Function cbm_build_cmdline

src/foundation/compat_fs.c:589–684

Build a properly-quoted Windows command line from an argv array. * Returns a heap-allocated wide string, or NULL on allocation failure. * Quoting follows the MSVC CRT convention: arguments containing spaces, * tabs, or double-quotes are wrapped in double-quotes, with backslashes * before a closing quote doubled and the quote itself escaped. Argument * bytes are treated as UTF-8 and converted

 * so non-ASCII arguments (e.g. a non-ASCII %USERPROFILE%) survive intact.
 * Declared in compat_fs_internal.h so the test suite can drive it. */
wchar_t *cbm_build_cmdline(const char *const *argv) {
    /* First pass: compute required buffer size. */
    size_t total = 1; /* NUL terminator */
    for (int i = 0; argv[i]; i++) {
        const char *arg = argv[i];
        bool needs_quote = (arg[0] == '\0');
        for (const char *p = arg; *p; p++) {
            if (*p == ' ' || *p == '\t' || *p == '"') {
                needs_quote = true;
            }
        }
        if (i > 0) {
            total++; /* space separator */
        }
        if (needs_quote) {
            total += 2; /* opening and closing quote */
            size_t backslashes = 0;
            for (const char *p = arg; *p; p++) {
                if (*p == '\\') {
                    backslashes++;
                } else if (*p == '"') {
                    total += backslashes + 1; /* double backslashes + escape backslash */
                    backslashes = 0;
                } else {
                    backslashes = 0;
                }
                total++;
            }
            /* Trailing backslashes before closing quote must be doubled. */
            total += backslashes;
        } else {
            total += strlen(arg);
        }
    }

    /* Build the quoted command line in UTF-8 first, then widen it as a
     * whole via cbm_utf8_to_wide. Every character the quoting logic acts
     * on (space, tab, '"', '\\') is ASCII and, by UTF-8's design, never
     * appears inside a multibyte sequence, so operating on raw bytes here
     * is safe and keeps multibyte argument bytes intact for conversion. */
    char *buf = (char *)malloc(total);
    if (!buf) {
        return NULL;
    }

    /* Second pass: write the command line bytes. */
    char *w = buf;
    for (int i = 0; argv[i]; i++) {
        const char *arg = argv[i];
        bool needs_quote = (arg[0] == '\0');
        for (const char *p = arg; *p; p++) {
            if (*p == ' ' || *p == '\t' || *p == '"') {
                needs_quote = true;
                break;
            }
        }
        if (i > 0) {
            *w++ = ' ';