Fix timezone extraction to work in all locales

Followup to emscripten-core#21585

Author: sbc100

.circleci/config.yml

@@ -918,7 +918,9 @@ jobs:
           python: "$EMSDK_PYTHON"
       - run-tests:
           title: "crossplatform tests"
-          test_targets: "--crossplatform-only"
+          # Skipping test_strftime_zZ_gb_locale since this test forces the
+          # locale to one that is not necessarily available on macOS.
+          test_targets: "--crossplatform-only skip:other.test_strftime_zZ_gb_locale"
       - upload-test-results
 
   test-mac-arm64:
@@ -940,7 +942,9 @@ jobs:
       # are currently missing arm64 macos binaries.
       - run-tests:
           title: "crossplatform tests"
-          test_targets: "--crossplatform-only"
+          # Skipping test_strftime_zZ_gb_locale since this test forces the
+          # locale to one that is not necessarily available on macOS.
+          test_targets: "--crossplatform-only skip:other.test_strftime_zZ_gb_locale"
       - upload-test-results
 
 workflows:

src/generated_struct_info32.json

@@ -402,7 +402,7 @@
         "TIOCGWINSZ": 21523,
         "TIOCSPGRP": 21520,
         "TIOCSWINSZ": 21524,
-        "TZNAME_MAX": 6,
+        "TZNAME_MAX": 16,
         "UUID_TYPE_DCE_RANDOM": 4,
         "UUID_VARIANT_DCE": 1,
         "W_OK": 2,

src/generated_struct_info64.json

@@ -402,7 +402,7 @@
         "TIOCGWINSZ": 21523,
         "TIOCSPGRP": 21520,
         "TIOCSWINSZ": 21524,
-        "TZNAME_MAX": 6,
+        "TZNAME_MAX": 16,
         "UUID_TYPE_DCE_RANDOM": 4,
         "UUID_VARIANT_DCE": 1,
         "W_OK": 2,

src/library.js

@@ -649,10 +649,12 @@ addToLibrary({
 
     {{{ makeSetValue('daylight', '0', 'Number(winterOffset != summerOffset)', 'i32') }}};
 
-    var extractZone = (date) => date.toLocaleTimeString(undefined, {timeZoneName:'short'}).split(' ')[2];
+    var extractZone = (date) => date.toLocaleTimeString(undefined, {hour12:false, timeZoneName:'short'}).split(' ')[1];
     var winterName = extractZone(winter);
     var summerName = extractZone(summer);
 #if ASSERTIONS
+    assert(winterName);
+    assert(summerName);
     assert(lengthBytesUTF8(winterName) <= {{{ cDefs.TZNAME_MAX }}}, `timezone name truncated to fit in TZNAME_MAX (${winterName})`);
     assert(lengthBytesUTF8(summerName) <= {{{ cDefs.TZNAME_MAX }}}, `timezone name truncated to fit in TZNAME_MAX (${summerName})`);
 #endif

system/lib/libc/musl/include/limits.h

@@ -51,7 +51,14 @@
 #define SYMLOOP_MAX 40
 #define WORD_BIT 32
 #define SSIZE_MAX LONG_MAX
+#ifdef __EMSCRIPTEN__
+// We depend on the JS API to reteive the local name for the current
+// timezone and this can sometimes exceed 6 chats.  For example:
+// TZ='Asia/Kathmandu' yields 'GMT+5:45'.
+#define TZNAME_MAX 16
+#else
 #define TZNAME_MAX 6
+#endif
 #define TTY_NAME_MAX 32
 #define HOST_NAME_MAX 255
 

test/common.py

@@ -365,6 +365,29 @@ def modified(self, *args, **kwargs):
   return decorated
 
 
+# Decorator version of env_modify
+def also_with_env_modify(name_updates_mapping):
+
+  def decorated(f):
+    @wraps(f)
+    def metafunc(self, updates, *args, **kwargs):
+      if updates:
+        with env_modify(updates):
+          return f(self, *args, **kwargs)
+      else:
+        return f(self, *args, **kwargs)
+
+    parameterize = {'': (None,)}
+    for name, updates in name_updates_mapping.items():
+      parameterize[name] = (updates,)
+
+    metafunc._parameterize = parameterize
+
+    return metafunc
+
+  return decorated
+
+
 def also_with_minimal_runtime(f):
   assert callable(f)
 

test/other/test_strftime_zZ.c

@@ -0,0 +1,65 @@
+#include <errno.h>
+#include <stdio.h>
+#include <string.h>
+#include <time.h>
+
+int main() {
+  void tzset(void);
+
+  // Buffer to hold the current hour of the day.  Format is HH + nul
+  // character.
+  char hour[3];
+
+  // Buffer to hold our ISO 8601 formatted UTC offset for the current
+  // timezone.  Format is [+-]hhmm + nul character.
+  char utcOffset[6];
+
+  // Buffer to hold the timezone name or abbreviation.  Just make it
+  // sufficiently large to hold most timezone names.
+  char timezone[128];
+
+  struct tm tm;
+
+  // Get the current timestamp.
+  const time_t now = time(NULL);
+
+  // What time is that here?
+  if (localtime_r(&now, &tm) == NULL) {
+    const int error = errno;
+    printf("Failed to get localtime for timestamp=%lld; errno=%d; %s", now, errno, strerror(error));
+    return 1;
+  }
+
+  size_t result = 0;
+
+  // Get the formatted hour of the day.
+  if ((result = strftime(hour, 3, "%H", &tm)) != 2) {
+    const int error = errno;
+    printf("Failed to format hour for timestamp=%lld; result=%zu; errno=%d; %s\n",
+           now, result, error, strerror(error));
+    return 1;
+  }
+  printf("The current hour of the day is: %s\n", hour);
+
+  // Get the formatted UTC offset in ISO 8601 format.
+  if ((result = strftime(utcOffset, 6, "%z", &tm)) != 5) {
+    const int error = errno;
+    printf("Failed to format UTC offset for timestamp=%lld; result=%zu; errno=%d; %s\n",
+           now, result, error, strerror(error));
+    return 1;
+  }
+  printf("The current timezone offset is: %s\n", utcOffset);
+
+  // Get the formatted timezone name or abbreviation.  We don't know how long
+  // this will be, so just expect some data to be written to the buffer.
+  if ((result = strftime(timezone, 128, "%Z", &tm)) == 0) {
+    const int error = errno;
+    printf("Failed to format timezone for timestamp=%lld; result=%zu; errno=%d; %s\n",
+           now, result, error, strerror(error));
+    return 1;
+  }
+  printf("The current timezone is: %s\n", timezone);
+
+  printf("ok!\n");
+  return 0;
+}

test/runner.py

@@ -466,6 +466,7 @@ def prepend_default(arg):
   all_tests = get_all_tests(modules)
   if options.crossplatform_only:
     tests = get_crossplatform_tests(modules)
+    skip_requested_tests(options.tests, modules)
   else:
     tests = tests_with_expanded_wildcards(tests, all_tests)
     tests = skip_requested_tests(tests, modules)

test/test_other.py

@@ -34,7 +34,8 @@
 from common import env_modify, no_mac, no_windows, only_windows, requires_native_clang, with_env_modify
 from common import create_file, parameterized, NON_ZERO, node_pthreads, TEST_ROOT, test_file
 from common import compiler_for, EMBUILDER, requires_v8, requires_node, requires_wasm64, requires_node_canary
-from common import requires_wasm_eh, crossplatform, with_both_eh_sjlj, with_both_sjlj, also_with_standalone_wasm
+from common import requires_wasm_eh, crossplatform, with_both_eh_sjlj, with_both_sjlj
+from common import also_with_standalone_wasm, also_with_env_modify
 from common import also_with_minimal_runtime, also_with_wasm_bigint, also_with_wasm64, flaky
 from common import EMTEST_BUILD_VERBOSE, PYTHON, WEBIDL_BINDER
 from common import requires_network
@@ -5717,73 +5718,10 @@ def test_only_force_stdlibs_2(self):
       self.run_process([EMXX, 'src.cpp', '-sDISABLE_EXCEPTION_CATCHING=0'])
     self.assertContained('Caught exception: std::exception', self.run_js('a.out.js'))
 
+  @crossplatform
+  @also_with_env_modify({'gb_locale': {'LC_ALL': 'en_GB'}, 'long_tz': {'TZ': 'Asia/Kathmandu'}})
   def test_strftime_zZ(self):
-    create_file('src.c', r'''
-#include <errno.h>
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-
-int main() {
-  // Buffer to hold the current hour of the day.  Format is HH + nul
-  // character.
-  char hour[3];
-
-  // Buffer to hold our ISO 8601 formatted UTC offset for the current
-  // timezone.  Format is [+-]hhmm + nul character.
-  char utcOffset[6];
-
-  // Buffer to hold the timezone name or abbreviation.  Just make it
-  // sufficiently large to hold most timezone names.
-  char timezone[128];
-
-  struct tm tm;
-
-  // Get the current timestamp.
-  const time_t now = time(NULL);
-
-  // What time is that here?
-  if (localtime_r(&now, &tm) == NULL) {
-    const int error = errno;
-    printf("Failed to get localtime for timestamp=%lld; errno=%d; %s", now, errno, strerror(error));
-    return 1;
-  }
-
-  size_t result = 0;
-
-  // Get the formatted hour of the day.
-  if ((result = strftime(hour, 3, "%H", &tm)) != 2) {
-    const int error = errno;
-    printf("Failed to format hour for timestamp=%lld; result=%zu; errno=%d; %s\n",
-           now, result, error, strerror(error));
-    return 1;
-  }
-  printf("The current hour of the day is: %s\n", hour);
-
-  // Get the formatted UTC offset in ISO 8601 format.
-  if ((result = strftime(utcOffset, 6, "%z", &tm)) != 5) {
-    const int error = errno;
-    printf("Failed to format UTC offset for timestamp=%lld; result=%zu; errno=%d; %s\n",
-           now, result, error, strerror(error));
-    return 1;
-  }
-  printf("The current timezone offset is: %s\n", utcOffset);
-
-  // Get the formatted timezone name or abbreviation.  We don't know how long
-  // this will be, so just expect some data to be written to the buffer.
-  if ((result = strftime(timezone, 128, "%Z", &tm)) == 0) {
-    const int error = errno;
-    printf("Failed to format timezone for timestamp=%lld; result=%zu; errno=%d; %s\n",
-           now, result, error, strerror(error));
-    return 1;
-  }
-  printf("The current timezone is: %s\n", timezone);
-
-  printf("ok!\n");
-  return 0;
-}
-''')
-    self.do_runf('src.c', 'ok!')
+    self.do_runf('other/test_strftime_zZ.c', 'ok!')
 
   def test_strptime_symmetry(self):
     self.do_runf('strptime_symmetry.cpp', 'TEST PASSED')