X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/e47b75737d714a11e22cf3f5bceadddbf340e3b3..6835539fb69c791e27d536c8136a7ee199237754:/apps/lesmianator/models.py?ds=sidebyside diff --git a/apps/lesmianator/models.py b/apps/lesmianator/models.py index 5103ebbf1..7fceda2fe 100644 --- a/apps/lesmianator/models.py +++ b/apps/lesmianator/models.py @@ -18,8 +18,7 @@ from django.contrib.contenttypes.models import ContentType from django.contrib.contenttypes import generic from django.conf import settings -from librarian import text -from catalogue.fields import JSONField +from jsonfield import JSONField from catalogue.models import Book, Tag @@ -126,31 +125,32 @@ class Continuations(models.Model): def for_book(cls, book, length=3): # count from this book only output = StringIO() - f = open(book.xml_file.path) - text.transform(f, output, False, ('raw-text',)) - f.close() + wldoc = book.wldocument(parse_dublincore=False) + output = wldoc.as_text(('raw-text',)).get_string() + del wldoc + conts = {} last_word = '' - for letter in output.getvalue().decode('utf-8').strip().lower(): + for letter in output.decode('utf-8').strip().lower(): mydict = conts.setdefault(last_word, {}) mydict.setdefault(letter, 0) mydict[letter] += 1 last_word = last_word[-length+1:] + letter # add children return reduce(cls.join_conts, - (cls.get(child) for child in book.children.all()), + (cls.get(child) for child in book.children.all().iterator()), conts) @classmethod def for_set(cls, tag): # book contains its descendants, we don't want them twice books = Book.tagged.with_any((tag,)) - l_tags = Tag.objects.filter(category='book', slug__in=[book.book_tag_slug() for book in books]) - descendants_keys = [book.pk for book in Book.tagged.with_any(l_tags)] + l_tags = Tag.objects.filter(category='book', slug__in=[book.book_tag_slug() for book in books.iterator()]) + descendants_keys = [book.pk for book in Book.tagged.with_any(l_tags).iterator()] if descendants_keys: books = books.exclude(pk__in=descendants_keys) - cont_tabs = (cls.get(b) for b in books) + cont_tabs = (cls.get(b) for b in books.iterator()) return reduce(cls.join_conts, cont_tabs) @classmethod @@ -158,7 +158,7 @@ class Continuations(models.Model): object_type = ContentType.objects.get_for_model(sth) should_keys = set([sth.id]) if isinstance(sth, Tag): - should_keys = set(b.pk for b in Book.tagged.with_any((sth,))) + should_keys = set(b.pk for b in Book.tagged.with_any((sth,)).iterator()) try: obj = cls.objects.get(content_type=object_type, object_id=sth.id) if not obj.pickle: