void encode () {
codeCount = 1;
Tree codeTrie;
createT(&codeTrie);
if (!Escaped){
initializeTrie(&codeTrie);
}
int firstChar = getchar();
while (firstChar != EOF) {
firstChar = getLargestSequence(firstChar, &codeTrie);
}
freeTrie(&codeTrie);
}
int main()
{
Trie Root;
Root =initializeTrie(Root);
Root= addwords(Root ,"sairam");
Root =addwords(Root,"surfeit");
Root =addwords(Root,"hello");
if(!searchWords(Root,"sairam"))
printf("does not exist\n");
else
printf("it does\n");
return 0;
}
Trie addwords(Trie Root ,char *words)
{
int t;
if(words[0]=='\0')
{
}
else
{
t = words[0]-'a';
if(Root->edges[t]==NULL)
Root->edges[t] =initializeTrie(Root->edges[t]);
else
Root->edges[t] =addwords(Root->edges[t],++words);
}
return Root;
}
void insert_i(Trie * node, const char * key, const char * val, int pos, int n, int idx)
{
if( pos == n)
{
/* Se achou o elemento */
int size = strlen(val);
/* Se ainda está nulo, aloca o espaço */
if(node->elem == NULL)
node->elem = (char*) malloc(sizeof(char) * size);
/* Copia o identificador para a posição correta */
strcpy(node->elem, val);
node->idx = idx;
} else {
/* Se o nó que precisa ser inserido está nulo, alocar */
if(node->branch[mapChar(val[pos])] == NULL) {
node->branch[mapChar(val[pos])] = (Trie*) malloc(sizeof(Trie));
initializeTrie(node->branch[mapChar(val[pos])]);
}
insert_i(node->branch[mapChar(val[pos])], key, val, pos+1, n, idx);
}
}
int main(){
//int indexer() {
FILE *files,*fp,*conf;
char temp[MAXIMUM],tempFileName[MAXIMUM],filesName[MAXIMUM],textFileName[MAXIMUM];
int i,docId = 1;
int fileType;
int ret;
int totalErrors = 0;
initializeStopWordsArray();
initializeTrie();
/* conf = fopen("CONFIG","r");
readLine(conf,filesName);
fclose(conf);*/
files = fopen("INDEX/files.txt","r");
if(files == NULL){
goto createStatusFile;
}
while(readLine(files,tempFileName) != FAIL) {
printf("%d \n",docId);
printf("%s\n",tempFileName);
fileType = checkFileExtension(tempFileName);
strcpy(textFileName,"temp.txt");
if(fileType == PDF){
ret = convertPdfToTxt(tempFileName,textFileName);
if(ret != SUCCESS){
++totalErrors;
++docId;
continue;
}
}
else if(fileType == DOC){
ret = convertDocToTxt(tempFileName,textFileName);
if(ret != SUCCESS){
++totalErrors;
++docId;
continue;
}
}
else if(fileType == ODT){
ret = convertOdtToTxt(tempFileName,textFileName);
if(ret != SUCCESS){
++totalErrors;
++docId;
continue;
}
}else if(fileType == TXT){
strcpy(textFileName,tempFileName);
}else if(fileType == 0){
++totalErrors;
++docId;
continue;
}
fp = fopen(textFileName,"r");
if(fp == NULL){
++totalErrors;
++docId;
continue;
}
// Read each filename from the file 'files.txt'
// For each filename, read each word and add to index trie.
while(readWord(fp,temp) != FAIL) {
convertToLowerCase(temp);
removeSpecialSymbols(temp);
// printf("%s ",temp);
if(temp[0] == '\0' || temp[0] == '\t' || temp[0] == ' ' || temp[0] == '\n')
continue;
if(checkForStopWords(temp) != SUCCESS) {
int i = 0;
// Presently on the words containing letters are allowed
while(temp[i] != '\0') {
if(temp[i] < 97 || temp[i] > 122)
break;
++i;
}
if(temp[i] == '\0') {
fileLen[docId]++;
// printf("%d %s\n",docId,temp);
addTerm(temp,docId);
}
//if(temp[0] >= 97 && temp[0]<=122) {
// addTerm(temp,docId);
//}
//printf("%s\n",temp);
}
}
++docId;
fclose(fp);
if(fileType != TXT){
system("rm temp.txt");
system("touch temp.txt");
}
printf("...Done\n");
}
//printf("\nTrie Traversal\n");
//traverseTerms();
numOfDocs = docId - 1;
printf("Number of docs : %d\n",numOfDocs);
printf("Number of Errors : %d\n",totalErrors);
// traverseTerms();
createIndexFile();
createStatusFile:
system("echo \"Index Status\n==================\nIndex created on :\n\" > INDEX/STATUS.txt");
system("date >> INDEX/STATUS.txt");
system("echo \"Directory indexed : \"#// >> INDEX/STATUS.txt");
sprintf(temp,"echo \"No of Documents indexed : %d\" >> INDEX/STATUS.txt",numOfDocs);
system(temp);
return 0;
}
void decode () {
int possition = 0;
int code = 9;
Tree codeTrie;
createT(&codeTrie);
int codeCount = 1;
int *outputString = malloc(sizeof(int));
int *lastString = NULL;
outputString[0] = -1;
int * nextString = NULL;
int bitsCount = 8;
int getMaxBits = getBits(bitsCount);
setMaxBits(getMaxBits);
int Escaped = getBits(1);
if (Escaped) {
setEscaped();
}
else {
initializeTrie(&codeTrie);
}
int shouldPrune = getBits(1);
if (shouldPrune) {
setShouldPrune();
}
while ((code = getBits(bitsCount)) != EOF) {
if (code == 0 && Escaped) {
int singleLetter = getBits(8);
nextString = malloc(2 * sizeof(int));
nextString[0] = singleLetter;
nextString[1] = -1;
}
else {
nextString = checkForNextCode(&codeTrie, code);
}
if (!nextString) {
nextString = addLetterToStringNoFree(lastString, lastString[0]);
}
printArray(nextString);
int *output = combine(outputString, nextString, possition);
free(outputString);
outputString = output;
possition = addStringToTrie(&codeTrie, outputString, 0, &codeCount);
int stringLength = arrayLen(nextString);
if (lastString) {
free(lastString);
lastString = NULL;
}
lastString = malloc((arrayLen(nextString) + 1) * sizeof(int));
for (int i = 0; nextString[i] != -1; i++) {
lastString[i] = nextString[i];
}
lastString[stringLength] = -1;
bitsCount = getBitsCount();
free(nextString);
nextString = NULL;
}
if (lastString) {
free(lastString);
}
freeTrie(&codeTrie);
free(outputString);
}
