Separate before/after in diff; use better differ

app.py

@@ -9,6 +9,10 @@ from langchain.prompts import PromptTemplate
 from langchain.prompts.chat import ChatPromptTemplate, HumanMessagePromptTemplate
 from langchain.schema import HumanMessage
 
+from diff_match_patch import diff_match_patch
+import collections
+from itertools import islice
+
 def load_chain(api_key, prompt_template):
   api_key = openai.api_key if api_key == "" or api_key.isspace() else api_key
   if api_key:
@@ -26,29 +30,54 @@ def load_chain(api_key, prompt_template):
 
 def run(content, chain):
   edited = "\n".join([(chain.run(x) if should_check else x) for x, should_check in split_paragraphs(content)])
-  return diff_words(content, edited), edited
+  return diff_words(content, edited) + (edited,)
 
 def split_paragraphs(text):
   return [(x, x != "" and not x.startswith("#") and not x.isspace()) for x in text.split("\n")]
 
+def sliding_window(iterable, n): # from python itertools recipes
+    # sliding_window('ABCDEFG', 4) --> ABCD BCDE CDEF DEFG
+    it = iter(iterable)
+    window = collections.deque(islice(it, n), maxlen=n)
+    if len(window) == n:
+        yield tuple(window)
+    for x in it:
+        window.append(x)
+        yield tuple(window)
+
+dmp = diff_match_patch()
 def diff_words(content, edited):
-  content = re.split(r'(\s+|[,.!?"\':;]+)', content)
-  edited = re.split(r'(\s+|[,.!?"\':;]+)', edited)
-
-  output = []
-  for change, i1, i2, j1, j2 in SequenceMatcher(a=content, b=edited).get_opcodes():
-    if change == "equal":
-      output.append((get_parts(content, i1, i2), None))
-    elif change == "replace":
-      output.append((get_parts(content, i1, i2) + "→" + get_parts(edited, j1, j2), "→"))
-    elif change == "delete":
-      output.append((get_parts(content, i1, i2), "-"))
-    elif change == "insert":
-      output.append((get_parts(edited, j1, j2), "+"))
+  before = []
+  after = []
+  changes = []
+  change_count = 0
+  changed = False
+  diff = dmp.diff_main(content, edited)
+  dmp.diff_cleanupSemantic(diff)
+  diff += [(None, None)]
+  # print(diff)
+  for [(change, text), (next_change, next_text)] in sliding_window(diff, 2):
+    if change == 0:
+      before.append((text, None))
+      after.append((text, None))
     else:
-      raise Exception("Unknown change type: " + change)
-
-  return output
+      if change == -1 and next_change == 1:
+        change_count += 1
+        before.append((text, str(change_count)))
+        after.append((next_text, str(change_count)))
+        changes.append((text, next_text))
+        changed = True
+      elif change == -1:
+        before.append((text, "-"))
+      elif change == 1:
+        if changed:
+          changed = False
+        else:
+          after.append((text, "+"))
+      else:
+        raise Exception("Unknown change type: " + change)
+
+  return before, after, changes
 
 def get_parts(arr, start, end):
   return "".join(arr[start:end])
@@ -63,17 +92,26 @@ def run2(followup_question, chain1, llm):
 CHANGES = {
    "-": "remove",
    "+": "add",
-   "→": "change"
+  #  "→": "change"
 }
-def select_diff(evt: gr.SelectData):
+def select_diff(evt: gr.SelectData, changes):
   text, change = evt.value
   if not change:
     return
-  if change == "→":
-    original, edited = text.split("→")
-    return f"Why {CHANGES[change]} [{original}] to [{edited}]?"
-  return f"Why {CHANGES[change]} [{text}]?"
-
+  change_text = CHANGES.get(change, None)
+  if change_text:
+    return f"Why is it better to {change_text} [{text}]?"
+  # if change == "→":
+  else:
+    # clicked = evt.target
+    # if clicked.label == "Before":
+    #   original = text
+    # else:
+    #   edited = text
+
+    original, edited = changes[int(change) - 1]
+    # original, edited = text.split("→")
+    return f"Why is it better to change [{original}] to [{edited}]?"
 
 block = gr.Blocks()
 with block:
@@ -95,14 +133,21 @@ with block:
       # variant="secondary"
   ).style(full_width=False)
 
-  output = gr.HighlightedText(
-      label="Output",
-      color_map={
-          "-": "red",
-          "+": "green",
-          "→": "yellow",
-      }
-  )
+  with gr.Row():
+    output_before = gr.HighlightedText(
+        label="Before",
+        color_map={
+            "-": "red",
+            # "→": "yellow",
+        }
+    )
+    output_after = gr.HighlightedText(
+        label="After",
+        color_map={
+            "+": "green",
+            # "→": "yellow",
+        }
+    )
 
   followup_question = gr.Textbox(
       label="Follow-up Question",
@@ -114,6 +159,7 @@ with block:
       label="Answer"
   )
 
+  changes = gr.State()
   edited = gr.State()
   chain = gr.State()
   llm = gr.State()
@@ -122,16 +168,16 @@ with block:
   prompt.change(load_chain, [api_key, prompt], [chain, llm])
 
   inputs = [content, chain]
-  outputs = [output, edited]
+  outputs = [output_before, output_after, changes, edited]
   content.submit(run, inputs=inputs, outputs=outputs)
   submit.click(run, inputs=inputs, outputs=outputs)
 
-  output.select(select_diff, None, followup_question)
+  output_before.select(select_diff, changes, followup_question)
+  output_after.select(select_diff, changes, followup_question)
 
   inputs2 = [followup_question, chain, llm]
   outputs2 = followup_answer
   followup_question.submit(run2, inputs2, outputs2)
   followup_submit.click(run2, inputs2, outputs2)
 
-
 block.launch(debug=True)