// Parse a file from the starting to the end.
void parseFile(void) {
// Where we are paring now?
char * current;
char * wordStart;
int insideWord = 0;
current = fileStart;
while(current <= fileEnd) {
if(insideWord == 1) {
// If we are inside a word, we only care about
// the end of this word.
if(*current == ' ' || current==fileEnd) {
if(isGoodWord(wordStart, current-1)) {
addWordToList(wordStart, current-1);
}
insideWord = 0;
wordStart = NULL;
}
}
else {
if(*current != ' ') {
wordStart = current;
insideWord = 1;
// printf("wordStart %c\n", *current);
}
}
current++;
}
}
int main(int argc, char **argv){
int fd;
char *fname;
struct stat finfo;
char *fdata;
char *fstart;
char *fend;
char *fcurrent;
validWords *list=NULL;
if(argc<=2){
if(argc==2){
printf("Third argument is missing\n");
exit(-1);
} else {
printf("File location not specified\n");
exit(-1);
}
}
//reading file name from command line argument
fname=argv[1];
//Open system call will open the file for you. It returns the file descripter
fd=open(fname, O_RDONLY);
if(fd<0){
perror("Opening the file failed");
exit(-1);
}
//initializing fstats with the statistics of the file to retrieve the file size
if(fstat(fd, &finfo) < 0){
perror("couldn't fetch the stats of the file");
exit(-1);
}
//creating a private memory of size equal to size of the file and mapping it to the address space of this process
fdata = (char *) mmap(0, finfo.st_size + 1, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
if(fdata==MAP_FAILED){
perror("couldn't map a memory of file content with process address space");
exit(-1);
}
fstart=fdata;
fend=fdata+finfo.st_size;
fcurrent=fstart;
//Reaching to the first valid word
while((toupper(*fcurrent)<'A' || toupper(*fcurrent)>'Z') && *fcurrent!='-' && fcurrent<fend){
fcurrent++;
}
char *wordStartChar=fcurrent;
char *wordEndChar=NULL;
int isGoodWord=1;
// int count=0;
char c;
while(fcurrent<fend){
c = toupper(*fcurrent);
if((c<'A' || c>'Z') && c!='-' && c!=' ' && c!='\n'){
isGoodWord=0;
}else if(c==' ' || c=='\n'){
if(isGoodWord){
//add semaphore here
list = addWordToList(wordStartChar, wordEndChar, list);
//leaveSemaphore lock here
// count++;
}
while((c<'A' || c>'Z') && c!='-' && fcurrent<fend){
fcurrent++;
c = toupper(*fcurrent);
}
if(fcurrent>=fend){
break;
}else {
wordStartChar=fcurrent;
wordEndChar=NULL;
isGoodWord=1;
}
}
wordEndChar=fcurrent;
fcurrent++;
}
// printList(list);
list=mergeSort(list);
// printf("%d\n",count);
while(list!=NULL){
printf("%s\t%d\n",list->word,list->occurences);
list=list->next;
}
}
int main()
{
//Performance Calculation Variables
clock_t begin, end;
double processTime;
/////////////////////////
hashTable* hTable = NULL;
wList* wordList = NULL;
int i=0;
char** fileNames = getCWDtxtFiles();
FILE* file[NumOftxtFiles];
for(i=0;i<NumOftxtFiles;++i)
{
file[i] = fopen(fileNames[i], "r");
}
FILE* out = fopen("Analysis/output.txt", "a");
char lol[50] = {'\0'};
begin = clock();
for(i=0;i<NumOftxtFiles;++i)
{
while(!feof(file[i]))
{
fscanf(file[i], "%s", lol);
trimWord(lol);
if(lol[0] == '\0')
continue;
wordList = addWordToList(wordList, lol, i+1);
}
}
storeBST();
//Performance Calculation--
end = clock();
processTime = (double)(end-begin)/CLOCKS_PER_SEC;
//Performance Calculation--
int numWords = countWords(wordList);
hTable = createHashTable(numWords);
addToTableFromList(hTable, wordList);
//printHashTable(hTable);
//printHashTableToFile(hTable, out);
fclose(out);
fclose(out);
for(i=0;i<2;++i)
{
fclose(file[i]);
}
FILE* record = fopen("Analysis/record.txt", "a");
int emptyCount = countEmpty(hTable);
fprintf(record, "WordCount : %d Empty : %d ProcessTime : %f s\n", numWords, emptyCount, processTime);
fclose(record);
printf("\nNo in list = %d %d", countWords(wordList), numWords);
int c = 0;
for(i=0;i<numWords;i++){
bucket* b = hTable->table[i];
while(b != NULL){
++c;
b = b->next;
}
}
//Driver Loop
char ans = 'y';
while(ans == 'y'){
printBannner(processTime);
printf("\n\t"); printf("__________________________________________________________\n\n");
printf("\t"); printf(" Enter the Query \n\n");
printf("\t\t\t\t"); scanf("%s",lol);
printf("\t"); printf("__________________________________________________________\n");
if(lol[strlen(lol)-1] == '*'){
makeList(wordList, lol);
char* word = NULL;
while((word = getNextWord())!= NULL){
printf("\n\t\t");printf("%s : \n", word);
searchText(word, hTable, fileNames, NumOftxtFiles);
}
}
else{
trimWord(lol);
char* word = getCorrectWord(lol);
if(strcmp(word, lol) != 0)
{
printf("\n\t\t\t");printf("Corrected Word : %s\n", word);
searchText(word, hTable, fileNames, NumOftxtFiles);
printf("\n\t\t\t");printf("Do you want to search for original word? y/n : ");
char a = 'y';
while(getchar() != '\n');
scanf("%c", &a);
if(a == 'y')
searchText(lol, hTable, fileNames, NumOftxtFiles);
}
else
searchText(lol, hTable, fileNames, NumOftxtFiles);
}
printf("\n\n\t\t\t");printf("Search Again? y/n : ");
while(getchar() != '\n');
scanf("%c", &ans);
}
return 0;
}
