X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/357027375ff8867f42ca34bcbfb5a78b5b185fc3..2e45f239552a9e378a9e1f65274dfb7c13df220d:/src/search/index.py diff --git a/src/search/index.py b/src/search/index.py index 31417ca87..bfb1739b1 100644 --- a/src/search/index.py +++ b/src/search/index.py @@ -6,22 +6,21 @@ from django.conf import settings import os import re -import errno from librarian import dcparser from librarian.parser import WLDocument from lxml import etree import catalogue.models from pdcounter.models import Author as PDCounterAuthor, BookStub as PDCounterBook from itertools import chain -import traceback -import logging -log = logging.getLogger('search') import sunburnt import custom import operator +import logging +from wolnelektury.utils import makedirs log = logging.getLogger('search') + class SolrIndex(object): def __init__(self, mode=None): self.index = custom.CustomSolrInterface(settings.SOLR, mode=mode) @@ -36,20 +35,18 @@ class Snippets(object): SNIPPET_DIR = "snippets" def __init__(self, book_id, revision=None): - try: - os.makedirs(os.path.join(settings.SEARCH_INDEX, self.SNIPPET_DIR)) - except OSError as exc: - if exc.errno == errno.EEXIST: - pass - else: raise + makedirs(os.path.join(settings.SEARCH_INDEX, self.SNIPPET_DIR)) self.book_id = book_id self.revision = revision self.file = None + self.position = None @property def path(self): - if self.revision: fn = "%d.%d" % (self.book_id, self.revision) - else: fn = "%d" % self.book_id + if self.revision: + fn = "%d.%d" % (self.book_id, self.revision) + else: + fn = "%d" % self.book_id return os.path.join(settings.SEARCH_INDEX, self.SNIPPET_DIR, fn) @@ -57,7 +54,7 @@ class Snippets(object): """ Open the snippet file. Call .close() afterwards. """ - if not 'b' in mode: + if 'b' not in mode: mode += 'b' if 'w' in mode: @@ -173,8 +170,9 @@ class Index(SolrIndex): if not remove_only: # then add them [all or just one passed] if not tags: - tags = chain(catalogue.models.Tag.objects.exclude(category='set'), \ - PDCounterAuthor.objects.all(), \ + tags = chain( + catalogue.models.Tag.objects.exclude(category='set'), + PDCounterAuthor.objects.all(), PDCounterBook.objects.all()) for tag in tags: @@ -211,11 +209,9 @@ class Index(SolrIndex): """ Create a lucene document referring book id. """ - doc = { - 'book_id': int(book.id), - } + doc = {'book_id': int(book.id)} if book.parent is not None: - doc["parent_id"] = int(book.parent.id) + doc['parent_id'] = int(book.parent.id) return doc def remove_book(self, book_or_id, remove_snippets=True): @@ -244,7 +240,8 @@ class Index(SolrIndex): self.remove_book(book, remove_snippets=False) book_doc = self.create_book_doc(book) - meta_fields = self.extract_metadata(book, book_info, dc_only=['source_name', 'authors', 'translators', 'title']) + meta_fields = self.extract_metadata(book, book_info, dc_only=[ + 'source_name', 'authors', 'translators', 'title', 'epochs', 'kinds', 'genres']) # let's not index it - it's only used for extracting publish date if 'source_name' in meta_fields: del meta_fields['source_name'] @@ -261,8 +258,9 @@ class Index(SolrIndex): 'published_date': meta_fields['published_date'] } - if 'translators' in meta_fields: - book_fields['translators'] = meta_fields['translators'] + for tag_name in ('translators', 'epochs', 'kinds', 'genres'): + if tag_name in meta_fields: + book_fields[tag_name] = meta_fields[tag_name] self.index_content(book, book_fields=book_fields) @@ -273,18 +271,19 @@ class Index(SolrIndex): 'dramat_wierszowany_lp', 'dramat_wspolczesny', 'liryka_l', 'liryka_lp', 'wywiad', - ] + ] ignore_content_tags = [ - 'uwaga', 'extra', + 'uwaga', 'extra', 'nota_red', 'abstrakt', 'zastepnik_tekstu', 'sekcja_asterysk', 'separator_linia', 'zastepnik_wersu', 'didaskalia', 'naglowek_aktu', 'naglowek_sceny', 'naglowek_czesc', - ] + ] footnote_tags = ['pa', 'pt', 'pr', 'pe'] - skip_header_tags = ['autor_utworu', 'nazwa_utworu', 'dzielo_nadrzedne', '{http://www.w3.org/1999/02/22-rdf-syntax-ns#}RDF'] + skip_header_tags = ['autor_utworu', 'nazwa_utworu', 'dzielo_nadrzedne', + '{http://www.w3.org/1999/02/22-rdf-syntax-ns#}RDF'] published_date_re = re.compile("([0-9]+)[\]. ]*$") @@ -298,7 +297,7 @@ class Index(SolrIndex): book_info = dcparser.parse(open(book.xml_file.path)) fields['slug'] = book.slug - fields['tags'] = [t.name for t in book.tags] + fields['tags'] = [t.name for t in book.tags] fields['is_book'] = True # validator, name @@ -332,7 +331,8 @@ class Index(SolrIndex): match = self.published_date_re.search(book_info.source_name) if match is not None: pd = str(match.groups()[0]) - if not pd: pd = "" + if not pd: + pd = "" fields["published_date"] = pd return fields @@ -355,7 +355,7 @@ class Index(SolrIndex): if master.tag in self.master_tags: return master - def index_content(self, book, book_fields={}): + def index_content(self, book, book_fields): """ Walks the book XML and extract content from it. Adds parts for each header tag and for each fragment. @@ -367,9 +367,8 @@ class Index(SolrIndex): if master is None: return [] - def walker(node, ignore_tags=[]): - - if node.tag not in ignore_tags: + def walker(node): + if node.tag not in self.ignore_content_tags: yield node, None, None if node.text is not None: yield None, node.text, None @@ -383,7 +382,7 @@ class Index(SolrIndex): return def fix_format(text): - # separator = [u" ", u"\t", u".", u";", u","] + # separator = [u" ", u"\t", u".", u";", u","] if isinstance(text, list): # need to join it first text = filter(lambda s: s is not None, content) @@ -420,17 +419,10 @@ class Index(SolrIndex): if 'themes' in fields: doc['themes'] = fields['themes'] - doc['uid'] = "part%s%s%s" % (doc['header_index'], - doc['header_span'], - doc.get('fragment_anchor', '')) + doc['uid'] = "part%s-%s-%s-%s" % ( + book.id, doc['header_index'], doc['header_span'], doc.get('fragment_anchor', '')) return doc - def give_me_utf8(s): - if isinstance(s, unicode): - return s.encode('utf-8') - else: - return s - fragments = {} snippets = Snippets(book.id).open('w') try: @@ -451,7 +443,7 @@ class Index(SolrIndex): content.append(text) handle_text = [all_content] - for start, text, end in walker(header, ignore_tags=self.ignore_content_tags): + for start, text, end in walker(header): # handle footnotes if start is not None and start.tag in self.footnote_tags: footnote = [] @@ -471,12 +463,13 @@ class Index(SolrIndex): # handle fragments and themes. if start is not None and start.tag == 'begin': fid = start.attrib['id'][1:] - fragments[fid] = {'text': [], 'themes': [], 'start_section': position, 'start_header': header.tag} + fragments[fid] = { + 'text': [], 'themes': [], 'start_section': position, 'start_header': header.tag} # themes for this fragment elif start is not None and start.tag == 'motyw': fid = start.attrib['id'][1:] - handle_text.append(None) + handle_text.append(lambda text: None) if start.text is not None: fragments[fid]['themes'] += map(unicode.strip, map(unicode, (start.text.split(',')))) elif end is not None and end.tag == 'motyw': @@ -487,7 +480,7 @@ class Index(SolrIndex): if fid not in fragments: continue # a broken node, skip it frag = fragments[fid] - if frag['themes'] == []: + if not frag['themes']: continue # empty themes list. del fragments[fid] @@ -504,8 +497,7 @@ class Index(SolrIndex): if text is not None and handle_text is not []: hdl = handle_text[-1] - if hdl is not None: - hdl(text) + hdl(text) # in the end, add a section text. doc = add_part(snippets, header_index=position, @@ -518,13 +510,13 @@ class Index(SolrIndex): class SearchResult(object): - def __init__(self, doc, how_found=None, query=None, query_terms=None): - # self.search = search + def __init__(self, doc, how_found=None, query_terms=None): self.boost = 1.0 self._hits = [] self._processed_hits = None # processed hits self.snippets = [] self.query_terms = query_terms + self._book = None if 'score' in doc: self._score = doc['score'] @@ -561,7 +553,9 @@ class SearchResult(object): def __unicode__(self): return u"" % \ - (self.book_id, len(self._hits), self._processed_hits and len(self._processed_hits) or -1, self._score, len(self.snippets)) + (self.book_id, len(self._hits), + len(self._processed_hits) if self._processed_hits else -1, + self._score, len(self.snippets)) def __str__(self): return unicode(self).encode('utf-8') @@ -579,7 +573,7 @@ class SearchResult(object): return self def get_book(self): - if hasattr(self, '_book'): + if self._book is not None: return self._book self._book = catalogue.models.Book.objects.get(id=self.book_id) return self._book @@ -605,11 +599,8 @@ class SearchResult(object): # sections not covered by fragments sect = filter(lambda s: 0 == len(filter( - lambda f: s[self.POSITION][self.POSITION_INDEX] >= f[self.POSITION][self.POSITION_INDEX] - and s[self.POSITION][self.POSITION_INDEX] < f[self.POSITION][self.POSITION_INDEX] + f[self.POSITION][self.POSITION_SPAN], - frags)), sect) - - hits = [] + lambda f: f[self.POSITION][self.POSITION_INDEX] <= s[self.POSITION][self.POSITION_INDEX] < + f[self.POSITION][self.POSITION_INDEX] + f[self.POSITION][self.POSITION_SPAN], frags)), sect) def remove_duplicates(lst, keyfn, compare): els = {} @@ -679,7 +670,7 @@ class SearchResult(object): m.update(f[self.OTHER]) hits.append(m) - hits.sort(lambda a, b: cmp(a['score'], b['score']), reverse=True) + hits.sort(key=lambda h: h['score'], reverse=True) self._processed_hits = hits @@ -713,7 +704,7 @@ class SearchResult(object): def snippet_revision(self, idx=0): try: return self.hits[idx]['snippets_revision'] - except: + except (IndexError, KeyError): return None @@ -724,82 +715,36 @@ class Search(SolrIndex): def __init__(self, default_field="text"): super(Search, self).__init__(mode='r') - def make_term_query(self, query, field='text', modal=operator.or_): """ Returns term queries joined by boolean query. modal - applies to boolean query fuzzy - should the query by fuzzy. """ - if query is None: query = '' + if query is None: + query = '' q = self.index.Q() - q = reduce(modal, map(lambda s: self.index.Q(**{field: s}), - query.split(r" ")), q) + q = reduce(modal, map(lambda s: self.index.Q(**{field: s}), query.split(r" ")), q) return q - def search_phrase(self, searched, field='text', book=False, - filters=None, - snippets=False): - if filters is None: filters = [] - if book: filters.append(self.index.Q(is_book=True)) - - q = self.index.query(**{field: searched}) - q = self.apply_filters(q, filters).field_limit(score=True, all_fields=True) - res = q.execute() - return [SearchResult(found, how_found=u'search_phrase') for found in res] - - def search_some(self, searched, fields, book=True, - filters=None, snippets=True, query_terms=None): - assert isinstance(fields, list) - if filters is None: filters = [] - if book: filters.append(self.index.Q(is_book=True)) - - query = self.index.Q() - - for fld in fields: - query = self.index.Q(query | self.make_term_query(searched, fld)) - - query = self.index.query(query) + def search_words(self, words, fields, book=True): + filters = [] + for word in words: + word_filter = None + for field in fields: + q = self.index.Q(**{field: word}) + if word_filter is None: + word_filter = q + else: + word_filter |= q + filters.append(word_filter) + if book: + query = self.index.query(is_book=True) + else: + query = self.index.query() query = self.apply_filters(query, filters).field_limit(score=True, all_fields=True) - res = query.execute() - return [SearchResult(found, how_found='search_some', query_terms=query_terms) for found in res] - - - def search_everywhere(self, searched, query_terms=None): - """ - Tries to use search terms to match different fields of book (or its parts). - E.g. one word can be an author survey, another be a part of the title, and the rest - are some words from third chapter. - """ - books = [] - # content only query : themes x content - q = self.make_term_query(searched, 'text') - q_themes = self.make_term_query(searched, 'themes_pl') - - query = self.index.query(q).query(q_themes).field_limit(score=True, all_fields=True) - res = query.execute() - - for found in res: - books.append(SearchResult(found, how_found='search_everywhere_themesXcontent', query_terms=query_terms)) - - # query themes/content x author/title/tags - in_content = self.index.Q() - in_meta = self.index.Q() - - for fld in ['themes_pl', 'text']: - in_content |= self.make_term_query(searched, field=fld) - - for fld in ['tags', 'authors', 'title']: - in_meta |= self.make_term_query(searched, field=fld) - - q = in_content & in_meta - res = self.index.query(q).field_limit(score=True, all_fields=True).execute() - - for found in res: - books.append(SearchResult(found, how_found='search_everywhere', query_terms=query_terms)) - - return books + return [SearchResult(found, how_found='search_words', query_terms=words) for found in query.execute()] def get_snippets(self, searchresult, query, field='text', num=1): """ @@ -822,13 +767,18 @@ class Search(SolrIndex): text = snippets.get((int(position), int(length))) snip = self.index.highlight(text=text, field=field, q=query) - snips[idx] = snip - if snip: - num -= 1 + if snip not in snips: + snips[idx] = snip + if snip: + num -= 1 idx += 1 except IOError, e: - log.error("Cannot open snippet file for book id = %d [rev=%s], %s" % (book_id, revision, e)) + book = catalogue.models.Book.objects.filter(id=book_id) + if not book: + log.error("Book does not exist for book id = %d" % book_id) + elif not book.get().children.exists(): + log.error("Cannot open snippet file for book id = %d [rev=%s], %s" % (book_id, revision, e)) return [] finally: snippets.close() @@ -860,7 +810,8 @@ class Search(SolrIndex): """ Search for Tag objects using query. """ - if not filters: filters = [] + if not filters: + filters = [] if not pdcounter: filters.append(~self.index.Q(is_pdcounter=True)) res = self.apply_filters(query, filters).execute() @@ -872,25 +823,26 @@ class Search(SolrIndex): is_pdcounter = doc.get('is_pdcounter', False) category = doc.get('tag_category') try: - if is_pdcounter == True: + if is_pdcounter: if category == 'pd_author': tag = PDCounterAuthor.objects.get(id=doc.get('tag_id')) - elif category == 'pd_book': + else: # category == 'pd_book': tag = PDCounterBook.objects.get(id=doc.get('tag_id')) tag.category = 'pd_book' # make it look more lik a tag. - else: - print ("Warning. cannot get pdcounter tag_id=%d from db; cat=%s" % (int(doc.get('tag_id')), category)).encode('utf-8') pd_tags.append(tag) else: tag = catalogue.models.Tag.objects.get(id=doc.get("tag_id")) tags.append(tag) - except catalogue.models.Tag.DoesNotExist: pass - except PDCounterAuthor.DoesNotExist: pass - except PDCounterBook.DoesNotExist: pass + except catalogue.models.Tag.DoesNotExist: + pass + except PDCounterAuthor.DoesNotExist: + pass + except PDCounterBook.DoesNotExist: + pass tags_slugs = set(map(lambda t: t.slug, tags)) - tags = tags + filter(lambda t: not t.slug in tags_slugs, pd_tags) + tags = tags + filter(lambda t: t.slug not in tags_slugs, pd_tags) log.debug('search_tags: %s' % tags) @@ -906,8 +858,10 @@ class Search(SolrIndex): query = query.strip() if prefix: q |= self.index.Q(title=query + "*") + q |= self.index.Q(title_orig=query + "*") else: q |= self.make_term_query(query, field='title') + q |= self.make_term_query(query, field='title_orig') qu = self.index.query(q) only_books = self.index.Q(is_book=True) return self.search_books(qu, [only_books]) @@ -923,19 +877,20 @@ class Search(SolrIndex): for r in res: try: bid = r['book_id'] - if not bid in bks_found: + if bid not in bks_found: bks.append(catalogue.models.Book.objects.get(id=bid)) bks_found.add(bid) - except catalogue.models.Book.DoesNotExist: pass + except catalogue.models.Book.DoesNotExist: + pass return bks - @staticmethod def apply_filters(query, filters): """ Apply filters to a query """ - if filters is None: filters = [] + if filters is None: + filters = [] filters = filter(lambda x: x is not None, filters) for f in filters: query = query.query(f)