Merge pull request universal-ctags#256 from vhda/units/sed-r

units: Don't use extended regular expressions in sed
kleopatra999 · Mar 1, 2015 · ee66264 · ee66264
2 parents cc8e8f0 + e4b28d8
commit ee66264
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 7 deletions.
diff --git a/Units/php-simple.d/expected.tags b/Units/php-simple.d/expected.tags
@@ -7,5 +7,5 @@ add_item	input.php	/^    function add_item ($artnr, $num)$/;"	f	class:Cart	signa
 foo	input.php	/^function foo ($arg_1, $arg_2, ..., $arg_n)$/;"	f	signature:($arg_1, $arg_2, ..., $arg_n)
 items	input.php	/^    var $items;  \/\/ Items in our shopping cart$/;"	v	class:Cart	access:public
 remove_item	input.php	/^    function remove_item ($artnr, $num)$/;"	f	class:Cart	signature:($artnr, $num)
-t�yte	Units/php-simple.d/input.php	/^$t�yte = 'mansikka';    \/\/ valid; '�' is ASCII 228.$/;"	v
+t�yte	input.php	/^$t�yte = 'mansikka';    \/\/ valid; '�' is ASCII 228.$/;"	v
 var	input.php	/^$var = "Bob";$/;"	v
diff --git a/Units/simple.php.t/expected.tags b/Units/simple.php.t/expected.tags
@@ -7,5 +7,5 @@ add_item	input.php	/^    function add_item ($artnr, $num)$/;"	f	class:Cart
 foo	input.php	/^function foo ($arg_1, $arg_2, ..., $arg_n)$/;"	f
 items	input.php	/^    var $items;  \/\/ Items in our shopping cart$/;"	v	class:Cart
 remove_item	input.php	/^    function remove_item ($artnr, $num)$/;"	f	class:Cart
-t�yte	Units/simple.php.t/input.php	/^$t�yte = 'mansikka';    \/\/ valid; '�' is ASCII 228.$/;"	v
+t�yte	input.php	/^$t�yte = 'mansikka';    \/\/ valid; '�' is ASCII 228.$/;"	v
 var	input.php	/^$var = "Bob";$/;"	v
diff --git a/misc/units b/misc/units
@@ -291,10 +291,10 @@ run_shrink ()
 }
 
 # filters out the directory prefix in a ctags input
-tags_basename_filter_regex='s%(^[^\t]+\t)([^/\t]+/)*%\1%'
+tags_basename_filter_regex='s%\(^[^	]\{1,\}	\)\([^/	]\{1,\}/\)*%\1%'
 tags_basename_filter()
 {
-    sed -r "${tags_basename_filter_regex}"
+    sed "${tags_basename_filter_regex}"
 }
 
 run_record_cmdline ()
@@ -304,7 +304,7 @@ run_record_cmdline ()
 
     printf "%s \\\\\n| %s \\\\\n| %s\n"  \
 	"${_CMDLINE}" \
-	"sed -r '${tags_basename_filter_regex}'" \
+	"sed '${tags_basename_filter_regex}'" \
 	"${ffilter}" \
 	> "${ocmdline}"
 }
@@ -363,7 +363,7 @@ run_tcase ()
     #
     # Filtered by LANGUAGES
     #
-    guessed_lang=$( ${_CMDLINE} --print-language "$input" 2>/dev/null | sed -nr 's/^.*: (.+)$/\1/p')
+    guessed_lang=$( ${_CMDLINE} --print-language "$input" 2>/dev/null | sed -n 's/^.*: //p')
     if [ -n "${LANGUAGES}" ]; then
 	member_p "${guessed_lang}" ${LANGUAGES} || return 1
     fi
@@ -1459,7 +1459,7 @@ noise_lang_file ()
     local offset
     local guessed_lang
 
-    guessed_lang=$( ${_CMDLINE_FOR_SHRINKING} --print-language "${input}" 2>/dev/null | sed -nr 's/^.*: (.+)$/\1/p')
+    guessed_lang=$( ${_CMDLINE_FOR_SHRINKING} --print-language "${input}" 2>/dev/null | sed -n 's/^.*: //p')
     if [ "${lang}" !=  "${guessed_lang}" ]; then
 	return 0
     fi
@@ -1549,6 +1549,14 @@ $0 noise [OPTIONS] UNITS-DIR
 EOF
 }
 
+# Avoid issues between sed and the locale settings by overriding it using
+# LC_ALL, which takes precedence over all other locale configurations:
+# https://www.gnu.org/software/gettext/manual/html_node/Locale-Environment-Variables.html
+prepare_environment ()
+{
+    LC_ALL="C"; export LC_ALL
+}
+
 main ()
 {
     if [ $# = 0 ]; then
@@ -1587,5 +1595,6 @@ main ()
     esac
 }
 
+prepare_environment
 main "$@"
 exit $?