Algorithm-MinPerfHashTwoLevel

 view release on metacpan or  search on metacpan

MinPerfHashTwoLevel.xs  view on Meta::CPAN

    struct mph_bucket *buckets= (struct mph_bucket *) ((char *)mph + mph->table_ofs);
    struct mph_bucket *bucket;
    U8 *state= (char *)mph + mph->state_ofs;
    STRLEN key_len;
    U8 *key_pv;
    U64 h0;
    U32 h1;
    U32 h2;
    U32 index;
    U8 *got_key_pv;
    STRLEN got_key_len;

    if (SvUTF8(key_sv)) {
        SV *tmp= sv_2mortal(newSVsv(key_sv));
        sv_utf8_downgrade(tmp,1);
        key_sv= tmp;
    }
    key_pv= SvPV(key_sv,key_len);
    h0= mph_hash_with_state(state,key_pv,key_len);
    h1= h0 >> 32;
    index= h1 % mph->num_buckets;

    bucket= buckets + index;
    if (!bucket->xor_val)
        return 0;
    
    h2= h0 & 0xFFFFFFFF;
    if ( bucket->index < 0 ) {
        index = -bucket->index-1;
    } else {
        HASH2INDEX(index,h2,bucket->xor_val,mph->num_buckets);
    }
    bucket= buckets + index;
    got_key_pv= strs + bucket->key_ofs;
    if (bucket->key_len == key_len && memEQ(key_pv,got_key_pv,key_len)) {
        if (val_sv) {
            U64 gf= mph->general_flags;
            sv_set_from_bucket(aTHX_ val_sv,strs,bucket->val_ofs,bucket->val_len,index,((U8*)mph)+mph->val_flags_ofs,1,
                                 gf & MPH_VALS_ARE_SAME_UTF8NESS_MASK, MPH_VALS_ARE_SAME_UTF8NESS_SHIFT);
        }
        return 1;
    }
    return 0;
}

IV
mph_mmap(pTHX_ char *file, struct mph_obj *obj, SV *error, U32 flags) {
    struct stat st;
    struct mph_header *head;
    int fd = open(file, O_RDONLY, 0);
    void *ptr;
    U32 alignment;

    if (error)
        sv_setpvs(error,"");
    if (fd < 0) {
        if (error)
            sv_setpvf(error,"file '%s' could not be opened for read", file);
        return MPH_MOUNT_ERROR_OPEN_FAILED;
    }
    if (fstat(fd,&st)==-1) {
        if (error)
            sv_setpvf(error,"file '%s' could not be fstat()ed", file);
        return MPH_MOUNT_ERROR_FSTAT_FAILED;
    }
    if (st.st_size < sizeof(struct mph_header)) {
        if (error)
            sv_setpvf(error,"file '%s' is too small to be a valid PH2L file", file);
        return MPH_MOUNT_ERROR_TOO_SMALL;
    }
    ptr = mmap(NULL, st.st_size, PROT_READ, MAP_SHARED | MPH_MAP_POPULATE, fd, 0);
    close(fd); /* kernel holds its own refcount on the file, we do not need to keep it open */
    if (ptr == MAP_FAILED) {
        if (error)
            sv_setpvf(error,"failed to create mapping to file '%s'", file);
        return MPH_MOUNT_ERROR_MAP_FAILED;
    }

    obj->bytes= st.st_size;
    obj->header= head= (struct mph_header*)ptr;
    if (head->magic_num != MAGIC_DECIMAL) {
        if (head->magic_num == MAGIC_BIG_ENDIAN_DECIMAL) {
            if (error)
                sv_setpvf(error,"this is a big-endian machine, cant handle PH2L files here");
        }
        if (error)
            sv_setpvf(error,"file '%s' is not a PH2L file", file);
        return MPH_MOUNT_ERROR_BAD_MAGIC;
    }
    if (head->variant < MIN_VARIANT) {
        if (error)
            sv_setpvf(error,"unsupported old version '%d' in '%s'", head->variant, file);
        return MPH_MOUNT_ERROR_BAD_VERSION;
    }
    if (head->variant > MAX_VARIANT) {
        if (error)
            sv_setpvf(error,"unknown version '%d' in '%s'", head->variant, file);
        return MPH_MOUNT_ERROR_BAD_VERSION;
    }
    alignment = sizeof(U64);

    if (st.st_size % alignment) {
        if (error)
            sv_setpvf(error,"file '%s' does not have a size which is a multiple of 16 bytes", file);
        return MPH_MOUNT_ERROR_BAD_SIZE;
    }
    if (
        head->table_ofs < head->state_ofs           ||
        head->key_flags_ofs < head->table_ofs       ||
        head->val_flags_ofs < head->key_flags_ofs   ||
        head->str_buf_ofs < head->val_flags_ofs     ||
        st.st_size < head->str_buf_ofs
    ) {
        if (error)
            sv_setpvf(error,"corrupt header offsets in '%s'", file);
        return MPH_MOUNT_ERROR_BAD_OFFSETS;
    }
    if (flags & MPH_F_VALIDATE) {
        char *start= ptr;
        char *state_pv= start + head->state_ofs;
        char *str_buf_start= start + head->str_buf_ofs;
        char *str_buf_end= start + st.st_size;



( run in 1.529 second using v1.01-cache-2.11-cpan-df04353d9ac )