Function cbm_project_name_from_path

src/pipeline/fqn.c:416–517

}

char *cbm_project_name_from_path(const char *abs_path) {
    if (!abs_path || !abs_path[0]) {
        return strdup("root");
    }
    if (path_is_root_syntax(abs_path)) {
        return strdup("root");
    }

    char real[CBM_SZ_4K];
    const char *name_path = abs_path;
    /* Wide-path canonicalization — the ANSI _access/_fullpath pair corrupted
     * CJK paths on CJK-locale Windows (#973). */
    if (cbm_canonical_path(abs_path, real, sizeof(real))) {
        cbm_normalize_path_sep(real);
        name_path = real;
    }

    /* Work on mutable copy */
    char *path = strdup(name_path);
    if (!path) {
        return NULL;
    }
    size_t len = strlen(path);

    /* Normalize path separators */
    cbm_normalize_path_sep(path);

    /* Map every character that is unsafe for portable project DB names. We
     * keep derived names in [A-Za-z0-9._-], so anything else — path
     * separators, ':', spaces, '@', '+', … — must be normalized here.
     * Otherwise a repo like
     * "/home/u/my project" yields the name "home-u-my project": indexing
     * creates the DB and it shows in list_projects, but resolve_store rejects
     * the space and reports project-not-found (#349).
     *
     * Non-ASCII bytes (UTF-8 of CJK and other scripts, all >= 0x80) are NOT
     * dropped to '-' — that silently erased whole path segments and produced
     * unrecognizable / colliding names (#571). Instead each non-ASCII byte is
     * transliterated to its two lowercase hex digits, which use only [0-9a-f]
     * and therefore stay validator-safe while preserving the segment. */
    static const char hex_digits[] = "0123456789abcdef";
    char *mapped = malloc(len * 2 + 1); /* worst case: every byte → 2 hex chars */
    if (!mapped) {
        free(path);
        return strdup("root");
    }
    size_t mlen = 0;
    for (size_t i = 0; i < len; i++) {
        unsigned char c = (unsigned char)path[i];
        bool safe = (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9') ||
                    c == '.' || c == '_' || c == '-';
        if (safe) {
            mapped[mlen++] = (char)c;
        } else if (c >= 0x80) {
            mapped[mlen++] = hex_digits[(c >> 4) & 0xF];
            mapped[mlen++] = hex_digits[c & 0xF];
        } else {
            mapped[mlen++] = '-';