Function cbm_sem_tokenize

src/semantic/semantic.c:166–368

}

int cbm_sem_tokenize(const char *name, char **out, int max_out) {
    if (!name || !out || max_out <= 0) {
        return 0;
    }
    int count = 0;
    char buf[TOKEN_BUF_LEN];
    int blen = 0;

    for (int i = 0; name[i] && count < max_out; i++) {
        char c = name[i];
        bool split = is_token_delim(c);
        bool camel = is_camel_break(name, i);
        if (split || camel) {
            flush_token(buf, &blen, out, &count, max_out);
            if (split) {
                continue;
            }
        }
        if (blen < TOKEN_BUF_LEN - SKIP_ONE && isalnum((unsigned char)c)) {
            buf[blen++] = (char)tolower((unsigned char)c);
        }
    }
    flush_token(buf, &blen, out, &count, max_out);

    /* Abbreviation expansion: add expanded forms for common code abbreviations.
     * "err" → also add "error", "ctx" → "context", etc. */
    /* Cross-language abbreviation table — covers Go, Python, JS/TS, Rust,
     * Java, C/C++, Ruby, PHP, Kotlin, Swift, Scala, C#, and common patterns. */
    static const struct {
        const char *abbrev;
        const char *expanded;
    } abbrevs[] = {
        /* Error/exception handling */
        {"err", "error"},
        {"exc", "exception"},
        {"ex", "exception"},
        /* Context/config */
        {"ctx", "context"},
        {"cfg", "config"},
        {"conf", "configuration"},
        {"env", "environment"},
        {"opt", "option"},
        {"opts", "options"},
        /* Request/response (HTTP, RPC) */
        {"req", "request"},
        {"res", "response"},
        {"resp", "response"},
        {"rsp", "response"},
        {"hdr", "header"},
        {"hdrs", "headers"},
        /* Strings/formatting */
        {"str", "string"},
        {"fmt", "format"},
        {"msg", "message"},
        {"txt", "text"},
        {"lbl", "label"},
        {"desc", "description"},
        /* Data structures */