c_cpp/readscorr/sdleft.h

   1 // by Hu Xuesong
   2 #ifndef _G_SDLEFT_H
   3 #define _G_SDLEFT_H
   4
   5 #include "gtypendef.h"
   6 #include <stddef.h> //size_t
   7 #include <stdint.h> //uint64_t
   8 #include <stdio.h>  //FILE
   9 #include "gFileIO.h"
  10 #include "cfgparser.h"
  11
  12 #define HASH_LENB (128u)
  13 #define SDL_SUBARRAY_UNIT (8u)
  14 //#define SDLA_ITEMARRAY 32u
  15 #define SUBARRAY_SIZE (32u*1024u)
  16 //gcc -### -march=native -E /usr/include/stdlib.h 2>&1 | grep l1-cache-size
  17 #define ENCODEDBIT_ENTROPY_PAD (1u);
  18 /*
  19 http://math.stackexchange.com/questions/53353/is-there-some-reversible-mapping-that-as-uniform-as-a-hash
  20 Let $H$ be a hash function taking an arbitrary string as input and producing an $n$-bit string as output.
  21 Given an input string $s$, let $s_0$ denote the first $n$ bits of $s$ and $s_1$ the rest
  22 (i.e. $s = s_0\;||\;s_1$, where $s_0$ is $n$ bits long and $||$ denotes concatenation).
  23 Then define $$f(s) = (s_0 \oplus H(s_1))\;||\;s_1,$$ where $\oplus$ means bitwise XOR.
  24 It's easy to see that $f$ is its own inverse, i.e.$f(f(s)) = s$,
  25 so that the input string $s$ can be recovered from $f(s)$ just by running it through $f$ again.
  26
  27 f(s)=(s0 XOR H(s1))||s1, length(s0)=EncodedBit, length(s1)=rBit
  28
  29 EncodedBit < ArrayBit < rBit
  30 */
  31 typedef struct __SDLeftArray_t {
  32     unsigned char CountBit, rBit, ArrayBit, EncodedBit;
  33     unsigned char itemByte; //, HashCnt;
  34     uint16_t SubItemCount,SubItemByUnit;  //max(SubItemCount) should be 32768
  35     size_t ArraySize,SDLAbyte;
  36     //uint64_t maxCount; == Item_CountBitMask
  37     //unsigned char ArrayCount;
  38     uint64_t ItemInsideAll, CellOverflowCount, CountBitOverflow; // ItemInsideAll = ItemInsideArray + CellOverflowCount
  39     double FalsePositiveRatio;
  40     void *pDLA, *pextree;
  41     uint64_t maxCountSeen;
  42     //uint64_t *outhash;
  43     uint64_t outhash[2];    // both ArrayBit and rBit is (0,64], so HashCnt==1 for MurmurHash3_x64_128
  44     uint128_t Item_rBitMask;
  45     uint64_t Hash_ArrayBitMask, Hash_rBitMask, Item_CountBitMask;
  46 } SDLeftArray_t;
  47
  48 typedef struct __SDLdumpHead_t {
  49     char FileID[4]; //"GDSD"
  50     unsigned char FileVersion[2];    //0,1
  51     uint16_t kmersize;
  52     unsigned char CountBit, rBit;
  53     uint16_t SubItemCount;
  54     uint64_t ArraySize, SDLAbyte, extreebyte;
  55     uint64_t ItemInsideAll, CellOverflowCount, CountBitOverflow;
  56     uint64_t maxCountSeen;
  57     uint64_t HistMaxCntVal;
  58     uint64_t HistMaxHistVal;
  59     double HistMean;
  60     double HistSStd;
  61     uint32_t crc32c;
  62 } __attribute__ ((packed)) SDLdumpHead;
  63
  64 SDLeftArray_t *dleft_arraynew(unsigned char CountBit, const SDLConfig * const psdlcfg);
  65 SDLeftArray_t *dleft_arrayinit(unsigned char CountBit, size_t ArraySize, uint16_t SubItemCount);
  66 size_t dleft_insert_read(unsigned int k, char const *const inseq, size_t len, SDLeftArray_t *dleftobj);
  67
  68 unsigned char GETitemByte_PADrBit_trimSubItemCount(const unsigned char CountBit, unsigned char *prBit, uint16_t *pSubItemCount);
  69
  70 void fprintSDLAnfo(FILE *stream, const SDLeftArray_t * dleftobj);
  71 void dleft_arraydestroy(SDLeftArray_t * const dleftobj);
  72 void dleft_dump(const SDLeftArray_t * const, SDLdumpHead * const, FILE *);
  73
  74 //#include "sdleftTF.h"
  75 typedef struct __SDLeftStat_t {
  76     uint64_t HistMaxCntVal;
  77     uint64_t HistMaxHistVal;
  78     double HistMean;
  79     double HistSStd;
  80 } SDLeftStat_t;
  81
  82 typedef SDLeftStat_t *(G_SDLeftArray_IN)(SDLeftArray_t * const, FILE *);    //*G_SDLeftArray_IN() is OK,too .
  83
  84 #ifdef TEST
  85 SDLeftStat_t * dleft_stat(SDLeftArray_t * const dleftobj, FILE *stream, FILE *fpdat);
  86 #else
  87 SDLeftStat_t * dleft_stat(SDLeftArray_t * const dleftobj, FILE *stream);
  88 #endif
  89
  90 #endif /* sdleft.h */
  91