Fix case-insensitive parsing of non-ASCII (e.g. Cyrillic) unit names

.github/workflows/test-musl.yml

@@ -0,0 +1,38 @@
+name: Run tests with musl libc (Alpine Linux)
+
+on:
+  pull_request:
+    branches:
+     - main
+
+jobs:
+  build:
+    name: musl libc test
+    runs-on: ubuntu-latest
+    env:
+      CI_CONTAINER: libbytesize-ci-alpine
+
+    steps:
+      - name: Checkout libbytesize repository
+        uses: actions/checkout@v6
+
+      - name: Install podman
+        run: |
+          sudo apt -qq update
+          sudo apt -y -qq install podman
+
+      - name: Build the container
+        run: |
+          podman build --no-cache -t ${{ env.CI_CONTAINER }} -f misc/alpine.Dockerfile .
+
+      - name: Start the container
+        run: |
+          podman run -d -t --name ${{ env.CI_CONTAINER }} --volume "$(pwd):/app" --workdir "/app" ${{ env.CI_CONTAINER }}
+
+      - name: Build in the container
+        run: |
+          podman exec -it ${{ env.CI_CONTAINER }} bash -c "./autogen.sh && LIBS='-lintl' ./configure --with-python3 --without-gtk-doc --without-tools && make"
+
+      - name: Run tests in the container
+        run: |
+          podman exec -it ${{ env.CI_CONTAINER }} bash -c "top_srcdir=/app top_builddir=/app source tests/testenv.sh && python3 tests/libbytesize_unittest.py"

misc/alpine.Dockerfile

@@ -0,0 +1,6 @@
+FROM alpine:latest
+
+RUN apk add --no-cache \
+    gcc make autoconf automake libtool pkgconf musl-dev bash git \
+    gmp-dev pcre2-dev gettext-dev \
+    python3

po/libbytesize.pot

@@ -8,7 +8,7 @@ msgid ""
 msgstr ""
 "Project-Id-Version: libbytesize 2.12\n"
 "Report-Msgid-Bugs-To: vtrefny@redhat.com\n"
-"POT-Creation-Date: 2026-01-08 06:02-0800\n"
+"POT-Creation-Date: 2026-01-16 12:41+0100\n"
 "PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n"
 "Last-Translator: FULL NAME <EMAIL@ADDRESS>\n"
 "Language-Team: LANGUAGE <LL@li.org>\n"
@@ -18,86 +18,86 @@ msgstr ""
 "Content-Transfer-Encoding: 8bit\n"
 
 #. TRANSLATORS: 'B' for bytes
-#: src/bs_size.c:52 src/bs_size.c:73
+#: src/bs_size.c:54 src/bs_size.c:75
 msgid "B"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for kibibyte, 2**10 bytes
-#: src/bs_size.c:54
+#: src/bs_size.c:56
 msgid "KiB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for mebibyte, 2**20 bytes
-#: src/bs_size.c:56
+#: src/bs_size.c:58
 msgid "MiB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for gibibyte, 2**30 bytes
-#: src/bs_size.c:58
+#: src/bs_size.c:60
 msgid "GiB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for tebibyte, 2**40 bytes
-#: src/bs_size.c:60
+#: src/bs_size.c:62
 msgid "TiB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for pebibyte, 2**50 bytes
-#: src/bs_size.c:62
+#: src/bs_size.c:64
 msgid "PiB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for exbibyte, 2**60 bytes
-#: src/bs_size.c:64
+#: src/bs_size.c:66
 msgid "EiB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for zebibyte, 2**70 bytes
-#: src/bs_size.c:66
+#: src/bs_size.c:68
 msgid "ZiB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for yobibyte, 2**80 bytes
-#: src/bs_size.c:68
+#: src/bs_size.c:70
 msgid "YiB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for kilobyte, 10**3 bytes
-#: src/bs_size.c:75
+#: src/bs_size.c:77
 msgid "KB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for megabyte, 10**6 bytes
-#: src/bs_size.c:77
+#: src/bs_size.c:79
 msgid "MB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for gigabyte, 10**9 bytes
-#: src/bs_size.c:79
+#: src/bs_size.c:81
 msgid "GB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for terabyte, 10**12 bytes
-#: src/bs_size.c:81
+#: src/bs_size.c:83
 msgid "TB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for petabyte, 10**15 bytes
-#: src/bs_size.c:83
+#: src/bs_size.c:85
 msgid "PB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for exabyte, 10**18 bytes
-#: src/bs_size.c:85
+#: src/bs_size.c:87
 msgid "EB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for zettabyte, 10**21 bytes
-#: src/bs_size.c:87
+#: src/bs_size.c:89
 msgid "ZB"
 msgstr ""
 
 #. TRANSLATORS: abbreviation for yottabyte, 10**24 bytes
-#: src/bs_size.c:89
+#: src/bs_size.c:91
 msgid "YB"
 msgstr ""

src/bs_size.c

@@ -8,6 +8,7 @@
 #include <ctype.h>
 #include <limits.h>
 #include <assert.h>
+#include <wchar.h>
 
 /* set code unit width to 8 so we can use generic macros like 'pcre2_compile'
  * instead of 'pcre2_compile_8'
@@ -225,6 +226,36 @@ static void strstrip(char *str) {
     str[i-begin] = '\0';
 }
 
+/* Case-insensitive comparison that handles multibyte UTF-8 (e.g. Cyrillic) */
+static int u8_casecmp (const char *s1, const char *s2, size_t n1) {
+    wchar_t *w1 = NULL;
+    wchar_t *w2 = NULL;
+    size_t wlen1, wlen2;
+    int ret;
+
+    wlen1 = mbstowcs (NULL, s1, 0);
+    wlen2 = mbstowcs (NULL, s2, 0);
+    if (wlen1 == (size_t) -1 || wlen2 == (size_t) -1)
+        return strncasecmp (s1, s2, n1);
+
+    w1 = malloc ((wlen1 + 1) * sizeof (wchar_t));
+    w2 = malloc ((wlen2 + 1) * sizeof (wchar_t));
+    if (!w1 || !w2) {
+        free (w1);
+        free (w2);
+        return strncasecmp (s1, s2, n1);
+    }
+
+    mbstowcs (w1, s1, wlen1 + 1);
+    mbstowcs (w2, s2, wlen2 + 1);
+
+    ret = wcsncasecmp (w1, w2, wlen1);
+
+    free (w1);
+    free (w2);
+    return ret;
+}
+
 static bool multiply_size_by_unit (mpq_t size, char *unit_str) {
     BSBunit bunit = BS_BUNIT_UNDEF;
     BSDunit dunit = BS_DUNIT_UNDEF;
@@ -236,7 +267,7 @@ static bool multiply_size_by_unit (mpq_t size, char *unit_str) {
     unit_str_len = strlen (unit_str);
 
     for (bunit=BS_BUNIT_B; bunit < BS_BUNIT_UNDEF; bunit++)
-        if (strncasecmp (unit_str, b_units[bunit-BS_BUNIT_B], unit_str_len) == 0) {
+        if (u8_casecmp (unit_str, b_units[bunit-BS_BUNIT_B], unit_str_len) == 0) {
             pwr = (uint64_t) bunit - BS_BUNIT_B;
             mpz_mul_2exp (mpq_numref (size), mpq_numref (size), 10 * pwr);
             return true;
@@ -245,7 +276,7 @@ static bool multiply_size_by_unit (mpq_t size, char *unit_str) {
     mpq_init (dec_mul);
     mpz_init (pow_1000);
     for (dunit=BS_DUNIT_B; dunit < BS_DUNIT_UNDEF; dunit++)
-        if (strncasecmp (unit_str, d_units[dunit-BS_DUNIT_B], unit_str_len) == 0) {
+        if (u8_casecmp (unit_str, d_units[dunit-BS_DUNIT_B], unit_str_len) == 0) {
             pwr = (uint64_t) (dunit - BS_DUNIT_B);
             mpz_ui_pow_ui (pow_1000, 1000, pwr);
             mpq_set_z (dec_mul, pow_1000);
@@ -256,7 +287,7 @@ static bool multiply_size_by_unit (mpq_t size, char *unit_str) {
         }
 
     for (bunit=BS_BUNIT_B; bunit < BS_BUNIT_UNDEF; bunit++)
-        if (strncasecmp (unit_str, _(b_units[bunit-BS_BUNIT_B]), unit_str_len) == 0) {
+        if (u8_casecmp (unit_str, _(b_units[bunit-BS_BUNIT_B]), unit_str_len) == 0) {
             pwr = (uint64_t) bunit - BS_BUNIT_B;
             mpz_mul_2exp (mpq_numref (size), mpq_numref (size), 10 * pwr);
             mpz_clear (pow_1000);
@@ -265,7 +296,7 @@ static bool multiply_size_by_unit (mpq_t size, char *unit_str) {
         }
 
     for (dunit=BS_DUNIT_B; dunit < BS_DUNIT_UNDEF; dunit++)
-        if (strncasecmp (unit_str, _(d_units[dunit-BS_DUNIT_B]), unit_str_len) == 0) {
+        if (u8_casecmp (unit_str, _(d_units[dunit-BS_DUNIT_B]), unit_str_len) == 0) {
             pwr = (uint64_t) (dunit - BS_DUNIT_B);
             mpz_ui_pow_ui (pow_1000, 1000, pwr);
             mpq_set_z (dec_mul, pow_1000);

tests/libbytesize_unittest.py

@@ -5,6 +5,7 @@
 import unittest
 import sys
 import ctypes
+import os
 
 from locale_utils import get_avail_locales, missing_locales, requires_locales
 
@@ -17,7 +18,7 @@
 except ImportError:
     from bytesize.bytesize import SizeStruct
 
-DEFAULT_LOCALE = "en_US.utf8"
+DEFAULT_LOCALE = "C"
 
 class SizeTestCase(unittest.TestCase):
 
@@ -32,6 +33,7 @@ def setUp(self):
             self.skipTest("requires missing locales: %s" % missing)
         locale.setlocale(locale.LC_ALL, DEFAULT_LOCALE)
         self.addCleanup(self._clean_up)
+        os.environ["LANGUAGE"] = ""
 
     def _clean_up(self):
         locale.setlocale(locale.LC_ALL, DEFAULT_LOCALE)
@@ -140,6 +142,33 @@ def testNewFromStrLocalePsAF(self):
         expected = (1536, -1)
         self.assertEqual(actual, expected)
 
+    @requires_locales({'ru_RU.UTF-8'})
+    def testNewFromStrLocaleRuRU(self):
+        locale.setlocale(locale.LC_ALL, 'ru_RU.UTF-8')
+
+        # uppercase Cyrillic unit (canonical translation)
+        actual = SizeStruct.new_from_str('1 МиБ').get_bytes()
+        expected = (1048576, 1)
+        self.assertEqual(actual, expected)
+
+        # lowercase Cyrillic unit -- case-insensitive matching for non-ASCII
+        actual = SizeStruct.new_from_str('1 миб').get_bytes()
+        expected = (1048576, 1)
+        self.assertEqual(actual, expected)
+
+        actual = SizeStruct.new_from_str('2 гиб').get_bytes()
+        expected = (2147483648, 1)
+        self.assertEqual(actual, expected)
+
+        # ASCII units should still work under Russian locale
+        actual = SizeStruct.new_from_str('1 MiB').get_bytes()
+        expected = (1048576, 1)
+        self.assertEqual(actual, expected)
+
+        actual = SizeStruct.new_from_str('1 mib').get_bytes()
+        expected = (1048576, 1)
+        self.assertEqual(actual, expected)
+
     #enddef
 
     def testNewFromBytes(self):