cleanup solr schema

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

diff --git a/src/search/index.py b/src/search/index.py
index 4c278ea..b94d8f6 100644 (file)
--- a/src/search/index.py
+++ b/src/search/index.py
@@ -20,6 +20,13 @@ from wolnelektury.utils import makedirs
 
 log = logging.getLogger('search')
 
+if os.path.isfile(settings.SOLR_STOPWORDS):
+    stopwords = set(
+        line.decode('utf-8').strip()
+        for line in open(settings.SOLR_STOPWORDS) if not line.startswith('#'))
+else:
+    stopwords = set()
+
 
 class SolrIndex(object):
     def __init__(self, mode=None):
@@ -297,7 +304,6 @@ 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['is_book'] = True
 
         # validator, name
@@ -551,6 +557,17 @@ class SearchResult(object):
 
             self._hits.append(hit)
 
+    @classmethod
+    def from_book(cls, book, how_found=None, query_terms=None):
+        doc = {
+            'score': book.popularity.count,
+            'book_id': book.id,
+            'published_date': 0,
+        }
+        result = cls(doc, how_found=how_found, query_terms=query_terms)
+        result._book = book
+        return result
+
     def __unicode__(self):
         return u"<SR id=%d %d(%d) hits score=%f %d snippets>" % \
             (self.book_id, len(self._hits),
@@ -566,10 +583,9 @@ class SearchResult(object):
 
     def merge(self, other):
         if self.book_id != other.book_id:
-            raise ValueError("this search result is or book %d; tried to merge with %d" % (self.book_id, other.book_id))
+            raise ValueError("this search result is for 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 += max(other._score, 0)
         return self
 
     def get_book(self):
@@ -728,17 +744,29 @@ class Search(SolrIndex):
 
         return q
 
+    def search_by_author(self, words):
+        from catalogue.models import Book
+        books = Book.objects.filter(parent=None).order_by('-popularity__count')
+        for word in words:
+            books = books.filter(cached_author__iregex='\m%s\M' % word).select_related('popularity__count')
+        return [SearchResult.from_book(book, how_found='search_by_author', query_terms=words) for book in books[:30]]
+
     def search_words(self, words, fields, book=True):
+        if book and fields == ['authors']:
+            return self.search_by_author(words)
         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 or (word not in stopwords):
+                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 not filters:
+            return []
         if book:
             query = self.index.query(is_book=True)
         else: