u*-vasnprintf tests: Add more tests of %U, %lU, %llU directives

[Bruno Haible]

In the %ls and %U directives handling of the u*_vasnprintf functions, the
width handling is tricky. Let me add some more unit tests for this code.


2024-06-19  Bruno Haible  <bruno@clisp.org>

        u*-vasnprintf tests: Add more tests of %U, %lU, %llU directives.
        * tests/unistdio/test-u8-printf1.h (test_xfunction): Add more tests of
        %U, %lU, %llU directives with width and non-ASCII argument.
        * tests/unistdio/test-u16-printf1.h (test_xfunction): Likewise.
        * tests/unistdio/test-u32-printf1.h (test_xfunction): Likewise.
        * tests/unistdio/test-ulc-vasnprintf2.c (test_function): Likewise.
        * tests/unistdio/test-ulc-vasnprintf3.c (test_function): Likewise.

diff --git a/tests/unistdio/test-u16-printf1.h 
b/tests/unistdio/test-u16-printf1.h
index e501c9e462..d2a71153c6 100644
--- a/tests/unistdio/test-u16-printf1.h
+++ b/tests/unistdio/test-u16-printf1.h
@@ -120,6 +120,32 @@ test_xfunction (uint16_t * (*my_xasprintf) (const char *, 
...))
       free (result);
     }
   }
+  { /* Width with a non-ASCII argument.  */
+    static const uint8_t unicode_string[] = /* hétérogénéité */
+      "h\303\251t\303\251rog\303\251n\303\251it\303\251";
+    uint16_t *result =
+      my_xasprintf ("%20U %d", unicode_string, 33, 44, 55);
+    static const uint16_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', 'h', 0x00e9, 't',
+        0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9, 'i', 't', 0x00e9,
+        ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u16_strcmp (result, expected) == 0);
+    free (result);
+  }
+  { /* Width with a non-BMP argument.  */
+    static const uint8_t unicode_string[] = "\360\237\220\203"; /* 🐃 */
+    uint16_t *result =
+      my_xasprintf ("%10U %d", unicode_string, 33, 44, 55);
+    static const uint16_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', 0xd83d,
+        0xdc03, ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u16_strcmp (result, expected) == 0);
+    free (result);
+  }
 
   {
     static const uint16_t unicode_string[] = { 'H', 'e', 'l', 'l', 'o', 0 };
@@ -178,6 +204,34 @@ test_xfunction (uint16_t * (*my_xasprintf) (const char *, 
...))
       free (result);
     }
   }
+  { /* Width with a non-ASCII argument.  */
+    static const uint16_t unicode_string[] = /* hétérogénéité */
+      { 'h', 0x00e9, 't', 0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9,
+        'i', 't', 0x00e9, 0
+      };
+    uint16_t *result =
+      my_xasprintf ("%20lU %d", unicode_string, 33, 44, 55);
+    static const uint16_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', 'h', 0x00e9, 't',
+        0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9, 'i', 't', 0x00e9,
+        ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u16_strcmp (result, expected) == 0);
+    free (result);
+  }
+  { /* Width with a non-BMP argument.  */
+    static const uint16_t unicode_string[] = { 0xd83d, 0xdc03, 0 }; /* 🐃 */
+    uint16_t *result =
+      my_xasprintf ("%10lU %d", unicode_string, 33, 44, 55);
+    static const uint16_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', 0xd83d,
+        0xdc03, ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u16_strcmp (result, expected) == 0);
+    free (result);
+  }
 
   {
     static const uint32_t unicode_string[] = { 'H', 'e', 'l', 'l', 'o', 0 };
@@ -236,6 +290,34 @@ test_xfunction (uint16_t * (*my_xasprintf) (const char *, 
...))
       free (result);
     }
   }
+  { /* Width with a non-ASCII argument.  */
+    static const uint32_t unicode_string[] = /* hétérogénéité */
+      { 'h', 0x00e9, 't', 0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9,
+        'i', 't', 0x00e9, 0
+      };
+    uint16_t *result =
+      my_xasprintf ("%20llU %d", unicode_string, 33, 44, 55);
+    static const uint16_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', 'h', 0x00e9, 't',
+        0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9, 'i', 't', 0x00e9,
+        ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u16_strcmp (result, expected) == 0);
+    free (result);
+  }
+  { /* Width with a non-BMP argument.  */
+    static const uint32_t unicode_string[] = { 0x1f403, 0 }; /* 🐃 */
+    uint16_t *result =
+      my_xasprintf ("%10llU %d", unicode_string, 33, 44, 55);
+    static const uint16_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', 0xd83d,
+        0xdc03, ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u16_strcmp (result, expected) == 0);
+    free (result);
+  }
 
   /* Test the support of the 's' conversion specifier for strings.  */
 
diff --git a/tests/unistdio/test-u32-printf1.h 
b/tests/unistdio/test-u32-printf1.h
index ae0edc6a63..d7ba2e0998 100644
--- a/tests/unistdio/test-u32-printf1.h
+++ b/tests/unistdio/test-u32-printf1.h
@@ -120,6 +120,32 @@ test_xfunction (uint32_t * (*my_xasprintf) (const char *, 
...))
       free (result);
     }
   }
+  { /* Width with a non-ASCII argument.  */
+    static const uint8_t unicode_string[] = /* hétérogénéité */
+      "h\303\251t\303\251rog\303\251n\303\251it\303\251";
+    uint32_t *result =
+      my_xasprintf ("%20U %d", unicode_string, 33, 44, 55);
+    static const uint32_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', 'h', 0x00e9, 't',
+        0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9, 'i', 't', 0x00e9,
+        ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u32_strcmp (result, expected) == 0);
+    free (result);
+  }
+  { /* Width with a non-BMP argument.  */
+    static const uint8_t unicode_string[] = "\360\237\220\203"; /* 🐃 */
+    uint32_t *result =
+      my_xasprintf ("%10U %d", unicode_string, 33, 44, 55);
+    static const uint32_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', 0x1f403,
+        ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u32_strcmp (result, expected) == 0);
+    free (result);
+  }
 
   {
     static const uint16_t unicode_string[] = { 'H', 'e', 'l', 'l', 'o', 0 };
@@ -178,6 +204,34 @@ test_xfunction (uint32_t * (*my_xasprintf) (const char *, 
...))
       free (result);
     }
   }
+  { /* Width with a non-ASCII argument.  */
+    static const uint16_t unicode_string[] = /* hétérogénéité */
+      { 'h', 0x00e9, 't', 0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9,
+        'i', 't', 0x00e9, 0
+      };
+    uint32_t *result =
+      my_xasprintf ("%20lU %d", unicode_string, 33, 44, 55);
+    static const uint32_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', 'h', 0x00e9, 't',
+        0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9, 'i', 't', 0x00e9,
+        ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u32_strcmp (result, expected) == 0);
+    free (result);
+  }
+  { /* Width with a non-BMP argument.  */
+    static const uint16_t unicode_string[] = { 0xd83d, 0xdc03, 0 }; /* 🐃 */
+    uint32_t *result =
+      my_xasprintf ("%10lU %d", unicode_string, 33, 44, 55);
+    static const uint32_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', 0x1f403,
+        ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u32_strcmp (result, expected) == 0);
+    free (result);
+  }
 
   {
     static const uint32_t unicode_string[] = { 'H', 'e', 'l', 'l', 'o', 0 };
@@ -236,6 +290,34 @@ test_xfunction (uint32_t * (*my_xasprintf) (const char *, 
...))
       free (result);
     }
   }
+  { /* Width with a non-ASCII argument.  */
+    static const uint32_t unicode_string[] = /* hétérogénéité */
+      { 'h', 0x00e9, 't', 0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9,
+        'i', 't', 0x00e9, 0
+      };
+    uint32_t *result =
+      my_xasprintf ("%20llU %d", unicode_string, 33, 44, 55);
+    static const uint32_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', 'h', 0x00e9, 't',
+        0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9, 'i', 't', 0x00e9,
+        ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u32_strcmp (result, expected) == 0);
+    free (result);
+  }
+  { /* Width with a non-BMP argument.  */
+    static const uint32_t unicode_string[] = { 0x1f403, 0 }; /* 🐃 */
+    uint32_t *result =
+      my_xasprintf ("%10llU %d", unicode_string, 33, 44, 55);
+    static const uint32_t expected[] =
+      { ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', 0x1f403,
+        ' ', '3', '3', 0
+      };
+    ASSERT (result != NULL);
+    ASSERT (u32_strcmp (result, expected) == 0);
+    free (result);
+  }
 
   /* Test the support of the 's' conversion specifier for strings.  */
 
diff --git a/tests/unistdio/test-u8-printf1.h b/tests/unistdio/test-u8-printf1.h
index 3e8999b638..3963adb6b7 100644
--- a/tests/unistdio/test-u8-printf1.h
+++ b/tests/unistdio/test-u8-printf1.h
@@ -110,6 +110,26 @@ test_xfunction (uint8_t * (*my_xasprintf) (const char *, 
...))
       free (result);
     }
   }
+  { /* Width with a non-ASCII argument.  */
+    static const uint8_t unicode_string[] = /* hétérogénéité */
+      "h\303\251t\303\251rog\303\251n\303\251it\303\251";
+    uint8_t *result =
+      my_xasprintf ("%20U %d", unicode_string, 33, 44, 55);
+    static const uint8_t expected[] =
+      "       h\303\251t\303\251rog\303\251n\303\251it\303\251 33";
+    ASSERT (result != NULL);
+    ASSERT (u8_strcmp (result, expected) == 0);
+    free (result);
+  }
+  { /* Width with a non-BMP argument.  */
+    static const uint8_t unicode_string[] = "\360\237\220\203"; /* 🐃 */
+    uint8_t *result =
+      my_xasprintf ("%10U %d", unicode_string, 33, 44, 55);
+    static const uint8_t expected[] = "         \360\237\220\203 33";
+    ASSERT (result != NULL);
+    ASSERT (u8_strcmp (result, expected) == 0);
+    free (result);
+  }
 
   {
     static const uint16_t unicode_string[] = { 'H', 'e', 'l', 'l', 'o', 0 };
@@ -162,6 +182,28 @@ test_xfunction (uint8_t * (*my_xasprintf) (const char *, 
...))
       free (result);
     }
   }
+  { /* Width with a non-ASCII argument.  */
+    static const uint16_t unicode_string[] = /* hétérogénéité */
+      { 'h', 0x00e9, 't', 0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9,
+        'i', 't', 0x00e9, 0
+      };
+    uint8_t *result =
+      my_xasprintf ("%20lU %d", unicode_string, 33, 44, 55);
+    static const uint8_t expected[] =
+      "       h\303\251t\303\251rog\303\251n\303\251it\303\251 33";
+    ASSERT (result != NULL);
+    ASSERT (u8_strcmp (result, expected) == 0);
+    free (result);
+  }
+  { /* Width with a non-BMP argument.  */
+    static const uint16_t unicode_string[] = { 0xd83d, 0xdc03, 0 }; /* 🐃 */
+    uint8_t *result =
+      my_xasprintf ("%10lU %d", unicode_string, 33, 44, 55);
+    static const uint8_t expected[] = "         \360\237\220\203 33";
+    ASSERT (result != NULL);
+    ASSERT (u8_strcmp (result, expected) == 0);
+    free (result);
+  }
 
   {
     static const uint32_t unicode_string[] = { 'H', 'e', 'l', 'l', 'o', 0 };
@@ -214,6 +256,28 @@ test_xfunction (uint8_t * (*my_xasprintf) (const char *, 
...))
       free (result);
     }
   }
+  { /* Width with a non-ASCII argument.  */
+    static const uint32_t unicode_string[] = /* hétérogénéité */
+      { 'h', 0x00e9, 't', 0x00e9, 'r', 'o', 'g', 0x00e9, 'n', 0x00e9,
+        'i', 't', 0x00e9, 0
+      };
+    uint8_t *result =
+      my_xasprintf ("%20llU %d", unicode_string, 33, 44, 55);
+    static const uint8_t expected[] =
+      "       h\303\251t\303\251rog\303\251n\303\251it\303\251 33";
+    ASSERT (result != NULL);
+    ASSERT (u8_strcmp (result, expected) == 0);
+    free (result);
+  }
+  { /* Width with a non-BMP argument.  */
+    static const uint32_t unicode_string[] = { 0x1f403, 0 }; /* 🐃 */
+    uint8_t *result =
+      my_xasprintf ("%10llU %d", unicode_string, 33, 44, 55);
+    static const uint8_t expected[] = "         \360\237\220\203 33";
+    ASSERT (result != NULL);
+    ASSERT (u8_strcmp (result, expected) == 0);
+    free (result);
+  }
 
   /* Test the support of the 's' conversion specifier for strings.  */
 
diff --git a/tests/unistdio/test-ulc-vasnprintf2.c 
b/tests/unistdio/test-ulc-vasnprintf2.c
index d91fa86e5a..ba14e612bc 100644
--- a/tests/unistdio/test-ulc-vasnprintf2.c
+++ b/tests/unistdio/test-ulc-vasnprintf2.c
@@ -76,6 +76,16 @@ test_function (char * (*my_asnprintf) (char *, size_t *, 
const char *, ...))
       free (result);
     }
   }
+  { /* Width with a non-BMP argument.  */
+    static const uint8_t unicode_string[] = "\360\237\220\203"; /* 🐃 */
+    size_t length;
+    char *result =
+      my_asnprintf (NULL, &length, "%10U %d", unicode_string, 33, 44, 55);
+    ASSERT (result != NULL);
+    ASSERT (strcmp (result, "         ? 33") == 0);
+    ASSERT (length == strlen (result));
+    free (result);
+  }
 
   {
     static const uint16_t unicode_string[] = /* Rafał Maszkowski */
@@ -124,6 +134,16 @@ test_function (char * (*my_asnprintf) (char *, size_t *, 
const char *, ...))
       free (result);
     }
   }
+  { /* Width with a non-BMP argument.  */
+    static const uint16_t unicode_string[] = { 0xd83d, 0xdc03, 0 }; /* 🐃 */
+    size_t length;
+    char *result =
+      my_asnprintf (NULL, &length, "%10lU %d", unicode_string, 33, 44, 55);
+    ASSERT (result != NULL);
+    ASSERT (strcmp (result, "         ? 33") == 0);
+    ASSERT (length == strlen (result));
+    free (result);
+  }
 
   {
     static const uint32_t unicode_string[] = /* Rafał Maszkowski */
@@ -172,6 +192,16 @@ test_function (char * (*my_asnprintf) (char *, size_t *, 
const char *, ...))
       free (result);
     }
   }
+  { /* Width with a non-BMP argument.  */
+    static const uint32_t unicode_string[] = { 0x1f403, 0 }; /* 🐃 */
+    size_t length;
+    char *result =
+      my_asnprintf (NULL, &length, "%10llU %d", unicode_string, 33, 44, 55);
+    ASSERT (result != NULL);
+    ASSERT (strcmp (result, "         ? 33") == 0);
+    ASSERT (length == strlen (result));
+    free (result);
+  }
 
   /* Test the support of the 's' conversion specifier for strings.  */
 
diff --git a/tests/unistdio/test-ulc-vasnprintf3.c 
b/tests/unistdio/test-ulc-vasnprintf3.c
index 7fc6171952..f73b4539cd 100644
--- a/tests/unistdio/test-ulc-vasnprintf3.c
+++ b/tests/unistdio/test-ulc-vasnprintf3.c
@@ -72,6 +72,16 @@ test_function (char * (*my_asnprintf) (char *, size_t *, 
const char *, ...))
       free (result);
     }
   }
+  { /* Width with a non-BMP argument.  */
+    static const uint8_t unicode_string[] = "\360\237\220\203"; /* 🐃 */
+    size_t length;
+    char *result =
+      my_asnprintf (NULL, &length, "%10U %d", unicode_string, 33, 44, 55);
+    ASSERT (result != NULL);
+    ASSERT (strcmp (result, "         \360\237\220\203 33") == 0);
+    ASSERT (length == strlen (result));
+    free (result);
+  }
 
   {
     static const uint16_t unicode_string[] = /* Rafał Maszkowski */
@@ -116,6 +126,16 @@ test_function (char * (*my_asnprintf) (char *, size_t *, 
const char *, ...))
       free (result);
     }
   }
+  { /* Width with a non-BMP argument.  */
+    static const uint16_t unicode_string[] = { 0xd83d, 0xdc03, 0 }; /* 🐃 */
+    size_t length;
+    char *result =
+      my_asnprintf (NULL, &length, "%10lU %d", unicode_string, 33, 44, 55);
+    ASSERT (result != NULL);
+    ASSERT (strcmp (result, "         \360\237\220\203 33") == 0);
+    ASSERT (length == strlen (result));
+    free (result);
+  }
 
   {
     static const uint32_t unicode_string[] = /* Rafał Maszkowski */
@@ -160,6 +180,16 @@ test_function (char * (*my_asnprintf) (char *, size_t *, 
const char *, ...))
       free (result);
     }
   }
+  { /* Width with a non-BMP argument.  */
+    static const uint32_t unicode_string[] = { 0x1f403, 0 }; /* 🐃 */
+    size_t length;
+    char *result =
+      my_asnprintf (NULL, &length, "%10llU %d", unicode_string, 33, 44, 55);
+    ASSERT (result != NULL);
+    ASSERT (strcmp (result, "         \360\237\220\203 33") == 0);
+    ASSERT (length == strlen (result));
+    free (result);
+  }
 
   /* Test the support of the 's' conversion specifier for strings.  */