add use_top_result flag, and timestamp sorting

1 files changed, 18 insertions, 12 deletions

diff --git a/obsmirror.sh/obsmirror.sh b/obsmirror.sh/obsmirror.sh
index 86180f3..2e913fb 100755
--- a/obsmirror.sh/obsmirror.sh
+++ b/obsmirror.sh/obsmirror.sh
@@ -58,18 +58,24 @@ test -z "${thisuser}" && thisuser=obsmirror
 
    mkdir -p "${workdir}" ; chmod "0755" "${workdir}" ; chown "${thisuser}:$( id -G "${thisuser}" | awk '{print $1}' )" "${workdir}"
    cd "${workdir}"
-   # get mirrorlist of Packages.gz file and find the one that lists the most packages
-   step1="$( curl -s -L "${inurl}/Packages.gz.mirrorlist" )"
-   options="$( echo "${step1}" | grep -oE 'href="[^"]+">' | awk '!x[$0]++' | sed -r -e 's/^href="//;' -e 's/">$//;' | grep -iE '^(ht|f)tps?:\/\/.*Packages\.gz$' )"
-   results="$(
-   for entry in ${options} ; do
-      curl -s -L "${entry}" | zgrep -cE '^Package:' | sed -r -e "s@\$@ ${entry}@;"
-   done )"
-   topresult_line="$( echo "${results}" | sort -nr | head -n1 | sed -r -e 's/\/Packages\.gz$//;' )"
-   topresult_packagecount="$( echo "${topresult_line}" | awk '{print $1}' )"
-   topresult="$( echo "${topresult_line}" | awk '{print $2}' )"
-   echo "USING ${topresult} with ${topresult_packagecount} packages" 1>&2
-   inurl="${topresult}"
+   test "${use_top_result}" = "yes" && {
+      # get mirrorlist of Packages.gz file and find the one that lists the most packages
+      step1="$( curl -s -L "${inurl}/Packages.gz.mirrorlist" )"
+      options="$( echo "${step1}" | grep -oE 'href="[^"]+">' | awk '!x[$0]++' | sed -r -e 's/^href="//;' -e 's/">$//;' | grep -iE '^(ht|f)tps?:\/\/.*Packages\.gz$' )"
+      echo "${options}" 1>&2
+      results="$(
+      for entry in ${options} ; do
+         # use package count
+         #curl -s -L "${entry}" | zgrep -cE '^Package:' | sed -r -e "s@\$@ ${entry}@;"
+         # use last modified timestamp of Packages.gz file
+         wget --content-disposition --quiet "${entry}" -O tmpfile.$$ ; find tmpfile.$$ -printf "%T@ ${entry}\n"
+      done ; rm tmpfile.$$ )"
+      topresult_line="$( echo "${results}" | sort -nr | head -n1 | sed -r -e 's/\/Packages\.gz$//;' )"
+      topresult_packagecount="$( echo "${topresult_line}" | awk '{print $1}' )"
+      topresult="$( echo "${topresult_line}" | awk '{print $2}' )"
+      echo "USING ${topresult} with ${topresult_packagecount} packages" 1>&2
+      inurl="${topresult}"
+   }
 
    step1="$( curl -s -L "${inurl}" )"
    echo "${step1}" | parse_obs_page_and_subdirs "${inurl}" "${tmpfile}"