from catalogue.utils import split_tags
from catalogue.models import Book, Tag, Fragment
+from pdcounter.models import Author as PDCounterAuthor, BookStub as PDCounterBook
from catalogue.views import JSONResponse
-from search import Search, JVM, SearchResult
+from search import Search, SearchResult
from lucene import StringReader
from suggest.forms import PublishingSuggestForm
+from time import sleep
import re
-import enchant
-
-dictionary = enchant.Dict('pl_PL')
+#import enchant
+import json
def match_word_re(word):
def did_you_mean(query, tokens):
- change = {}
- for t in tokens:
- authors = Tag.objects.filter(category='author', name__iregex=match_word_re(t))
- if len(authors) > 0:
- continue
-
- if not dictionary.check(t):
- try:
- change_to = dictionary.suggest(t)[0].lower()
- if change_to != t.lower():
- change[t] = change_to
- except IndexError:
- pass
-
- if change == {}:
- return None
-
- for frm, to in change.items():
- query = query.replace(frm, to)
-
return query
+ # change = {}
+ # for t in tokens:
+ # authors = Tag.objects.filter(category='author', name__iregex=match_word_re(t))
+ # if len(authors) > 0:
+ # continue
+
+ # if False:
+ # if not dictionary.check(t):
+ # try:
+ # change_to = dictionary.suggest(t)[0].lower()
+ # if change_to != t.lower():
+ # change[t] = change_to
+ # except IndexError:
+ # pass
+
+ # if change == {}:
+ # return None
-JVM.attachCurrentThread()
-search = Search()
+ # for frm, to in change.items():
+ # query = query.replace(frm, to)
+
+ # return query
def hint(request):
prefix = request.GET.get('term', '')
if len(prefix) < 2:
return JSONResponse([])
- JVM.attachCurrentThread()
-
- hint = search.hint()
- try:
- tags = request.GET.get('tags', '')
- hint.tags(Tag.get_tag_list(tags))
- except:
- pass
+ search = Search()
# tagi beda ograniczac tutaj
# ale tagi moga byc na ksiazce i na fragmentach
# jezeli tagi dot tylko ksiazki, to wazne zeby te nowe byly w tej samej ksiazce
tags = search.hint_tags(prefix, pdcounter=True)
books = search.hint_books(prefix)
+ def is_dupe(tag):
+ if isinstance(tag, PDCounterAuthor):
+ if filter(lambda t: t.slug == tag.slug and t != tag, tags):
+ return True
+ elif isinstance(tag, PDCounterBook):
+ if filter(lambda b: b.slug == tag.slug, tags):
+ return True
+ return False
+
+ tags = filter(lambda t: not is_dupe(t), tags)
+
def category_name(c):
if c.startswith('pd_'):
c = c[len('pd_'):]
return _(c)
- return JSONResponse(
- [{'label': t.name,
- 'category': category_name(t.category),
- 'id': t.id,
- 'url': t.get_absolute_url()}
- for t in tags] + \
- [{'label': b.title,
- 'category': _('book'),
- 'id': b.id,
- 'url': b.get_absolute_url()}
- for b in books])
+ callback = request.GET.get('callback', None)
+ data = [{'label': t.name,
+ 'category': category_name(t.category),
+ 'id': t.id,
+ 'url': t.get_absolute_url()}
+ for t in tags] + \
+ [{'label': b.title,
+ 'category': _('book'),
+ 'id': b.id,
+ 'url': b.get_absolute_url()}
+ for b in books]
+ if callback:
+ return HttpResponse("%s(%s);" % (callback, json.dumps(data)),
+ content_type="application/json; charset=utf-8")
+ else:
+ return JSONResponse(data)
def main(request):
results = {}
- JVM.attachCurrentThread() # where to put this?
results = None
query = None
- fuzzy = False #0.8
-
- query = request.GET.get('q','')
- # book_id = request.GET.get('book', None)
- # book = None
- # if book_id is not None:
- # book = get_object_or_404(Book, id=book_id)
- # hint = search.hint()
- # try:
- # tag_list = Tag.get_tag_list(tags)
- # except:
- # tag_list = []
+ query = request.GET.get('q', '')
if len(query) < 2:
- return render_to_response('catalogue/search_too_short.html', {'prefix': query},
- context_instance=RequestContext(request))
+ return render_to_response('catalogue/search_too_short.html',
+ {'prefix': query},
+ context_instance=RequestContext(request))
+ search = Search()
- # hint.tags(tag_list)
- # if book:
- # hint.books(book)
- tags = search.hint_tags(query, pdcounter=True, prefix=False, fuzzy=fuzzy)
- tags = split_tags(tags)
+ theme_terms = search.index.analyze(text=query, field="themes_pl") \
+ + search.index.analyze(text=query, field="themes")
- toks = StringReader(query)
- tokens_cache = {}
+ # change hints
+ tags = search.hint_tags(query, pdcounter=True, prefix=False)
+ tags = split_tags(tags)
- author_results = search.search_phrase(toks, 'authors', fuzzy=fuzzy, tokens_cache=tokens_cache)
- title_results = search.search_phrase(toks, 'title', fuzzy=fuzzy, tokens_cache=tokens_cache)
+ author_results = search.search_phrase(query, 'authors', book=True)
+ title_results = search.search_phrase(query, 'title', book=True)
# Boost main author/title results with mixed search, and save some of its results for end of list.
# boost author, title results
- author_title_mixed = search.search_some(toks, ['authors', 'title', 'tags'], fuzzy=fuzzy, tokens_cache=tokens_cache)
+ author_title_mixed = search.search_some(query, ['authors', 'title', 'tags'], query_terms=theme_terms)
author_title_rest = []
+
for b in author_title_mixed:
- bks = filter(lambda ba: ba.book_id == b.book_id, author_results + title_results)
- for b2 in bks:
+ also_in_mixed = filter(lambda ba: ba.book_id == b.book_id, author_results + title_results)
+ for b2 in also_in_mixed:
b2.boost *= 1.1
- if bks is []:
+ if also_in_mixed is []:
author_title_rest.append(b)
# Do a phrase search but a term search as well - this can give us better snippets then search_everywhere,
# Because the query is using only one field.
text_phrase = SearchResult.aggregate(
- search.search_phrase(toks, 'content', fuzzy=fuzzy, tokens_cache=tokens_cache, snippets=True, book=False, slop=4),
- search.search_some(toks, ['content'], tokens_cache=tokens_cache, snippets=True, book=False))
+ search.search_phrase(query, 'text', snippets=True, book=False),
+ search.search_some(query, ['text'], snippets=True, book=False, query_terms=theme_terms))
- everywhere = search.search_everywhere(toks, fuzzy=fuzzy, tokens_cache=tokens_cache)
+ everywhere = search.search_everywhere(query, query_terms=theme_terms)
def already_found(results):
def f(e):
everywhere = SearchResult.aggregate(everywhere, author_title_rest)
- for res in [author_results, title_results, text_phrase, everywhere]:
+ for field, res in [('authors', author_results),
+ ('title', title_results),
+ ('text', text_phrase),
+ ('text', everywhere)]:
res.sort(reverse=True)
+ print "get snips %s, res size %d" % (field, len(res))
for r in res:
- for h in r.hits:
- h['snippets'] = map(lambda s:
- re.subn(r"(^[ \t\n]+|[ \t\n]+$)", u"",
- re.subn(r"[ \t\n]*\n[ \t\n]*", u"\n", s)[0])[0], h['snippets'])
+ print "Get snippets for %s" % r
+ search.get_snippets(r, query, field, 3)
+ # for r in res:
+ # for h in r.hits:
+ # h['snippets'] = map(lambda s:
+ # re.subn(r"(^[ \t\n]+|[ \t\n]+$)", u"",
+ # re.subn(r"[ \t\n]*\n[ \t\n]*", u"\n", s)[0])[0], h['snippets'])
- suggestion = did_you_mean(query, search.get_tokens(toks, field="SIMPLE"))
+ # suggestion = did_you_mean(query, search.get_tokens(toks, field="SIMPLE"))
+ suggestion = u''
def ensure_exists(r):
try:
return render_to_response('catalogue/search_multiple_hits.html',
{'tags': tags,
'prefix': query,
- 'results': { 'author': author_results,
- 'title': title_results,
- 'content': text_phrase,
- 'other': everywhere},
+ 'results': {'author': author_results,
+ 'title': title_results,
+ 'content': text_phrase,
+ 'other': everywhere},
'did_you_mean': suggestion},
context_instance=RequestContext(request))
fnp / wolnelektury.git / blobdiff