Spaces:
Running
on
CPU Upgrade
Running
on
CPU Upgrade
new
Browse files
app.py
CHANGED
@@ -40,6 +40,7 @@ CLASSIFIER_URL = (
|
|
40 |
)
|
41 |
ASSETS_URL = "https://d26smi9133w0oo.cloudfront.net/diffusers-gallery/"
|
42 |
|
|
|
43 |
|
44 |
s3 = boto3.client(
|
45 |
service_name="s3",
|
@@ -120,12 +121,22 @@ def get_yaml_data(text_content):
|
|
120 |
print(exc)
|
121 |
return {}
|
122 |
|
123 |
-
async def find_image_in_model_card(text):
|
124 |
-
|
125 |
-
|
126 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
127 |
return []
|
128 |
|
|
|
129 |
async with aiohttp.ClientSession() as session:
|
130 |
tasks = [
|
131 |
asyncio.ensure_future(upload_resize_image_url(session, image_url))
|
@@ -188,9 +199,10 @@ async def sync_data():
|
|
188 |
with open(DB_FOLDER / "models.json", "w") as f:
|
189 |
json.dump(all_models, f)
|
190 |
# with open(DB_FOLDER / "models.json", "r") as f:
|
191 |
-
#
|
192 |
|
193 |
new_models_ids = [model["id"] for model in all_models]
|
|
|
194 |
|
195 |
# get existing models
|
196 |
with database.get_db() as db:
|
@@ -212,7 +224,7 @@ async def sync_data():
|
|
212 |
print("Parsing model card")
|
213 |
model_card_data = get_yaml_data(model_card)
|
214 |
print("Finding images in model card")
|
215 |
-
images = await find_image_in_model_card(model_card)
|
216 |
|
217 |
classifier = run_classifier(images)
|
218 |
print(images, classifier)
|
@@ -257,7 +269,7 @@ async def sync_data():
|
|
257 |
print("Parsing model card")
|
258 |
model_card_data = get_yaml_data(model_card)
|
259 |
print("Finding images in model card")
|
260 |
-
images = await find_image_in_model_card(model_card)
|
261 |
classifier = run_classifier(images)
|
262 |
model_data["images"] = images
|
263 |
model_data["class"] = classifier
|
@@ -322,6 +334,7 @@ class Style(str, Enum):
|
|
322 |
s3D = "3d"
|
323 |
realistic = "realistic"
|
324 |
nsfw = "nsfw"
|
|
|
325 |
|
326 |
|
327 |
@app.get("/api/models")
|
@@ -344,9 +357,13 @@ def get_page(
|
|
344 |
style_query = "json_extract(data, '$.class.3d') > 0.1 AND isNFSW = false"
|
345 |
elif style == Style.realistic:
|
346 |
style_query = "json_extract(data, '$.class.real_life') > 0.1 AND isNFSW = false"
|
|
|
|
|
347 |
elif style == Style.nsfw:
|
348 |
style_query = "isNFSW = true"
|
349 |
|
|
|
|
|
350 |
with database.get_db() as db:
|
351 |
cursor = db.cursor()
|
352 |
cursor.execute(
|
@@ -359,7 +376,7 @@ def get_page(
|
|
359 |
json_extract(data, '$.class.explicit') > 0.3 OR json_extract(data, '$.class.suggestive') > 0.3 AS isNFSW
|
360 |
FROM models
|
361 |
) AS subquery
|
362 |
-
WHERE (? IS NULL AND likes >
|
363 |
AND {style_query}
|
364 |
AND (? IS NULL OR EXISTS (
|
365 |
SELECT 1
|
@@ -368,7 +385,7 @@ def get_page(
|
|
368 |
))
|
369 |
ORDER BY {sort_query}
|
370 |
LIMIT {MAX_PAGE_SIZE} OFFSET {(page - 1) * MAX_PAGE_SIZE};
|
371 |
-
""",
|
372 |
(tag, tag, tag, tag),
|
373 |
)
|
374 |
results = cursor.fetchall()
|
|
|
40 |
)
|
41 |
ASSETS_URL = "https://d26smi9133w0oo.cloudfront.net/diffusers-gallery/"
|
42 |
|
43 |
+
BLOCKED_MODELS_REGEX = re.compile(r"(CyberHarem)", re.IGNORECASE)
|
44 |
|
45 |
s3 = boto3.client(
|
46 |
service_name="s3",
|
|
|
121 |
print(exc)
|
122 |
return {}
|
123 |
|
124 |
+
async def find_image_in_model_card(text, model_id):
|
125 |
+
base_url = f"https://huggingface.co/{model_id}/resolve/main/"
|
126 |
+
image_regex = re.compile(r"!\[.*\]\((.*?\.(png|jpg|jpeg|gif|bmp|webp))\)|src=\"(.*?\.(png|jpg|jpeg|gif|bmp|webp))\">", re.IGNORECASE)
|
127 |
+
matches = image_regex.findall(text)
|
128 |
+
urls = []
|
129 |
+
for match in matches:
|
130 |
+
for url in match:
|
131 |
+
if url:
|
132 |
+
if not url.startswith("http") and not url.startswith("https"):
|
133 |
+
url = base_url + url
|
134 |
+
urls.append(url)
|
135 |
+
|
136 |
+
if len(urls) == 0:
|
137 |
return []
|
138 |
|
139 |
+
print(urls)
|
140 |
async with aiohttp.ClientSession() as session:
|
141 |
tasks = [
|
142 |
asyncio.ensure_future(upload_resize_image_url(session, image_url))
|
|
|
199 |
with open(DB_FOLDER / "models.json", "w") as f:
|
200 |
json.dump(all_models, f)
|
201 |
# with open(DB_FOLDER / "models.json", "r") as f:
|
202 |
+
# all_models = json.load(f)
|
203 |
|
204 |
new_models_ids = [model["id"] for model in all_models]
|
205 |
+
new_models_ids = [model_id for model_id in new_models_ids if not re.match(BLOCKED_MODELS_REGEX, model_id)]
|
206 |
|
207 |
# get existing models
|
208 |
with database.get_db() as db:
|
|
|
224 |
print("Parsing model card")
|
225 |
model_card_data = get_yaml_data(model_card)
|
226 |
print("Finding images in model card")
|
227 |
+
images = await find_image_in_model_card(model_card, model_id)
|
228 |
|
229 |
classifier = run_classifier(images)
|
230 |
print(images, classifier)
|
|
|
269 |
print("Parsing model card")
|
270 |
model_card_data = get_yaml_data(model_card)
|
271 |
print("Finding images in model card")
|
272 |
+
images = await find_image_in_model_card(model_card, model_id)
|
273 |
classifier = run_classifier(images)
|
274 |
model_data["images"] = images
|
275 |
model_data["class"] = classifier
|
|
|
334 |
s3D = "3d"
|
335 |
realistic = "realistic"
|
336 |
nsfw = "nsfw"
|
337 |
+
lora = "lora"
|
338 |
|
339 |
|
340 |
@app.get("/api/models")
|
|
|
357 |
style_query = "json_extract(data, '$.class.3d') > 0.1 AND isNFSW = false"
|
358 |
elif style == Style.realistic:
|
359 |
style_query = "json_extract(data, '$.class.real_life') > 0.1 AND isNFSW = false"
|
360 |
+
elif style == Style.lora:
|
361 |
+
style_query = "json_extract(data, '$.meta.tags') LIKE '%lora%' AND isNFSW = false"
|
362 |
elif style == Style.nsfw:
|
363 |
style_query = "isNFSW = true"
|
364 |
|
365 |
+
|
366 |
+
|
367 |
with database.get_db() as db:
|
368 |
cursor = db.cursor()
|
369 |
cursor.execute(
|
|
|
376 |
json_extract(data, '$.class.explicit') > 0.3 OR json_extract(data, '$.class.suggestive') > 0.3 AS isNFSW
|
377 |
FROM models
|
378 |
) AS subquery
|
379 |
+
WHERE (? IS NULL AND likes > 1 OR ? IS NOT NULL)
|
380 |
AND {style_query}
|
381 |
AND (? IS NULL OR EXISTS (
|
382 |
SELECT 1
|
|
|
385 |
))
|
386 |
ORDER BY {sort_query}
|
387 |
LIMIT {MAX_PAGE_SIZE} OFFSET {(page - 1) * MAX_PAGE_SIZE};
|
388 |
+
""",
|
389 |
(tag, tag, tag, tag),
|
390 |
)
|
391 |
results = cursor.fetchall()
|