Update app.py

app.py

@@ -12,7 +12,7 @@ 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 settings ---
-OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")  # HF Spaces: Settings → Variables & secrets
+OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")  # Add in HF Spaces: Settings → Variables & secrets
 PREFERRED_OPENAI_MODEL = os.getenv("OPENAI_MODEL", "gpt-5o")   # try this first
 FALLBACK_OPENAI_MODEL  = "gpt-4o-mini"                         # fallback
 OPENAI_CHAT_URL        = "https://api.openai.com/v1/chat/completions"
@@ -42,7 +42,10 @@ def normalize_url(url: str) -> str:
         return "https://" + url
     return url
 
-def get_text_blocks(url):
+def to_plain_text(html_or_text: str) -> str:
+    return BeautifulSoup(html_or_text, "html.parser").get_text(separator=" ", strip=True)
+
+def get_text_blocks(url: str):
     resp = requests.get(url, timeout=20, headers=UA)
     resp.raise_for_status()
     soup = BeautifulSoup(resp.text, "html.parser")
@@ -67,19 +70,22 @@ def embed(texts):
     return mean_pool(out.last_hidden_state, batch["attention_mask"])
 
 def wrap_existing_anchor(sentence: str, anchor_text: str, target_url: str) -> str:
-    """
-    If the sentence already contains the anchor text, wrap the FIRST occurrence with <a>.
-    Case-insensitive; keeps original casing in the sentence.
-    """
+    """If the sentence already contains the anchor text (case-insensitive), wrap the first match with <a>."""
     pattern = re.compile(re.escape(anchor_text), re.IGNORECASE)
     return pattern.sub(f'<a href="{target_url}">{anchor_text}</a>', sentence, count=1)
 
+def anchor_in_text(text: str, anchor_text: str) -> bool:
+    return re.search(re.escape(anchor_text), text or "", flags=re.IGNORECASE) is not None
+
+# =========================
+# LinkBERT selection
+# =========================
 def suggest_insertions(source_url, target_url, anchor_text, top_k=1):
     blocks = get_text_blocks(source_url)
     if not blocks:
         return [{"error":"No text blocks found on the page."}]
 
-    # Target context → helps similarity
+    # Target context to bias similarity
     try:
         tgt_html = requests.get(target_url, timeout=20, headers=UA).text
         tt = BeautifulSoup(tgt_html, "html.parser").title
@@ -89,7 +95,6 @@ def suggest_insertions(source_url, target_url, anchor_text, top_k=1):
 
     ext = tldextract.extract(target_url)
     tgt_domain = ".".join([p for p in [ext.domain, ext.suffix] if p])
-
     query = f"{anchor_text} — relevant to: {tgt_title} ({tgt_domain})"
     q_emb = embed([query])[0]
 
@@ -99,30 +104,28 @@ def suggest_insertions(source_url, target_url, anchor_text, top_k=1):
 
     results = []
     for idx in top_idx:
-        blk = blocks[idx]
-        sents = re.split(r'(?<=[.!?])\s+', blk)
+        paragraph = blocks[idx]
+        sents = re.split(r'(?<=[.!?])\s+', paragraph)
         s_embs = embed(sents)
         s_sims = F.cosine_similarity(s_embs, q_emb.repeat(len(sents),1))
         si = int(torch.argmax(s_sims))
         best_sent = sents[si]
 
-        # Check if anchor text already appears in the chosen sentence
-        if re.search(re.escape(anchor_text), best_sent, flags=re.IGNORECASE):
+        if anchor_in_text(best_sent, anchor_text):
             wrapped = wrap_existing_anchor(best_sent, anchor_text, target_url)
             results.append({
                 "anchor_was_present": True,
+                "best_paragraph": paragraph,
                 "best_sentence_original": best_sent,
-                "best_sentence_with_anchor": wrapped,
-                "best_paragraph": blk
+                "best_sentence_with_anchor": wrapped
             })
         else:
-            # No injection here — just return the raw original + paragraph;
-            # GPT will do the smart integration.
+            # No forced injection here. We will let GPT rebuild the sentence naturally.
             results.append({
                 "anchor_was_present": False,
+                "best_paragraph": paragraph,
                 "best_sentence_original": best_sent,
-                "best_sentence_with_anchor": None,  # GPT will produce
-                "best_paragraph": blk
+                "best_sentence_with_anchor": None
             })
     return results
 
@@ -146,49 +149,65 @@ def _openai_chat(model_name: str, system: str, user_json: dict):
     txt = r.json()["choices"][0]["message"]["content"]
     return json.loads(txt)
 
+def _mk_anchor_html(target_url: str, anchor_text: str) -> str:
+    return f'<a href="{target_url}">{anchor_text}</a>'
+
+def _ensure_placeholder(text: str) -> str:
+    """
+    If GPT forgets the [[ANCHOR]] placeholder,
+    add a neutral second sentence with it so we can safely inject the link.
+    """
+    text = (text or "").strip()
+    if "[[ANCHOR]]" not in text:
+        if text and not text.endswith((".", "!", "?")):
+            text += "."
+        text += " See [[ANCHOR]]."
+    return text
+
 # =========================
-# GPT editor (single step; may expand to 1–3 sentences)
+# GPT editor: integrate anchor with placeholder
 # =========================
 def gpt_integrate_anchor(paragraph_text: str,
                          original_sentence: str,
                          anchor_text: str,
                          target_url: str):
     """
-    Replace the original sentence with 1–3 natural sentences that integrate the anchor text.
-    Keep EXACT anchor text and SAME href. No em dashes; avoid clichés.
-    Returns: {"replacement_html": "<html>"}  (1–3 sentences)
+    Ask GPT to replace the original sentence with 1–3 natural sentences.
+    It must place EXACTLY ONE placeholder [[ANCHOR]] where the link should go.
+    Then we inject the exact <a href="...">anchor_text</a> ourselves.
     """
+    anchor_html = _mk_anchor_html(target_url, anchor_text)
+
     if not OPENAI_API_KEY:
-        # Fallback if no key: minimally append a neutral clause (not ideal)
+        # Fallback when no key
         base = original_sentence.rstrip(".!? ")
-        draft = f'{base} (see <a href="{target_url}">{anchor_text}</a>).'
-        return {"replacement_html": draft}
+        draft = f"{base}. See [[ANCHOR]]."
+        return {"replacement_html": draft.replace("[[ANCHOR]]", anchor_html)}
 
     system = (
-        "You are an elite human content editor.\n"
-        "INPUT:\n"
+        "You are an elite human content editor.\n\n"
+        "INPUT FIELDS:\n"
         "• paragraph_text: the full paragraph from an article\n"
-        "• original_sentence: the specific sentence inside that paragraph we will replace\n"
-        "• anchor_text and target_url: must appear EXACTLY once as <a href=\"URL\">ANCHOR</a>\n\n"
+        "• original_sentence: the specific sentence inside that paragraph we will replace\n\n"
         "TASK:\n"
-        "Rewrite the original sentence so the result reads 100% natural. You may expand to 1–3 sentences if that\n"
-        "improves clarity and flow. Integrate the anchor seamlessly (do NOT bolt it on awkwardly).\n\n"
-        "STRICT RULES:\n"
-        "• Keep the EXACT anchor_text inside the <a> tag; keep the SAME href.\n"
+        "Rewrite the original sentence so the result reads 100% natural IN CONTEXT. "
+        "You may expand to 1–3 sentences if that improves clarity and flow.\n\n"
+        "CRITICAL INSTRUCTION:\n"
+        "• Insert the literal placeholder [[ANCHOR]] EXACTLY ONCE where the link should appear. "
+        "Do NOT write any HTML for the link—only [[ANCHOR]].\n\n"
+        "STYLE RULES:\n"
         "• No em dashes or '--'. Avoid clichés like 'for details', 'click here', 'learn more', 'visit', 'read more'.\n"
         "• Prefer simple connectors ('with', 'from', 'in', 'at') over 'via' or 'through'.\n"
-        "• Tone: clear, professional blog prose. 1–3 sentences total.\n"
-        "• Output JSON with key 'replacement_html' only."
+        "• Tone: clear, professional blog prose. Keep it tight (1–3 sentences total).\n\n"
+        "OUTPUT:\n"
+        "Return JSON with key 'replacement_html' only, where [[ANCHOR]] appears exactly once."
     )
 
     user = {
         "paragraph_text": paragraph_text,
         "original_sentence": original_sentence,
-        "anchor_text": anchor_text,
-        "target_url": target_url
     }
 
-    # Try preferred model, then fallback
     try:
         obj = _openai_chat(PREFERRED_OPENAI_MODEL, system, user)
     except Exception:
@@ -196,38 +215,22 @@ def gpt_integrate_anchor(paragraph_text: str,
             obj = _openai_chat(FALLBACK_OPENAI_MODEL, system, user)
         except Exception:
             base = original_sentence.rstrip(".!? ")
-            draft = f'{base} (see <a href="{target_url}">{anchor_text}</a>).'
-            return {"replacement_html": draft}
-
-    out = obj.get("replacement_html", "").strip() or original_sentence
-
-    # Safety: ensure anchor text + href exist in output
-    soup = BeautifulSoup(out, "html.parser")
-    a = soup.find("a")
-    if not a:
-        return {"replacement_html": original_sentence}
-    if anchor_text.lower() not in a.get_text().lower():
-        return {"replacement_html": original_sentence}
-    if target_url.lower() not in (a.get("href") or "").lower():
-        return {"replacement_html": original_sentence}
-
-    # Keep result reasonably short (<= ~3 sentences)
-    plain = soup.get_text(" ", strip=True)
-    # (we don't hard block multiple sentences; model already instructed 1–3)
+            draft = f"{base}. See [[ANCHOR]]."
+            return {"replacement_html": draft.replace("[[ANCHOR]]", anchor_html)}
 
+    out = obj.get("replacement_html", "").strip()
+    out = _ensure_placeholder(out)
+    out = out.replace("[[ANCHOR]]", anchor_html)
     return {"replacement_html": out}
 
-# ---------- Plain-text helper (preserve spacing between tags)
-def to_plain_text(html_or_text):
-    return BeautifulSoup(html_or_text, "html.parser").get_text(separator=" ", strip=True)
-
 # =========================
-# Gradio UI
+# Gradio UI flow
 # =========================
 def run_tool(source_url, target_url, anchor_text, smart_rewrite, plain_text):
     if not source_url or not target_url or not anchor_text:
         return "❌ Please provide Source URL, Target URL, and Anchor Text."
 
+    # Auto-correct swapped inputs
     warn = ""
     if looks_like_url(anchor_text) and not looks_like_url(target_url):
         anchor_text, target_url = target_url, anchor_text
@@ -239,13 +242,13 @@ def run_tool(source_url, target_url, anchor_text, smart_rewrite, plain_text):
     if "error" in res:
         return f"❌ {res['error']}"
 
-    # Case A: anchor already present → just wrap it and show where to add
+    # Case A: anchor already present in the chosen sentence → just wrap and return
     if res.get("anchor_was_present", False):
         wrapped_html = res["best_sentence_with_anchor"]
         final_output = to_plain_text(wrapped_html) if plain_text else wrapped_html
         return warn + f"✅ Add link here:\n\n{final_output}"
 
-    # Case B: anchor not present → ask GPT to integrate naturally (may expand 1–3 sentences)
+    # Case B: anchor not present → GPT integrates naturally with paragraph context
     original_sentence = res["best_sentence_original"]
     paragraph = res.get("best_paragraph", original_sentence)
 
@@ -253,14 +256,16 @@ def run_tool(source_url, target_url, anchor_text, smart_rewrite, plain_text):
         edit = gpt_integrate_anchor(paragraph, original_sentence, anchor_text, target_url)
         final_html = edit["replacement_html"]
     else:
-        # deterministic fallback when GPT off
+        # deterministic fallback when GPT is off
         base = original_sentence.rstrip(".!? ")
-        final_html = f'{base} (see <a href="{target_url}">{anchor_text}</a>).'
+        final_html = f'{base}. See {_mk_anchor_html(target_url, anchor_text)}.'
 
     final_output = to_plain_text(final_html) if plain_text else final_html
     return warn + f"Change this sentence:\n\n{original_sentence}\n\nWith this one:\n\n{final_output}"
 
-# Show GPT status in the header
+# =========================
+# Gradio app
+# =========================
 gpt_status = "ON" if OPENAI_API_KEY else "OFF"
 title_model = PREFERRED_OPENAI_MODEL if OPENAI_API_KEY else "OFF"
 
@@ -275,7 +280,7 @@ demo = gr.Interface(
     ],
     outputs=gr.Textbox(label="Result", lines=12),
     title=f"Link Insertion Helper • GPT: {gpt_status} • Model: {title_model}",
-    description="Suggests the best place to add your link and returns one clean instruction. If the anchor isn't present, GPT rewrites the sentence (1–3 sentences) to integrate it naturally."
+    description="Finds the best place to add your link. If the anchor isn't present, GPT rewrites the sentence (1–3 sentences) using a placeholder and the app injects the exact link."
 )
 
 if __name__ == "__main__":