# ("diff") individual tests.
#
# Usage:
-# run_test.sh run TEST_FULL RUN_COMMAND
+# run_test.sh run TEST_FULL [COMMAND and ARGS]
#
-# run_test.sh diff TEST_FULL DIFF_EXECUTABLE DIFF_EXECUTABLE COMPARISON_FILE
+# run_test.sh diff TEST_FULL NUMDIFF_EXECUTABLE DIFF_EXECUTABLE \
+# COMPARISON_FILE
#
set -u
STAGE="$1"
TEST_FULL="$2"
-NUMDIFF_EXECUTABLE="$3"
-DIFF_EXECUTABLE="$4"
-COMPARISON_FILE="$5"
-shift 5
-
-save () {
- for i do printf %s\\n "$i" | sed "s/'/'\\\\''/g;1s/^/'/;\$s/\$/' \\\\/" ;
- done
- echo " "
-}
-
-RUN_COMMAND=$(save "$@")
+shift 2
# Ensure uniform sorting for pathname expansion
export LC_ALL=C
-#
-# Add a top level target to run and compare the test:
-#
-
-run(){
- rm -f failing_output
- rm -f output
- rm -f stdout
-
- eval "set -- ${RUN_COMMAND}"
- "$@" > stdout 2>&1
- RETURN_VALUE=$?
-
- [ -f output ] || mv stdout output
-
- if [ $RETURN_VALUE -ne 0 ]; then
- mv output failing_output
- echo "${TEST_FULL}: BUILD successful."
- echo "${TEST_FULL}: RUN failed. ------ Return code $RETURN_VALUE"
- echo "${TEST_FULL}: RUN failed. ------ Result: `pwd`/failing_output"
- echo "${TEST_FULL}: RUN failed. ------ Partial output:"
- cat failing_output
- if [ -f stdout ]; then
- echo ""
- echo "${TEST_FULL}: RUN failed. ------ Additional output on stdout/stderr:"
- echo ""
- cat stdout
+case $STAGE in
+ run)
+ ##
+ # run stage:
+ # - run specified command and parameters given by $@
+ # - creates file "output" on success containing stdout and stderr of
+ # the test
+ # - if test exits with non-zero return value, output is renamed to
+ # failing_output
+ ##
+
+ rm -f failing_output
+ rm -f output
+ rm -f stdout
+
+ "$@" > stdout 2>&1
+ RETURN_VALUE=$?
+
+ [ -f output ] || mv stdout output
+
+ if [ $RETURN_VALUE -ne 0 ]; then
+ mv output failing_output
+ echo "${TEST_FULL}: BUILD successful."
+ echo "${TEST_FULL}: RUN failed. ------ Return code $RETURN_VALUE"
+ echo "${TEST_FULL}: RUN failed. ------ Result: `pwd`/failing_output"
+ echo "${TEST_FULL}: RUN failed. ------ Partial output:"
+ cat failing_output
+ if [ -f stdout ]; then
+ echo ""
+ echo "${TEST_FULL}: RUN failed. ------ Additional output on stdout/stderr:"
+ echo ""
+ cat stdout
+ fi
+ exit 1
fi
- exit 1
- fi
-}
-diff() {
- rm -f failing_diff*
- rm -f diff*
- touch diff
+ ;;
+
+ diff)
+
+ ##
+ # diff stage:
+ # - compares the file "output" against various comparison files.
+ ##
+
+ NUMDIFF_EXECUTABLE="$1"
+ DIFF_EXECUTABLE="$2"
+ COMPARISON_FILE="$3"
- test_successful=false
+ rm -f failing_diff*
+ rm -f diff*
+ touch diff
- #
- # Pick up main comparison file and all variants. A valid variant name is
- # of the form [...].output.[STRING]
- #
- for file in "${COMPARISON_FILE}"*; do
- # determine variant name (empty string for main comparison file):
- variant="${file#*.output}"
+ test_successful=false
#
- # Run diff or numdiff (if available) to determine whether files are the
- # same. Create a diff file "diff${variant}" for each variant file that
- # is found (including the main comparison file).
+ # Pick up main comparison file and all variants. A valid variant name is
+ # of the form [...].output.[STRING]
#
- case "${NUMDIFF_EXECUTABLE}" in
- *numdiff*)
- "${NUMDIFF_EXECUTABLE}" -a 1e-6 -r 1e-8 -s ' \t\n:<>=,;' \
- "${file}" output > diff${variant}
- ;;
- *)
- "${DIFF_EXECUTABLE}" "${file}" output > diff${variant}
- ;;
- esac
-
- if [ $? -eq 0 ]; then
+ for file in "${COMPARISON_FILE}"*; do
+ # determine variant name (empty string for main comparison file):
+ variant="${file#*.output}"
+
#
- # Ensure that only a single diff file with no contents remains
- # (numdiff has the bad habit of being very verbose...):
+ # Run diff or numdiff (if available) to determine whether files are the
+ # same. Create a diff file "diff${variant}" for each variant file that
+ # is found (including the main comparison file).
#
- rm -f diff*
- touch diff
-
- if [ -n "${variant}" ]; then
+ case "${NUMDIFF_EXECUTABLE}" in
+ *numdiff*)
+ "${NUMDIFF_EXECUTABLE}" -a 1e-6 -r 1e-8 -s ' \t\n:<>=,;' \
+ "${file}" output > diff${variant}
+ ;;
+ *)
+ "${DIFF_EXECUTABLE}" "${file}" output > diff${variant}
+ ;;
+ esac
+
+ if [ $? -eq 0 ]; then
#
- # In case of a successful comparison against a variant, store the
- # fact that we compared against a variant in the diff file.
+ # Ensure that only a single diff file with no contents remains
+ # (numdiff has the bad habit of being very verbose...):
#
- echo "${TEST_FULL}: DIFF successful. - Variant: ${file}" > diff
+ rm -f diff*
+ touch diff
+
+ if [ -n "${variant}" ]; then
+ #
+ # In case of a successful comparison against a variant, store the
+ # fact that we compared against a variant in the diff file.
+ #
+ echo "${TEST_FULL}: DIFF successful. - Variant: ${file}" > diff
+ fi
+
+ test_successful=true
+ break
fi
+ done
- test_successful=true
- break
+ #
+ # If none of the diffs succeeded, use the diff against the main comparison
+ # file. Output the first few lines of the output of numdiff, followed by
+ # the results of regular diff since the latter is just more readable.
+ #
+ if [ $test_successful = false ] ; then
+ for file in diff*; do
+ mv "$file" failing_"$file"
+ done
+ echo "${TEST_FULL}: BUILD successful."
+ echo "${TEST_FULL}: RUN successful."
+ echo "${TEST_FULL}: DIFF failed. ------ Source: ${COMPARISON_FILE}"
+ echo "${TEST_FULL}: DIFF failed. ------ Result: `pwd`/output"
+ echo "Check `pwd`/output ${COMPARISON_FILE}"
+ echo "${TEST_FULL}: DIFF failed. ------ Diff: `pwd`/failing_diff"
+ echo "${TEST_FULL}: DIFF failed. ------ First 8 lines of numdiff/diff output:"
+ cat failing_diff | head -n 8
+ echo "${TEST_FULL}: DIFF failed. ------ First 50 lines diff output:"
+ "${DIFF_EXECUTABLE}" -c "${COMPARISON_FILE}" output | head -n 50
+ exit 1
fi
- done
-
- #
- # If none of the diffs succeeded, use the diff against the main comparison
- # file. Output the first few lines of the output of numdiff, followed by
- # the results of regular diff since the latter is just more readable.
- #
- if [ $test_successful = false ] ; then
- for file in diff*; do
- mv "$file" failing_"$file"
- done
- echo "${TEST_FULL}: BUILD successful."
- echo "${TEST_FULL}: RUN successful."
- echo "${TEST_FULL}: DIFF failed. ------ Source: ${COMPARISON_FILE}"
- echo "${TEST_FULL}: DIFF failed. ------ Result: `pwd`/output"
- echo "Check `pwd`/output ${COMPARISON_FILE}"
- echo "${TEST_FULL}: DIFF failed. ------ Diff: `pwd`/failing_diff"
- echo "${TEST_FULL}: DIFF failed. ------ First 8 lines of numdiff/diff output:"
- cat failing_diff | head -n 8
- echo "${TEST_FULL}: DIFF failed. ------ First 50 lines diff output:"
- "${DIFF_EXECUTABLE}" -c "${COMPARISON_FILE}" output | head -n 50
- exit 1
- fi
- exit 0
-}
+ exit 0
-case $STAGE in
- run)
- run;;
- diff)
- diff;;
+ ;;
*)
- exit 1;;
+
+ exit 1
+ ;;
esac
projects / dealii.git / commitdiff