Prevent override the autoinst-log context within handle_unreachable

.github/workflows/ci.yml

@@ -12,7 +12,7 @@ jobs:
           python --version
           pytest --version
       - run: |
-          sudo apt-get install cpanminus
+          sudo apt-get install cpanminus html-xml-utils
           sudo cpanm -n -M https://cpan.metacpan.org --installdeps .
       - name: unit- and integration tests
         run: |

openqa-label-known-issues

@@ -29,14 +29,19 @@ client_args=(api --header 'User-Agent: openqa-label-known-issues (https://github
 
 out="${REPORT_FILE:-$(mktemp -t openqa-label-known-issues--output-XXXX)}"
 trap 'error-handler "$LINENO"' ERR
-trap 'test "$KEEP_REPORT_FILE" == "1" || rm "$out"' EXIT
+trap '_cleanup' EXIT
+
+_cleanup() {
+    test "$KEEP_REPORT_FILE" == "1" || rm "$out"
+    test "$KEEP_JOB_HTML_FILE" == "1" || rm -f "$html_out"
+}
 
 echoerr() { echo "$@" >&2; }
 
 handle_unreachable() {
     local testurl=$1
-    local out=$2
-
+    local timeago
+    html_out=${JOB_HTML_FILE:-$(mktemp -t openqa-label-known-issues--job-details-XXXX)}
     if ! curl "${curl_args[@]}" -s --head "$testurl" -o /dev/null; then
         # the page might be gone, try the scheme+host we configured (might be different one though)
         if ! grep -q "$host_url" <<< "$testurl"; then
@@ -51,18 +56,27 @@ handle_unreachable() {
         return 1
     fi
     # resorting to downloading the job details page instead of the
-    # log, overwrites $out
-    if ! curl "${curl_args[@]}" -s "$testurl" -o "$out"; then
+    # log
+    if ! curl "${curl_args[@]}" -s "$testurl" -o "${html_out}"; then
         echoerr "'$testurl' can be reached but not downloaded, bailing out"
         curl "${curl_args[@]}" "$testurl"
         exit 2
     fi
 
-    if hxnormalize -x "$out" | hxselect -s '\n' -c '.links_a .resborder' | grep -qPzo '(?s)Gru job failed.*connection error.*Inactivity timeout'; then
+    if hxnormalize -x "${html_out}" | hxselect -s '\n' -c '.links_a .resborder' | grep -qPzo '(?s)Gru job failed.*connection error.*Inactivity timeout'; then
         "${client_call[@]}" -X POST jobs/"$id"/comments text='poo#62456 test incompletes after failing in GRU download task on "Inactivity timeout" with no logs'
         "${client_call[@]}" -X POST jobs/"$id"/restart
         return 1
     fi
+
+    # Checking timestamp given by job details page
+    timeago=$(grep timeago "$html_out" | hxselect -s '\n' -c '.timeago::attr(title)')
+    if [[ $(date -uIs -d '-14days') > $timeago ]]; then
+        # if the page is there but not even an autoinst-log.txt exists
+        # then the job might be too old and logs are already deleted.
+        echoerr "'$testurl' job#${id} without autoinst-log.txt older than 14 days. Do not label"
+        return 1
+    fi
 }
 
 label_on_issues_from_issue_tracker() {
@@ -136,15 +150,9 @@ investigate_issue() {
     echo "$reason" >> "$out"
     if [[ "$curl_output" != "200" ]] && [[ "$curl_output" != "301" ]]; then
         # if we can not even access the page it is something more critical
-        handle_unreachable "$testurl" "$out" || return
+        handle_unreachable "$testurl" "$out" || return 0
+
         [[ $curl_output != 404 ]] && return
-        # Checking timestamp
-        if [[ $(date -uIs -d '-14days') > $(grep timeago "$out" | hxselect -s '\n' -c '.timeago::attr(title)') ]]; then
-            # if the page is there but not even an autoinst-log.txt exists
-            # then the job might be too old and logs are already deleted.
-            echoerr "'$testurl' job#${id} without autoinst-log.txt older than 14 days. Do not label"
-            return
-        fi
         # not unreachable, no log, no reason, not too old
         if [[ -z $reason ]] || [[ $reason = null ]]; then
             echoerr "'$testurl' does not have autoinst-log.txt or reason, cannot label"

test/07-openqa-label-known-issues.t

@@ -9,6 +9,7 @@ source openqa-label-known-issues
 client_args=(api --host "$host_url")
 
 export KEEP_REPORT_FILE=1
+export KEEP_JOB_HTML_FILE=1
 
 client_output=''
 mock-client-output() {
@@ -39,12 +40,6 @@ comment_on_job() {
     echo "$comment"
 }
 out=''
-hxnormalize() {
-    cat "$2"
-}
-hxselect() {
-    cat -
-}
 
 try investigate_issue
 is "$rc" 1 'id required'
@@ -60,47 +55,54 @@ action
 Job incompletes with reason auto_review:\"(?m)api failure$\" (and no further details)
 action"
 
+setup() {
+  local id=$1
+  declare -g id=$id
+  declare -g testurl="https://openqa.opensuse.org/tests/${id}"
+  declare -g tmplog=$(mktemp)
+  declare -g out=$tmplog
+  declare -g tmpjobpage=$(mktemp)
+}
+
 # test data with reason but 404 in curl respond
-id=404
-testurl="https://openqa.opensuse.org/tests/${id}"
+setup 404
+older40d_date=$(date -uIs -d '-40days')
+echo -n "Result:<b>incomplete</b>finished<abbr class=\"timeago\" title=\"${older40d_date}\"</abbr>>" > $tmpjobpage
+export JOB_HTML_FILE=$tmpjobpage
 try-client-output investigate_issue $id
-is "$rc" 0 'investigate_issue with missing autoinst-log and with reason in job_data'
+is "$rc" 0 'investigate_issue with missing autoinst-log and with reason in job_data' #ok 3
 has "$got" "without autoinst-log.txt older than 14 days. Do not label" "exits succefully when is old job without autoinst-log.txt"
 
 # all assets are missing
-id=101
-testurl="https://openqa.opensuse.org/tests/${id}"
-# `cur_date` is used on the following 4 test cases
-cur_date=$(date +%F)
-sed -i "s/yyyy-mm-dd/${cur_date}/" "$dir/data/${id}.json"
-+fs:mktemp
-tmplog=$temp
-echo -n "Result: <b>incomplete</b>, finished <abbr class=\"timeago\" title=\"${cur_date}T08:06:42Z\"</abbr>>" > $tmplog
+setup 101
+# `older1d_date` is used on the following 4 test cases
+older1d_date=$(date -uIs -d '-1days')
+sed -i "s/yyyy-mm-dd/${older1d_date}/" "$dir/data/${id}.json"
+echo -n "Result:<b>incomplete</b>finished<abbr class=\"timeago\" title=\"${older1d_date}\"</abbr>>" > $tmpjobpage
+html_out=$tmpjobpage
+export JOB_HTML_FILE=$tmpjobpage
+echo > $tmplog
 out=$tmplog
 try-client-output investigate_issue $id
-is "$rc" 0 'investigate_issue with missing autoinst-log but with reason in job_data'
+is "$rc" 0 'investigate_issue with missing autoinst-log but with reason in job_data' # ok 4
 has "$got" "does not have autoinst-log.txt or reason, cannot label" "investigation exits when no reason and autoinst-log"
 # Cleanup 404.json
-sed -i "s/${cur_date}/yyyy-mm-dd/" "$dir/data/${id}.json"
+sed -i "s/${older1d_date}/yyyy-mm-dd/" "$dir/data/${id}.json"
 
 # Unknown reason - not included in issues
-id=102
-testurl="https://openqa.opensuse.org/tests/${id}"
-echo -n "Result: <b>incomplete</b>, finished <abbr class=\"timeago\" title=\"${cur_date}T08:06:42Z\"</abbr>>" > $tmplog
+setup 102
 echo -n "\nthe reason is whatever" >> $tmplog
 out=$tmplog
 try-client-output investigate_issue $id
 is "$rc" 0 'investigate no old issue with missing autoinst-log and unknown reason in job_data'
 has "$got" "Unknown test issue, to be reviewed" "investigation still label Unknown reason"
 
-id=414
-testurl="https://openqa.opensuse.org/tests/${id}"
+setup 414
 try-client-output investigate_issue $id
 is "$rc" 0 'investigate_issue with missing old autoinst-log and without reason in job_data'
 has "$got" "does not have autoinst-log.txt or reason, cannot label" "investigation exits successfully when no reason and no autoinst-log"
 
-id=200
-testurl="https://openqa.opensuse.org/tests/${id}"
+setup 200
 cp $autoinst_log $tmplog
 out=$tmplog
 try-client-output investigate_issue $id