Spaces:
Running
Running
Update requirements.txt
Browse files- requirements.txt +7 -4
requirements.txt
CHANGED
@@ -9,12 +9,12 @@ httpx[http2]>=0.25.0,<0.28.0
|
|
9 |
# ASGI Framework (Core)
|
10 |
starlette>=0.35.0,<0.38.0
|
11 |
|
12 |
-
# Web Scraping (HTML Parsing)
|
13 |
beautifulsoup4>=4.11,<4.13
|
14 |
-
lxml>=4.9.0,<6.0.0
|
15 |
|
16 |
# Web Scraping (Primary Method - Headless Browser)
|
17 |
-
crawl4ai>=0.5.0,<0.6.0
|
18 |
|
19 |
# YouTube Transcript Library (Primary method)
|
20 |
youtube-transcript-api>=0.6,<0.7
|
@@ -31,4 +31,7 @@ anyio>=3.6,<5.0
|
|
31 |
asgiref>=3.5,<3.9
|
32 |
|
33 |
# Retry Logic
|
34 |
-
tenacity>=8.0,<9.0
|
|
|
|
|
|
|
|
9 |
# ASGI Framework (Core)
|
10 |
starlette>=0.35.0,<0.38.0
|
11 |
|
12 |
+
# Web Scraping (HTML Parsing - Fallback 1)
|
13 |
beautifulsoup4>=4.11,<4.13
|
14 |
+
lxml>=4.9.0,<6.0.0 # Optional but recommended BS4 parser
|
15 |
|
16 |
# Web Scraping (Primary Method - Headless Browser)
|
17 |
+
crawl4ai>=0.5.0,<0.6.0 # <<< ADDED crawl4ai
|
18 |
|
19 |
# YouTube Transcript Library (Primary method)
|
20 |
youtube-transcript-api>=0.6,<0.7
|
|
|
31 |
asgiref>=3.5,<3.9
|
32 |
|
33 |
# Retry Logic
|
34 |
+
tenacity>=8.0,<9.0
|
35 |
+
|
36 |
+
# Crawl4AI requires Playwright, but we install it via Dockerfile command
|
37 |
+
# playwright>=1.30 # Not strictly needed here if installed via RUN command
|