#!/bin/bash
#
# Plugin to graph response times of the specified websites/URLs.
#
# Parameters:
#
#       config   (required)
#       autoconf (optional - used by lrrd-config)
#
# Configuration example:
#
# [wget_page]
# timeout 30
# env.names url1 url2
# env.timeout 20
# env.error_value 60
# env.max 120
#
# env.url_url1 http://www1.example.com/path1/page1
# env.label_url1 Example URL#1
# env.timeout_url1 10
# env.warning_url1 5
# env.critical_url1 8
#
# env.url_url2 https://www2.example.com/path2/page2
# env.label_url2 Example URL#2
# env.timeout_url2 30
# env.warning_url2 15
# env.critical_url2 20
# env.wget_opts_url2 --no-cache --tries=1 --no-check-certificate
#
# URL options:
#
# You can define the following options for each specified URL
# as seen in the above example.
#
# - url: the URL to be downloaded with Wget
# - label: the label assigned to the line of the given in URL in the graph
# - timeout: the value passed to Wget through the "--timeout" option.
# - warning: the value for the given URL that stands for the warning level
# - critical: the value for the given URL that stands for the critical level
# - max: the maximum value for the given URL (values above this will be
#     discarded)
# - error_value: the value for the given URL that will be used to mark when
#     Wget returned an error. A zero error_value causes the plugin to ignore
#     Wget's return value.
# - regex_error_value: the value for the given URL that will be used to mark
#     when a regular expression match failed (either for the HTTP response
#     headers or the body).
# - regex_header_<n>: a regular expression that the HTTP response header must
#     match or the plugin will return regex_error_value for the given URL.
#     By default the plugin uses egrep, thus extended regexps are expected.
#     You can define any number of regexps, but you've to start the index at
#     "1" and increase it sequentially.
#     I.e. regex_header_1, regex_header_2, ...
# - regex_body_<n>: same as regex_header_<n>, but matches the HTTP response
#     body.
# - grep_opts: various options supplied to grep for regexp matches for the
#     given URL. By default these are: -E (use of extended regexps)
#     and -i (case insensitive regexp matching)
# - wget_opts: various options supplied to the Wget command for the given URL.
# - join_lines: if "true" and regexp matching is applied, the HTTP response body
#     is stripped of newline ("\n") characters. This helps with complex regexps
#     since grep can match only in a single line at a time and it would not be
#     possible to match on complex HTML/XML structures otherwise.
#     This is enabled by default.
#
# $Log$
#
# Revision 1.0  2006/07/11 08:49:43 cipixul@gmail.com
# Initial version
#
# Revision 2.0  2010/03/25 13:46:13 muzso@muzso.hu
# Rewrote most of the code. Added multips-like options.
#
# Revision 2.1  2010/04/22 11:43:53 muzso@muzso.hu
# Added regular expression matching against the contents of the checked URL.
#
# Revision 2.2  2010/04/23 15:21:12 muzso@muzso.hu
# Bugfix. Regexp matching on HTTP response bodies with a trailing newline
# was flawed.
#
#%# family=auto
#%# capabilities=autoconf

[ -n "${wget_bin}" ] || wget_bin=$(which wget)
[ -n "${time_bin}" ] || time_bin=$(which time)
[ -n "${mktemp_bin}" ] || mktemp_bin=$(which mktemp)
[ -n "${grep_bin}" ] || grep_bin=$(which grep)
[ -n "${tail_bin}" ] || tail_bin=$(which tail)

default_error_value=30
default_regex_error_value=40
default_grep_opts="-E -i"
default_wget_opts="--no-cache --tries=1"
default_timeout=20
default_join_lines=true

if [ "${1}" = "autoconf" ]; then
  if [ -z "$wget_bin" ] || [ ! -f "$wget_bin" ] || [ ! -x "$wget_bin" ]; then
    echo "no (missing 'wget' executable)"
  elif [ -z "$time_bin" ] || [ ! -f "$time_bin" ] || [ ! -x "$time_bin" ]; then
    echo "no (missing 'time' executable)"
  elif [ -z "$mktemp_bin" ] || [ ! -f "$mktemp_bin" ] || [ ! -x "$mktemp_bin" ]; then
    echo "no (missing 'mktemp' executable)"
  elif [ -z "$grep_bin" ] || [ ! -f "$grep_bin" ] || [ ! -x "$grep_bin" ]; then
    echo "no (missing 'grep' executable)"
  elif [ -z "$tail_bin" ] || [ ! -f "$tail_bin" ] || [ ! -x "$tail_bin" ]; then
    echo "no (missing 'tail' executable)"
  else
    echo "yes"
  fi
  exit 0
fi

if [ -z "${names}" ]; then
  echo "Configuration required"
  exit 1
fi

[ -n "${error_value}" ] || error_value=${default_error_value}
[ -n "${regex_error_value}" ] || regex_error_value=${default_regex_error_value}
[ -n "${grep_opts}" ] || grep_opts=${default_grep_opts}
[ -n "${wget_opts}" ] || wget_opts=${default_wget_opts}
[ -n "${timeout}" ] || timeout=${default_timeout}
[ -n "${join_lines}" ] || join_lines=${default_join_lines}
[ -n "${warning}" ] || warning=$((timeout/2))
[ -n "${critical}" ] || critical=${timeout}
[ -n "${max}" ] || max=$((timeout*2))

if [ "${1}" = "config" ]; then
  echo "graph_title wget loadtime of webpages"
  echo "graph_args --base 1000 -l 0"
  echo "graph_scale no"
  echo "graph_vlabel Load time in seconds"
  echo "graph_category webserver"
  echo "graph_info This graph shows load time in seconds of one or more urls"
  I=1
  for name in ${names}; do
    eval iurl='${url_'${name}'}'
    if [ -n "${iurl}" ]; then
      eval ilabel='${label_'${name}':-url${I}}'
      eval iwarning='${warning_'${name}':-${warning}}'
      eval icritical='${critical_'${name}':-${critical}}'
      eval imax='${max_'${name}':-${max}}'
      cat << EOH
loadtime${I}.label ${ilabel}
loadtime${I}.info Load time for ${iurl}
loadtime${I}.min 0
loadtime${I}.max ${imax}
EOH
      [ ${iwarning} -gt 0 ] && echo "loadtime${I}.warning ${iwarning}"
      [ ${icritical} -gt 0 ] && echo "loadtime${I}.critical ${icritical}"
      I=$((I+1))
    fi
  done
  exit 0
fi

I=1
for name in ${names}; do
  eval iurl='${url_'${name}'}'
  if [ -n "${iurl}" ]; then
    eval ierror_value='${error_value_'${name}':-${error_value}}'
    eval iregex_error_value='${regex_error_value_'${name}':-${regex_error_value}}'
    eval igrep_opts='${grep_opts_'${name}':-${grep_opts}}'
    eval iwget_opts='${wget_opts_'${name}':-${wget_opts}}'
    eval iwget_post_data='${wget_post_data_'${name}':-${wget_post_data}}'
    eval ijoin_lines='${join_lines_'${name}':-${join_lines}}'
    eval itimeout='${timeout_'${name}':-${timeout}}'
    loadtime=""
    tempfile=$(mktemp)
    if [ -z "${iwget_post_data}" ]; then
      timing=$(${time_bin} -p ${wget_bin} --save-headers --no-directories --output-document "${tempfile}" --timeout ${itimeout} ${iwget_opts} "${iurl}" 2>&1)
    else
      timing=$(${time_bin} -p ${wget_bin} --post-data "${iwget_post_data}" --save-headers --no-directories --output-document "${tempfile}" --timeout ${itimeout} ${iwget_opts} "${iurl}" 2>&1)
    fi
    wget_result=$?
    if [ -f "${tempfile}" ]; then
      if [ ${wget_result} -ne 0 -a ${ierror_value} -gt 0 ]; then
        loadtime=${ierror_value}
      else
        tempheader=""
        tempbody=""
        K=0
        while [ -z "${loadtime}" ]; do
          K=$((K+1))
          eval iregex_header='${regex_header_'${K}'_'${name}':-${regex_header_'${K}'}}'
          eval iregex_body='${regex_body_'${K}'_'${name}':-${regex_body_'${K}'}}'
          [ -z "${iregex_header}" -a -z "${iregex_body}" ] && break
          if [ ${K} -eq 1 ]; then
            OIFS="${IFS}"
            # we skip carrige return characters from the end of header lines
            IFS=$(echo -en "\r")
            inheader=0
            # The "read" command reads only lines terminated by a specific
            # character (which by default the newline char).
            # To read the end of the file (the bytes after the last newline) too
            # we append a newline.
            echo "" >> "${tempfile}"
            while read -r line; do
              if [ -z "${line}" ]; then
                inheader=1
                # We reached the border of the header and the body.
                # Setting IFS to an empty string puts the entire read lines from
                # the body into our "line" variable.
                IFS=""
              else
                if [ ${inheader} -eq 0 ]; then
                  tempheader="${tempheader}${line}
"
                else
                  if [ "${ijoin_lines}" = "true" ]; then
                    tempbody="${tempbody}${line}"
                  else
                    tempbody="${tempbody}${line}
"
                  fi
                fi
              fi
            done < "${tempfile}"
            IFS="${OIFS}"
          fi
          if [ -n "${iregex_header}" ] && ! echo "${tempheader}" | ${grep_bin} -qs ${igrep_opts} "${iregex_header}" 2> /dev/null; then
            loadtime=${iregex_error_value}
          else
            if [ -n "${iregex_body}" ] && ! echo "${tempbody}" | ${grep_bin} -qs ${igrep_opts} "${iregex_body}" 2> /dev/null; then
              loadtime=${iregex_error_value}
            fi
          fi
        done
        if [ -z "${loadtime}" ]; then
          loadtime=$(echo "${timing}" | grep "^real *[0-9]" | cut -d ' ' -f 2)
        fi
      fi
      rm -f "${tempfile}" > /dev/null 2>&1
    else
      loadtime=$((ierror_value*2))
    fi
    echo "loadtime${I}.value ${loadtime}"
    I=$((I+1))
  fi
done