
@@ -274,13 +274,14 @@ class Book(models.Model):
         cover_changed = old_cover != book.cover_info()
         obsolete_children = set(b for b in book.children.all()
                                 if b not in children)
+        notify_cover_changed = []
         for n, child_book in enumerate(children):
             new_child = child_book.parent != book
             child_book.parent = book
             child_book.parent_number = n
             child_book.save()
             if new_child or cover_changed:
-                child_book.parent_cover_changed()
+                notify_cover_changed.append(child_book)
         # Disown unfaithful children and let them cope on their own.
         for child in obsolete_children:
             child.parent = None
@@ -288,7 +289,7 @@ class Book(models.Model):
             child.save()
             tasks.fix_tree_tags.delay(child)
             if old_cover:
-                child.parent_cover_changed()
+                notify_cover_changed.append(child)
 
         # delete old fragments when overwriting
         book.fragments.all().delete()
@@ -312,6 +313,9 @@ class Book(models.Model):
             book.search_index(index_tags=search_index_tags, reuse_index=search_index_reuse)
             #index_book.delay(book.id, book_info)
 
+        for child in notify_cover_changed:
+            child.parent_cover_changed()
+
         cls.published.send(sender=book)
         return book
 

diff --git a/lib/librarian b/lib/librarian
index cbe81ee..fb1bbfe 160000 (submodule)
--- a/lib/librarian
+++ b/lib/librarian
@@ -1 +1 @@
-Subproject commit cbe81ee35b07783b4f52c3d3dda83db7aaf82d34
+Subproject commit fb1bbfe622abeaaea2a2dfc459d8b779b2d369a2