summaryrefslogtreecommitdiff
path: root/localwebstats
blob: 819e77720bdc9901301004add688b4b0fc75f185 (plain)
  1. #!/bin/bash
  2. # /etc/cron.daily/localstats: Webstats maintenance script
  3. # Written by Jonas Smedegaard <dr@jones.dk>
  4. # halt on errors (NB! this is a bashism...)
  5. set -e
  6. function usage() {
  7. echo "Usage: $(basename $0) init|update|prelogrotate|postlogrotate|ignore <website> [<website>...]"
  8. echo " If no website is given, all are attempted"
  9. echo " Tip: Automagically runs when symlinked to /etc/cron.{daily,weekly,monthly}/"
  10. exit 1
  11. }
  12. function exit1() {
  13. echo "Error: $1"
  14. echo "Exiting..."
  15. exit 1
  16. }
  17. # automagically configure when run from cron dirs
  18. case $(dirname $0) in
  19. /etc/cron.daily)
  20. stamp=update
  21. ;;
  22. /etc/cron.weekly)
  23. stamp=ignore
  24. ;;
  25. /etc/cron.monthly)
  26. stamp=ignore
  27. ;;
  28. *)
  29. stamp=$1
  30. shift || usage
  31. ;;
  32. esac
  33. case "$stamp" in
  34. init|update|prelogrotate|postlogrotate|ignore)
  35. ;;
  36. *)
  37. usage
  38. ;;
  39. esac
  40. if [ "$stamp" = "ignore" ]; then
  41. if [ $DEBUG ]; then
  42. echo "Asked to ignore - exiting silently..."
  43. fi
  44. exit 0
  45. fi
  46. WEBROOT=/var/www
  47. LOGROOT=/var/log/apache
  48. # Options: $1=LOGROOT, $2=WEBROOT, $3=WEBSITE
  49. function statsdir() { echo $2/VIRTUAL/stats.$(dnsdomainname)/www/$3; }
  50. function webdirs() { find $1 -type d -mindepth 1 -maxdepth 1 | grep '\.*\.' | sed 's!$1!!'; }
  51. #function logfiles() { $(ls -r $LOGDIR/*-access*.gz) $(ls -r $LOGDIR/access*.??.gz) $(ls -r $LOGDIR/access*.?.gz) $(ls -r $LOGDIR/access*.?); }
  52. function logcontentresolved() { for file in $(find $1/$3 -name '????.??.00.gz' -type f -mindepth 1 -maxdepth 1 -follow); do zcat $file; done; for file in $(find $1/$3 -name '????.??.00' -type f -mindepth 1 -maxdepth 1 -follow); do cat $file; done; }
  53. function logcontent() { for file in $(find $1/$3 -name '????.??.??.gz' ! -name '*00.gz' -type f -mindepth 1 -maxdepth 1 -follow); do zcat $file; done; for file in $(find $1/$3 -name '????.??.??' ! -name '*00' -type f -mindepth 1 -maxdepth 1 -follow); do cat $file; done; }
  54. function host() { cat $2/VIRTUAL/$3/hostname || exit1 "Unable to get hostname for virtual host."; }
  55. function domain() { cat $2/VIRTUAL/$3/domainname || exit1 "Unable to get domainname for virtual host."; }
  56. function analog_cfg() { echo /etc/analog_$3.conf; }
  57. function rmagic_cfg() { echo /etc/rmagic/rmagic_$3.conf; }
  58. function webalizer_cfg() { echo /etc/webalizer_$3.conf; }
  59. function pre_init() { true; }
  60. function post_init() { true; }
  61. function pre_update() { true; }
  62. function post_update() { true; }
  63. # The above can be overridden in /etc/local/www
  64. . /etc/local/www || exit1 "Unable to read prefs file"
  65. # variables and functions too boring to be configurable
  66. JDRESOLVE_BIN="/usr/bin/jdresolve"
  67. JDRESOLVE_DB="/var/cache/jdresolve/hosts.db"
  68. #JDRESOLVE_OPTIONS="-r -t 5 --database=$JDRESOLVE_DB --dbfirst --expiredb=48"
  69. # DB access doesn't work currently (no output...)
  70. #JDRESOLVE_OPTIONS="-r --database=$JDRESOLVE_DB --dbfirst --expiredb=48"
  71. JDRESOLVE_OPTIONS="-r"
  72. JDRESOLVE_MERGE_OPTIONS="--mergedb --database=$JDRESOLVE_DB"
  73. LOGRESOLVE_BIN="/usr/sbin/logresolve"
  74. ANALOG_BIN="/usr/bin/analog"
  75. RMAGIC_BIN="/usr/bin/rmagic"
  76. WEBALIZER_BIN="/usr/bin/webalizer"
  77. AWSTATS_BIN="/usr/lib/cgi-bin/awstats.pl"
  78. function awstats_setlog() { sed -e "s!^\(LogFile=\).*\$!\\1$2!" $1 > $1.tmp; mv $1.tmp $1; }
  79. # Webiste/independent checks
  80. test -d $WEBROOT || exit1 "Webroot \"$WEBROOT\" doesn't exist"
  81. test -d $LOGROOT || exit1 "Logroot \"$LOGROOT\" doesn't exist"
  82. if [ -x $JDRESOLVE_BIN ]; then
  83. touch $JDRESOLVE_DB || exit1 "Couldn't touch JDRESOLVE_DB"
  84. fi
  85. # Generate stats for websites from stdin or all default sites
  86. WEBSITES=$@
  87. if [ "$WEBSITES" = "" ]; then
  88. WEBSITES=$(webdirs $LOGROOT $WEBROOT $WEBSITE)
  89. fi
  90. for WEBSITE in $WEBSITES; do
  91. STATSDIR=$(statsdir $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to resolve STATSDIR."
  92. # FIXME test -d $STATSDIR/.. || exit1 "Directory above STATSDIR doesn't exist."
  93. ANALOG_CFG=$(analog_cfg $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to resolve ANALOG_CFG."
  94. RMAGIC_CFG=$(rmagic_cfg $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to resolve RMAGIC_CFG."
  95. WEBALIZER_CFG=$(webalizer_cfg $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to resolve WEBALIZER_CFG."
  96. if [ $stamp = "init" ]; then
  97. pre_init $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing PRE_INIT."
  98. fi
  99. if [ $stamp = "update" ]; then
  100. pre_update $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing PRE_UPDATE."
  101. fi
  102. HOST=$(host $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to get hostname for virtual host."
  103. DOMAIN=$(domain $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to get domainname for virtual host."
  104. ANALOG_OPTIONS="-G +g/etc/analog.conf +g$ANALOG_CFG +A -a"
  105. RMAGIC_OPTIONS=""
  106. WEBALIZER_OPTIONS="-c /etc/webalizer.conf -c $WEBALIZER_CFG -o $STATSDIR/webalizer -f"
  107. AWSTATS_OPTIONS="-update"
  108. FQDN_ESC=`echo "$HOST.$DOMAIN" | sed -e 's/\./\\\./g'` # needed for awstats config
  109. if [ $DEBUG ]; then
  110. echo "Making stats for $WEBSITE in $STATSDIR:"
  111. JDRESOLVE_OPTIONS="$JDRESOLVE_OPTIONS -p"
  112. ANALOG_OPTIONS="$ANALOG_OPTIONS +q"
  113. RMAGIC_OPTIONS="$RMAGIC_OPTIONS"
  114. WEBALIZER_OPTIONS="$WEBALIZER_OPTIONS -T"
  115. AWSTATS_OPTIONS="$AWSTATS_OPTIONS -showsteps"
  116. else
  117. JDRESOLVE_OPTIONS="$JDRESOLVE_DB -n"
  118. ANALOG_OPTIONS="$ANALOG_OPTIONS -q"
  119. RMAGIC_OPTIONS="$RMAGIC_OPTIONS -statistics_Verbose=NONE"
  120. WEBALIZER_OPTIONS="$WEBALIZER_OPTIONS -Q"
  121. AWSTATS_OPTIONS="$AWSTATS_OPTIONS"
  122. fi
  123. if [ $stamp = "init" -o $stamp = "init" ]; then
  124. [ $DEBUG ] && echo "$WEBSITE: Purge STATSDIR"
  125. rm -rf $STATSDIR
  126. mkdir $STATSDIR
  127. fi
  128. LOGDATARESOLVED="$STATSDIR/rawlog_old.txt"
  129. LOGDATA="$STATSDIR/rawlog_new.txt"
  130. LOGDATATMP="$STATSDIR/rawlog_incoming.txt"
  131. touch $LOGDATARESOLVED $LOGDATA $LOGDATATMP || exit1 "Couldn't touch LOGDATA files."
  132. if [ $stamp = "init" ]; then
  133. logcontentresolved $LOGROOT $WEBROOT $WEBSITE >> $LOGDATARESOLVED
  134. if [ -x $ANALOG_BIN ]; then
  135. [ $DEBUG ] && echo "$WEBSITE: Create/update analog config"
  136. echo "\
  137. # NB! This file is automatically generated. Do not edit directly!
  138. # Instead, put additions/overrides in $ANALOG_CFG.local
  139. HOSTNAME $HOST.$DOMAIN
  140. HOSTURL http://$HOST.$DOMAIN/
  141. BASEURL http://$HOST.$DOMAIN
  142. LANGUAGE DANISH
  143. "\
  144. > $ANALOG_CFG
  145. [ -s $ANALOG_CFG.local ] && cat $ANALOG_CFG.local >> $ANALOG_CFG
  146. mkdir $STATSDIR/analog
  147. [ $DEBUG ] && echo "$WEBSITE: Create initial analog stats"
  148. $ANALOG_BIN $ANALOG_OPTIONS -C"LOGFILE none" -C"LOGFILE $LOGDATARESOLVED" -C"CACHEOUTFILE $STATSDIR/analog/cache.data" -C"OUTFILE $STATSDIR/analog/index.html"
  149. fi
  150. if [ -x $RMAGIC_BIN -a -x $ANALOG_BIN ]; then
  151. [ $DEBUG ] && echo "$WEBSITE: Create/update Report Magic config"
  152. echo "\
  153. # NB! This file is automatically generated. Do not edit directly!
  154. # Instead, put additions/overrides in $RMAGIC_CFG.local
  155. [statistics]
  156. File_In = $STATSDIR/rmagic/report.dat
  157. Frame_File_Out = $STATSDIR/rmagic/index.html
  158. Language = en
  159. [reports]
  160. File_Out = $STATSDIR/rmagic/
  161. [QUICK]
  162. Rows = ALL
  163. [navigation]
  164. File_Out = navfile.html
  165. "\
  166. > $RMAGIC_CFG
  167. [ -s $RMAGIC_CFG.local ] && $RMAGIC_OPTIONS="$RMAGIC_OPTIONS -statistics_Include=$RMAGIC_CFG.local"
  168. mkdir $STATSDIR/rmagic
  169. [ $DEBUG ] && echo "$WEBSITE: Create initial Report Magic stats"
  170. $ANALOG_BIN $ANALOG_OPTIONS -C"LOGFILE none" -C"CACHEFILE $STATSDIR/analog/cache.data" -C"LANGUAGE ENGLISH" -C"OUTPUT COMPUTER" -C"OUTFILE $STATSDIR/rmagic/report.dat"
  171. $RMAGIC_BIN $RMAGIC_OPTIONS $RMAGIC_CFG
  172. fi
  173. if [ -x $WEBALIZER_BIN ]; then
  174. [ $DEBUG ] && echo "$WEBSITE: Create/update Webalizer config"
  175. echo "\
  176. # NB! This file is automatically generated. Do not edit directly!
  177. # Instead, put additions/overrides in $WEBALIZER_CFG.local
  178. HostName $HOST.$DOMAIN
  179. HideSite *$DOMAIN
  180. HideReferrer $DOMAIN/
  181. "\
  182. > $WEBALIZER_CFG
  183. [ -s $WEBALIZER_CFG.local ] && cat $WEBALIZER_CFG.local >> $WEBALIZER_CFG
  184. mkdir $STATSDIR/webalizer
  185. [ $DEBUG ] && echo "$WEBSITE: Create initial Webalizer stats"
  186. cat $LOGDATARESOLVED | $WEBALIZER_BIN $WEBALIZER_OPTIONS -N 0 - || true
  187. fi
  188. if [ -x $AWSTATS_BIN ]; then
  189. [ $DEBUG ] && echo "$WEBSITE: Create/update AWStats config"
  190. echo "\
  191. # NB! This file is automatically generated. Do not edit directly!
  192. # Instead, put additions/overrides in $AWSTATS_CFG.local
  193. LogFile=\"$LOGDATATMP\"
  194. LogFormat=4
  195. DNSLookup=0
  196. DirData=\"$STATSDIR/awstats\"
  197. AllowToUpdateStatsFromBrowser=0
  198. DirCgi=\"http://cgi.jones.dk/cgi-bin\"
  199. DirIcons=\"http://stats.jones.dk/awstats-icon\"
  200. SiteDomain=\"$FQDN_ESC\"
  201. HostAliases=\"$FQDN_ESC\"
  202. Lang=\"dk\"
  203. DirLang=\"/usr/share/awstats/lang\"
  204. DefaultFile=\"index.html\"
  205. SkipHosts=\"\"
  206. SkipFiles=\"\"
  207. ShowLinksOnUrl=1
  208. ShowFlagLinks=0
  209. "\
  210. >/etc/awstats/awstats.$WEBSITE.conf
  211. [ -s $WEBALIZER_CFG.local ] && cat $WEBALIZER_CFG.local >> $WEBALIZER_CFG
  212. mkdir $STATSDIR/awstats
  213. [ $DEBUG ] && echo "$WEBSITE: Create initial AWStats stats"
  214. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATARESOLVED
  215. $AWSTATS_BIN -config=$WEBSITE $AWSTATS_OPTIONS -output > $STATSDIR/awstats/index.html
  216. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATATMP
  217. fi
  218. [ $DEBUG ] && echo "$WEBSITE: Compress DNS-resolved logdata"
  219. gzip -9 $LOGDATARESOLVED
  220. fi
  221. if [ -x $JDRESOLVE_BIN ]; then
  222. [ $DEBUG ] && echo "$WEBSITE: DNS-resolve new logdata using jdresolve"
  223. # DB access doesn't work currently (no output...)
  224. # logcontent $LOGROOT $WEBROOT $WEBSITE | $JDRESOLVE_BIN $JDRESOLVE_OPTIONS - >> $LOGDATA
  225. # jdresolve $JDRESOLVE_MERGE_OPTIONS $LOGDATA
  226. logcontent $LOGROOT $WEBROOT $WEBSITE | $JDRESOLVE_BIN $JDRESOLVE_OPTIONS - >> $LOGDATA
  227. elif [ -x $LOGRESOLVE_BIN ]; then
  228. [ $DEBUG ] && echo "$WEBSITE: DNS-resolve new logdata using logresolve"
  229. logcontent $LOGROOT $WEBROOT $WEBSITE | $LOGRESOLVE_BIN >> $LOGDATA
  230. else
  231. [ $DEBUG ] && echo "$WEBSITE: Merging new logdata without DNS-resolving (no resolver found)"
  232. logcontent $LOGROOT $WEBROOT $WEBSITE >> $LOGDATA
  233. fi
  234. if [ -s $LOGDATA ]; then
  235. if [ -x $ANALOG_BIN ]; then
  236. [ $DEBUG ] && echo "$WEBSITE: Update analog stats"
  237. $ANALOG_BIN $ANALOG_OPTIONS -C"LOGFILE none" -C"CACHEFILE $STATSDIR/analog/cache.data" -C"LOGFILE $LOGDATA" -C"OUTFILE $STATSDIR/analog/index.html"
  238. fi
  239. if [ -x $RMAGIC_BIN -a -x $ANALOG_BIN ]; then
  240. [ $DEBUG ] && echo "$WEBSITE: Update Report Magic stats"
  241. $ANALOG_BIN $ANALOG_OPTIONS -C"LOGFILE none" -C"CACHEFILE $STATSDIR/analog/cache.data" -C"LOGFILE $LOGDATA" -C"LANGUAGE ENGLISH" -C"OUTPUT COMPUTER" -C"OUTFILE $STATSDIR/rmagic/report.dat"
  242. $RMAGIC_BIN $RMAGIC_OPTIONS $RMAGIC_CFG
  243. fi
  244. if [ -x $WEBALIZER_BIN ]; then
  245. [ $DEBUG ] && echo "$WEBSITE: Update Webalog stats"
  246. zcat -f $LOGDATA | $WEBALIZER_BIN $WEBALIZER_OPTIONS -N 0 -
  247. fi
  248. if [ -x $AWSTATS_BIN ]; then
  249. [ $DEBUG ] && echo "$WEBSITE: Update AWStats stats"
  250. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATA
  251. $AWSTATS_BIN -config=$WEBSITE $AWSTATS_OPTIONS -output > $STATSDIR/awstats/index.html
  252. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATATMP
  253. fi
  254. gzip -f9 $LOGDATA
  255. fi
  256. if [ $stamp = "init" ]; then
  257. post_init $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing POST_INIT."
  258. fi
  259. if [ $stamp = "update" ]; then
  260. post_update $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing POST_UPDATE."
  261. fi
  262. done