1 |
################################################### |
################################################### |
2 |
|
|
|
#IndexDir http://www.rot13.org |
|
|
|
|
|
#IndexFile /data/swish/index/rot13 |
|
|
|
|
|
#IndexName "rot13.org" |
|
|
#IndexDescription "Internet web pages" |
|
|
#IndexPointer "http://www.rot13.org/" |
|
|
#IndexAdmin "dpavlin@rot13.org" |
|
|
|
|
3 |
# don't index files with this suffix (filesys only?) |
# don't index files with this suffix (filesys only?) |
4 |
NoContents .gif .xbm .au .mov .mpg ..ps .bin .exe .wmv |
NoContents .gif .xbm .au .mov .mpg ..ps .bin .exe .wmv |
5 |
|
|
6 |
# MetaNames automatic |
# MetaNames automatic |
7 |
MetaNames first author |
#MetaNames first author |
8 |
|
|
9 |
# 0 to 3 - 0 is totally silent, 3 is the most verbose. |
# 0 to 3 - 0 is totally silent, 3 is the most verbose. |
10 |
#IndexReport 2 |
#IndexReport 2 |
87 |
DefaultContents HTML |
DefaultContents HTML |
88 |
|
|
89 |
# use 100 chars from html |
# use 100 chars from html |
90 |
StoreDescription HTML <body> 300 |
StoreDescription HTML <body> 3000 |
91 |
StoreDescription TXT 300 |
StoreDescription TXT 3000 |
92 |
|
|
93 |
|
|
94 |
# index pdf |
# index pdf |
95 |
IndexContents HTML .pdf |
IndexContents HTML .pdf |
96 |
FileFilter .pdf pdftotext "-htmlmeta '%p' -" |
FileFilter .pdf pdftotext "-htmlmeta '%p' -" |
97 |
|
|
98 |
# various other |
# various other |
99 |
FileFilter .gz gzip "-cd '%p'" |
FileFilter .tar.gz tar "tvfz '%p'" |
100 |
FileFilter .bz2 bzip2 "-cd '%p'" |
FileFilter .gz gzip "-cd '%p'" |
101 |
|
FileFilter .bz2 bzip2 "-cd '%p'" |
102 |
|
|
103 |
|
# regex version of above |
104 |
|
#FileFilterMatch tar "tvfz '%p'" /\.tar\.gz/i |
105 |
|
#FileFilterMatch gzip "-cd '%p'" /\.gz/i |
106 |
|
#FileFilterMatch bzip2 "-cd '%p'" /\.bz2/i |
107 |
|
|
108 |
|
# store path of document, title |
109 |
|
MetaNames swishdocpath swishtitle |
110 |
|
|
111 |
# debug |
# debug |
112 |
ParserWarnLevel 3 |
ParserWarnLevel 3 |