Function read_file

src/pipeline/pass_parallel.c:233–281

Read file into a malloc'd buffer (= mimalloc in production). * *out_size receives the on-disk size and *out_status the failure reason so the * caller can attribute a skip to the right phase (read vs oversized) instead of * a silent drop. Both out params may be NULL. */

 * caller can attribute a skip to the right phase (read vs oversized) instead of
 * a silent drop. Both out params may be NULL. */
static char *read_file(const char *path, int *out_len, long *out_size,
                       cbm_read_status_t *out_status) {
    if (out_size) {
        *out_size = 0;
    }
    if (out_status) {
        *out_status = CBM_READ_OK;
    }
    FILE *f = cbm_fopen(path, "rb");
    if (!f) {
        if (out_status) {
            *out_status = CBM_READ_OPEN_FAIL;
        }
        return NULL;
    }
    (void)fseek(f, 0, SEEK_END);
    long size = ftell(f);
    (void)fseek(f, 0, SEEK_SET);
    if (out_size) {
        *out_size = size;
    }
    if (size <= 0) {
        (void)fclose(f);
        if (out_status) {
            *out_status = CBM_READ_EMPTY;
        }
        return NULL;
    }
    if (size > cbm_max_file_bytes()) { /* generous, env-configurable cap (B4) */
        (void)fclose(f);
        if (out_status) {
            *out_status = CBM_READ_OVERSIZED;
        }
        return NULL;
    }
    char *buf = (char *)malloc((size_t)size + SKIP_ONE);
    if (!buf) {
        (void)fclose(f);
        if (out_status) {
            *out_status = CBM_READ_OOM;
        }
        return NULL;
    }
    size_t nread = fread(buf, SKIP_ONE, (size_t)size, f);
    (void)fclose(f);
    buf[nread] = '\0';
    *out_len = (int)nread;
    return buf;
}

/* ── Per-worker skip list (Stage 2 / Track B) ───────────────────────
 * Each extract worker appends read/extract/oversized skips into its OWN list