Spaces:
Running
Running
Commit
·
a24c8fd
1
Parent(s):
709e431
adding action
Browse files- .github/workflows/main.yaml +19 -0
- app.py +6 -1
.github/workflows/main.yaml
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
name: Sync to Hugging Face Space
|
| 2 |
+
on:
|
| 3 |
+
push:
|
| 4 |
+
branches: [main]
|
| 5 |
+
|
| 6 |
+
jobs:
|
| 7 |
+
sync-to-hub:
|
| 8 |
+
runs-on: ubuntu-latest
|
| 9 |
+
steps:
|
| 10 |
+
- uses: actions/checkout@v3
|
| 11 |
+
with:
|
| 12 |
+
fetch-depth: 0
|
| 13 |
+
lfs: true
|
| 14 |
+
- name: Push to Hugging Face Space
|
| 15 |
+
env:
|
| 16 |
+
HF_TOKEN: ${{ secrets.HF_TOKEN }}
|
| 17 |
+
run: |
|
| 18 |
+
git remote add space https://huggingface.co/spaces/patrickacraig/docs-scraper
|
| 19 |
+
git push space main
|
app.py
CHANGED
|
@@ -7,10 +7,14 @@ from urllib.parse import urlparse
|
|
| 7 |
|
| 8 |
load_dotenv()
|
| 9 |
|
| 10 |
-
base_url = os.getenv('
|
| 11 |
api_key = os.getenv('FIRECRAWL_API_KEY')
|
| 12 |
limit_rate = os.getenv('LIMIT_RATE', 'False').lower() == 'true'
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
# Get Firecrawl App instance
|
| 15 |
def get_firecrawl_app(api_key):
|
| 16 |
return FirecrawlApp(api_key=api_key)
|
|
@@ -33,6 +37,7 @@ async def async_scrape_url(app, url):
|
|
| 33 |
def map_website(app, url):
|
| 34 |
try:
|
| 35 |
map_status = app.map_url(url)
|
|
|
|
| 36 |
if isinstance(map_status, list):
|
| 37 |
return map_status
|
| 38 |
else:
|
|
|
|
| 7 |
|
| 8 |
load_dotenv()
|
| 9 |
|
| 10 |
+
base_url = os.getenv('TARGET_URL')
|
| 11 |
api_key = os.getenv('FIRECRAWL_API_KEY')
|
| 12 |
limit_rate = os.getenv('LIMIT_RATE', 'False').lower() == 'true'
|
| 13 |
|
| 14 |
+
print(f"base_url: {base_url}")
|
| 15 |
+
print(f"api_key: {api_key}")
|
| 16 |
+
print(f"limit_rate: {limit_rate}")
|
| 17 |
+
|
| 18 |
# Get Firecrawl App instance
|
| 19 |
def get_firecrawl_app(api_key):
|
| 20 |
return FirecrawlApp(api_key=api_key)
|
|
|
|
| 37 |
def map_website(app, url):
|
| 38 |
try:
|
| 39 |
map_status = app.map_url(url)
|
| 40 |
+
print(f"Map status for {url}: {map_status}")
|
| 41 |
if isinstance(map_status, list):
|
| 42 |
return map_status
|
| 43 |
else:
|