Update app.py

app.py

@@ -11,9 +11,10 @@ import gradio as gr
 MODEL = "michiyasunaga/LinkBERT-base"
 UA = {"User-Agent": "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0 Safari/537.36"}
 
-OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")  # set in HF Spaces: Settings → Variables & secrets
-OPENAI_URL = "https://api.openai.com/v1/responses"     # OpenAI Responses API
-OPENAI_MODEL = "gpt-4o"
+# OpenAI (Chat Completions)
+OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")  # add in HF Spaces: Settings → Variables & secrets
+OPENAI_MODEL = "gpt-4o-mini"                  # fast & good; you can switch to "gpt-4o"
+OPENAI_CHAT_URL = "https://api.openai.com/v1/chat/completions"
 
 # =========================
 # Load LinkBERT
@@ -57,8 +58,9 @@ def inject_anchor_into_sentence(sentence, anchor_text, target_url):
         pattern = re.compile(re.escape(anchor_text), re.IGNORECASE)
         return pattern.sub(f'<a href="{target_url}">{anchor_text}</a>', sentence), True
 
-    # add short clause non-spammy
+    # add short clause (non-spammy)
     base, punct = (sentence[:-1], sentence[-1]) if sentence[-1:] in ".!?" else (sentence, ".")
+    # NOTE: spaces around <a> so plain-text stripping preserves spacing
     rewritten = f'{base} — see <a href="{target_url}">{anchor_text}</a> for details{punct}'
     return rewritten, False
 
@@ -101,20 +103,19 @@ def suggest_insertions(source_url, target_url, anchor_text, top_k=1):
         })
     return results
 
-# ---------- GPT rewrite (optional)
+# ---------- GPT rewrite (Chat Completions)
 def gpt_rewrite(sentence_html, anchor_text, target_url, style="neutral"):
     """Improve HTML sentence via OpenAI; fail-soft to the provided sentence."""
     if not OPENAI_API_KEY:
-        return {
-            "sentence_html": sentence_html,
-            "micro_paragraph_html": f'For more detail, see <a href="{target_url}">{anchor_text}</a>.'
-        }
+        print("[GPT] No OPENAI_API_KEY found → using fallback.")
+        return {"sentence_html": sentence_html}
 
+    print("[GPT] Calling OpenAI Chat Completions...")
+    headers = {"Authorization": f"Bearer {OPENAI_API_KEY}", "Content-Type": "application/json"}
     system = (
         "You are a precise editor. Rewrite minimally for clarity and flow. "
-        "Output valid HTML only (no markdown). "
-        "Must include an <a href> with the exact anchor text provided. "
-        "Return a compact JSON object with keys sentence_html and micro_paragraph_html."
+        "Output valid HTML only (no markdown). Must include an <a href> tag using the exact anchor text. "
+        "Return a compact JSON object with key sentence_html only."
     )
     user = {
         "task": "rewrite_for_link_insertion",
@@ -124,34 +125,30 @@ def gpt_rewrite(sentence_html, anchor_text, target_url, style="neutral"):
         "style": style,
         "constraints": {"max_extra_words": 20, "avoid": ["spammy phrasing", "overclaiming"]}
     }
-    headers = {"Authorization": f"Bearer {OPENAI_API_KEY}", "Content-Type": "application/json"}
     body = {
         "model": OPENAI_MODEL,
-        "input": [
-            {"role":"system", "content": system},
-            {"role":"user", "content": json.dumps(user)}
+        "response_format": {"type": "json_object"},
+        "messages": [
+            {"role": "system", "content": system},
+            {"role": "user", "content": json.dumps(user)}
         ],
-        "response_format": {"type":"json_object"}
+        "temperature": 0.2
     }
     try:
-        r = requests.post(OPENAI_URL, headers=headers, data=json.dumps(body), timeout=60)
+        r = requests.post(OPENAI_CHAT_URL, headers=headers, json=body, timeout=60)
+        print(f"[GPT] HTTP {r.status_code}")
         r.raise_for_status()
-        text = r.json().get("output", "")
-        obj = json.loads(text)
-        return {
-            "sentence_html": obj.get("sentence_html", sentence_html),
-            "micro_paragraph_html": obj.get("micro_paragraph_html", f'For more detail, see <a href="{target_url}">{anchor_text}</a>.')
-        }
-    except Exception:
-        return {
-            "sentence_html": sentence_html,
-            "micro_paragraph_html": f'For more detail, see <a href="{target_url}">{anchor_text}</a>.'
-        }
-
-# ---------- Plain-text helper (strip HTML tags)
+        txt = r.json()["choices"][0]["message"]["content"]
+        obj = json.loads(txt)
+        return {"sentence_html": obj.get("sentence_html", sentence_html)}
+    except Exception as e:
+        print(f"[GPT] Error: {e}")
+        return {"sentence_html": sentence_html}
+
+# ---------- Plain-text helper (fixes spacing)
 def to_plain_text(html_or_text):
-    # Converts HTML to plain text (keeps inner anchor text, removes <a ...>)
-    return BeautifulSoup(html_or_text, "html.parser").get_text(separator="", strip=True)
+    # IMPORTANT: separator=" " inserts spaces between tag boundaries -> avoids "seeUFABETfor"
+    return BeautifulSoup(html_or_text, "html.parser").get_text(separator=" ", strip=True)
 
 # =========================
 # Gradio UI
@@ -164,28 +161,27 @@ def run_tool(source_url, target_url, anchor_text, smart_rewrite, plain_text):
     if "error" in res:
         return f"❌ {res['error']}"
 
-    # Choose sentence to polish (already has <a> if anchor existed)
     draft_html = res["best_sentence_with_anchor"]
 
+    # Optionally pass through GPT for a cleaner sentence
     if smart_rewrite:
         g = gpt_rewrite(draft_html, anchor_text, target_url, style="neutral")
         final_html = g["sentence_html"]
     else:
         final_html = draft_html
 
-    # Optionally return plain text (no <a>, no tags)
+    # Optionally convert to plain text (no <a>, no tags), with proper spacing
     final_output = to_plain_text(final_html) if plain_text else final_html
 
     if res.get("anchor_was_present", False):
         return f"✅ Add link here:\n\n{final_output}"
     else:
-        # show original (plain text, since it never had a link)
         original_sentence = res['best_sentence_original']
-        if plain_text:
-            # ensure the replacement is also plain
-            return f"Change this sentence:\n\n{original_sentence}\n\nWith this one:\n\n{final_output}"
-        else:
-            return f"Change this sentence:\n\n{original_sentence}\n\nWith this one:\n\n{final_output}"
+        # show original (never had link) + final replacement (HTML or plain, per toggle)
+        return f"Change this sentence:\n\n{original_sentence}\n\nWith this one:\n\n{final_output}"
+
+# Show GPT status in the header
+gpt_status = "ON" if OPENAI_API_KEY else "OFF"
 
 demo = gr.Interface(
     fn=run_tool,
@@ -197,7 +193,7 @@ demo = gr.Interface(
         gr.Checkbox(label="Plain text (no URL)", value=False)
     ],
     outputs=gr.Textbox(label="Result", lines=12),
-    title="Link Insertion Helper",
+    title=f"Link Insertion Helper • GPT: {gpt_status}",
     description="Suggests the best place to add your link and returns one clean instruction. Toggle GPT and Plain text (no URL) as needed."
 )