summaryrefslogtreecommitdiff
path: root/localwebstats
blob: 2dec780923cd3d4e880cf3253f61920e3d53deca (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 webalizer_cfg() { echo /etc/webalizer_$2.conf; }
  62. # The above can be overridden in /etc/local/www
  63. . /etc/local/www || exit1 "Unable to read prefs file"
  64. # variables and functions too boring to be configurable
  65. JDRESOLVE_BIN="/usr/bin/jdresolve"
  66. JDRESOLVE_DB="/var/cache/jdresolve/hosts.db"
  67. #JDRESOLVE_OPTIONS="-r -t 5 --database=$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. WEBALIZER_BIN="/usr/bin/webalizer"
  75. AWSTATS_BIN="/usr/lib/cgi-bin/awstats.pl"
  76. function awstats_setlog() { sed -e "s!^\(LogFile=\).*\$!\\1$2!" $1 > $1.tmp; mv $1.tmp $1; }
  77. # Webiste/independent checks
  78. test -d $WEBROOT || exit1 "Webroot \"$WEBROOT\" doesn't exist"
  79. test -d $LOGROOT || exit1 "Logroot \"$LOGROOT\" doesn't exist"
  80. if [ -x $JDRESOLVE_BIN ]; then
  81. touch $JDRESOLVE_DB || exit1 "Couldn't touch JDRESOLVE_DB"
  82. fi
  83. # Generate stats for websites from stdin or all default sites
  84. WEBSITES=$@
  85. if [ "$WEBSITES" = "" ]; then
  86. WEBSITES=$(webdirs)
  87. fi
  88. for WEBSITE in $WEBSITES; do
  89. STATSDIR=$(statsdir $WEBROOT $WEBSITE) || exit1 "Unable to resolve STATSDIR."
  90. # FIXME test -d $STATSDIR/.. || exit1 "Directory above STATSDIR doesn't exist."
  91. ANALOG_CFG=$(analog_cfg $WEBROOT $WEBSITE) || exit1 "Unable to resolve ANALOG_CFG."
  92. WEBALIZER_CFG=$(webalizer_cfg $WEBROOT $WEBSITE) || exit1 "Unable to resolve WEBALIZER_CFG."
  93. HOST=$(host $WEBROOT $WEBSITE) || exit1 "Unable to get hostname for virtual host."
  94. DOMAIN=$(domain $WEBROOT $WEBSITE) || exit1 "Unable to get domainname for virtual host."
  95. LOGFILESRESOLVEDGZ=$(logfilesresolvedgz $LOGROOT $WEBSITE)
  96. ANALOG_OPTIONS="-G +g/etc/analog.conf +g$ANALOG_CFG +A -a"
  97. WEBALIZER_OPTIONS="-c /etc/webalizer.conf -c $WEBALIZER_CFG -o $STATSDIR/webalizer -f"
  98. AWSTATS_OPTIONS="-update"
  99. FQDN_ESC=`echo "$HOST.$DOMAIN" | sed -e 's/\./\\\./g'` # needed for awstats config
  100. if [ $DEBUG ]; then
  101. echo "Making stats for $WEBSITE in $STATSDIR:"
  102. JDRESOLVE_OPTIONS="$JDRESOLVE_OPTIONS -p"
  103. ANALOG_OPTIONS="$ANALOG_OPTIONS +q"
  104. WEBALIZER_OPTIONS="$WEBALIZER_OPTIONS -T"
  105. AWSTATS_OPTIONS="$AWSTATS_OPTIONS -showsteps"
  106. else
  107. JDRESOLVE_OPTIONS="$JDRESOLVE_DB -n"
  108. ANALOG_OPTIONS="$ANALOG_OPTIONS -q"
  109. WEBALIZER_OPTIONS="$WEBALIZER_OPTIONS -Q"
  110. AWSTATS_OPTIONS="$AWSTATS_OPTIONS"
  111. fi
  112. if [ $stamp = "init" -o $stamp = "init" ]; then
  113. [ $DEBUG ] && echo "$WEBSITE: Purge STATSDIR"
  114. rm -rf $STATSDIR
  115. mkdir $STATSDIR
  116. fi
  117. LOGDATARESOLVED="$STATSDIR/rawlog_old.txt"
  118. LOGDATA="$STATSDIR/rawlog_new.txt"
  119. LOGDATATMP="$STATSDIR/rawlog_incoming.txt"
  120. touch $LOGDATARESOLVED $LOGDATA $LOGDATATMP || exit1 "Couldn't touch LOGDATA files."
  121. if [ $stamp = "init" ]; then
  122. logcontentresolved $LOGROOT $WEBSITE >> $LOGDATARESOLVED
  123. if [ -x $ANALOG_BIN ]; then
  124. [ $DEBUG ] && echo "$WEBSITE: Create/update analog config"
  125. echo "\
  126. # NB! This file is automatically generated. Do not edit directly!
  127. # Instead, put additions/overrides in $ANALOG_CFG.local
  128. HOSTNAME $HOST.$DOMAIN
  129. HOSTURL http://$HOST.$DOMAIN/
  130. BASEURL http://$HOST.$DOMAIN
  131. LANGUAGE DANISH
  132. "\
  133. > $ANALOG_CFG
  134. [ -s $ANALOG_CFG.local ] && cat $ANALOG_CFG.local >> $ANALOG_CFG
  135. mkdir $STATSDIR/analog
  136. [ $DEBUG ] && echo "$WEBSITE: Create initial analog stats"
  137. $ANALOG_BIN $ANALOG_OPTIONS -C"LOGFILE none" -C"LOGFILE $LOGDATARESOLVED" -C"CACHEOUTFILE $STATSDIR/analog/cache.data" -C"OUTFILE $STATSDIR/analog/index.html"
  138. fi
  139. if [ -x $WEBALIZER_BIN ]; then
  140. [ $DEBUG ] && echo "$WEBSITE: Create/update Webalizer config"
  141. echo "\
  142. # NB! This file is automatically generated. Do not edit directly!
  143. # Instead, put additions/overrides in $WEBALIZER_CFG.local
  144. HostName $HOST.$DOMAIN
  145. HideSite *$DOMAIN
  146. HideReferrer $DOMAIN/
  147. "\
  148. > $WEBALIZER_CFG
  149. [ -s $WEBALIZER_CFG.local ] && cat $WEBALIZER_CFG.local >> $WEBALIZER_CFG
  150. mkdir $STATSDIR/webalizer
  151. [ $DEBUG ] && echo "$WEBSITE: Create initial Webalizer stats"
  152. cat $LOGDATARESOLVED | $WEBALIZER_BIN $WEBALIZER_OPTIONS -N 0 -
  153. fi
  154. if [ -x $AWSTATS_BIN ]; then
  155. [ $DEBUG ] && echo "$WEBSITE: Create/update AWStats config"
  156. echo "\
  157. # NB! This file is automatically generated. Do not edit directly!
  158. # Instead, put additions/overrides in $AWSTATS_CFG.local
  159. LogFile=\"$LOGDATATMP\"
  160. LogFormat=4
  161. DNSLookup=0
  162. DirData=\"/var/cache/awstats\"
  163. AllowToUpdateStatsFromBrowser=0
  164. DirCgi=\"http://cgi.jones.dk/cgi-bin\"
  165. DirIcons=\"http://stats.jones.dk/awstats-icon\"
  166. SiteDomain=\"$FQDN_ESC\"
  167. HostAliases=\"$FQDN_ESC\"
  168. Lang=\"dk\"
  169. DirLang=\"/usr/share/awstats/lang\"
  170. DefaultFile=\"index.html\"
  171. SkipHosts=\"\"
  172. SkipFiles=\"\"
  173. ShowLinksOnUrl=1
  174. ShowFlagLinks=0
  175. "\
  176. >/etc/awstats/awstats.$WEBSITE.conf
  177. [ -s $WEBALIZER_CFG.local ] && cat $WEBALIZER_CFG.local >> $WEBALIZER_CFG
  178. mkdir $STATSDIR/awstats
  179. [ $DEBUG ] && echo "$WEBSITE: Create initial AWStats stats"
  180. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATARESOLVED
  181. $AWSTATS_BIN -config=$WEBSITE $AWSTATS_OPTIONS -output > $STATSDIR/awstats/index.html
  182. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATATMP
  183. fi
  184. [ $DEBUG ] && echo "$WEBSITE: Compress DNS-resolved logdata"
  185. gzip -9 $LOGDATARESOLVED
  186. fi
  187. if [ -x $JDRESOLVE_BIN ]; then
  188. [ $DEBUG ] && echo "$WEBSITE: DNS-resolve new logdata using jdresolve"
  189. # DB access doesn't work currently (no output...)
  190. # logcontent $LOGROOT $WEBSITE | $JDRESOLVE_BIN $JDRESOLVE_OPTIONS - >> $LOGDATA
  191. # jdresolve $JDRESOLVE_MERGE_OPTIONS $LOGDATA
  192. logcontent $LOGROOT $WEBSITE | $JDRESOLVE_BIN $JDRESOLVE_OPTIONS - >> $LOGDATA
  193. elif [ -x $LOGRESOLVE_BIN ]; then
  194. [ $DEBUG ] && echo "$WEBSITE: DNS-resolve new logdata using logresolve"
  195. logcontent $LOGROOT $WEBSITE | $LOGRESOLVE_BIN >> $LOGDATA
  196. else
  197. [ $DEBUG ] && echo "$WEBSITE: Merging new logdata without DNS-resolving (no resolver found)"
  198. logcontent $LOGROOT $WEBSITE >> $LOGDATA
  199. fi
  200. if [ -s $LOGDATA ]; then
  201. if [ -x $ANALOG_BIN ]; then
  202. [ $DEBUG ] && echo "$WEBSITE: Update analog stats"
  203. $ANALOG_BIN $ANALOG_OPTIONS -C"LOGFILE none" -C"LOGFILE $LOGDATA" -C"CACHEFILE $STATSDIR/analog/cache.data" -C"OUTFILE $STATSDIR/analog/index.html"
  204. fi
  205. if [ -x $WEBALIZER_BIN ]; then
  206. [ $DEBUG ] && echo "$WEBSITE: Update Webalog stats"
  207. zcat -f $LOGDATA | $WEBALIZER_BIN $WEBALIZER_OPTIONS -N 0 -
  208. fi
  209. if [ -x $AWSTATS_BIN ]; then
  210. [ $DEBUG ] && echo "$WEBSITE: Update AWStats stats"
  211. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATA
  212. $AWSTATS_BIN -config=$WEBSITE $AWSTATS_OPTIONS -output > $STATSDIR/awstats/index.html
  213. awstats_setlog /etc/awstats/awstats.$WEBSITE.conf $LOGDATATMP
  214. fi
  215. gzip -9 $LOGDATA
  216. fi
  217. done