X-Git-Url: https://git.llucax.com/z.facultad/75.06/jacu.git/blobdiff_plain/dff2f7bc143397a35e8c67bbf2848283b6764dda..2ee6f95a8db79ede018e0f15124dd486451c4170:/src/statichuff/statichuff.c diff --git a/src/statichuff/statichuff.c b/src/statichuff/statichuff.c index b6da6f4..000f0ba 100644 --- a/src/statichuff/statichuff.c +++ b/src/statichuff/statichuff.c @@ -1,21 +1,41 @@ -#include +#include "statichuff.h" +#include +#include -typedef unsigned short int t_freq; - -typedef struct t_freqnode { - unsigned short int symbol; - unsigned short int freq; - struct t_freqnode *lchild; - struct t_freqnode *rchild; -} HUFFNODE; +void putbit(char bit, char restart, char flush, VFILE *fp) +{ + static unsigned long int bits_buffer = 0; + static unsigned char bits_used = 0; -typedef struct t_code { - unsigned short int code; - unsigned char codelength; -} CODE; + /* me obligan a emitir el output */ + if ((flush == 1) && (bits_used > 0)) { + bits_buffer = bits_buffer << ((sizeof(unsigned long int)*8) - bits_used); + vfwrite(&bits_buffer,sizeof(unsigned long int),1,fp); + bits_buffer = 0; + bits_used = 0; + return; + } + /* me indican que comienza un nuevo output */ + if (restart) { + bits_buffer = 0; + bits_used = 0; + } + /* inserto el bit en el buffer */ + bits_buffer = bits_buffer << 1; + bits_buffer |= bit; + bits_used++; + + /* lleno el buffer, escribo */ + if (bits_used == 32) { + vfwrite(&bits_buffer,sizeof(unsigned long int),1,fp); + bits_buffer = 0; + bits_used = 0; + } + return; +} -void cpynode(HUFFNODE *node1, HUFFNODE *node2) +void shuff_cpynode(SHUFFNODE *node1, SHUFFNODE *node2) { node1->symbol = node2->symbol; node1->freq = node2->freq; @@ -23,63 +43,64 @@ void cpynode(HUFFNODE *node1, HUFFNODE *node2) node1->rchild = node2->rchild; } -int compnode(HUFFNODE *node1, HUFFNODE *node2) +int shuff_compnode(const void *node1, const void *node2) { - if (node1->freq < node2->freq) return 1; - if (node1->freq > node2->freq) return -11; + if (((SHUFFNODE*)node1)->freq < ((SHUFFNODE*)node2)->freq) return 1; + if (((SHUFFNODE*)node1)->freq > ((SHUFFNODE*)node2)->freq) return -1; return 0; } -HUFFNODE *buildlist(t_freq *freqtable, int *nonzerofreqs) -{ - int i,j = 0,nonzero = 0; - HUFFNODE *inputlist; - - /* Calculo cuantas frequencias > 0 hay y creo la tabla */ - for (i = 0; i < 256; ++i) if (freqtable[i] > 0) nonzero++; - inputlist = (HUFFNODE*)malloc(sizeof(HUFFNODE)*nonzero); - - /* Cargo la inputlist del huffman solo con freqs > 0 */ - for (i = 0; i < 256; ++i) - if (freqtable[i] > 0) { - inputlist[j].symbol = i; - inputlist[j].freq = freqtable[i]; - inputlist[j].lchild = NULL; - inputlist[j].rchild = NULL; - j++; - } - - *nonzerofreqs = nonzero; - return inputlist; -} - -int rescalefreq(t_freq *freqtable) +int shuff_rescalefreq(t_freq *freqtable) { int i; - int totalfreq = 0; + t_freq totalfreq = 0; /* Divido por la mitad las frecuencias, asegurando de no perder */ - /* frequencias en 1, por ello le sumo 1 antes de partir */ for (i = 0; i < 256; i++) { - freqtable[i] = (freqtable[i]+1)/2; + freqtable[i] = (freqtable[i] >> 2) | 1; totalfreq += freqtable[i]; } return totalfreq; } -int scanfreq(char *inputfile, t_freq *freqtable) +int shuff_scanfreq_chunk(HUFF_STATE *chunkshuff, char* chunk, int chunksize) +{ + /* Locals */ + int i = 0; + unsigned char symbol = 0; + + /* Contamos las frecuencias del chunk a menos que se use un canonico */ + if (!chunkshuff->canonic) { + for (i = 0; i < chunksize; ++i) { + symbol = chunk[i]; + chunkshuff->freqtable[symbol] += 1; + chunkshuff->sumfreq += 1; + + /* Si llegue al tope de freq acumulada, halve em */ + if (chunkshuff->sumfreq == 14930352) + chunkshuff->sumfreq = shuff_rescalefreq(chunkshuff->freqtable); + } + } + + /* Dumpeamos el chunk en el temporal homero */ + fwrite(chunk,chunksize,1,chunkshuff->coderfp); + + return 1; +} + +int shuff_scanfreq(char *inputfile, t_freq *freqtable) { /* Locals */ FILE *fp; - int sumfreq = 0,auxsum = 0; + t_freq sumfreq = 0; int i,symbol; - /* Inicializamos la tabla de frecuencias */ + /* Inicializamos la tabla de frecuencias */ for (i = 0; i < 256; ++i) freqtable[i] = 0; /* Abrimos el file */ - if ((fp = fopen(inputfile,"rb")) == NULL) return 0; + if ((fp = fopen(inputfile,"r")) == NULL) return 0; while (!feof(fp)) { /* Contamos las frecuencias */ symbol = fgetc(fp); @@ -89,15 +110,77 @@ int scanfreq(char *inputfile, t_freq *freqtable) ++sumfreq; /* Si llegue al tope de freq acumulada, halve em */ - if (sumfreq == 4181) - sumfreq = rescalefreq(freqtable); + if (sumfreq == 14930352) + sumfreq = shuff_rescalefreq(freqtable); } fclose(fp); - return 1; + return 1; } -void printcodes(CODE *codetable,t_freq *freqtable) +SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs) +{ + int i,j = 0,nonzero = 0; + SHUFFNODE *inputlist; + + /* Calculo cuantas frequencias > 0 hay y creo la tabla */ + for (i = 0; i < 256; ++i) if (freqtable[i] > 0) nonzero++; + inputlist = (SHUFFNODE*)malloc(sizeof(SHUFFNODE)*nonzero); + + /* Cargo la inputlist del huffman solo con freqs > 0 */ + for (i = 0; i < 256; ++i) + if (freqtable[i] > 0) { + inputlist[j].symbol = i; + inputlist[j].freq = freqtable[i]; + inputlist[j].lchild = NULL; + inputlist[j].rchild = NULL; + j++; + } + + *nonzerofreqs = nonzero; + return inputlist; +} + +SHUFFNODE *shuff_buildtree(t_freq *ftable) +{ + SHUFFNODE *lastsymbol; + SHUFFNODE *node1,*node2,*root; + SHUFFNODE *inputlist; + int freqcount = 0; + + /* Genero la input list en base a la cual genera el arbol */ + inputlist = shuff_buildlist(ftable, &freqcount); + lastsymbol = inputlist+(freqcount-1); + + while (lastsymbol > inputlist) { + /* Ordeno la lista por frecuencia descendente */ + qsort(inputlist,freqcount,sizeof(SHUFFNODE),shuff_compnode); + /* Tomo los ultimos dos elementos, generando dos nodos del arbol */ + node1 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE)); + node2 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE)); + shuff_cpynode(node1,lastsymbol-1); + shuff_cpynode(node2,lastsymbol); + lastsymbol -= 1; + /* Nodo ficticio con la suma de las probs y los ptros a childs */ + lastsymbol->symbol = 256; + lastsymbol->freq = node1->freq + node2->freq; + lastsymbol->lchild = node1; + lastsymbol->rchild = node2; + --freqcount; + } + + /* Copio la raiz para poder liberar la lista sin perderla */ + root = (SHUFFNODE*)malloc(sizeof(SHUFFNODE)); + shuff_cpynode(root,lastsymbol); + + /* Free up mem */ + free(inputlist); + + /* Devuelvo el puntero a la raiz del arbol de huffman */ + return root; +} + +void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable) { int i,j; unsigned short int auxcode; @@ -106,7 +189,7 @@ void printcodes(CODE *codetable,t_freq *freqtable) for (i = 0; i < 256; ++i) { if (codetable[i].codelength > 0) { auxcode = codetable[i].code; - printf("Symbol:%i Freq: %i Code:",i,freqtable[i]); + printf("Symbol:%i Freq: %li Code:",i,freqtable[i]); for (j = codetable[i].codelength-1; j >= 0; --j) { auxcode = codetable[i].code; auxcode = auxcode >> j; @@ -118,82 +201,320 @@ void printcodes(CODE *codetable,t_freq *freqtable) } } -void zerocodes(CODE *table) +void shuff_zerocodes(SHUFFCODE *table) { int i; - /* Inicializo los codigos prefijos */ + /* Inicializo los codigos prefijos */ for (i = 0; i < 256; ++i) { table[i].code = 0; table[i].codelength = 0; } } -void buildcodes(CODE *table, HUFFNODE *node, int level, int code) +void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code) { if (node->symbol < 256) { /* Guardo el codigo en la tabla */ table[node->symbol].code = code; - table[node->symbol].codelength = level; - /*printf("Found symbol %i with freq %i at depth %i\n",node->symbol,node->freq,level);*/ + table[node->symbol].codelength = level; } else { code = code << 1; - buildcodes(table,node->lchild,level+1,code); + shuff_buildcodes(table,node->lchild,level+1,code); code |= 1; - buildcodes(table,node->rchild,level+1,code); + shuff_buildcodes(table,node->rchild,level+1,code); + } +} + +int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable) +{ + FILE *fpsource; + VFILE *fpdest; + int symbol,i; + unsigned long int sourcesize; + char bit; + SHUFFCODE symbolcode; + + /* Abrimos el source y el destino */ + if (shuff->coderfp != NULL) fclose(shuff->coderfp); /* close bychunk temp file */ + if ((fpsource = fopen(shuff->sourcefile,"r")) == NULL) return 0; + if ((fpdest = vfopen(shuff->targetfile,"w",shuff->volsize)) == NULL) return 0; + + /* Guardamos el size el archivo original e inputlist como header */ + fseek(fpsource,0,SEEK_END); + sourcesize = ftell(fpsource); + vfwrite(&sourcesize,sizeof(unsigned long int),1,fpdest); + vfwrite(shuff->freqtable,sizeof(t_freq),256,fpdest); + + /* Encodeo */ + fseek(fpsource,0,SEEK_SET); + while (!feof(fpsource)) { + /* Levanto un symbolo (byte) */ + symbol = fgetc(fpsource); + if (symbol == EOF) continue; + + /* Cargamos el codigo y lo emitimos */ + symbolcode = ctable[symbol]; + for (i = symbolcode.codelength; i > 0; --i) { + bit = (symbolcode.code >> (i-1)) & 1; + putbit(bit,0,0,fpdest); + } } + + /* Hacemos un flush de lo que haya quedado en el buffer de salida */ + putbit(0,0,1,fpdest); + fclose(fpsource); + vfclose(fpdest); + return 1; } -HUFFNODE *buildtree(HUFFNODE *list, int listcount) +int shuff_encode_file(HUFF_STATE *shuff) { - HUFFNODE *lastsymbol = list+(listcount-1); - HUFFNODE *node1,*node2,*fictnode; + /* Locals */ + SHUFFCODE *codetable = (SHUFFCODE*)malloc(sizeof(SHUFFCODE)*256); + + /* Veo si debo armar una freqtable o si esta preloaded */ + if ((!shuff->canonic) && (!shuff->bychunk)) + if (!shuff_scanfreq(shuff->sourcefile,shuff->freqtable)) return 0; + + /* Genero el arbol de huffman */ + shuff->codetree = shuff_buildtree(shuff->freqtable); - /* Ordenamos inicialmente la inputlist para tomar las dos freqs min */ - while (lastsymbol > list) { - /* Ordeno la lista por frecuencia descendente */ - qsort(list,listcount,sizeof(HUFFNODE),compnode); - /* Tomo los ultimos dos elementos, generando dos nodos del arbol */ - node1 = (HUFFNODE*)malloc(sizeof(HUFFNODE)); - node2 = (HUFFNODE*)malloc(sizeof(HUFFNODE)); - cpynode(node1,lastsymbol-1); - cpynode(node2,lastsymbol); - lastsymbol -= 1; - /* Nodo ficticio con la suma de las probs y los ptros a childs */ - lastsymbol->symbol = 256; - lastsymbol->freq = node1->freq + node2->freq; - lastsymbol->lchild = node1; - lastsymbol->rchild = node2; - --listcount; + /* Armo la tabla de codigos prefijos para el encoder */ + shuff_zerocodes(codetable); + shuff_buildcodes(codetable,shuff->codetree,0,0); + /*shuff_printcodes(codetable,shuff->freqtable);*/ + + /* Encodeo byte per byte */ + shuff_encode_symbols(shuff,codetable); + + /* Free up memory baby yeah */ + free(codetable); + + return 1; +} + +SHUFFNODE *shuff_decode_symbols(SHUFFNODE *entrynode, unsigned long int buffer, + int *bitsleft, unsigned short int *symbol) +{ + char bit = 0; + + /* Levanto el symbolo y si es uno valido, devuelvo */ + *symbol = entrynode->symbol; + if (*symbol != 256) return entrynode; + if (*bitsleft == 0) return entrynode; + + /* Obtengo otro bit a procesar y me muevo en el arbol */ + bit = (buffer >> ((*bitsleft)-1)) & 1; + --(*bitsleft); + if (bit == 0) return shuff_decode_symbols(entrynode->lchild,buffer,bitsleft,symbol); + else return shuff_decode_symbols(entrynode->rchild,buffer,bitsleft,symbol); +} + +int shuff_decode_chunk(HUFF_STATE *shuff, char *chunk, int chunksize, int *decodedbytes) +{ + SHUFFNODE *currnode = shuff->codetree; + unsigned short int decoded_symbol; + *decodedbytes = 0; + + while (!vfeof(shuff->decoderfp) && (shuff->bytesleft > 0) && (*decodedbytes < chunksize)) { + + /* Leo un buffer de 32 bits si es que quedo vacio el anterior */ + if (shuff->bitsleft == 0) { + if (vfread(&(shuff->codebuffer),sizeof(unsigned long int),1,shuff->decoderfp) != 1) continue; + shuff->bitsleft = sizeof(unsigned long int) * 8; + } + + /* Proceso el buffer sacando simbolos till se me agote el buffer, file o chunk */ + while ((shuff->bitsleft > 0) && (shuff->bytesleft > 0) && (*decodedbytes < chunksize)) { + currnode = shuff_decode_symbols(currnode,shuff->codebuffer,&(shuff->bitsleft),&decoded_symbol); + /* Si obtuve un symbolo valido lo emito*/ + if (decoded_symbol != 256) { + chunk[(*decodedbytes)++] = decoded_symbol; + currnode = shuff->codetree; + --(shuff->bytesleft); + } + } } + + if (shuff->bytesleft == 0) return 0; + else return 1; +} + +int shuff_decode_file(HUFF_STATE *shuff) +{ + SHUFFNODE *currnode; + unsigned long int codebuffer; + FILE *fpdest; + unsigned short int decoded_symbol; + int bitsleft; + + /* Comienzo a decodificar, pues la tabla ya la levante en el decinit */ + if ((fpdest = fopen(shuff->targetfile,"w")) == NULL) return 0; + currnode = shuff->codetree; + + while (!vfeof(shuff->decoderfp) && (shuff->bytesleft > 0)) { - /* Devuelvo el puntero a la raiz del arbol de huffman */ - return lastsymbol; + /* Leo un buffer de 32 bits */ + if (vfread(&codebuffer,sizeof(unsigned long int),1,shuff->decoderfp) != 1) continue; + bitsleft = sizeof(unsigned long int) * 8; + + /* Proceso el buffer sacando simbolos hasta que se me agote */ + while ((bitsleft > 0) && (shuff->bytesleft > 0)) { + currnode = shuff_decode_symbols(currnode,codebuffer,&bitsleft,&decoded_symbol); + /* Si obtuve un symbolo valido lo emito*/ + if (decoded_symbol != 256) { + fputc(decoded_symbol,fpdest); + currnode = shuff->codetree; + --(shuff->bytesleft); + } + } + } + + /* Close destination */ + fclose(fpdest); + + return 1; +} + +HUFF_STATE *shuff_init_decoder(char *inputfile, char *outputfile) +{ + /* Locals */ + HUFF_STATE *shuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE)); + shuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256); + + /* Init fields */ + shuff->codebuffer = 0; + shuff->bitsleft = 0; + shuff->coderfp = NULL; + shuff->targetfile = NULL; + shuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(inputfile)+1)); + strcpy(shuff->sourcefile,inputfile); + if (outputfile != NULL) { + shuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1)); + strcpy(shuff->targetfile,outputfile); + } + + /* Levanto cuantos bytes debo decodificar y la freqtable */ + if ((shuff->decoderfp = vfopen(shuff->sourcefile,"r",0)) == NULL) return NULL; + vfread(&(shuff->bytesleft),sizeof(unsigned long int),1,shuff->decoderfp); + vfread(shuff->freqtable,sizeof(t_freq),256,shuff->decoderfp); + /* Armo el arbol de huffman que uso para decodificar */ + shuff->codetree = shuff_buildtree(shuff->freqtable); + + return shuff; +} + +HUFF_STATE *shuff_init_encoder_byfile(char *inputfile, char *outputfile, long volsize) +{ + /* Locals */ + HUFF_STATE *fshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE)); + int i; + + /* Inicializo la estructura para trabajar con Huff Static by File */ + fshuff->coderfp = NULL; + fshuff->decoderfp = NULL; + fshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(inputfile)+1)); + fshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1)); + strcpy(fshuff->sourcefile,inputfile); + strcpy(fshuff->targetfile,outputfile); + fshuff->volsize = volsize; + fshuff->bychunk = 0; + fshuff->canonic = 0; + fshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256); + for (i = 0; i < 256; ++i) fshuff->freqtable[i] = 0; + fshuff->sumfreq = 0; + fshuff->codetree = NULL; + + return fshuff; } -int main(int argc, char* argv[]) +HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long volsize) { /* Locals */ - t_freq *freqtable = (t_freq*)malloc(sizeof(t_freq)*256); - HUFFNODE *inputlist; - HUFFNODE *codetree; - CODE *codetable = (CODE*)malloc(sizeof(CODE)*256); - int freqcount = 0,i; + HUFF_STATE *cshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE)); + int i; - if (argc == 1) return -1; + /* Inicializo la estructura para trabajar con Huff Static by Chunks */ + cshuff->decoderfp = NULL; + cshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(outputfile)+2)); + cshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1)); + strcpy(cshuff->targetfile,outputfile); + strcpy(cshuff->sourcefile,outputfile); + strcat(cshuff->sourcefile,"~"); + cshuff->volsize = volsize; + cshuff->bychunk = 1; + cshuff->canonic = 0; + cshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256); + for (i = 0; i < 256; ++i) cshuff->freqtable[i] = 0; + cshuff->sumfreq = 0; + cshuff->codetree = NULL; - /* Armamos la tabla de frecuencias */ - if (!scanfreq(argv[1],freqtable)) return -1; + /* Abrimos un archivo temporal para ir tirando los chunks */ + if ((cshuff->coderfp = fopen(cshuff->sourcefile,"w")) == NULL) return NULL; + + return cshuff; +} + +int shuff_loadmodel(HUFF_STATE *shuff, char *modelfile) { + + FILE *fp; - /* Armo el input list y genero el arbol de huffman */ - inputlist = buildlist(freqtable, &freqcount); - codetree = buildtree(inputlist,freqcount); - zerocodes(codetable); - buildcodes(codetable,codetree,0,0); - printcodes(codetable,freqtable); - /*encode(codetable)*/ + if ((shuff) && (shuff->freqtable) && (modelfile)) { + /* Cargo el modelo de disco */ + if ((fp = fopen(modelfile,"r")) == NULL) return 0; + if (fread(shuff->freqtable,sizeof(t_freq),256,fp) != 256) return 0; + shuff->canonic = 1; + if (fp) fclose(fp); + return 1; + } + return 0; +} + +int shuff_savemodel(HUFF_STATE *shuff) { + + FILE *fp; + char *auxfilename; + char *stopchar; + if ((shuff) && (shuff->targetfile) && (shuff->freqtable)) { + /* Preparo el nombre del archivo con la tabla */ + auxfilename = (char*)malloc(strlen(shuff->targetfile)+1); + stopchar = strrchr(shuff->targetfile,'.'); + strncpy(auxfilename,shuff->targetfile,stopchar - shuff->targetfile); + auxfilename[stopchar - shuff->targetfile] = 0; + strcat(auxfilename,".ftb"); + + /* Lo creamos y dumpeamos la tabla de frecuencias (modelo) */ + if ((fp = fopen(auxfilename,"w")) == NULL) return 0; + fwrite(shuff->freqtable,sizeof(t_freq),256,fp); + if (fp) fclose(fp); + + return 1; + } return 0; +} +void shuff_deinit_encoder(HUFF_STATE *shuff) +{ + /* Libero mallocs y cierro archivos */ + if (shuff->freqtable) free(shuff->freqtable); + if (shuff->coderfp) fclose(shuff->coderfp); + if (shuff->bychunk) unlink(shuff->sourcefile); + if (shuff->sourcefile) free(shuff->sourcefile); + if (shuff->targetfile) free(shuff->targetfile); + + /* Destruyo recursivamente el arbol de codigos */ +} + +void shuff_deinit_decoder(HUFF_STATE *shuff) +{ + /* Libero mallocs y cierro archivos */ + if (shuff->freqtable) free(shuff->freqtable); + if (shuff->sourcefile != NULL) free(shuff->sourcefile); + if (shuff->targetfile != NULL) free(shuff->targetfile); + if (shuff->decoderfp != NULL) vfclose(shuff->decoderfp); + + /* Destruyo recursivamente el arbol de codigos */ }