This will create list of precompiled regex object to find cloud URLs within the content. Returns -------- list a list of precompiled regex objects.
()
| 423 | |
| 424 | |
| 425 | def PreCompiledRegexCloud(): |
| 426 | """ |
| 427 | |
| 428 | This will create list of precompiled regex object to find cloud URLs within the content. |
| 429 | |
| 430 | Returns |
| 431 | -------- |
| 432 | list |
| 433 | a list of precompiled regex objects. |
| 434 | """ |
| 435 | cfreg = re.compile(r'([\w]+\.cloudfront\.net)', |
| 436 | re.MULTILINE | re.IGNORECASE) |
| 437 | gbureg = re.compile(r'([\w\-.]+\.appspot\.com)', |
| 438 | re.MULTILINE | re.IGNORECASE) |
| 439 | s3bucketreg = re.compile( |
| 440 | r'([\w\-.]*s3[\w\-.]*\.?amazonaws\.com/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 441 | doreg = re.compile( |
| 442 | r'([\w\-.]*\.?digitaloceanspaces\.com/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 443 | gsreg1 = re.compile( |
| 444 | r'(storage\.cloud\.google\.com/[\w\-.]+)', re.MULTILINE | re.IGNORECASE) |
| 445 | gsreg2 = re.compile( |
| 446 | r'([\w\-.]*\.?storage.googleapis.com/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 447 | gsreg3 = re.compile( |
| 448 | r'([\w\-.]*\.?storage-download.googleapis.com/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 449 | gsreg4 = re.compile( |
| 450 | r'([\w\-.]*\.?content-storage-upload.googleapis.com/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 451 | gsreg5 = re.compile( |
| 452 | r'([\w\-.]*\.?content-storage-download.googleapis.com/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 453 | azureg1 = re.compile( |
| 454 | r'([\w\-.]*\.?1drv\.com/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 455 | azureg2 = re.compile( |
| 456 | r'(onedrive.live.com/[\w.\-]+)', re.MULTILINE | re.IGNORECASE) |
| 457 | azureg3 = re.compile( |
| 458 | r'([\w\-.]*\.?blob\.core\.windows\.net/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 459 | rackcdnreg = re.compile( |
| 460 | r'([\w\-.]*\.?rackcdn.com/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 461 | dreamhostreg1 = re.compile( |
| 462 | r'([\w\-.]*\.?objects\.cdn\.dream\.io/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 463 | dreamhostreg2 = re.compile( |
| 464 | r'([\w\-.]*\.?objects-us-west-1.dream.io/?[\w\-.]*)', re.MULTILINE | re.IGNORECASE) |
| 465 | firebase = re.compile(r'([\w\-.]+\.firebaseio\.com)', |
| 466 | re.MULTILINE | re.IGNORECASE) |
| 467 | |
| 468 | cloudlist = [cfreg, s3bucketreg, doreg, gsreg1, gsreg2, gsreg3, gsreg4, gsreg5, |
| 469 | azureg1, azureg2, azureg3, rackcdnreg, dreamhostreg1, dreamhostreg2, firebase, gbureg] |
| 470 | |
| 471 | return cloudlist |
| 472 | |
| 473 | |
| 474 | def PreCompiledRegexDomain(url): |