Mercurial > hg > egg-tcls
annotate fetch_feeds.tcl @ 415:ff932030a9b3
weather: Bump version.
| author | Matti Hamalainen <ccr@tnsp.org> |
|---|---|
| date | Sat, 07 Jan 2017 22:41:07 +0200 |
| parents | 51c08336d7b1 |
| children | 880a07485275 |
| rev | line source |
|---|---|
| 0 | 1 #!/usr/bin/tclsh |
| 1 | 2 # |
| 3 # NOTICE! Change above path to correct tclsh binary path! | |
| 4 # | |
|
268
96310b1c88fa
feeds: Improve config resiliency.
Matti Hamalainen <ccr@tnsp.org>
parents:
265
diff
changeset
|
5 ############################################################################## |
| 0 | 6 # |
| 323 | 7 # FeedCheck fetcher v1.0 by Matti 'ccr' Hamalainen <ccr@tnsp.org> |
|
265
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
8 # (C) Copyright 2008-2015 Tecnic Software productions (TNSP) |
| 0 | 9 # |
| 10 # This script is freely distributable under GNU GPL (version 2) license. | |
| 11 # | |
|
268
96310b1c88fa
feeds: Improve config resiliency.
Matti Hamalainen <ccr@tnsp.org>
parents:
265
diff
changeset
|
12 ############################################################################## |
|
139
3305e142eecc
Change feed fetcher to use SQLite3 backend.
Matti Hamalainen <ccr@tnsp.org>
parents:
114
diff
changeset
|
13 package require sqlite3 |
|
265
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
14 source [file dirname [info script]]/utillib.tcl |
| 0 | 15 |
|
265
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
16 ### The configuration should be in config.feeds in same directory |
|
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
17 ### as this script. Or change the line below to point where ever |
|
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
18 ### you wish. See "config.feeds.example" for an example config file. |
|
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
19 source [file dirname [info script]]/config.feeds |
| 0 | 20 |
| 21 | |
|
268
96310b1c88fa
feeds: Improve config resiliency.
Matti Hamalainen <ccr@tnsp.org>
parents:
265
diff
changeset
|
22 ############################################################################## |
|
139
3305e142eecc
Change feed fetcher to use SQLite3 backend.
Matti Hamalainen <ccr@tnsp.org>
parents:
114
diff
changeset
|
23 |
| 0 | 24 package require http |
| 271 | 25 |
|
265
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
26 if {[info exists http_user_agent] && $http_user_agent != ""} { |
| 296 | 27 ::http::config -urlencoding utf8 -useragent $http_user_agent |
|
265
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
28 } else { |
| 322 | 29 ::http::config -urlencoding utf8 -useragent "Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Firefox/38.0" |
|
265
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
30 } |
| 271 | 31 |
|
268
96310b1c88fa
feeds: Improve config resiliency.
Matti Hamalainen <ccr@tnsp.org>
parents:
265
diff
changeset
|
32 if {[info exists http_use_proxy] && $http_use_proxy != 0} { |
| 63 | 33 ::http::config -proxyhost $http_proxy_host -proxyport $http_proxy_port |
| 0 | 34 } |
| 35 | |
|
268
96310b1c88fa
feeds: Improve config resiliency.
Matti Hamalainen <ccr@tnsp.org>
parents:
265
diff
changeset
|
36 if {[info exists http_tls_support] && $http_tls_support != 0} { |
|
265
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
37 package require tls |
|
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
38 ::http::register https 443 [list ::tls::socket -request 1 -require 1 -tls1 1 -cadir $http_tls_cadir] |
|
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
39 } |
|
908edc54005a
feeds: Move configuration to separate file.
Matti Hamalainen <ccr@tnsp.org>
parents:
159
diff
changeset
|
40 |
| 0 | 41 |
|
268
96310b1c88fa
feeds: Improve config resiliency.
Matti Hamalainen <ccr@tnsp.org>
parents:
265
diff
changeset
|
42 ############################################################################## |
|
96310b1c88fa
feeds: Improve config resiliency.
Matti Hamalainen <ccr@tnsp.org>
parents:
265
diff
changeset
|
43 |
|
321
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
44 proc fetch_dorequest { urlStr urlStatus urlSCode urlCode urlData urlMeta } { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
45 upvar 1 $urlStatus ustatus |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
46 upvar 1 $urlSCode uscode |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
47 upvar 1 $urlCode ucode |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
48 upvar 1 $urlData udata |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
49 upvar 1 $urlMeta umeta |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
50 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
51 if {[catch {set utoken [::http::geturl $urlStr -timeout 6000 -binary 1 -headers {Accept-Encoding identity}]} uerrmsg]} { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
52 puts "HTTP request failed: $uerrmsg" |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
53 return 0 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
54 } |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
55 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
56 set ustatus [::http::status $utoken] |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
57 if {$ustatus == "timeout"} { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
58 puts "HTTP request timed out ($urlStr)" |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
59 return 0 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
60 } |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
61 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
62 if {$ustatus != "ok"} { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
63 puts "Error in HTTP transaction: [::http::error $utoken] ($urlStr)" |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
64 return 0 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
65 } |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
66 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
67 set ustatus [::http::status $utoken] |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
68 set uscode [::http::code $utoken] |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
69 set ucode [::http::ncode $utoken] |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
70 set udata [::http::data $utoken] |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
71 array set umeta [::http::meta $utoken] |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
72 ::http::cleanup $utoken |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
73 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
74 return 1 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
75 } |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
76 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
77 |
|
139
3305e142eecc
Change feed fetcher to use SQLite3 backend.
Matti Hamalainen <ccr@tnsp.org>
parents:
114
diff
changeset
|
78 proc add_entry {uname uprefix uurl utitle} { |
| 142 | 79 global currclock feeds_db nitems |
|
292
9f90d6918626
feeds: Also use the html entity conversion from utillib here.
Matti Hamalainen <ccr@tnsp.org>
parents:
271
diff
changeset
|
80 set utmp [utl_convert_html_ent $uurl] |
| 147 | 81 if {[string match "http://*" $utmp] || [string match "https://*" $utmp]} { |
| 82 set utest "$utmp" | |
| 83 } else { | |
|
321
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
84 if {[string range $uprefix end end] != "/" && [string range $utmp 0 0] != "/"} { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
85 set utest "$uprefix/$utmp" |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
86 } else { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
87 set utest "$uprefix$utmp" |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
88 } |
| 147 | 89 } |
|
139
3305e142eecc
Change feed fetcher to use SQLite3 backend.
Matti Hamalainen <ccr@tnsp.org>
parents:
114
diff
changeset
|
90 |
| 296 | 91 set usql "SELECT title FROM feeds WHERE url='[utl_escape $utest]' AND feed='[utl_escape $uname]'" |
|
140
b0648e05c855
Change some variable names, etc.
Matti Hamalainen <ccr@tnsp.org>
parents:
139
diff
changeset
|
92 if {![feeds_db exists $usql]} { |
|
321
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
93 # puts "NEW: $utest : $utitle" |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
94 set usql "INSERT INTO feeds (feed,utime,url,title) VALUES ('[utl_escape $uname]', $currclock, '[utl_escape $utest]', '[utl_escape [utl_convert_html_ent $utitle]]')" |
| 142 | 95 incr nitems |
|
140
b0648e05c855
Change some variable names, etc.
Matti Hamalainen <ccr@tnsp.org>
parents:
139
diff
changeset
|
96 if {[catch {feeds_db eval $usql} uerrmsg]} { |
|
139
3305e142eecc
Change feed fetcher to use SQLite3 backend.
Matti Hamalainen <ccr@tnsp.org>
parents:
114
diff
changeset
|
97 puts "\nError: $uerrmsg on:\n$usql" |
|
3305e142eecc
Change feed fetcher to use SQLite3 backend.
Matti Hamalainen <ccr@tnsp.org>
parents:
114
diff
changeset
|
98 exit 15 |
|
3305e142eecc
Change feed fetcher to use SQLite3 backend.
Matti Hamalainen <ccr@tnsp.org>
parents:
114
diff
changeset
|
99 } |
| 63 | 100 } |
| 0 | 101 } |
| 102 | |
| 103 | |
| 104 proc add_rss_feed {datauri dataname dataprefix} { | |
|
321
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
105 if {[catch {set utoken [::http::geturl $datauri -binary 1 -timeout 6000 -headers {Accept-Encoding identity}]} uerrmsg]} { |
| 63 | 106 puts "Error getting $datauri: $uerrmsg" |
| 107 return 1 | |
| 108 } | |
| 109 set upage [::http::data $utoken] | |
| 110 ::http::cleanup $utoken | |
| 111 | |
| 112 set umatches [regexp -all -nocase -inline -- "<item>.\*\?<title><..CDATA.(.\*\?)\\\]\\\]></title>.\*\?<link>(http.\*\?)</link>.\*\?</item>" $upage] | |
| 113 set nmatches [llength $umatches] | |
| 114 for {set n 0} {$n < $nmatches} {incr n 3} { | |
| 115 add_entry $dataname $dataprefix [lindex $umatches [expr $n+2]] [lindex $umatches [expr $n+1]] | |
| 116 } | |
| 117 | |
| 118 if {$nmatches == 0} { | |
| 119 set umatches [regexp -all -nocase -inline -- "<item>.\*\?<title>(.\*\?)</title>.\*\?<link>(http.\*\?)</link>.\*\?</item>" $upage] | |
| 120 set nmatches [llength $umatches] | |
| 121 for {set n 0} {$n < $nmatches} {incr n 3} { | |
| 122 add_entry $dataname $dataprefix [lindex $umatches [expr $n+2]] [lindex $umatches [expr $n+1]] | |
| 123 } | |
| 124 } | |
| 0 | 125 |
| 63 | 126 if {$nmatches == 0} { |
| 127 set umatches [regexp -all -nocase -inline -- "<item \[^>\]*>.\*\?<title>(.\*\?)</title>.\*\?<link>(http.\*\?)</link>.\*\?</item>" $upage] | |
| 128 set nmatches [llength $umatches] | |
| 129 for {set n 0} {$n < $nmatches} {incr n 3} { | |
| 130 add_entry $dataname $dataprefix [lindex $umatches [expr $n+2]] [lindex $umatches [expr $n+1]] | |
| 131 } | |
| 132 } | |
| 143 | 133 |
| 63 | 134 return 0 |
| 0 | 135 } |
| 136 | |
| 137 | |
| 138 ############################################################################## | |
|
69
df3230f8aa46
Translate some comments to english and cosmetic fixes.
Matti Hamalainen <ccr@tnsp.org>
parents:
63
diff
changeset
|
139 ### Fetch and parse Halla-aho's blog page data |
|
321
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
140 proc fetch_halla_aho { } { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
141 set datauri "http://www.halla-aho.com/scripta/"; |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
142 set dataname "Mestari" |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
143 if {![fetch_dorequest $datauri ustatus uscode ucode upage umeta]} { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
144 return 0 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
145 } |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
146 |
| 63 | 147 set umatches [regexp -all -nocase -inline -- "<a href=\"(\[^\"\]+\.html)\"><b>(\[^<\]+)</b>" $upage] |
| 148 set nmatches [llength $umatches] | |
| 149 for {set n 0} {$n < $nmatches} {incr n 3} { | |
| 150 add_entry $dataname $datauri [lindex $umatches [expr $n+1]] [lindex $umatches [expr $n+2]] | |
| 151 } | |
| 0 | 152 |
| 63 | 153 set umatches [regexp -all -nocase -inline -- "<a href=\"(\[^\"\]+\.html)\">(\[^<\]\[^b\]\[^<\]+)</a>" $upage] |
| 154 set nmatches [llength $umatches] | |
| 155 for {set n 0} {$n < $nmatches} {incr n 3} { | |
| 156 add_entry $dataname $datauri [lindex $umatches [expr $n+1]] [lindex $umatches [expr $n+2]] | |
| 157 } | |
| 0 | 158 } |
| 159 | |
| 160 | |
| 161 ### The Adventurers | |
|
321
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
162 proc fetch_adventurers { } { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
163 set datauri "http://www.peldor.com/chapters/index_sidebar.html"; |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
164 set dataname "The Adventurers" |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
165 if {![fetch_dorequest $datauri ustatus uscode ucode upage umeta]} { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
166 return 0 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
167 } |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
168 |
| 63 | 169 set umatches [regexp -all -nocase -inline -- "<a href=\"(\[^\"\]+)\">(\[^<\]+)</a>" $upage] |
| 170 set nmatches [llength $umatches] | |
| 171 for {set n 0} {$n < $nmatches} {incr n 3} { | |
| 172 add_entry $dataname "http://www.peldor.com/" [lindex $umatches [expr $n+1]] [lindex $umatches [expr $n+2]] | |
| 173 } | |
| 0 | 174 } |
| 175 | |
| 176 | |
| 177 ### Order of the Stick | |
|
321
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
178 proc fetch_oots { } { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
179 set datauri "http://www.giantitp.com/comics/oots.html"; |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
180 set dataname "OOTS" |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
181 if {![fetch_dorequest $datauri ustatus uscode ucode upage umeta]} { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
182 return 0 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
183 } |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
184 |
| 63 | 185 set umatches [regexp -all -nocase -inline -- "<a href=\"(/comics/oots\[0-9\]+\.html)\">(\[^<\]+)</a>" $upage] |
| 186 set nmatches [llength $umatches] | |
| 187 for {set n 0} {$n < $nmatches} {incr n 3} { | |
| 188 add_entry $dataname "http://www.giantitp.com" [lindex $umatches [expr $n+1]] [lindex $umatches [expr $n+2]] | |
| 189 } | |
| 0 | 190 } |
| 191 | |
| 192 | |
|
350
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
193 ### Poliisi tiedotteet |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
194 proc fetch_poliisi { datauri dataname dataprefix } { |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
195 if {![fetch_dorequest $datauri ustatus uscode ucode upage umeta]} { |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
196 return 0 |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
197 } |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
198 |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
199 set umatches [regexp -all -nocase -inline -- "<div class=\"channelitem\"><div class=\"date\">(.*?)</div><a class=\"article\" href=\"(\[^\"\]+)\">(\[^<\]+)</a>" $upage] |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
200 set nmatches [llength $umatches] |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
201 for {set n 0} {$n < $nmatches} {incr n 4} { |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
202 set stmp [string trim [lindex $umatches [expr $n+3]]] |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
203 add_entry $dataname $dataprefix [lindex $umatches [expr $n+2]] "[lindex $umatches [expr $n+1]]: $stmp" |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
204 } |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
205 } |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
206 |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
207 |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
208 |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
209 |
|
321
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
210 ### Open database, etc |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
211 set nitems 0 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
212 set currclock [clock seconds] |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
213 global feeds_db |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
214 if {[catch {sqlite3 feeds_db $feeds_dbfile} uerrmsg]} { |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
215 puts "Could not open SQLite3 database '$feeds_dbfile': $uerrmsg." |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
216 exit 2 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
217 } |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
218 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
219 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
220 ### Fetch the feeds |
|
350
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
221 fetch_poliisi "http://www.poliisi.fi/oulu/tiedotteet/1/0?all1/0" "Poliisi/Oulu" "http://www.poliisi.fi" |
|
51c08336d7b1
feeds: Add support for Poliisi.fi information reports.
Matti Hamalainen <ccr@tnsp.org>
parents:
323
diff
changeset
|
222 |
|
321
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
223 fetch_halla_aho |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
224 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
225 fetch_adventurers |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
226 |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
227 fetch_oots |
|
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
228 |
| 143 | 229 #add_rss_feed "http://www.kaleva.fi/rss/145.xml" "Kaleva/Tiede" "" |
| 0 | 230 |
| 231 add_rss_feed "http://www.effi.org/xml/uutiset.rss" "EFFI" "" | |
| 232 | |
| 143 | 233 add_rss_feed "http://static.mtv3.fi/rss/uutiset_rikos.rss" "MTV3/Rikos" "" |
| 0 | 234 |
|
321
d8b957796121
feeds: Refactor the feeds fetching.
Matti Hamalainen <ccr@tnsp.org>
parents:
296
diff
changeset
|
235 #add_rss_feed "http://www.blastwave-comic.com/rss/blastwave.xml" "Blastwave" "" |
| 0 | 236 |
| 237 #add_rss_feed "http://lehti.samizdat.info/feed/" "Lehti" "" | |
| 238 | |
|
139
3305e142eecc
Change feed fetcher to use SQLite3 backend.
Matti Hamalainen <ccr@tnsp.org>
parents:
114
diff
changeset
|
239 |
|
3305e142eecc
Change feed fetcher to use SQLite3 backend.
Matti Hamalainen <ccr@tnsp.org>
parents:
114
diff
changeset
|
240 ### Close database |
|
140
b0648e05c855
Change some variable names, etc.
Matti Hamalainen <ccr@tnsp.org>
parents:
139
diff
changeset
|
241 feeds_db close |
| 142 | 242 |
| 243 puts "$nitems new items." |