Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -16,7 +16,7 @@ def log(msg):
|
|
| 16 |
print(msg, flush=True)
|
| 17 |
|
| 18 |
# ======================
|
| 19 |
-
# MODEL 1 β RATING & KATEGORI (ASLI
|
| 20 |
# ======================
|
| 21 |
classifier = pipeline(
|
| 22 |
"zero-shot-classification",
|
|
@@ -40,7 +40,7 @@ abusive_model.to(device)
|
|
| 40 |
id2label = {0: "Tidak Abusif", 1: "Abusif"}
|
| 41 |
|
| 42 |
# ======================
|
| 43 |
-
# CONFIG (ASLI
|
| 44 |
# ======================
|
| 45 |
SEMANTIC_EVENT = {
|
| 46 |
"sadisme": [
|
|
@@ -99,6 +99,22 @@ def split_kalimat(text):
|
|
| 99 |
if len(k.strip()) >= 20
|
| 100 |
]
|
| 101 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
def rating_usia(kategori):
|
| 103 |
if "sadisme" in kategori:
|
| 104 |
return 21
|
|
@@ -109,13 +125,21 @@ def rating_usia(kategori):
|
|
| 109 |
return 0
|
| 110 |
|
| 111 |
# ======================
|
| 112 |
-
#
|
| 113 |
# ======================
|
| 114 |
def analyze_text(judul, isi):
|
| 115 |
kalimat = split_kalimat(isi)
|
|
|
|
|
|
|
| 116 |
detected = set()
|
| 117 |
|
| 118 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 119 |
for kategori, desc in SEMANTIC_EVENT.items():
|
| 120 |
res = classifier(
|
| 121 |
w,
|
|
@@ -123,14 +147,26 @@ def analyze_text(judul, isi):
|
|
| 123 |
hypothesis_template="Teks ini menggambarkan {}."
|
| 124 |
)
|
| 125 |
score = max(res["scores"])
|
| 126 |
-
|
| 127 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 128 |
|
| 129 |
usia = rating_usia(detected)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
return usia, ", ".join(sorted(detected))
|
| 131 |
|
| 132 |
# ======================
|
| 133 |
-
#
|
| 134 |
# ======================
|
| 135 |
def filter_abusive(isi):
|
| 136 |
paragraphs = re.split(r'\n+', isi)
|
|
@@ -170,9 +206,9 @@ def filter_abusive(isi):
|
|
| 170 |
return "\n".join(output)
|
| 171 |
|
| 172 |
# ======================
|
| 173 |
-
# ROUTER
|
| 174 |
# ======================
|
| 175 |
-
def
|
| 176 |
if mode == "rating":
|
| 177 |
usia, kategori = analyze_text(judul, isi)
|
| 178 |
return {
|
|
@@ -190,46 +226,26 @@ def api_router(judul, isi, mode):
|
|
| 190 |
}
|
| 191 |
|
| 192 |
# ======================
|
| 193 |
-
#
|
| 194 |
# ======================
|
| 195 |
demo = gr.Interface(
|
| 196 |
-
fn=
|
| 197 |
-
inputs=[
|
| 198 |
-
gr.Textbox(label="Judul"),
|
| 199 |
-
gr.Textbox(label="Isi Cerita", lines=10)
|
| 200 |
-
],
|
| 201 |
-
outputs=[
|
| 202 |
-
gr.Number(label="Rating Usia"),
|
| 203 |
-
gr.Textbox(label="Kategori")
|
| 204 |
-
],
|
| 205 |
-
title="READEAR",
|
| 206 |
-
api_name="analyze"
|
| 207 |
-
)
|
| 208 |
-
|
| 209 |
-
# ======================
|
| 210 |
-
# API UNTUK RAILWAY
|
| 211 |
-
# ======================
|
| 212 |
-
api = gr.Interface(
|
| 213 |
-
fn=api_router,
|
| 214 |
inputs=[
|
| 215 |
gr.Textbox(),
|
| 216 |
gr.Textbox(),
|
| 217 |
gr.Radio(["rating", "abusive"])
|
| 218 |
],
|
| 219 |
outputs=gr.JSON(),
|
| 220 |
-
api_name="
|
| 221 |
)
|
| 222 |
|
| 223 |
# ======================
|
| 224 |
# ENTRY POINT
|
| 225 |
# ======================
|
| 226 |
if __name__ == "__main__":
|
| 227 |
-
|
| 228 |
-
[demo, api],
|
| 229 |
-
["UI", "API"]
|
| 230 |
-
).launch(
|
| 231 |
server_name="0.0.0.0",
|
| 232 |
server_port=7860,
|
| 233 |
ssr_mode=False,
|
| 234 |
show_error=True
|
| 235 |
-
)
|
|
|
|
| 16 |
print(msg, flush=True)
|
| 17 |
|
| 18 |
# ======================
|
| 19 |
+
# MODEL 1 β RATING & KATEGORI (ASLI)
|
| 20 |
# ======================
|
| 21 |
classifier = pipeline(
|
| 22 |
"zero-shot-classification",
|
|
|
|
| 40 |
id2label = {0: "Tidak Abusif", 1: "Abusif"}
|
| 41 |
|
| 42 |
# ======================
|
| 43 |
+
# CONFIG (ASLI)
|
| 44 |
# ======================
|
| 45 |
SEMANTIC_EVENT = {
|
| 46 |
"sadisme": [
|
|
|
|
| 99 |
if len(k.strip()) >= 20
|
| 100 |
]
|
| 101 |
|
| 102 |
+
def sliding_window(kalimat, window=4):
|
| 103 |
+
if len(kalimat) < window:
|
| 104 |
+
return kalimat
|
| 105 |
+
return [
|
| 106 |
+
" ".join(kalimat[i:i+window])
|
| 107 |
+
for i in range(len(kalimat) - window + 1)
|
| 108 |
+
]
|
| 109 |
+
|
| 110 |
+
def event_completeness(text):
|
| 111 |
+
text = text.lower()
|
| 112 |
+
return sum(any(w in text for w in words) for words in EVENT_UNITS.values())
|
| 113 |
+
|
| 114 |
+
def is_metaphor(text):
|
| 115 |
+
text = text.lower()
|
| 116 |
+
return any(a in text and b in text for a, b in METAPHOR_PAIRS)
|
| 117 |
+
|
| 118 |
def rating_usia(kategori):
|
| 119 |
if "sadisme" in kategori:
|
| 120 |
return 21
|
|
|
|
| 125 |
return 0
|
| 126 |
|
| 127 |
# ======================
|
| 128 |
+
# CORE ANALYSIS + DEBUG (JANGAN DIUBAH)
|
| 129 |
# ======================
|
| 130 |
def analyze_text(judul, isi):
|
| 131 |
kalimat = split_kalimat(isi)
|
| 132 |
+
windows = sliding_window(kalimat)
|
| 133 |
+
|
| 134 |
detected = set()
|
| 135 |
|
| 136 |
+
log("\n" + "=" * 80)
|
| 137 |
+
log(f"JUDUL: {judul}")
|
| 138 |
+
|
| 139 |
+
for w in windows:
|
| 140 |
+
log("\n[WINDOW]")
|
| 141 |
+
log(w)
|
| 142 |
+
|
| 143 |
for kategori, desc in SEMANTIC_EVENT.items():
|
| 144 |
res = classifier(
|
| 145 |
w,
|
|
|
|
| 147 |
hypothesis_template="Teks ini menggambarkan {}."
|
| 148 |
)
|
| 149 |
score = max(res["scores"])
|
| 150 |
+
threshold = THRESHOLD[kategori]
|
| 151 |
+
|
| 152 |
+
log(f" - {kategori} | score={score:.3f} | threshold={threshold}")
|
| 153 |
+
|
| 154 |
+
if score < threshold:
|
| 155 |
+
log(" β FAIL: threshold")
|
| 156 |
+
continue
|
| 157 |
+
|
| 158 |
+
log(" β
ACCEPTED")
|
| 159 |
+
detected.add(kategori)
|
| 160 |
|
| 161 |
usia = rating_usia(detected)
|
| 162 |
+
log(f"\nFINAL KATEGORI: {list(detected)}")
|
| 163 |
+
log(f"RATING USIA: {usia}")
|
| 164 |
+
log("=" * 80)
|
| 165 |
+
|
| 166 |
return usia, ", ".join(sorted(detected))
|
| 167 |
|
| 168 |
# ======================
|
| 169 |
+
# FILTER ABUSIVE (TERPISAH, AMAN)
|
| 170 |
# ======================
|
| 171 |
def filter_abusive(isi):
|
| 172 |
paragraphs = re.split(r'\n+', isi)
|
|
|
|
| 206 |
return "\n".join(output)
|
| 207 |
|
| 208 |
# ======================
|
| 209 |
+
# ROUTER /ANALYZE (MODE-BASED)
|
| 210 |
# ======================
|
| 211 |
+
def analyze_router(judul, isi, mode):
|
| 212 |
if mode == "rating":
|
| 213 |
usia, kategori = analyze_text(judul, isi)
|
| 214 |
return {
|
|
|
|
| 226 |
}
|
| 227 |
|
| 228 |
# ======================
|
| 229 |
+
# GRADIO API (ROUTE TETAP /analyze)
|
| 230 |
# ======================
|
| 231 |
demo = gr.Interface(
|
| 232 |
+
fn=analyze_router,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 233 |
inputs=[
|
| 234 |
gr.Textbox(),
|
| 235 |
gr.Textbox(),
|
| 236 |
gr.Radio(["rating", "abusive"])
|
| 237 |
],
|
| 238 |
outputs=gr.JSON(),
|
| 239 |
+
api_name="analyze"
|
| 240 |
)
|
| 241 |
|
| 242 |
# ======================
|
| 243 |
# ENTRY POINT
|
| 244 |
# ======================
|
| 245 |
if __name__ == "__main__":
|
| 246 |
+
demo.launch(
|
|
|
|
|
|
|
|
|
|
| 247 |
server_name="0.0.0.0",
|
| 248 |
server_port=7860,
|
| 249 |
ssr_mode=False,
|
| 250 |
show_error=True
|
| 251 |
+
)
|