Update app.py

app.py

@@ -1,54 +1,76 @@
 import re
+from flask import Flask, request, jsonify
 import gradio as gr
 from transformers import pipeline
 
-# =========================
-# LOAD MODEL (ONCE)
-# =========================
+# ======================
+# MODEL
+# ======================
 classifier = pipeline(
     "zero-shot-classification",
     model="joeddav/xlm-roberta-large-xnli",
-    device=-1,
+    framework="pt",
     use_fast=False
 )
 
-# =========================
-# SEMANTIC EVENT DEFINITIONS
-# =========================
+# ======================
+# CONFIG
+# ======================
 SEMANTIC_EVENT = {
     "sadisme": [
         "kekerasan ekstrem yang menyebabkan penderitaan fisik parah pada manusia",
-        "penyiksaan tubuh manusia hingga kematian",
-        "pengorbanan manusia secara kejam"
+        "tubuh manusia dipotong atau dimutilasi",
+        "orang tua membunuh anaknya sendiri",
+        "manusia dikorbankan dengan cara kejam"
     ],
-    "kekerasan_fisik": [
+    "kekerasan fisik": [
         "tindakan fisik yang melukai tubuh manusia",
-        "serangan fisik yang menyebabkan cedera atau kematian"
+        "kekerasan yang menyebabkan cedera atau kematian"
     ],
-    "kekerasan_verbal": [
-        "ancaman serius terhadap seseorang",
-        "penghinaan keras dan intimidasi verbal"
+    "kekerasan verbal": [
+        "ancaman serius atau penghinaan keras"
     ],
     "seksual": [
-        "aktivitas seksual eksplisit antara manusia",
-        "tindakan seksual yang digambarkan secara jelas"
+        "aktivitas seksual eksplisit"
     ],
     "perjudian": [
-        "aktivitas perjudian dengan taruhan uang atau barang"
+        "aktivitas perjudian dengan taruhan"
     ],
     "narkoba": [
-        "penggunaan atau peredaran narkotika atau obat terlarang"
+        "penggunaan atau peredaran narkoba"
     ]
 }
 
-# =========================
-# UTILITIES
-# =========================
+EVENT_UNITS = {
+    "aktor": ["ayah", "ibu", "orang", "dia", "mereka"],
+    "korban": ["anak", "putri", "manusia"],
+    "aksi": ["membunuh", "menusuk", "memukul", "menyiksa", "memotong"],
+    "dampak": ["mati", "meninggal", "darah", "luka"]
+}
+
+METAPHOR_PAIRS = [
+    ("menusuk", "hati"),
+    ("terbunuh", "rindu"),
+    ("terluka", "perasaan")
+]
+
+THRESHOLD = {
+    "sadisme": 0.48,
+    "kekerasan fisik": 0.50,
+    "kekerasan verbal": 0.55,
+    "seksual": 0.58,
+    "perjudian": 0.60,
+    "narkoba": 0.60
+}
+
+# ======================
+# UTIL
+# ======================
 def split_kalimat(text):
     return [
-        s.strip()
-        for s in re.split(r'(?<=[.!?])\s+', text)
-        if len(s.strip()) >= 30
+        k.strip()
+        for k in re.split(r'(?<=[.!?])\s+', text)
+        if len(k.strip()) >= 30
     ]
 
 def sliding_window(kalimat, window=4):
@@ -57,62 +79,90 @@ def sliding_window(kalimat, window=4):
         for i in range(len(kalimat) - window + 1)
     ]
 
+def event_completeness(text):
+    score = 0
+    text = text.lower()
+    for words in EVENT_UNITS.values():
+        if any(w in text for w in words):
+            score += 1
+    return score
+
+def is_metaphor(text):
+    text = text.lower()
+    return any(a in text and b in text for a, b in METAPHOR_PAIRS)
+
 def rating_usia(kategori):
     if "sadisme" in kategori:
         return 21
     if any(k in kategori for k in ["seksual", "perjudian", "narkoba"]):
         return 17
-    if any(k in kategori for k in ["kekerasan_fisik", "kekerasan_verbal"]):
+    if any(k in kategori for k in ["kekerasan fisik", "kekerasan verbal"]):
         return 13
     return 0
 
-# =========================
-# CORE ANALYSIS FUNCTION
-# =========================
-def analyze_story(judul, isi):
-    if not isi or len(isi) < 100:
-        return {
-            "categories": [],
-            "age_rating": 0
-        }
-
+# ======================
+# CORE ANALYSIS
+# ======================
+def analyze_text(judul, isi):
     kalimat = split_kalimat(isi)
     windows = sliding_window(kalimat)
 
     detected = set()
 
-    for window in windows:
+    for w in windows:
         for kategori, desc in SEMANTIC_EVENT.items():
-            result = classifier(
-                window,
-                desc,
-                hypothesis_template="Teks ini mengandung {}."
-            )
-            if max(result["scores"]) >= 0.80:
-                detected.add(kategori)
+            res = classifier(w, desc, hypothesis_template="Teks ini menggambarkan {}.")
+            score = max(res["scores"])
+
+            if score < THRESHOLD[kategori]:
+                continue
+
+            if score < 0.60:
+                if event_completeness(w) < 2:
+                    continue
+                if is_metaphor(w):
+                    continue
+
+            detected.add(kategori)
 
     return {
-        "categories": sorted(list(detected)),
-        "age_rating": rating_usia(detected)
+        "rating_usia": rating_usia(detected),
+        "kategori": sorted(list(detected))
     }
 
-# =========================
-# GRADIO INTERFACE
-# =========================
-def api_wrapper(judul, isi):
-    result = analyze_story(judul, isi)
-    return result
+# ======================
+# FLASK API (FlutterFlow)
+# ======================
+app = Flask(__name__)
+
+@app.route("/analyze", methods=["POST"])
+def analyze_api():
+    data = request.json
+    result = analyze_text(data.get("judul", ""), data.get("isi", ""))
+    return jsonify(result)
+
+# ======================
+# GRADIO DEMO (JANGAN DIHILANGKAN)
+# ======================
+def gradio_analyze(judul, isi):
+    r = analyze_text(judul, isi)
+    return r["rating_usia"], ", ".join(r["kategori"])
 
 demo = gr.Interface(
-    fn=api_wrapper,
+    fn=gradio_analyze,
     inputs=[
-        gr.Textbox(label="Judul", lines=1),
-        gr.Textbox(label="Isi Cerita", lines=12)
+        gr.Textbox(label="Judul"),
+        gr.Textbox(label="Isi Cerita", lines=10)
+    ],
+    outputs=[
+        gr.Number(label="Rating Usia"),
+        gr.Textbox(label="Kategori")
     ],
-    outputs=gr.JSON(label="Hasil Analisis"),
-    title="READEAR — Semantic Content Rating",
-    description="Analisis konten cerita berbasis semantic (zero-shot NLP)"
+    title="Semantic Content Rating Demo"
 )
 
+# ======================
+# ENTRY POINT
+# ======================
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch(server_name="0.0.0.0", server_port=8080)