Spaces:
Running
Running
pushhyyyyyy Merge branch 'main' of https://huggingface.co/spaces/Ezi/dataset_PII_checker
Browse files- __pycache__/app.cpython-310.pyc +0 -0
- __pycache__/main.cpython-310.pyc +0 -0
- __pycache__/server.cpython-310.pyc +0 -0
- main.py +0 -152
- server.py +0 -27
__pycache__/app.cpython-310.pyc
DELETED
Binary file (4.76 kB)
|
|
__pycache__/main.cpython-310.pyc
DELETED
Binary file (4.76 kB)
|
|
__pycache__/server.cpython-310.pyc
DELETED
Binary file (1.2 kB)
|
|
main.py
DELETED
@@ -1,152 +0,0 @@
|
|
1 |
-
import json
|
2 |
-
import os
|
3 |
-
from difflib import SequenceMatcher
|
4 |
-
from typing import Any, Dict, Optional, Tuple
|
5 |
-
|
6 |
-
from fastapi import FastAPI, Request, Response
|
7 |
-
from huggingface_hub import (DatasetCard, HfApi, ModelCard, comment_discussion,
|
8 |
-
create_discussion, get_discussion_details,
|
9 |
-
get_repo_discussions, login)
|
10 |
-
from huggingface_hub.utils import EntryNotFoundError
|
11 |
-
from tabulate import tabulate
|
12 |
-
|
13 |
-
KEY = os.environ.get("WEBHOOK_SECRET")
|
14 |
-
#HF_TOKEN = os.environ.get("HF_ACCESS_TOKEN")
|
15 |
-
|
16 |
-
#api = HfApi(token=HF_TOKEN)
|
17 |
-
#login(HF_TOKEN)
|
18 |
-
|
19 |
-
#app = FastAPI()
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
def similar(a, b):
|
25 |
-
"""Check similarity of two sequences"""
|
26 |
-
return SequenceMatcher(None, a, b).ratio()
|
27 |
-
|
28 |
-
|
29 |
-
def create_metadata_key_dict(card_data, repo_type: str):
|
30 |
-
shared_keys = ["tags", "license"]
|
31 |
-
if repo_type == "model":
|
32 |
-
model_keys = ["library_name", "datasets", "metrics", "co2", "pipeline_tag"]
|
33 |
-
shared_keys.extend(model_keys)
|
34 |
-
keys = shared_keys
|
35 |
-
return {key: card_data.get(key) for key in keys}
|
36 |
-
if repo_type == "dataset":
|
37 |
-
data_keys = [
|
38 |
-
"pretty_name",
|
39 |
-
"size_categories",
|
40 |
-
"task_categories",
|
41 |
-
"task_ids",
|
42 |
-
"source_datasets",
|
43 |
-
]
|
44 |
-
shared_keys.extend(data_keys)
|
45 |
-
keys = shared_keys
|
46 |
-
return {key: card_data.get(key) for key in keys}
|
47 |
-
|
48 |
-
|
49 |
-
def create_metadata_breakdown_table(desired_metadata_dictionary):
|
50 |
-
data = {k:v or "Field Missing" for k,v in desired_metadata_dictionary.items()}
|
51 |
-
metadata_fields_column = list(data.keys())
|
52 |
-
metadata_values_column = list(data.values())
|
53 |
-
table_data = list(zip(metadata_fields_column, metadata_values_column))
|
54 |
-
return tabulate(
|
55 |
-
table_data, tablefmt="github", headers=("Metadata Field", "Provided Value")
|
56 |
-
)
|
57 |
-
|
58 |
-
|
59 |
-
def calculate_grade(desired_metadata_dictionary):
|
60 |
-
metadata_values = list(desired_metadata_dictionary.values())
|
61 |
-
score = sum(1 if field else 0 for field in metadata_values) / len(metadata_values)
|
62 |
-
return round(score, 2)
|
63 |
-
|
64 |
-
|
65 |
-
def create_markdown_report(
|
66 |
-
desired_metadata_dictionary, repo_name, repo_type, score, update: bool = False
|
67 |
-
):
|
68 |
-
report = f"""# {repo_type.title()} metadata report card {"(updated)" if update else ""}
|
69 |
-
\n
|
70 |
-
This is an automatically produced metadata quality report card for {repo_name}. This report is meant as a POC!
|
71 |
-
\n
|
72 |
-
## Breakdown of metadata fields for your{repo_type}
|
73 |
-
\n
|
74 |
-
{create_metadata_breakdown_table(desired_metadata_dictionary)}
|
75 |
-
\n
|
76 |
-
You scored a metadata coverage grade of: **{score}**% \n {f"We're not angry we're just disappointed! {repo_type.title()} metadata is super important. Please try harder..."
|
77 |
-
if score <= 0.5 else f"Not too shabby! Make sure you also fill in a {repo_type} card too!"}
|
78 |
-
"""
|
79 |
-
return report
|
80 |
-
|
81 |
-
|
82 |
-
def parse_webhook_post(data: Dict[str, Any]) -> Optional[Tuple[str, str]]:
|
83 |
-
event = data["event"]
|
84 |
-
if event["scope"] != "repo":
|
85 |
-
return None
|
86 |
-
repo = data["repo"]
|
87 |
-
repo_name = repo["name"]
|
88 |
-
repo_type = repo["type"]
|
89 |
-
if repo_type not in {"model", "dataset"}:
|
90 |
-
raise ValueError("Unknown hub type")
|
91 |
-
return repo_type, repo_name
|
92 |
-
|
93 |
-
|
94 |
-
def load_repo_card_metadata(repo_type, repo_name):
|
95 |
-
if repo_type == "dataset":
|
96 |
-
try:
|
97 |
-
return DatasetCard.load(repo_name).data.to_dict()
|
98 |
-
except EntryNotFoundError:
|
99 |
-
return {}
|
100 |
-
if repo_type == "model":
|
101 |
-
try:
|
102 |
-
return ModelCard.load(repo_name).data.to_dict()
|
103 |
-
except EntryNotFoundError:
|
104 |
-
return {}
|
105 |
-
|
106 |
-
|
107 |
-
def create_or_update_report(data):
|
108 |
-
if parsed_post := parse_webhook_post(data):
|
109 |
-
repo_type, repo_name = parsed_post
|
110 |
-
else:
|
111 |
-
return Response("Unable to parse webhook data", status_code=400)
|
112 |
-
card_data = load_repo_card_metadata(repo_type, repo_name)
|
113 |
-
desired_metadata_dictionary = create_metadata_key_dict(card_data, repo_type)
|
114 |
-
score = calculate_grade(desired_metadata_dictionary)
|
115 |
-
report = create_markdown_report(
|
116 |
-
desired_metadata_dictionary, repo_name, repo_type, score, update=False
|
117 |
-
)
|
118 |
-
repo_discussions = get_repo_discussions(
|
119 |
-
repo_name,
|
120 |
-
repo_type=repo_type,
|
121 |
-
)
|
122 |
-
for discussion in repo_discussions:
|
123 |
-
if (
|
124 |
-
discussion.title == "Metadata Report Card" and discussion.status == "open"
|
125 |
-
): # An existing open report card thread
|
126 |
-
discussion_details = get_discussion_details(
|
127 |
-
repo_name, discussion.num, repo_type=repo_type
|
128 |
-
)
|
129 |
-
last_comment = discussion_details.events[-1].content
|
130 |
-
if similar(report, last_comment) <= 0.999:
|
131 |
-
report = create_markdown_report(
|
132 |
-
desired_metadata_dictionary,
|
133 |
-
repo_name,
|
134 |
-
repo_type,
|
135 |
-
score,
|
136 |
-
update=True,
|
137 |
-
)
|
138 |
-
comment_discussion(
|
139 |
-
repo_name,
|
140 |
-
discussion.num,
|
141 |
-
comment=report,
|
142 |
-
repo_type=repo_type,
|
143 |
-
)
|
144 |
-
return True
|
145 |
-
create_discussion(
|
146 |
-
repo_name,
|
147 |
-
"Metadata Report Card",
|
148 |
-
description=report,
|
149 |
-
repo_type=repo_type,
|
150 |
-
)
|
151 |
-
return True
|
152 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
server.py
DELETED
@@ -1,27 +0,0 @@
|
|
1 |
-
import os
|
2 |
-
from fastapi import FastAPI, Request, Response
|
3 |
-
from main import create_or_update_report
|
4 |
-
from tabulate import tabulate
|
5 |
-
|
6 |
-
KEY = os.environ.get("WEBHOOK_SECRET")
|
7 |
-
|
8 |
-
app = FastAPI()
|
9 |
-
|
10 |
-
@app.get("/")
|
11 |
-
def read_root():
|
12 |
-
data = """
|
13 |
-
<h2 style="text-align:center">Metadata Review Bot</h2>
|
14 |
-
<p style="text-align:center">This is a demo app showing how to use webhooks to automate metadata review for models and datasets shared on the Hugging Face Hub.</p>
|
15 |
-
"""
|
16 |
-
return Response(content=data, media_type="text/html")
|
17 |
-
|
18 |
-
@app.post("/webhook")
|
19 |
-
async def webhook(request: Request):
|
20 |
-
if request.method == "POST":
|
21 |
-
if request.headers.get("X-Webhook-Secret") != KEY:
|
22 |
-
return Response("Invalid secret", status_code=401)
|
23 |
-
data = await request.json()
|
24 |
-
result = create_or_update_report(data)
|
25 |
-
return "Webhook received!" if result else result
|
26 |
-
|
27 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|