Update app.py

app.py

@@ -7,6 +7,7 @@ from keybert import KeyBERT
 from textblob import TextBlob
 from transformers import pipeline
 import textstat
+import trafilatura
 
 # constants
 MAX_TEXT_LENGTH = 1500
@@ -36,6 +37,18 @@ def _load_nlp_models() -> typing.Dict[str, typing.Any]:
         aggregation_strategy="simple"
     )
     keyword_extractor = KeyBERT(model="all-mpnet-base-v2")
+    nli_classifier = pipeline(
+        "text-classification", 
+        model="roberta-large-mnli" 
+    )
+
+    return {
+        "sentiment": sentiment_analyzer,
+        "emotion": emotion_classifier,
+        "ner": ner_extractor,
+        "keyword": keyword_extractor,
+        "nli": nli_classifier,
+    }
 
     return {
         "sentiment": sentiment_analyzer,
@@ -165,6 +178,27 @@ def _highlight_keywords(text: str, keywords: typing.List[str]) -> str:
     return highlighted_text
 
 
+def fetch_article_text(url: str) -> str:
+    """Scrapes clean article text from a given URL."""
+    downloaded = trafilatura.fetch_url(url)
+    if downloaded:
+        text = trafilatura.extract(downloaded)
+        return text if text else "Error: Could not extract text."
+    return "Error: Could not fetch URL. It might be protected."
+
+
+def check_contradiction(text_a: str, text_b: str) -> dict:
+    """Uses NLI to see if the arguments of the articles contradict."""
+    models = _load_nlp_models()
+    premise = text_a[:400]
+    hypothesis = text_b[:400]
+    
+    nli_input = f"{premise} </s></s> {hypothesis}"
+    result = models["nli"](nli_input)[0]
+    
+    return {"relationship": result["label"], "confidence": result["score"]}
+    
+
 # STATE MANAGEMENT
 if "results_a" not in st.session_state:
     st.session_state.results_a = None
@@ -181,13 +215,23 @@ st.divider()
 with st.spinner("Starting NLP models."):
     _load_nlp_models()
 
+input_method = st.radio("Input Method", ["Paste Text", "Paste URL"], horizontal=True)
+
 col1, col2 = st.columns(2)
 
 with col1:
-    user_article_a = st.text_area("Data Source A", value=ARTICLE_A.strip(), height=220)
+    if input_method == "Paste Text":
+        user_article_a = st.text_area("Data Source A", value=ARTICLE_A.strip(), height=220)
+    else:
+        url_a = st.text_input("Source A URL")
+        user_article_a = fetch_article_text(url_a) if url_a else ""
 
 with col2:
-    user_article_b = st.text_area("Data Source B", value=ARTICLE_B.strip(), height=220)
+    if input_method == "Paste Text":
+        user_article_b = st.text_area("Data Source B", value=ARTICLE_B.strip(), height=220)
+    else:
+        url_b = st.text_input("Source B URL")
+        user_article_b = fetch_article_text(url_b) if url_b else ""
 
 st.write("")
 
@@ -202,7 +246,15 @@ if st.session_state.results_a and st.session_state.results_b:
     st.divider()
     st.markdown("### Framing Analytics & Comparison")
     
-    # Radar Chart spans the top
+    # Display Contradictions
+    nli_result = check_contradiction(user_article_a, user_article_b)
+    if nli_result["relationship"] == "CONTRADICTION":
+        st.error(f"**NARRATIVE CONTRADICTION** (Confidence: {nli_result['confidence']:.2f}) - These sources are actively disputing each other's foundational facts.")
+    elif nli_result["relationship"] == "ENTAILMENT":
+        st.success(f"**NARRATIVE ALIGNMENT** (Confidence: {nli_result['confidence']:.2f}) - These sources agree on the core premise.")
+    else:
+        st.info(f"**NEUTRAL RELATIONSHIP** - These sources are discussing the topic without direct contradiction or alignment.")
+    
     st.plotly_chart(_create_comparison_radar_chart(st.session_state.results_a, st.session_state.results_b), use_container_width=True)
 
     res_col1, res_col2 = st.columns(2)