summaryrefslogtreecommitdiff
path: root/localwebstats
blob: c94fcb22eb64909d6e11a8001e13b632ce7b1722 (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. [ $DEBUG ] && echo "Asked to ignore - exiting silently..."
  42. exit 0
  43. fi
  44. LOGROOT=/var/log/apache
  45. WEBROOT=/var/www
  46. # Options: $1=LOGROOT, $2=WEBROOT, $3=WEBSITE
  47. function statsdir() { echo $2/VIRTUAL/stats.$(dnsdomainname)/www/$3; }
  48. function webdirs() { find $1 -type d -mindepth 1 -maxdepth 1 | grep '\.*\.' | sed 's!$1!!'; }
  49. #function logfiles() { $(ls -r $LOGDIR/*-access*.gz) $(ls -r $LOGDIR/access*.??.gz) $(ls -r $LOGDIR/access*.?.gz) $(ls -r $LOGDIR/access*.?); }
  50. 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; }
  51. 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; }
  52. function host() { cat $2/VIRTUAL/$3/hostname || exit1 "Unable to get hostname for virtual host."; }
  53. function domain() { cat $2/VIRTUAL/$3/domainname || exit1 "Unable to get domainname for virtual host."; }
  54. function analog_cfg() { echo /etc/analog_$3.conf; }
  55. function rmagic_cfg() { echo /etc/rmagic/rmagic_$3.conf; }
  56. function webalizer_cfg() { echo /etc/webalizer_$3.conf; }
  57. function pre_init() { true; }
  58. function post_init() { true; }
  59. function pre_update() { true; }
  60. function post_update() { true; }
  61. # The above can be overridden
  62. LOCALCONFIG=/etc/local/webstats.conf
  63. . $LOCALCONFIG || exit1 "Unable to read local config file $LOCALCONFIG"
  64. # variables and functions too boring to be configurable
  65. JDRESOLVE_BIN="/usr/bin/jdresolve"
  66. JDRESOLVE_DB="/var/cache/jdresolve/hosts.db"
  67. #JDRESOLVE_OPTIONS="-r -t 5 --database=$JDRESOLVE_DB --dbfirst --expiredb=48"
  68. # DB access doesn't work currently (no output...)
  69. #JDRESOLVE_OPTIONS="-r --database=$JDRESOLVE_DB --dbfirst --expiredb=48"
  70. JDRESOLVE_OPTIONS="-r"
  71. JDRESOLVE_MERGE_OPTIONS="--mergedb --database=$JDRESOLVE_DB"
  72. LOGRESOLVE_BIN="/usr/sbin/logresolve"
  73. ANALOG_BIN="/usr/bin/analog"
  74. RMAGIC_BIN="/usr/bin/rmagic"
  75. WEBALIZER_BIN="/usr/bin/webalizer"
  76. AWSTATS_BIN="/usr/lib/cgi-bin/awstats.pl"
  77. function awstats_setlog() { sed -e "s!^\(LogFile=\).*\$!\\1$2!" $1 > $1.tmp; mv $1.tmp $1; }
  78. # Webiste/independent checks
  79. test -d $WEBROOT || exit1 "Webroot \"$WEBROOT\" doesn't exist"
  80. test -d $LOGROOT || exit1 "Logroot \"$LOGROOT\" doesn't exist"
  81. if [ -x $JDRESOLVE_BIN ]; then
  82. touch $JDRESOLVE_DB || exit1 "Couldn't touch JDRESOLVE_DB"
  83. fi
  84. # Generate stats for websites from stdin or all default sites
  85. WEBSITES=$@
  86. if [ "$WEBSITES" = "" ]; then
  87. WEBSITES=$(webdirs $LOGROOT $WEBROOT $WEBSITE)
  88. fi
  89. for WEBSITE in $WEBSITES; do
  90. STATSDIR=$(statsdir $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to resolve STATSDIR."
  91. # FIXME test -d $STATSDIR/.. || exit1 "Directory above STATSDIR doesn't exist."
  92. ANALOG_CFG=$(analog_cfg $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to resolve ANALOG_CFG."
  93. RMAGIC_CFG=$(rmagic_cfg $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to resolve RMAGIC_CFG."
  94. WEBALIZER_CFG=$(webalizer_cfg $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to resolve WEBALIZER_CFG."
  95. if [ $stamp = "init" ]; then
  96. [ $DEBUG ] && echo "Execute $WEBSITE PRE_INIT"
  97. pre_init $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing PRE_INIT."
  98. fi
  99. if [ $stamp = "update" ]; then
  100. [ $DEBUG ] && echo "Execute $WEBSITE PRE_UPDATE"
  101. pre_update $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing PRE_UPDATE."
  102. fi
  103. HOST=$(host $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to get hostname for virtual host."
  104. DOMAIN=$(domain $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to get domainname for virtual host."
  105. ANALOG_OPTIONS="-G +g/etc/analog.conf +g$ANALOG_CFG +A -a"
  106. RMAGIC_OPTIONS=""
  107. WEBALIZER_OPTIONS="-c /etc/webalizer.conf -c $WEBALIZER_CFG -o $STATSDIR/webalizer -f"
  108. AWSTATS_OPTIONS="-update"
  109. FQDN_ESC=`echo "$HOST.$DOMAIN" | sed -e 's/\./\\\./g'` # needed for awstats config
  110. if [ $DEBUG ]; then
  111. echo "Making stats for $WEBSITE in $STATSDIR:"
  112. JDRESOLVE_OPTIONS="$JDRESOLVE_OPTIONS -p"
  113. ANALOG_OPTIONS="$ANALOG_OPTIONS +q"
  114. RMAGIC_OPTIONS="$RMAGIC_OPTIONS"
  115. WEBALIZER_OPTIONS="$WEBALIZER_OPTIONS -T"
  116. AWSTATS_OPTIONS="$AWSTATS_OPTIONS -showsteps"
  117. else
  118. JDRESOLVE_OPTIONS="$JDRESOLVE_DB -n"
  119. ANALOG_OPTIONS="$ANALOG_OPTIONS -q"
  120. RMAGIC_OPTIONS="$RMAGIC_OPTIONS -statistics_Verbose=NONE"
  121. WEBALIZER_OPTIONS="$WEBALIZER_OPTIONS -Q"
  122. AWSTATS_OPTIONS="$AWSTATS_OPTIONS"
  123. fi
  124. if [ $stamp = "init" ]; then
  125. [ $DEBUG ] && echo "$WEBSITE: Purge STATSDIR"
  126. rm -rf $STATSDIR
  127. mkdir $STATSDIR
  128. fi
  129. LOGDATARESOLVED="$STATSDIR/rawlog_old.txt"
  130. LOGDATA="$STATSDIR/rawlog_new.txt"
  131. LOGDATATMP="$STATSDIR/rawlog_incoming.txt"
  132. touch $LOGDATARESOLVED $LOGDATA $LOGDATATMP || exit1 "Couldn't touch LOGDATA files."
  133. if [ $stamp = "init" ]; then
  134. logcontentresolved $LOGROOT $WEBROOT $WEBSITE >> $LOGDATARESOLVED
  135. if [ -x $ANALOG_BIN ]; then
  136. [ $DEBUG ] && echo "$WEBSITE: Create/update analog config"
  137. echo "\
  138. # NB! This file is automatically generated. Do not edit directly!
  139. # Instead, put additions/overrides in $ANALOG_CFG.local
  140. HOSTNAME $HOST.$DOMAIN
  141. HOSTURL http://$HOST.$DOMAIN/
  142. BASEURL http://$HOST.$DOMAIN
  143. LANGUAGE DANISH
  144. "\
  145. > $ANALOG_CFG
  146. [ -s $ANALOG_CFG.local ] && cat $ANALOG_CFG.local >> $ANALOG_CFG
  147. mkdir $STATSDIR/analog
  148. [ $DEBUG ] && echo "$WEBSITE: Create initial analog stats"
  149. $ANALOG_BIN $ANALOG_OPTIONS -C"LOGFILE none" -C"LOGFILE $LOGDATARESOLVED" -C"CACHEOUTFILE $STATSDIR/analog/cache.data" -C"OUTFILE $STATSDIR/analog/index.html"
  150. fi
  151. if [ -x $RMAGIC_BIN -a -x $ANALOG_BIN ]; then
  152. [ $DEBUG ] && echo "$WEBSITE: Create/update Report Magic config"
  153. echo "\
  154. # NB! This file is automatically generated. Do not edit directly!
  155. # Instead, put additions/overrides in $RMAGIC_CFG.local
  156. [statistics]
  157. File_In = $STATSDIR/rmagic/report.dat
  158. Frame_File_Out = $STATSDIR/rmagic/index.html
  159. Language = en
  160. [reports]
  161. File_Out = $STATSDIR/rmagic/
  162. [QUICK]
  163. Rows = ALL
  164. [navigation]
  165. File_Out = navfile.html
  166. "\
  167. > $RMAGIC_CFG
  168. [ -s $RMAGIC_CFG.local ] && $RMAGIC_OPTIONS="$RMAGIC_OPTIONS -statistics_Include=$RMAGIC_CFG.local"
  169. mkdir $STATSDIR/rmagic
  170. [ $DEBUG ] && echo "$WEBSITE: Create initial Report Magic stats"
  171. $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"
  172. $RMAGIC_BIN $RMAGIC_OPTIONS $RMAGIC_CFG
  173. fi
  174. if [ -x $WEBALIZER_BIN ]; then
  175. [ $DEBUG ] && echo "$WEBSITE: Create/update Webalizer config"
  176. echo "\
  177. # NB! This file is automatically generated. Do not edit directly!
  178. # Instead, put additions/overrides in $WEBALIZER_CFG.local
  179. HostName $HOST.$DOMAIN
  180. HideSite *$DOMAIN
  181. HideReferrer $DOMAIN/
  182. "\
  183. > $WEBALIZER_CFG
  184. [ -s $WEBALIZER_CFG.local ] && cat $WEBALIZER_CFG.local >> $WEBALIZER_CFG
  185. mkdir $STATSDIR/webalizer
  186. [ $DEBUG ] && echo "$WEBSITE: Create initial Webalizer stats"
  187. cat $LOGDATARESOLVED | $WEBALIZER_BIN $WEBALIZER_OPTIONS -N 0 - || true
  188. fi
  189. if [ -x $AWSTATS_BIN ]; then
  190. [ $DEBUG ] && echo "$WEBSITE: Create/update AWStats config"
  191. echo "\
  192. # NB! This file is automatically generated. Do not edit directly!
  193. # Instead, put additions/overrides in $AWSTATS_CFG.local
  194. LogFile=\"$LOGDATATMP\"
  195. LogFormat=4
  196. DNSLookup=0
  197. DirData=\"$STATSDIR/awstats\"
  198. AllowToUpdateStatsFromBrowser=0
  199. DirCgi=\"http://cgi.jones.dk/cgi-bin\"
  200. DirIcons=\"http://stats.jones.dk/awstats-icon\"
  201. SiteDomain=\"$FQDN_ESC\"
  202. HostAliases=\"$FQDN_ESC\"
  203. Lang=\"dk\"
  204. DirLang=\"/usr/share/awstats/lang\"
  205. DefaultFile=\"index.html\"
  206. SkipHosts=\"\"
  207. SkipFiles=\"\"
  208. ShowLinksOnUrl=1
  209. ShowFlagLinks=0
  210. "\
  211. >/etc/awstats/awstats.$WEBSITE.conf
  212. [ -s $WEBALIZER_CFG.local ] && cat $WEBALIZER_CFG.local >> $WEBALIZER_CFG
  213. mkdir $STATSDIR/awstats
  214. [ $DEBUG ] && echo "$WEBSITE: Create initial AWStats stats"
  215. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATARESOLVED
  216. $AWSTATS_BIN -config=$WEBSITE $AWSTATS_OPTIONS -output > $STATSDIR/awstats/index.html
  217. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATATMP
  218. fi
  219. [ $DEBUG ] && echo "$WEBSITE: Compress DNS-resolved logdata"
  220. gzip -9 $LOGDATARESOLVED
  221. fi
  222. if [ -x $JDRESOLVE_BIN ]; then
  223. [ $DEBUG ] && echo "$WEBSITE: DNS-resolve new logdata using jdresolve"
  224. # DB access doesn't work currently (no output...)
  225. # logcontent $LOGROOT $WEBROOT $WEBSITE | $JDRESOLVE_BIN $JDRESOLVE_OPTIONS - >> $LOGDATA
  226. # jdresolve $JDRESOLVE_MERGE_OPTIONS $LOGDATA
  227. logcontent $LOGROOT $WEBROOT $WEBSITE | $JDRESOLVE_BIN $JDRESOLVE_OPTIONS - >> $LOGDATA
  228. elif [ -x $LOGRESOLVE_BIN ]; then
  229. [ $DEBUG ] && echo "$WEBSITE: DNS-resolve new logdata using logresolve"
  230. logcontent $LOGROOT $WEBROOT $WEBSITE | $LOGRESOLVE_BIN >> $LOGDATA
  231. else
  232. [ $DEBUG ] && echo "$WEBSITE: Merging new logdata without DNS-resolving (no resolver found)"
  233. logcontent $LOGROOT $WEBROOT $WEBSITE >> $LOGDATA
  234. fi
  235. if [ -s $LOGDATA ]; then
  236. if [ -x $ANALOG_BIN ]; then
  237. [ $DEBUG ] && echo "$WEBSITE: Update analog stats"
  238. $ANALOG_BIN $ANALOG_OPTIONS -C"LOGFILE none" -C"CACHEFILE $STATSDIR/analog/cache.data" -C"LOGFILE $LOGDATA" -C"OUTFILE $STATSDIR/analog/index.html"
  239. fi
  240. if [ -x $RMAGIC_BIN -a -x $ANALOG_BIN ]; then
  241. [ $DEBUG ] && echo "$WEBSITE: Update Report Magic stats"
  242. $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"
  243. $RMAGIC_BIN $RMAGIC_OPTIONS $RMAGIC_CFG
  244. fi
  245. if [ -x $WEBALIZER_BIN ]; then
  246. [ $DEBUG ] && echo "$WEBSITE: Update Webalog stats"
  247. zcat -f $LOGDATA | $WEBALIZER_BIN $WEBALIZER_OPTIONS -N 0 -
  248. fi
  249. if [ -x $AWSTATS_BIN ]; then
  250. [ $DEBUG ] && echo "$WEBSITE: Update AWStats stats"
  251. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATA
  252. $AWSTATS_BIN -config=$WEBSITE $AWSTATS_OPTIONS -output > $STATSDIR/awstats/index.html
  253. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATATMP
  254. fi
  255. gzip -f9 $LOGDATA
  256. fi
  257. if [ $stamp = "init" ]; then
  258. [ $DEBUG ] && echo "Execute $WEBSITE POST_INIT"
  259. post_init $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing POST_INIT."
  260. fi
  261. if [ $stamp = "update" ]; then
  262. [ $DEBUG ] && echo "Execute $WEBSITE POST_UPDATE"
  263. post_update $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing POST_UPDATE."
  264. fi
  265. done