| skipped 88 lines |
89 | 89 | | else: |
90 | 90 | | content_type = '-' |
91 | 91 | | |
| 92 | + | if len(url) > 1: |
| 93 | + | url = url.strip('_') |
| 94 | + | |
92 | 95 | | match = title_regexp.search( r.text ) |
93 | 96 | | title = match.group(1) if match else '-' |
94 | 97 | | |
| skipped 15 lines |
110 | 113 | | # filename = t_urlparse.path.strip('/') |
111 | 114 | | filename = t_urlparse.path.strip('/') + '_' + t_urlparse.query |
112 | 115 | | filename = re.sub( '[^0-9a-zA-Z_\-\.]', '_', filename ) |
| 116 | + | if len(filename) > 1: |
| 117 | + | filename = filename.strip('_') |
| 118 | + | |
113 | 119 | | d_output = d_output + '/' + t_urlparse.netloc |
114 | 120 | | f_output = d_output + '/' + filename |
115 | 121 | | # print(f_output) |
| skipped 244 lines |