librarian bump with proper merge

[wolnelektury.git] / apps / search / index.py

diff --git a/apps/search/index.py b/apps/search/index.py
index df44888..6d97047 100644 (file)
--- a/apps/search/index.py
+++ b/apps/search/index.py
@@ -256,7 +256,7 @@ class Index(BaseIndex):
         self.index.addDocument(book_doc)
         del book_doc
 
         self.index.addDocument(book_doc)
         del book_doc
 

-        self.index_content(book, book_fields=[meta_fields['title'], meta_fields['authors']])
+        self.index_content(book, book_fields=[meta_fields['title'], meta_fields['authors'], meta_fields['published_date']])

 
     master_tags = [
         'opowiadanie',
 
     master_tags = [
         'opowiadanie',


@@ -264,10 +264,21 @@ class Index(BaseIndex):
         'dramat_wierszowany_l',
         'dramat_wierszowany_lp',
         'dramat_wspolczesny', 'liryka_l', 'liryka_lp',
         'dramat_wierszowany_l',
         'dramat_wierszowany_lp',
         'dramat_wspolczesny', 'liryka_l', 'liryka_lp',

-        'wywiad'
+        'wywiad',

         ]
 
         ]
 

-    skip_header_tags = ['autor_utworu', 'nazwa_utworu', 'dzielo_nadrzedne']
+    ignore_content_tags = [
+        'uwaga', 'extra',
+        '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']
+
+    published_date_re = re.compile("([0-9]+)[\]. ]*$")

 
     def extract_metadata(self, book, book_info=None):
         """
 
     def extract_metadata(self, book, book_info=None):
         """


@@ -309,6 +320,12 @@ class Index(BaseIndex):
                     fields[field.name] = Field(field.name, "%04d%02d%02d" %\
                                                (dt.year, dt.month, dt.day), Field.Store.NO, Field.Index.NOT_ANALYZED)
 
                     fields[field.name] = Field(field.name, "%04d%02d%02d" %\
                                                (dt.year, dt.month, dt.day), Field.Store.NO, Field.Index.NOT_ANALYZED)
 

+        # get published date
+        source = book_info.source_name
+        match = self.published_date_re.search(source)
+        if match is not None:
+            fields["published_date"] = Field("published_date", str(match.groups()[0]), Field.Store.YES, Field.Index.NOT_ANALYZED)
+

         return fields
 
     def add_gaps(self, fields, fieldname):
         return fields
 
     def add_gaps(self, fields, fieldname):


@@ -341,15 +358,26 @@ class Index(BaseIndex):
         if master is None:
             return []
 
         if master is None:
             return []
 

-        def walker(node):
+        def walker(node, ignore_tags=[]):

             yield node, None
             yield node, None

-            for child in list(node):
+            for child in filter(lambda n: n.tag not in ignore_tags, list(node)):

                 for b, e in walker(child):
                     yield b, e
             yield None, node
             return
 
         def fix_format(text):
                 for b, e in walker(child):
                     yield b, e
             yield None, node
             return
 
         def fix_format(text):

+            #            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)
+                text = u' '.join(text)
+                # for i in range(len(text)):
+                #     if i > 0:
+                #         if text[i][0] not in separator\
+                #             and text[i - 1][-1] not in separator:
+                #          text.insert(i, u" ")
+

             return re.sub("(?m)/$", "", text)
 
         def add_part(snippets, **fields):
             return re.sub("(?m)/$", "", text)
 
         def add_part(snippets, **fields):


@@ -407,9 +435,21 @@ class Index(BaseIndex):
 
                 # section content
                 content = []
 
                 # section content
                 content = []

+                footnote = None
+
+                for start, end in walker(header, ignore_tags=self.ignore_content_tags):
+                    # handle footnotes
+                    # if start is not None and start.tag in self.footnote_tags:
+                    #     footnote = ' '.join(start.itertext())
+                    # elif end is not None and footnote is not None and end.tag in self.footnote_tags:
+                    #     doc = add_part(snippets, header_index=position, header_type=header.tag,
+                    #                    content=footnote)
+
+                    #     self.index.addDocument(doc)

 
 

-                for start, end in walker(header):
-                        # handle fragments and themes.
+                    #     footnote = None
+
+                    # handle fragments and themes.

                     if start is not None and start.tag == 'begin':
                         fid = start.attrib['id'][1:]
                         fragments[fid] = {'content': [], 'themes': [], 'start_section': position, 'start_header': header.tag}
                     if start is not None and start.tag == 'begin':
                         fid = start.attrib['id'][1:]
                         fragments[fid] = {'content': [], 'themes': [], 'start_section': position, 'start_header': header.tag}


@@ -429,17 +469,12 @@ class Index(BaseIndex):
                             continue  # empty themes list.
                         del fragments[fid]
 
                             continue  # empty themes list.
                         del fragments[fid]
 

-                        def jstr(l):
-                            return u' '.join(map(
-                                lambda x: x == None and u'(none)' or unicode(x),
-                                l))
-

                         doc = add_part(snippets,
                                        header_type=frag['start_header'],
                                        header_index=frag['start_section'],
                                        header_span=position - frag['start_section'] + 1,
                                        fragment_anchor=fid,
                         doc = add_part(snippets,
                                        header_type=frag['start_header'],
                                        header_index=frag['start_section'],
                                        header_span=position - frag['start_section'] + 1,
                                        fragment_anchor=fid,

-                                       content=u' '.join(filter(lambda s: s is not None, frag['content'])),
+                                       content=fix_format(frag['content']),

                                        themes=frag['themes'])
 
                         self.index.addDocument(doc)
                                        themes=frag['themes'])
 
                         self.index.addDocument(doc)


@@ -456,7 +491,7 @@ class Index(BaseIndex):
 
                         # in the end, add a section text.
                 doc = add_part(snippets, header_index=position, header_type=header.tag,
 
                         # in the end, add a section text.
                 doc = add_part(snippets, header_index=position, header_type=header.tag,

-                               content=fix_format(u' '.join(filter(lambda s: s is not None, content))))
+                               content=fix_format(content))

 
                 self.index.addDocument(doc)
 
 
                 self.index.addDocument(doc)
 


@@ -553,20 +588,20 @@ class JoinSearch(object):
 
 
 class SearchResult(object):
 
 
 class SearchResult(object):

-    def __init__(self, searcher, scoreDocs, score=None, how_found=None, snippets=None, searched=None, tokens_cache=None):
+    def __init__(self, search, scoreDocs, score=None, how_found=None, snippets=None, searched=None, tokens_cache=None):

         if tokens_cache is None: tokens_cache = {}
         if tokens_cache is None: tokens_cache = {}

-            
+

         if score:
             self._score = score
         else:
             self._score = scoreDocs.score
         if score:
             self._score = score
         else:
             self._score = scoreDocs.score

-            
+

         self.boost = 1.0
 
         self._hits = []
         self.boost = 1.0
 
         self._hits = []

-        self.hits = None  # processed hits
+        self._processed_hits = None  # processed hits

 
 

-        stored = searcher.doc(scoreDocs.doc)
+        stored = search.searcher.doc(scoreDocs.doc)

         self.book_id = int(stored.get("book_id"))
 
         header_type = stored.get("header_type")
         self.book_id = int(stored.get("book_id"))
 
         header_type = stored.get("header_type")


@@ -579,13 +614,18 @@ class SearchResult(object):
 
         fragment = stored.get("fragment_anchor")
 
 
         fragment = stored.get("fragment_anchor")
 

+        pd = stored.get("published_date")
+        if pd is None:
+            pd = 0
+        self.published_date = int(pd)
+

         if snippets:
             snippets = snippets.replace("/\n", "\n")
         hit = (sec + (header_span,), fragment, scoreDocs.score, {'how_found': how_found, 'snippets': snippets and [snippets] or []})
 
         self._hits.append(hit)
 
         if snippets:
             snippets = snippets.replace("/\n", "\n")
         hit = (sec + (header_span,), fragment, scoreDocs.score, {'how_found': how_found, 'snippets': snippets and [snippets] or []})
 
         self._hits.append(hit)
 

-        self.searcher = searcher
+        self.search = search

         self.searched = searched
         self.tokens_cache = tokens_cache
 
         self.searched = searched
         self.tokens_cache = tokens_cache
 


@@ -598,7 +638,7 @@ class SearchResult(object):
             raise ValueError("this search result is or book %d; tried to merge with %d" % (self.book_id, other.book_id))
         self._hits += other._hits
         if other.score > self.score:
             raise ValueError("this search result is or book %d; tried to merge with %d" % (self.book_id, other.book_id))
         self._hits += other._hits
         if other.score > self.score:

-            self.score = other.score
+            self._score = other._score

         return self
 
     def get_book(self):
         return self
 
     def get_book(self):


@@ -606,7 +646,11 @@ class SearchResult(object):
 
     book = property(get_book)
 
 
     book = property(get_book)
 

-    def process_hits(self):
+    @property
+    def hits(self):
+        if self._processed_hits is not None:
+            return self._processed_hits
+

         POSITION = 0
         FRAGMENT = 1
         POSITION_INDEX = 1
         POSITION = 0
         FRAGMENT = 1
         POSITION_INDEX = 1


@@ -653,17 +697,21 @@ class SearchResult(object):
         hits = sections.values()
 
         for f in frags:
         hits = sections.values()
 
         for f in frags:

-            frag = catalogue.models.Fragment.objects.get(anchor=f[FRAGMENT])
+            try:
+                frag = catalogue.models.Fragment.objects.get(anchor=f[FRAGMENT])
+            except catalogue.models.Fragment.DoesNotExist:
+                # stale index
+                continue

 
             # Figure out if we were searching for a token matching some word in theme name.
             themes = frag.tags.filter(category='theme')
             themes_hit = []
             if self.searched is not None:
 
             # Figure out if we were searching for a token matching some word in theme name.
             themes = frag.tags.filter(category='theme')
             themes_hit = []
             if self.searched is not None:

-                tokens = self.searcher.get_tokens(self.searched, 'POLISH', tokens_cache=self.tokens_cache)
+                tokens = self.search.get_tokens(self.searched, 'POLISH', cached=self.tokens_cache)

                 for theme in themes:
                 for theme in themes:

-                    name_tokens = self.searcher.get_tokens(theme.name, 'POLISH')
+                    name_tokens = self.search.get_tokens(theme.name, 'POLISH')

                     for t in tokens:
                     for t in tokens:

-                        if name_tokens.index(t):
+                        if t in name_tokens:

                             if not theme in themes_hit:
                                 themes_hit.append(theme)
                             break
                             if not theme in themes_hit:
                                 themes_hit.append(theme)
                             break


@@ -679,9 +727,9 @@ class SearchResult(object):
 
         hits.sort(lambda a, b: cmp(a['score'], b['score']), reverse=True)
 
 
         hits.sort(lambda a, b: cmp(a['score'], b['score']), reverse=True)
 

-        self.hits = hits
+        self._processed_hits = hits

 
 

-        return self
+        return hits

 
     def __unicode__(self):
         return u'SearchResult(book_id=%d, score=%d)' % (self.book_id, self.score)
 
     def __unicode__(self):
         return u'SearchResult(book_id=%d, score=%d)' % (self.book_id, self.score)


@@ -699,7 +747,12 @@ class SearchResult(object):
         return books.values()
 
     def __cmp__(self, other):
         return books.values()
 
     def __cmp__(self, other):

-        return cmp(self.score, other.score)
+        c = cmp(self.score, other.score)
+        if c == 0:
+            # this is inverted, because earlier date is better
+            return cmp(other.published_date, self.published_date)
+        else:
+            return c

 
 
 class Hint(object):
 
 
 class Hint(object):


@@ -909,21 +962,21 @@ class Search(IndexStore):
         return q
 
     def search_phrase(self, searched, field, book=True, max_results=20, fuzzy=False,
         return q
 
     def search_phrase(self, searched, field, book=True, max_results=20, fuzzy=False,

-                      filters=None, tokens_cache=None, boost=None):
+                      filters=None, tokens_cache=None, boost=None, snippets=False, slop=2):

         if filters is None: filters = []
         if tokens_cache is None: tokens_cache = {}
 
         tokens = self.get_tokens(searched, field, cached=tokens_cache)
 
         if filters is None: filters = []
         if tokens_cache is None: tokens_cache = {}
 
         tokens = self.get_tokens(searched, field, cached=tokens_cache)
 

-        query = self.make_phrase(tokens, field=field, fuzzy=fuzzy)
+        query = self.make_phrase(tokens, field=field, fuzzy=fuzzy, slop=slop)

         if book:
             filters.append(self.term_filter(Term('is_book', 'true')))
         top = self.searcher.search(query, self.chain_filters(filters), max_results)
 
         if book:
             filters.append(self.term_filter(Term('is_book', 'true')))
         top = self.searcher.search(query, self.chain_filters(filters), max_results)
 

-        return [SearchResult(self.searcher, found) for found in top.scoreDocs]
+        return [SearchResult(self, found, snippets=(snippets and self.get_snippets(found, query) or None), searched=searched) for found in top.scoreDocs]

 
     def search_some(self, searched, fields, book=True, max_results=20, fuzzy=False,
 
     def search_some(self, searched, fields, book=True, max_results=20, fuzzy=False,

-                    filters=None, tokens_cache=None, boost=None):
+                    filters=None, tokens_cache=None, boost=None, snippets=True):

         if filters is None: filters = []
         if tokens_cache is None: tokens_cache = {}
 
         if filters is None: filters = []
         if tokens_cache is None: tokens_cache = {}
 


@@ -940,7 +993,8 @@ class Search(IndexStore):
 
         top = self.searcher.search(query, self.chain_filters(filters), max_results)
 
 
         top = self.searcher.search(query, self.chain_filters(filters), max_results)
 

-        return [SearchResult(self.searcher, found, searched=searched, tokens_cache=tokens_cache) for found in top.scoreDocs]
+        return [SearchResult(self, found, searched=searched, tokens_cache=tokens_cache,
+                             snippets=(snippets and self.get_snippets(found, query) or None)) for found in top.scoreDocs]

 
     def search_perfect_book(self, searched, max_results=20, fuzzy=False, hint=None):
         """
 
     def search_perfect_book(self, searched, max_results=20, fuzzy=False, hint=None):
         """


@@ -963,7 +1017,7 @@ class Search(IndexStore):
                 self.chain_filters([only_in, self.term_filter(Term('is_book', 'true'))]),
                 max_results)
             for found in top.scoreDocs:
                 self.chain_filters([only_in, self.term_filter(Term('is_book', 'true'))]),
                 max_results)
             for found in top.scoreDocs:

-                books.append(SearchResult(self.searcher, found, how_found="search_perfect_book"))
+                books.append(SearchResult(self, found, how_found="search_perfect_book"))

         return books
 
     def search_book(self, searched, max_results=20, fuzzy=False, hint=None):
         return books
 
     def search_book(self, searched, max_results=20, fuzzy=False, hint=None):


@@ -989,7 +1043,7 @@ class Search(IndexStore):
                                    self.chain_filters([only_in, self.term_filter(Term('is_book', 'true'))]),
             max_results)
         for found in top.scoreDocs:
                                    self.chain_filters([only_in, self.term_filter(Term('is_book', 'true'))]),
             max_results)
         for found in top.scoreDocs:

-            books.append(SearchResult(self.searcher, found, how_found="search_book"))
+            books.append(SearchResult(self, found, how_found="search_book"))

 
         return books
 
 
         return books
 


@@ -1011,7 +1065,7 @@ class Search(IndexStore):
                                                            flt]),
                                        max_results)
             for found in top.scoreDocs:
                                                            flt]),
                                        max_results)
             for found in top.scoreDocs:

-                books.append(SearchResult(self.searcher, found, snippets=self.get_snippets(found, q), how_found='search_perfect_parts'))
+                books.append(SearchResult(self, found, snippets=self.get_snippets(found, q), how_found='search_perfect_parts'))

 
         return books
 
 
         return books
 


@@ -1044,7 +1098,7 @@ class Search(IndexStore):
 
         topDocs = self.searcher.search(q, only_in, max_results)
         for found in topDocs.scoreDocs:
 
         topDocs = self.searcher.search(q, only_in, max_results)
         for found in topDocs.scoreDocs:

-            books.append(SearchResult(self.searcher, found, how_found='search_everywhere_themesXcontent'))
+            books.append(SearchResult(self, found, how_found='search_everywhere_themesXcontent', searched=searched))

             print "* %s theme x content: %s" % (searched, books[-1]._hits)
 
         # query themes/content x author/title/tags
             print "* %s theme x content: %s" % (searched, books[-1]._hits)
 
         # query themes/content x author/title/tags


@@ -1063,7 +1117,7 @@ class Search(IndexStore):
 
         topDocs = self.searcher.search(q, only_in, max_results)
         for found in topDocs.scoreDocs:
 
         topDocs = self.searcher.search(q, only_in, max_results)
         for found in topDocs.scoreDocs:

-            books.append(SearchResult(self.searcher, found, how_found='search_everywhere'))
+            books.append(SearchResult(self, found, how_found='search_everywhere', searched=searched))

             print "* %s scatter search: %s" % (searched, books[-1]._hits)
 
         return books
             print "* %s scatter search: %s" % (searched, books[-1]._hits)
 
         return books


@@ -1119,11 +1173,15 @@ class Search(IndexStore):
 
         stored = self.searcher.doc(scoreDoc.doc)
 
 
         stored = self.searcher.doc(scoreDoc.doc)
 

+        position = stored.get('snippets_position')
+        length = stored.get('snippets_length')
+        if position is None or length is None:
+            return None

         # locate content.
         snippets = Snippets(stored.get('book_id')).open()
         try:
         # locate content.
         snippets = Snippets(stored.get('book_id')).open()
         try:

-            text = snippets.get((int(stored.get('snippets_position')),
-                                 int(stored.get('snippets_length'))))
+            text = snippets.get((int(position),
+                                 int(length)))

         finally:
             snippets.close()
 
         finally:
             snippets.close()