Update ppt_chunker.py

ppt_chunker.py

@@ -5,7 +5,7 @@ from unstructured.partition.pptx import partition_pptx
 
 from ordered_multimap import OrderedMultiIndexMapWeakRef
 
-WRONG_NOUNS = {"BY", "IN", "ON", "INTO", "A", "AT", "WITH", "FROM", "TO", "AND", "OR", "BUT", "OWN", 'BEHIND', 'THROUGH', 'FIERCE', "'S"}
+WRONG_NOUNS = {"BY", "IN", "ON", "INTO", "A", "AT", "WITH", "FROM", "TO", "AND", "OR", "BUT", "OWN", 'BEHIND', 'THROUGH', 'FIERCE', "'S", '&'}
 NON_ENDING_PUNCT = {',', ':', ';', "'", '/', '-'}
 
 def process_chunk(chunk, nlp):
@@ -92,21 +92,27 @@ def ppt_chunk(file_like, nlp):
     while j < len(chunks):       
         new_sub_chunks = []
         only_tables = True
-        print(f'TEST: {chunks[j]}')
-        for sub_chunk in chunk[1]:
-            print(f'TEST 2: {sub_chunk}')
-            if (type(sub_chunk) is list) and (sub_chunk[0] == 'Table'):
-                print('HEEEERE')
-                tables.append([chunk[0], sub_chunk])
+        title = ''
+        
+        for i, sub_chunk in enumerate(chunk[1]):
+            print(f'TEST : {sub_chunk}')
+            if (i == 0) and ((sub_chunk[0] == 'Title') or (sub_chunk[0] == 'UncategorizedText')):
+                title = sub_chunk[1]
+                
+            if sub_chunk[0] == 'Table':
+                if title != '':
+                    tables.append([chunk[0], title, sub_chunk])
+                else:
+                    tables.append([chunk[0], sub_chunk])
             else:
                 new_sub_chunks.append(sub_chunk)
                 only_tables = False
-                
+        
         if only_tables:
             del chunks[j]
+            j += 1
         else:
             chunks[j] = [chunk[0], new_sub_chunks]
-            j += 1
         
     er = time.time()
     fr = er - s