|
@@ -56,7 +56,7 @@ class Comments(qwarc.Item): |
|
|
if not sessionToken: |
|
|
if not sessionToken: |
|
|
self.logger.error('Could not find session token') |
|
|
self.logger.error('Could not find session token') |
|
|
return |
|
|
return |
|
|
if sessionToken.lstrip(b'0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_-') not in (b'', b'=', b'=='): |
|
|
|
|
|
|
|
|
if sessionToken.lstrip(b'0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_-') not in (b'', b'=', b'==', b'%3D', b'%3D%3D'): |
|
|
self.logger.error(f'Unexpected session token value: {sessionToken!r}') |
|
|
self.logger.error(f'Unexpected session token value: {sessionToken!r}') |
|
|
return |
|
|
return |
|
|
sessionToken = sessionToken.decode('ascii') |
|
|
sessionToken = sessionToken.decode('ascii') |
|
@@ -71,12 +71,12 @@ class Comments(qwarc.Item): |
|
|
return |
|
|
return |
|
|
section = content[continuationStartPos:sectionIdentifierPos] |
|
|
section = content[continuationStartPos:sectionIdentifierPos] |
|
|
continuationToken = qwarc.utils.str_get_between(section, b'"continuation":"', b'"') |
|
|
continuationToken = qwarc.utils.str_get_between(section, b'"continuation":"', b'"') |
|
|
if continuationToken.lstrip(b'0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_-') not in (b'', b'=', b'=='): |
|
|
|
|
|
|
|
|
if continuationToken.lstrip(b'0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_-') not in (b'', b'=', b'==', b'%3D', b'%3D%3D'): |
|
|
self.logger.error(f'Unexpected continuation token value: {continuationToken!r}') |
|
|
self.logger.error(f'Unexpected continuation token value: {continuationToken!r}') |
|
|
return |
|
|
return |
|
|
continuationToken = continuationToken.decode('ascii') |
|
|
continuationToken = continuationToken.decode('ascii') |
|
|
itct = qwarc.utils.str_get_between(section, b'"clickTrackingParams":"', b'"') |
|
|
itct = qwarc.utils.str_get_between(section, b'"clickTrackingParams":"', b'"') |
|
|
if itct.lstrip(b'0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_-') not in (b'', b'=', b'=='): |
|
|
|
|
|
|
|
|
if itct.lstrip(b'0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_-') not in (b'', b'=', b'==', b'%3D', b'%3D%3D'): |
|
|
self.logger.error(f'Unexpected itct value: {itct!r}') |
|
|
self.logger.error(f'Unexpected itct value: {itct!r}') |
|
|
return |
|
|
return |
|
|
itct = itct.decode('ascii') |
|
|
itct = itct.decode('ascii') |
|
|