Upload output_formatter.py

evaluation/output_formatter.py

@@ -0,0 +1,93 @@
+"""
+formats output text file from llampia 
+into markdown tables  
+"""
+
+import os 
+
+def format_edu(turn_str):
+    no = turn_str.split('<')[0].strip()
+    text = turn_str.split('>')[1].strip()
+    speaker = 'BUIL'
+    if 'Arch' in turn_str.split('>')[0]:
+        speaker = 'ARCH'
+    edu_str = ' ' + '**' + no + ' ' + speaker + '**' + ' ' + text
+    return edu_str
+
+current_folder=os.getcwd()
+
+output_path = '/path/to/test-output-file.txt'
+table_path = current_folder + '/test_predictions.md'
+
+with open(output_path, 'r') as txt:
+    text = txt.read().split('\n')
+
+group = 0
+pairs = []
+pair = []
+for t in text:
+    if t.startswith('New Turn'):   
+        group = 1
+        pair.append(t)
+    elif t.startswith(' ### DS:'):
+        pair.append(t)
+        pairs.append(pair)
+        pair = []
+        group = 0
+    elif group == 1:
+        pair.append(t)
+
+tables = []
+rows = []
+dial_str = ''
+struct_str = ''
+for pair in pairs:
+    #print(pair)
+    if pair[0].startswith('New Turn: 1 <'):
+        if len(rows) > 0:
+            tables.append(rows)
+            rows = []
+        dial_str = '**0 BUIL** Mission has started.' 
+        dial_str += format_edu(pair[0].lstrip('New Turn: '))
+        for p in pair[1:]:
+            if p.startswith(' ###'):
+                struct_str = p.lstrip(' ### DS:')
+                rows.append([dial_str, struct_str])
+                # dial_str = ''
+                # struct_str = ''
+            else:
+                dial_str += format_edu(p)
+    elif pair[0].startswith('New Turn: '):
+        dial_str = format_edu(pair[0].lstrip('New Turn: '))
+        #print(dial_str)
+        for p in pair[1:]:
+            if p.startswith(' ###'):
+                struct_str = p.lstrip(' ### DS:')
+                # print(struct_str)
+                # print(dial_str)
+                rows.append([dial_str, struct_str])
+                # dial_str = ''
+                # struct_str = ''
+            else:
+                dial_str += format_edu(p)
+
+all_md_tables = []
+
+for table in tables:
+    table_rows = ['| Dialogue | Structure |', '| ----- | ----- |']
+    for tr in table:
+        st = '| ' + tr[0] + ' | ' + tr[1] + ' |'
+        table_rows.append(st)
+    all_md_tables.extend(table_rows)
+    all_md_tables.extend(' ')
+
+md_tables_string = '\n'.join(all_md_tables)
+
+
+f = open(table_path, 'w')
+for r in all_md_tables:
+    print(r)
+    print(r, file=f)
+print("markdown printed")
+
+