@@ -785,3 +785,76 @@ function kill_port {
785785 kill -9 $PID
786786 fi
787787}
788+
789+ # Kills the process listening on a port if the name matches the first argument.
790+ # usage:
791+ # kill_listener_port program_name port
792+ function kill_listener_port {
793+ CMDLINE=" $1 "
794+ PORT=" $2 "
795+ kill -9 $( lsof -t -i TCP:${PORT} -s TCP:LISTEN -a -c " /^${CMDLINE} $/" ) || true
796+ }
797+
798+ # Performs timed reads on the output from a command passed via $1. The stream
799+ # will be interpreted as a chunked http encoding. Each chunk will be allowed
800+ # at most threshold_sec ($2) seconds to be read or the function will fail. When
801+ # the stream is fully read, the funcion will compare the total number of http
802+ # chunks read with expect_chunk_count ($3) and fail on mismatch.
803+ # Usage:
804+ # check_flushing "curl -N --raw --silent --proxy $SECONDARY_HOSTNAME $URL" 5 1
805+ # This will check if the curl command resulted in single chunk which was read
806+ # within one second or less.
807+ function check_flushing() {
808+ local command=" $1 "
809+ local threshold_sec=" $2 "
810+ local expect_chunk_count=" $3 "
811+ local output=" "
812+ local start=$( date +%s%N)
813+ local chunk_count=0
814+
815+ echo " Command: $command "
816+
817+ if [ " ${USE_VALGRIND:- } " = true ]; then
818+ # We can't say much about correctness of timings under valgrind, so relax
819+ # the test for that.
820+ threshold_sec=$( echo " scale=2; $threshold_sec *10" | bc)
821+ fi
822+
823+ while true ; do
824+ start=$( date +%s%N)
825+ # Read the http chunk size from the stream. This is also the read which
826+ # checks timings.
827+ check read -t $threshold_sec line
828+ echo " Chunk number [$chunk_count ] has size: $line "
829+ line=$( echo $line | tr -d ' \n' | tr -d ' \r' )
830+ # If we read 0 that means we have finished reading the stream.
831+ if [ $(( 16 #$line )) -eq " 0" ; then
832+ check [ $expect_chunk_count -le $chunk_count ]
833+ return
834+ fi
835+ let chunk_count=chunk_count+1
836+ # read the actual data from the stream, using the amount indicated in
837+ # the previous read. This read should be fast.
838+ # Note that we need to clear IFS for read since otherwise it can get
839+ # confused by whitespace-only chunks.
840+ IFS= check read -N $(( 16 #$line ))
841+ echo " Chunk data: $line "
842+ # Read the trailing \r\n - should be fast.
843+ check read -N 2 line
844+ done < <( $command )
845+ check 0
846+ }
847+
848+ # Given the output of a page with ?PageSpeedFilters=+debug, print the section of
849+ # the page where it lists what filters are enabled.
850+ function extract_filters_from_debug_html() {
851+ local debug_output=" $1 "
852+
853+ # Pull out the non-blank lines between "Filters:" and Options:". First
854+ # convert newlines to % so sed can operate on the whole file, then put them
855+ # back again.
856+ check_from -q " $debug_output " " ^Filters:$"
857+ check_from -q " $debug_output " " ^Options:$"
858+ echo " $debug_output " | tr ' \n' ' %' | sed ' s~.*%Filters:%~~'
859+ | sed " s~%Options:.*~~" | tr ' %' ' \n'
860+ }
0 commit comments