1 | ; make sure this directory exsist!
|
---|
2 | (setf save_path "/fart/d/jc/")
|
---|
3 |
|
---|
4 | ; pages to start traversing from
|
---|
5 | (add_page "http://personal.nbnet.nb.ca/jkeller/MP3.HTM")
|
---|
6 | (add_page "http://www.geocities.com/Area51/Vault/8821/mp3.html")
|
---|
7 | (add_page "http://personal.nbnet.nb.ca/jkeller/MP3.HTM")
|
---|
8 | (add_page "http://hem2.passagen.se/lethagen/Prodigy/")
|
---|
9 | (add_page "http://thebigsite.com/index.html")
|
---|
10 | (add_page "http://www.wgn.net/~eburrow/mp3.html")
|
---|
11 |
|
---|
12 | ; add thread takes x & y offset for window position
|
---|
13 | (add_thread 0 20) (add_thread 320 20)
|
---|
14 | (add_thread 0 70) (add_thread 320 70)
|
---|
15 | (add_thread 0 120) (add_thread 320 120)
|
---|
16 | (add_thread 0 170) (add_thread 320 170)
|
---|
17 | (add_thread 0 220) (add_thread 320 220)
|
---|
18 | (add_thread 0 270) (add_thread 320 270)
|
---|
19 | (add_thread 0 320) (add_thread 320 320)
|
---|
20 | (add_thread 0 370) (add_thread 320 370)
|
---|
21 | (add_thread 0 420)
|
---|
22 |
|
---|
23 |
|
---|
24 | (setf process_extensions '(".html" ".htm"))
|
---|
25 | (setf save_extensions '(".mp3"))
|
---|
26 |
|
---|
27 | (setf include_sub_strings '(
|
---|
28 | (".html") ; follow any html page
|
---|
29 | (".htm")
|
---|
30 | (".com")
|
---|
31 | (".net")
|
---|
32 | (".edu")
|
---|
33 | (".htm")
|
---|
34 | (".mp3"
|
---|
35 | "rem."
|
---|
36 | "weezer"
|
---|
37 | "ADAM_SANDLER" "adamsandler"
|
---|
38 | "u2"
|
---|
39 | "willsmith" "will_smith"
|
---|
40 | "offspring"
|
---|
41 | "sarah_mclachlan" "sarahmclachlan"
|
---|
42 | "No_Doubt" "nodoubt"
|
---|
43 | "Notorious" "puffdaddy"
|
---|
44 | "james_bond" "jamesbond"
|
---|
45 | "mariah_carey" "Mariahcarey"
|
---|
46 | "matchbox20"
|
---|
47 | "pinkfloyd"
|
---|
48 | "aerosmith"
|
---|
49 |
|
---|
50 | "batman" "movie" "jurpark" "indianajones" "indiana_jones"
|
---|
51 |
|
---|
52 | "Robert_Miles"
|
---|
53 | "No_Doubt"
|
---|
54 | "depeche"
|
---|
55 | "queen"
|
---|
56 | "simpsons"
|
---|
57 | "simon"
|
---|
58 |
|
---|
59 | "CHUMBAWAMBA"
|
---|
60 | "SirenPossession"
|
---|
61 | "Spice"
|
---|
62 | "Karma Police" "karampolice"
|
---|
63 | "prodigy"
|
---|
64 | )))
|
---|
65 |
|
---|
66 |
|
---|
67 |
|
---|
68 | ; don't process any page/file that has these sub strings
|
---|
69 | (setf exclude_sub_strings '("toriamos" ; enough already!
|
---|
70 | "jacky1" ; this guy has crap
|
---|
71 | "mailto"
|
---|
72 | ))
|
---|
73 |
|
---|
74 | ; don't save files smaller than this (700k)
|
---|
75 | (setf save_min_size 700000)
|
---|
76 |
|
---|
77 | ; load up our history of previously saved files
|
---|
78 | (load "saved_files.scm")
|
---|
79 |
|
---|
80 | ; load up history of visited urls
|
---|
81 | (load "visited_urls.scm")
|
---|
82 |
|
---|
83 |
|
---|
84 | ; timeout depends on how much data we have already, if we we a lot then
|
---|
85 | ; give a little longer to respond so we have less chance of wasting time
|
---|
86 |
|
---|
87 | (setf timeouts '((100000000 1000) ;
|
---|
88 | (1000000 500)
|
---|
89 | (500000 360)
|
---|
90 | (200000 120)
|
---|
91 | (100000 60)
|
---|
92 | (50000 45)
|
---|
93 | (10000 30) ; if we have 10K give them 30 seconds to respond
|
---|
94 | ))
|
---|