| skipped 88 lines |
89 | 89 | | else: |
90 | 90 | | content_type = '-' |
91 | 91 | | |
92 | | - | if len(url) > 1: |
93 | | - | url = url.strip('_') |
| 92 | + | # if len(url) > 1: |
| 93 | + | # url = url.strip('_') |
94 | 94 | | |
95 | 95 | | match = title_regexp.search( r.text ) |
96 | 96 | | title = match.group(1) if match else '-' |
| skipped 22 lines |
119 | 119 | | # filename = t_urlparse.path.strip('/') |
120 | 120 | | filename = t_urlparse.path.strip('/') + '_' + t_urlparse.query |
121 | 121 | | filename = re.sub( '[^0-9a-zA-Z_\-\.]', '_', filename ) |
122 | | - | if len(filename) > 1: |
123 | | - | filename = filename.strip('_') |
| 122 | + | # if len(filename) > 1: |
| 123 | + | # filename = filename.strip('_') |
124 | 124 | | |
125 | 125 | | d_output = d_output + '/' + t_urlparse.netloc |
126 | 126 | | f_output = d_output + '/' + filename |
| skipped 250 lines |