abstract snip

app.py

@@ -107,11 +107,25 @@ def find_source(text, docs, matched):
             if matched and remove_html(doc[3]).strip() != matched.strip():
                     continue
             new_text = text
-            for sent in nltk.sent_tokenize(remove_html(doc[3])):
+            sent_loc = None
+            sents = nltk.sent_tokenize(remove_html(doc[3]))
+            for i, sent in enumerate(sents):
                 if text in sent:
                     new_text = sent
+                    sent_loc = i
+
+            context = remove_html(doc[3]).replace('<strong class="highlight">', '').replace('</strong>', '')
+            if sent_loc:
+                context_len = 2
+                sent_beg = sent_loc - context_len
+                if sent_beg <= 0: sent_beg = 0
+                sent_end = sent_loc + context_len
+                if sent_end >= len(sents):
+                    sent_end = len(sents)
+                context = ''.join(sents[sent_beg:sent_end])
+
             return {
-                    'citation_statement': new_text if text != new_text else remove_html(doc[3]).replace('<strong class="highlight">', '').replace('</strong>', ''),
+                    'citation_statement': context,
                     'text': new_text,
                     'from': doc[0],
                     'supporting': doc[0],