Spaces:
Sleeping
Sleeping
Nikhil Pravin Pise commited on
Commit ·
aa69bc4
1
Parent(s): e98cc10
fix: restore block_resources function in utils.py
Browse files- src/utils.py +16 -0
src/utils.py
CHANGED
|
@@ -19,6 +19,22 @@ from urllib.parse import urlparse, urljoin
|
|
| 19 |
T = TypeVar("T")
|
| 20 |
|
| 21 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
# =============================================================================
|
| 23 |
# TIMING UTILITIES
|
| 24 |
# =============================================================================
|
|
|
|
| 19 |
T = TypeVar("T")
|
| 20 |
|
| 21 |
|
| 22 |
+
# =============================================================================
|
| 23 |
+
# PLAYWRIGHT UTILITIES
|
| 24 |
+
# =============================================================================
|
| 25 |
+
|
| 26 |
+
|
| 27 |
+
async def block_resources(route) -> None:
|
| 28 |
+
"""
|
| 29 |
+
Blocks unnecessary resources to speed up scraping.
|
| 30 |
+
Used with Playwright's page.route() to block images, stylesheets, etc.
|
| 31 |
+
"""
|
| 32 |
+
if route.request.resource_type in ["image", "stylesheet", "font", "media"]:
|
| 33 |
+
await route.abort()
|
| 34 |
+
else:
|
| 35 |
+
await route.continue_()
|
| 36 |
+
|
| 37 |
+
|
| 38 |
# =============================================================================
|
| 39 |
# TIMING UTILITIES
|
| 40 |
# =============================================================================
|