summaryrefslogtreecommitdiff
path: root/localwebstats
blob: ee1cb8dcdfea1f560e1dbb767a46971d02895c20 (plain)
  1. #!/bin/bash
  2. # /etc/cron.daily/localstats: Webstats maintenance script
  3. # Written by Jonas Smedegaard <dr@jones.dk>
  4. # halt on errors
  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. function statsdir() { echo $1/VIRTUAL/stats.$(dnsdomainname)/www/$2; }
  49. function webdirs() { find $LOGROOT -type d -mindepth 1 -maxdepth 1 | grep '\.*\.' | sed 's!$LOGROOT!!'; }
  50. #function logdir() { echo $LOGROOT/$WEBSITE; }
  51. #function logfiles() { $(ls -r $LOGDIR/*-access*.gz) $(ls -r $LOGDIR/access*.??.gz) $(ls -r $LOGDIR/access*.?.gz) $(ls -r $LOGDIR/access*.?); }
  52. function logfilesresolvedgz() { ls $1/$2/????.??.00.gz; }
  53. function logfilesresolved() { ls $1/$2/????.??.00; }
  54. function logfilesgz() { ls $1/$2/????.??.??.gz | grep -v '\.00\.gz$'; }
  55. function logfiles() { ls $1/$2/????.??.?? | grep -v '\.00$'; }
  56. function logcontentresolved() { for file in `logfilesresolvedgz $1 $2`; do zcat $file; done; for file in `logfilesresolved $1 $2`; do cat $file; done; }
  57. function logcontent() { for file in `logfilesgz $1 $2`; do zcat $file; done; for file in `logfiles $1 $2`; do cat $file; done; }
  58. function host() { cat $1/VIRTUAL/$2/hostname || exit1 "Unable to get hostname for virtual host."; }
  59. function domain() { cat $1/VIRTUAL/$2/domainname || exit1 "Unable to get domainname for virtual host."; }
  60. function analog_cfg() { echo /etc/analog_$2.conf; }
  61. function rmagic_cfg() { echo /etc/rmagic/rmagic_$2.conf; }
  62. function webalizer_cfg() { echo /etc/webalizer_$2.conf; }
  63. # The above can be overridden in /etc/local/www
  64. . /etc/local/www || exit1 "Unable to read prefs file"
  65. # variables and functions too boring to be configurable
  66. JDRESOLVE_BIN="/usr/bin/jdresolve"
  67. JDRESOLVE_DB="/var/cache/jdresolve/hosts.db"
  68. #JDRESOLVE_OPTIONS="-r -t 5 --database=$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)
  89. fi
  90. for WEBSITE in $WEBSITES; do
  91. STATSDIR=$(statsdir $WEBROOT $WEBSITE) || exit1 "Unable to resolve STATSDIR."
  92. # FIXME test -d $STATSDIR/.. || exit1 "Directory above STATSDIR doesn't exist."
  93. ANALOG_CFG=$(analog_cfg $WEBROOT $WEBSITE) || exit1 "Unable to resolve ANALOG_CFG."
  94. RMAGIC_CFG=$(rmagic_cfg $WEBROOT $WEBSITE) || exit1 "Unable to resolve RMAGIC_CFG."
  95. WEBALIZER_CFG=$(webalizer_cfg $WEBROOT $WEBSITE) || exit1 "Unable to resolve WEBALIZER_CFG."
  96. HOST=$(host $WEBROOT $WEBSITE) || exit1 "Unable to get hostname for virtual host."
  97. DOMAIN=$(domain $WEBROOT $WEBSITE) || exit1 "Unable to get domainname for virtual host."
  98. LOGFILESRESOLVEDGZ=$(logfilesresolvedgz $LOGROOT $WEBSITE)
  99. ANALOG_OPTIONS="-G +g/etc/analog.conf +g$ANALOG_CFG +A -a"
  100. RMAGIC_OPTIONS=""
  101. WEBALIZER_OPTIONS="-c /etc/webalizer.conf -c $WEBALIZER_CFG -o $STATSDIR/webalizer -f"
  102. AWSTATS_OPTIONS="-update"
  103. FQDN_ESC=`echo "$HOST.$DOMAIN" | sed -e 's/\./\\\./g'` # needed for awstats config
  104. if [ $DEBUG ]; then
  105. echo "Making stats for $WEBSITE in $STATSDIR:"
  106. JDRESOLVE_OPTIONS="$JDRESOLVE_OPTIONS -p"
  107. ANALOG_OPTIONS="$ANALOG_OPTIONS +q"
  108. RMAGIC_OPTIONS="$RMAGIC_OPTIONS"
  109. WEBALIZER_OPTIONS="$WEBALIZER_OPTIONS -T"
  110. AWSTATS_OPTIONS="$AWSTATS_OPTIONS -showsteps"
  111. else
  112. JDRESOLVE_OPTIONS="$JDRESOLVE_DB -n"
  113. ANALOG_OPTIONS="$ANALOG_OPTIONS -q"
  114. RMAGIC_OPTIONS="$RMAGIC_OPTIONS -statistics_Verbose=NONE"
  115. WEBALIZER_OPTIONS="$WEBALIZER_OPTIONS -Q"
  116. AWSTATS_OPTIONS="$AWSTATS_OPTIONS"
  117. fi
  118. if [ $stamp = "init" -o $stamp = "init" ]; then
  119. [ $DEBUG ] && echo "$WEBSITE: Purge STATSDIR"
  120. rm -rf $STATSDIR
  121. mkdir $STATSDIR
  122. fi
  123. LOGDATARESOLVED="$STATSDIR/rawlog_old.txt"
  124. LOGDATA="$STATSDIR/rawlog_new.txt"
  125. LOGDATATMP="$STATSDIR/rawlog_incoming.txt"
  126. touch $LOGDATARESOLVED $LOGDATA $LOGDATATMP || exit1 "Couldn't touch LOGDATA files."
  127. if [ $stamp = "init" ]; then
  128. logcontentresolved $LOGROOT $WEBSITE >> $LOGDATARESOLVED
  129. if [ -x $ANALOG_BIN ]; then
  130. [ $DEBUG ] && echo "$WEBSITE: Create/update analog config"
  131. echo "\
  132. # NB! This file is automatically generated. Do not edit directly!
  133. # Instead, put additions/overrides in $ANALOG_CFG.local
  134. HOSTNAME $HOST.$DOMAIN
  135. HOSTURL http://$HOST.$DOMAIN/
  136. BASEURL http://$HOST.$DOMAIN
  137. LANGUAGE DANISH
  138. "\
  139. > $ANALOG_CFG
  140. [ -s $ANALOG_CFG.local ] && cat $ANALOG_CFG.local >> $ANALOG_CFG
  141. mkdir $STATSDIR/analog
  142. [ $DEBUG ] && echo "$WEBSITE: Create initial analog stats"
  143. $ANALOG_BIN $ANALOG_OPTIONS -C"LOGFILE none" -C"LOGFILE $LOGDATARESOLVED" -C"CACHEOUTFILE $STATSDIR/analog/cache.data" -C"OUTFILE $STATSDIR/analog/index.html"
  144. fi
  145. if [ -x $RMAGIC_BIN -a -x $ANALOG_BIN ]; then
  146. [ $DEBUG ] && echo "$WEBSITE: Create/update Report Magic config"
  147. echo "\
  148. # NB! This file is automatically generated. Do not edit directly!
  149. # Instead, put additions/overrides in $RMAGIC_CFG.local
  150. [statistics]
  151. File_In = $STATSDIR/rmagic/report.dat
  152. Frame_File_Out = $STATSDIR/rmagic/index.html
  153. Language = en
  154. [reports]
  155. File_Out = $STATSDIR/rmagic/
  156. [QUICK]
  157. Rows = ALL
  158. [navigation]
  159. File_Out = navfile.html
  160. "\
  161. > $RMAGIC_CFG
  162. [ -s $RMAGIC_CFG.local ] && $RMAGIC_OPTIONS="$RMAGIC_OPTIONS -statistics_Include=$RMAGIC_CFG.local"
  163. mkdir $STATSDIR/rmagic
  164. [ $DEBUG ] && echo "$WEBSITE: Create initial Report Magic stats"
  165. $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"
  166. $RMAGIC_BIN $RMAGIC_OPTIONS $RMAGIC_CFG
  167. fi
  168. if [ -x $WEBALIZER_BIN ]; then
  169. [ $DEBUG ] && echo "$WEBSITE: Create/update Webalizer config"
  170. echo "\
  171. # NB! This file is automatically generated. Do not edit directly!
  172. # Instead, put additions/overrides in $WEBALIZER_CFG.local
  173. HostName $HOST.$DOMAIN
  174. HideSite *$DOMAIN
  175. HideReferrer $DOMAIN/
  176. "\
  177. > $WEBALIZER_CFG
  178. [ -s $WEBALIZER_CFG.local ] && cat $WEBALIZER_CFG.local >> $WEBALIZER_CFG
  179. mkdir $STATSDIR/webalizer
  180. [ $DEBUG ] && echo "$WEBSITE: Create initial Webalizer stats"
  181. cat $LOGDATARESOLVED | $WEBALIZER_BIN $WEBALIZER_OPTIONS -N 0 -
  182. fi
  183. if [ -x $AWSTATS_BIN ]; then
  184. [ $DEBUG ] && echo "$WEBSITE: Create/update AWStats config"
  185. echo "\
  186. # NB! This file is automatically generated. Do not edit directly!
  187. # Instead, put additions/overrides in $AWSTATS_CFG.local
  188. LogFile=\"$LOGDATATMP\"
  189. LogFormat=4
  190. DNSLookup=0
  191. DirData=\"/var/cache/awstats\"
  192. AllowToUpdateStatsFromBrowser=0
  193. DirCgi=\"http://cgi.jones.dk/cgi-bin\"
  194. DirIcons=\"http://stats.jones.dk/awstats-icon\"
  195. SiteDomain=\"$FQDN_ESC\"
  196. HostAliases=\"$FQDN_ESC\"
  197. Lang=\"dk\"
  198. DirLang=\"/usr/share/awstats/lang\"
  199. DefaultFile=\"index.html\"
  200. SkipHosts=\"\"
  201. SkipFiles=\"\"
  202. ShowLinksOnUrl=1
  203. ShowFlagLinks=0
  204. "\
  205. >/etc/awstats/awstats.$WEBSITE.conf
  206. [ -s $WEBALIZER_CFG.local ] && cat $WEBALIZER_CFG.local >> $WEBALIZER_CFG
  207. mkdir $STATSDIR/awstats
  208. [ $DEBUG ] && echo "$WEBSITE: Create initial AWStats stats"
  209. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATARESOLVED
  210. $AWSTATS_BIN -config=$WEBSITE $AWSTATS_OPTIONS -output > $STATSDIR/awstats/index.html
  211. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATATMP
  212. fi
  213. [ $DEBUG ] && echo "$WEBSITE: Compress DNS-resolved logdata"
  214. gzip -9 $LOGDATARESOLVED
  215. fi
  216. if [ -x $JDRESOLVE_BIN ]; then
  217. [ $DEBUG ] && echo "$WEBSITE: DNS-resolve new logdata using jdresolve"
  218. # DB access doesn't work currently (no output...)
  219. # logcontent $LOGROOT $WEBSITE | $JDRESOLVE_BIN $JDRESOLVE_OPTIONS - >> $LOGDATA
  220. # jdresolve $JDRESOLVE_MERGE_OPTIONS $LOGDATA
  221. logcontent $LOGROOT $WEBSITE | $JDRESOLVE_BIN $JDRESOLVE_OPTIONS - >> $LOGDATA
  222. elif [ -x $LOGRESOLVE_BIN ]; then
  223. [ $DEBUG ] && echo "$WEBSITE: DNS-resolve new logdata using logresolve"
  224. logcontent $LOGROOT $WEBSITE | $LOGRESOLVE_BIN >> $LOGDATA
  225. else
  226. [ $DEBUG ] && echo "$WEBSITE: Merging new logdata without DNS-resolving (no resolver found)"
  227. logcontent $LOGROOT $WEBSITE >> $LOGDATA
  228. fi
  229. if [ -s $LOGDATA ]; then
  230. if [ -x $ANALOG_BIN ]; then
  231. [ $DEBUG ] && echo "$WEBSITE: Update analog stats"
  232. $ANALOG_BIN $ANALOG_OPTIONS -C"LOGFILE none" -C"CACHEFILE $STATSDIR/analog/cache.data" -C"LOGFILE $LOGDATA" -C"OUTFILE $STATSDIR/analog/index.html"
  233. fi
  234. if [ -x $RMAGIC_BIN -a -x $ANALOG_BIN ]; then
  235. [ $DEBUG ] && echo "$WEBSITE: Update Report Magic stats"
  236. $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"
  237. $RMAGIC_BIN $RMAGIC_OPTIONS $RMAGIC_CFG
  238. fi
  239. if [ -x $WEBALIZER_BIN ]; then
  240. [ $DEBUG ] && echo "$WEBSITE: Update Webalog stats"
  241. zcat -f $LOGDATA | $WEBALIZER_BIN $WEBALIZER_OPTIONS -N 0 -
  242. fi
  243. if [ -x $AWSTATS_BIN ]; then
  244. [ $DEBUG ] && echo "$WEBSITE: Update AWStats stats"
  245. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATA
  246. $AWSTATS_BIN -config=$WEBSITE $AWSTATS_OPTIONS -output > $STATSDIR/awstats/index.html
  247. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATATMP
  248. fi
  249. gzip -f9 $LOGDATA
  250. fi
  251. done