|
9 | 9 | from urllib.parse import urlparse |
10 | 10 |
|
11 | 11 |
|
| 12 | +INVALID_HOMEPAGES = { |
| 13 | + 'https://', 'http://', 'n/a', 'na', '-', '--', '---', '....', '...', |
| 14 | + 'tbd', 'coming', 'coming soon', 'in process', 'no webside', 'no website', |
| 15 | + 'none', 'null', 'undefined', 'unknown' |
| 16 | +} |
| 17 | + |
| 18 | +INVALID_HOMEPAGE_SUBSTRINGS = [ |
| 19 | + 'foo.com', 'example.com', 'invalidurl', 'test.com', |
| 20 | + 'localhost', '127.0.0.1', 'yourdomain', 'yoursite', |
| 21 | + 'mysite.com', 'mypool.com', 'poolname.com' |
| 22 | +] |
| 23 | + |
| 24 | + |
12 | 25 | def get_pool_data_BQ(force_query): |
13 | 26 | """ |
14 | 27 | Queries the BigQuery database for pool data and writes it to a json file. |
@@ -269,27 +282,17 @@ def filter_homepage(homepage): |
269 | 282 | """ |
270 | 283 | if not homepage: |
271 | 284 | return None |
| 285 | + |
272 | 286 | homepage = homepage.strip() |
273 | 287 | if not homepage: |
274 | 288 | return None |
275 | 289 |
|
276 | 290 | homepage_lower = homepage.lower() |
277 | 291 |
|
278 | | - INVALID_EXACT = { |
279 | | - 'https://', 'http://', 'n/a', 'na', '-', '--', '---', '....', '...', |
280 | | - 'tbd', 'coming', 'coming soon', 'in process', 'no webside', 'no website', |
281 | | - 'none', 'null', 'undefined', 'unknown' |
282 | | - } |
283 | | - if homepage_lower in INVALID_EXACT: |
| 292 | + if homepage_lower in INVALID_HOMEPAGES: |
284 | 293 | return None |
285 | 294 |
|
286 | | - INVALID_SUBSTRINGS = [ |
287 | | - 'foo.com', 'example.com', 'invalidurl', 'test.com', |
288 | | - 'localhost', '127.0.0.1', 'yourdomain', 'yoursite', |
289 | | - 'mysite.com', 'mypool.com', 'poolname.com' |
290 | | - ] |
291 | | - |
292 | | - if any(kw in homepage_lower for kw in INVALID_SUBSTRINGS): |
| 295 | + if any(kw in homepage_lower for kw in INVALID_HOMEPAGE_SUBSTRINGS): |
293 | 296 | return None |
294 | 297 |
|
295 | 298 | return homepage |
|
0 commit comments