Merge branch 'master' of github.com:wo/opp-tools

wo · Jul 7, 2016 · e7f0c47 · e7f0c47
2 parents 43fb194 + 4deb747
commit e7f0c47
Showing 1 changed file with 9 additions and 8 deletions.
diff --git a/opp/blogpostprocessor.py b/opp/blogpostprocessor.py
@@ -53,14 +53,15 @@ def process_blogpost(doc):
         doc.hidden = True
 
     # categorize:
-    for (cat_id, cat) in categories():
-        clf = classifier.get_classifier(cat)
-        try:
-            strength = int(clf.classify(doc) * 100)
-            debug(3, "%s score %s", cat, strength)
-        except UntrainedClassifierException as e:
-            continue 
-        doc.assign_category(cat_id, strength)
+    if doc.numwords > 1000:
+        for (cat_id, cat) in categories():
+            clf = classifier.get_classifier(cat)
+            try:
+                strength = int(clf.classify(doc) * 100)
+                debug(3, "%s score %s", cat, strength)
+            except UntrainedClassifierException as e:
+                continue 
+            doc.assign_category(cat_id, strength)
 
     doc.found_date = datetime.now()
     doc.status = 1