simplify, and fix, setting the current locale

Fix test setup bogosities. Specifically, they weren't hermetic with respect to
locale env vars.

Rewrite the handling of locale vars to simplify the code and make it more like
the pattern most programs employ.

Fixes #3110

src/builtin_printf.cpp

@@ -197,24 +197,6 @@ static int octal_to_bin(wchar_t c) {
     }
 }
 
-double C_STRTOD(wchar_t const *nptr, wchar_t **endptr) {
-    double r;
-
-    const wcstring saved_locale = wsetlocale(LC_NUMERIC, NULL);
-
-    if (!saved_locale.empty()) {
-        wsetlocale(LC_NUMERIC, L"C");
-    }
-
-    r = wcstod(nptr, endptr);
-
-    if (!saved_locale.empty()) {
-        wsetlocale(LC_NUMERIC, saved_locale.c_str());
-    }
-
-    return r;
-}
-
 void builtin_printf_state_t::fatal_error(const wchar_t *fmt, ...) {
     // Don't error twice.
     if (early_exit) return;
@@ -283,7 +265,12 @@ uintmax_t raw_string_to_scalar_type(const wchar_t *s, wchar_t **end) {
 
 template <>
 long double raw_string_to_scalar_type(const wchar_t *s, wchar_t **end) {
-    return C_STRTOD(s, end);
+    // Forcing the locale to C is questionable but it's what the old C_STRTOD() that I inlined here
+    // as part of changing how locale management is done by fish.
+    char * old_locale = setlocale(LC_NUMERIC, "C");
+    double val = wcstod(s, end);
+    setlocale(LC_NUMERIC, old_locale);
+    return val;
 }
 
 template <typename T>

src/common.cpp

@@ -6,7 +6,6 @@
 #include <dlfcn.h>
 #include <errno.h>
 #include <limits.h>
-#include <locale.h>
 #include <math.h>
 #include <signal.h>
 #include <stdarg.h>
@@ -470,19 +469,12 @@ wchar_t *quote_end(const wchar_t *pos) {
     return 0;
 }
 
-wcstring wsetlocale(int category, const wchar_t *locale) {
+void fish_setlocale() {
-    char *lang = locale ? wcs2str(locale) : NULL;
-    char *res = setlocale(category, lang);
-    free(lang);
-
     // Use ellipsis if on known unicode system, otherwise use $.
     ellipsis_char = (wcwidth(L'\x2026') > 0) ? L'\x2026' : L'$';
 
     // U+23CE is the "return" character
     omitted_newline_char = (wcwidth(L'\x23CE') > 0) ? L'\x23CE' : L'~';
-
-    if (!res) return wcstring();
-    return format_string(L"%s", res);
 }
 
 bool contains_internal(const wchar_t *a, int vararg_handle, ...) {

src/common.h

@@ -627,10 +627,9 @@ wchar_t *quote_end(const wchar_t *in);
 /// interactive command executes, to allow new messages to be printed.
 void error_reset();
 
-/// This function behaves exactly like a wide character equivalent of the C function setlocale,
+/// This function should be called after calling `setlocale()` to perform fish specific locale
-/// except that it will also try to detect if the user is using a Unicode character set, and if so,
+/// initialization.
-/// use the unicode ellipsis character as ellipsis, instead of '$'.
+void fish_setlocale();
-wcstring wsetlocale(int category, const wchar_t *locale);
 
 /// Checks if \c needle is included in the list of strings specified. A warning is printed if needle
 /// is zero.

src/env.cpp

@@ -9,6 +9,7 @@
 #include <stdbool.h>
 #include <stddef.h>
 #include <stdlib.h>
+#include <string.h>
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <unistd.h>
@@ -168,37 +169,33 @@ static bool var_is_locale(const wcstring &key) {
 }
 
 /// Properly sets all locale information.
-static void handle_locale() {
+static void handle_locale(const wchar_t *env_var_name) {
-    const env_var_t lc_all = env_get_string(L"LC_ALL");
+    debug(2, L"handle_locale() called in response to '%ls' changing", env_var_name);
-    const wcstring old_locale = wsetlocale(LC_MESSAGES, NULL);
+    const char *old_msg_locale = setlocale(LC_MESSAGES, NULL);
 
-    // Array of locale constants corresponding to the local variable names defined in
+    for (size_t i = 0; locale_variable[i]; i++) {
-    // locale_variable.
+        const wchar_t *key = locale_variable[i];
-    static const int cat[] = {0,           LC_ALL,      LC_COLLATE, LC_CTYPE,
+        const env_var_t var = env_get_string(key);
-                              LC_MESSAGES, LC_MONETARY, LC_NUMERIC, LC_TIME};
+        if (!var.empty()) {
-
+            const std::string &name = wcs2string(key);
-    if (!lc_all.missing()) {
+            const std::string &value = wcs2string(var);
-        wsetlocale(LC_ALL, lc_all.c_str());
+            setenv(name.c_str(), value.c_str(), 1);
-    } else {
+            debug(3, L"locale var %s='%s'", name.c_str(), value.c_str());
-        const env_var_t lang = env_get_string(L"LANG");
-        if (!lang.missing()) {
-            wsetlocale(LC_ALL, lang.c_str());
-        }
-
-        for (int i = 2; locale_variable[i]; i++) {
-            const env_var_t val = env_get_string(locale_variable[i]);
-
-            if (!val.missing()) {
-                wsetlocale(cat[i], val.c_str());
-            }
         }
     }
 
-    const wcstring new_locale = wsetlocale(LC_MESSAGES, NULL);
+    char *locale = setlocale(LC_ALL, "");
-    if (old_locale != new_locale) {
+    fish_setlocale();
+    debug(2, L"handle_locale() setlocale(): '%s'", locale);
+
+    const char *new_msg_locale = setlocale(LC_MESSAGES, NULL);
+    debug(3, L"old LC_MESSAGES locale: '%s'", old_msg_locale);
+    debug(3, L"new LC_MESSAGES locale: '%s'", new_msg_locale);
+    if (strcmp(old_msg_locale, new_msg_locale)) {
         // Try to make change known to gettext. Both changing _nl_msg_cat_cntr and calling dcgettext
         // might potentially tell some gettext implementation that the translation strings should be
         // reloaded. We do both and hope for the best.
+        debug(2, L"changing message locale from '%s' to '%s'", old_msg_locale, new_msg_locale);
         extern int _nl_msg_cat_cntr;
         _nl_msg_cat_cntr++;
         fish_dcgettext("fish", "Changing language to English", LC_MESSAGES);
@@ -217,7 +214,8 @@ static bool var_is_curses(const wcstring &key) {
 
 /// Push all curses/terminfo env vars into the global environment where they can be found by those
 /// libraries.
-static void handle_curses() {
+static void handle_curses(const wchar_t *env_var_name) {
+    debug(2, L"handle_curses() called in response to '%ls' changing", env_var_name);
     for (size_t i = 0; curses_variable[i]; i++) {
         const wchar_t *key = curses_variable[i];
         const env_var_t var = env_get_string(key);
@@ -225,6 +223,7 @@ static void handle_curses() {
             const std::string &name = wcs2string(key);
             const std::string &value = wcs2string(var);
             setenv(name.c_str(), value.c_str(), 1);
+            debug(3, L"curses var %s='%s'", name.c_str(), value.c_str());
         }
     }
     // TODO: Modify input_init() to allow calling it when the terminfo env vars are dynamically
@@ -236,9 +235,9 @@ static void handle_curses() {
 /// React to modifying the given variable.
 static void react_to_variable_change(const wcstring &key) {
     if (var_is_locale(key)) {
-        handle_locale();
+        handle_locale(key.c_str());
     } else if (var_is_curses(key)) {
-        handle_curses();
+        handle_curses(key.c_str());
     } else if (key == L"fish_term256" || key == L"fish_term24bit") {
         update_fish_color_support();
         reader_react_to_color_change();
@@ -852,14 +851,13 @@ void env_push(bool new_scope) {
 void env_pop() {
     if (&top->env != global) {
         int i;
-        int locale_changed = 0;
+        const wchar_t *locale_changed = NULL;
-
         env_node_t *killme = top;
 
         for (i = 0; locale_variable[i]; i++) {
             var_table_t::iterator result = killme->env.find(locale_variable[i]);
             if (result != killme->env.end()) {
-                locale_changed = 1;
+                locale_changed = locale_variable[i];
                 break;
             }
         }
@@ -881,7 +879,7 @@ void env_pop() {
 
         delete killme;
 
-        if (locale_changed) handle_locale();
+        if (locale_changed) handle_locale(locale_changed);
 
     } else {
         debug(0, _(L"Tried to pop empty environment stack."));

src/fish.cpp

@@ -418,11 +418,12 @@ int main(int argc, char **argv) {
     assert(ANY_SENTINAL >= WILDCARD_RESERVED_BASE && ANY_SENTINAL <= WILDCARD_RESERVED_END);
     assert(R_SENTINAL >= INPUT_COMMON_BASE && R_SENTINAL <= INPUT_COMMON_END);
 
+    program_name = L"fish";
     set_main_thread();
     setup_fork_guards();
 
-    wsetlocale(LC_ALL, L"");
+    setlocale(LC_ALL, "");
-    program_name = L"fish";
+    fish_setlocale();
 
     // struct stat tmp;
     // stat("----------FISH_HIT_MAIN----------", &tmp);

src/fish_indent.cpp

@@ -328,12 +328,16 @@ static std::string html_colorize(const wcstring &text,
 static std::string no_colorize(const wcstring &text) { return wcs2string(text); }
 
 int main(int argc, char *argv[]) {
+    program_name = L"fish_indent";
     set_main_thread();
     setup_fork_guards();
-
+    // Using the user's default locale could be a problem if it doesn't use UTF-8 encoding. That's
-    wsetlocale(LC_ALL, L"");
+    // because the fish project assumes Unicode UTF-8 encoding in all of its scripts.
-    program_name = L"fish_indent";
+    //
-
+    // TODO: Auto-detect the encoding of the script. We should look for a vim style comment
+    // (e.g., "# vim: set fileencoding=<encoding-name>:") or an emacs style comment
+    // (e.g., "# -*- coding: <encoding-name> -*-").
+    setlocale(LC_ALL, "");
     env_init();
     input_init();
 

src/fish_key_reader.cpp

@@ -148,8 +148,7 @@ void setup_and_process_keys(bool continuous_mode) {
     is_interactive_session = 1;  // by definition this is interactive
     set_main_thread();
     setup_fork_guards();
-    wsetlocale(LC_ALL, L"POSIX");
+    setlocale(LC_ALL, "POSIX");
-    program_name = L"fish_key_reader";
     env_init();
     reader_init();
     input_init();
@@ -178,6 +177,7 @@ void setup_and_process_keys(bool continuous_mode) {
 }
 
 int main(int argc, char **argv) {
+    program_name = L"fish_key_reader";
     bool continuous_mode = false;
     const char *short_opts = "+c";
     const struct option long_opts[] = {{"continuous", no_argument, NULL, 'd'}, {NULL, 0, NULL, 0}};

tests/test_util.fish

@@ -1,4 +1,4 @@
-# vim: set ts=4 sw=4 et:
+# vim: set ts=4 sw=4 tw=100 et:
 # Utilities for the test runners
 
 if test "$argv[1]" = (status -f)
@@ -19,14 +19,12 @@ function die
     exit 1
 end
 
-# Check if we're running in the test environment.
+# Check if we're running in the test environment.  If not, set it up and rerun fish with exec.  The
-# If not, set it up and rerun fish with exec.
+# test is whether the special var __fish_is_running_tests exists and contains the same value as
-# The test is whether the special var __fish_is_running_tests
+# XDG_CONFIG_HOME. It checks the value and not just the presence because we're going to delete the
-# exists and contains the same value as XDG_CONFIG_HOME. It checks
+# config directory later if we're exiting successfully.
-# the value and not just the presence because we're going to delete
-# the config directory later if we're exiting successfully.
 if not set -q __fish_is_running_tests
-    # set up our test environment and re-run the original script
+    # Set up our test environment and re-run the original script.
     set -l script $argv[1]
     switch $script
     case '/*'
@@ -35,8 +33,11 @@ if not set -q __fish_is_running_tests
         # path is relative, make it absolute
         set script $PWD/$script
     end
+
+    begin
         set -l IFS  # clear IFS so cmd substitution doesn't split
         cd (dirname $script); or die
+    end
 
     set -lx XDG_DATA_HOME ../test/data
     rm -rf $XDG_DATA_HOME/fish
@@ -52,13 +53,21 @@ if not set -q __fish_is_running_tests
     printf 'set fish_function_path \'%s/functions\' \'%s/share/functions\'\n' $escaped_config $escaped_parent > $XDG_CONFIG_HOME/fish/config.fish; or die
     set -xl __fish_is_running_tests $XDG_CONFIG_HOME
 
-    # set locale information to be consistent
+    # Set locale information for consistent tests. Fish should work with a lot of locales but the
-    set -lx LANG C
+    # tests assume an english UTF-8 locale unless they explicitly override this default. We do not
-    set -lx LC_ALL ''
+    # want the users locale to affect the tests since they might, for example, change the wording of
-    for var in ALL COLLATE MESSAGES MONETARY NUMERIC TIME
+    # logged messages.
-        set -lx LC_$var ''
+    #
+    # TODO: set LANG to en_US.UTF-8 so we test the locale message conversions (i.e., gettext).
+    set -e LANGUAGE
+    set -x LANG C
+    # Remove "LC_" env vars from the test environment.
+    for var in (set -xn)
+        string match -q 'LC_*' $var
+        and set -e $var
     end
-    set -lx LC_CTYPE en_US.UTF-8
+    set -x LC_CTYPE en_US.UTF-8
+
     exec ../test/root/bin/fish $script $args_for_test_script
     die 'exec failed'
 else if test "$__fish_is_running_tests" != "$XDG_CONFIG_HOME"