copyfiles() {
writefile(TMPFILE);
int page_number1 = -1;
int page_number2 = -1;
int *buf1, *buf2;
int fd_inpfile, fd_finalfile;
if ((fd_inpfile=PF_OpenFile(TMPFILE))<0){
PF_PrintError("Error in opening file");
exit(1);
}
if ((fd_finalfile=PF_OpenFile(FILE))<0){
PF_PrintError("Error in opening file");
exit(1);
}
//transferring contents of finalfile to tmpfile
int i, error;
for(i = 0 ; i < B_R ; i++) {
// get the page page_number of each file
if((error=PF_GetNextPage(fd_finalfile,&page_number1,&buf1))!= PFE_OK) {
PF_PrintError("Error");
exit(1);
}
if((error=PF_GetNextPage(fd_inpfile,&page_number2,&buf2))!= PFE_OK) {
PF_PrintError("Error");
exit(1);
}
*((int *)buf2) = *((int *)buf1);
//printf("%d ", *((int *)buf2));
if ((error=PF_UnfixPage(fd_finalfile,page_number1,TRUE))!= PFE_OK){
PF_PrintError("unfix buffer\n");
exit(1);
}
if ((error=PF_UnfixPage(fd_inpfile,page_number2,TRUE))!= PFE_OK){
PF_PrintError("unfix buffer\n");
exit(1);
}
}
if ((error=PF_CloseFile(fd_inpfile))!= PFE_OK){
PF_PrintError("Error in closing input file\n");
exit(1);
}
if ((error=PF_CloseFile(fd_finalfile))!= PFE_OK){
PF_PrintError("Error in closing temporary file\n");
exit(1);
}
}
selectByName(char* fname, char* name)
{
int page_hdr = 4;
int i;
int fd,pagenum;
int *buf;
int error;
//printf("opening %s\n",fname);
if ((fd=PF_OpenFile(fname))<0){
PF_PrintError("open file");
exit(1);
}
pagenum = -1;
while ((error=PF_GetNextPage(fd,&pagenum,&buf))== PFE_OK){
struct Record r;
int n_tuppp = buf[0];
int i = 0;
for(; i < n_tuppp; ++i)
{
readRecord(&r, page_hdr + i * sizeof(r), sizeof(r), buf);
if(strcmp(r.name, name)){
// Do something
PF_PrintError(":)");
}
}
if ((error=PF_UnfixPage(fd,pagenum,FALSE))!= PFE_OK){
PF_PrintError("unfix");
exit(1);
}
}
if ((error=PF_CloseFile(fd))!= PFE_OK){
PF_PrintError("close file");
exit(1);
}
}
join(char *f1, char *f2) {
int page_hdr1 = 4, page_hdr2 = 4;
int fd1, pagenum1, *buf1, error1, fd2, pagenum2, *buf2, error2;
//printf("opening %s\n", f1);
if ((fd1=PF_OpenFile(f1))<0){
PF_PrintError("open file");
exit(1);
}
//printf("opening %s\n", f2);
if ((fd2=PF_OpenFile(f2))<0) {
PF_PrintError("open file");
exit(1);
}
pagenum1 = -1;
while ((error1=PF_GetNextPage(fd1,&pagenum1,&buf1))== PFE_OK) {
pagenum2 = -1;
while ((error2=PF_GetNextPage(fd2,&pagenum2,&buf2))== PFE_OK) {
struct Record r1;
struct Record r2;
int n_tuppp1 = buf1[0], n_tuppp2 = buf2[0], i1 = 0, i2 = 0;
for(; i1 < n_tuppp1; ++ i1) {
i2 = 0;
for(; i2 < n_tuppp2; ++ i2) {
readRecord(&r1, page_hdr1 + i1 * sizeof(r1), sizeof(r1), buf1);
readRecord(&r2, page_hdr2 + i2 * sizeof(r2), sizeof(r2), buf2);
// Can implement join condition and print output to stdout/file here. Will not affect buffer performance analysis.
}
}
if ((error2=PF_UnfixPage(fd2,pagenum2,FALSE))!= PFE_OK) {
PF_PrintError("unfix");
exit(1);
}
}
if ((error1=PF_UnfixPage(fd1,pagenum1,FALSE))!= PFE_OK) {
PF_PrintError("unfix");
exit(1);
}
}
if ((error2=PF_CloseFile(fd2))!= PFE_OK){
PF_PrintError("close file");
exit(1);
}
if ((error1=PF_CloseFile(fd1))!= PFE_OK){
PF_PrintError("close file");
exit(1);
}
}
main(int argc, char *argv[])
{
int error;
int i,fd;
int pagenum,*buf;
int *buf1,*buf2;
int fd1,fd2;
int nowrites;
//int stale_file[2][];
if( argc == 3 )
{
printf("output The argument supplied are %s, %s\n", argv[1], argv[2]);
//char temp[] = *argv[1];
//nowrites = *argv[2];
sscanf(argv[1], "%d", &filesize);
sscanf(argv[2], "%d", &nowrites);
filesize = filesize*PF_MAX_BUFS;
printf("output final The argument supplied are %d, %d\n", filesize, nowrites);
}
else
{
printf("One argument expected.\n");
filesize = 2*PF_MAX_BUFS;
nowrites = 20;
}
/* create a few files */
if ((error=PF_CreateFile(FILE1))!= PFE_OK){
PF_PrintError("file1");
exit(1);
}
printf("file size - %d", filesize);
int stale_file1[20000];
for (i=0; i<filesize; i++)
{
stale_file1[i] = 0;
}
printf("file1 created\n");
if ((error=PF_CreateFile(FILE2))!= PFE_OK){
PF_PrintError("file2");
exit(1);
}
int stale_file2[20000];
for (i=0; i<filesize; i++)
{
stale_file2[i] = 0;
}
printf("file2 created\n");
/* write to file1 */
writefile(FILE1,stale_file1);
/* print it out */
readfile(FILE1,stale_file1);
//Sequential Updates
reads = 0;
writes = 0;
erases = 0;
if ((fd=PF_OpenFile(FILE1))<0){
PF_PrintError("open file");
exit(1);
}
pagenum = -1;
int temp = 0;
printf("\noutput final Sequential Updates\n");
while (temp<nowrites && (error=PF_GetNextPage(fd,&pagenum,&buf))== PFE_OK)
//for(i=0; i<nowrites; i++)
{
i = pagenum;
temp++;
if ((error=PF_UnfixPage(fd,pagenum,FALSE))!= PFE_OK){
PF_PrintError("unfix");
exit(1);
}
updatefile(FILE1,stale_file1,i);
}
printf("reads - %d\n", reads);
printf("erases - %d\n", erases);
printf("writes - %d\n", writes);
garbagecollect(FILE1,stale_file1);
printf("output file size - %d\n", filesize);
printf("output Number of updates - %d\n", nowrites);
printf("output Number of pages in block - %d\n", PAGEINBLOCK);
printf("output After garbage collection\n");
printf("output final reads - %d\n", reads);
printf("output final erases(in units of blocks) - %d\n", erases/PAGEINBLOCK);
printf("output final writes - %d\n", writes);
if ((error=PF_CloseFile(fd))!= PFE_OK){
PF_PrintError("close file");
exit(1);
}
readfile(FILE1, stale_file1);
/* write to file2 */
writefile(FILE2,stale_file2);
/* print it out */
readfile(FILE2,stale_file2);
//Random Updates
reads = 0;
writes = 0;
erases = 0;
if ((fd=PF_OpenFile(FILE2))<0){
PF_PrintError("open file");
exit(1);
}
pagenum = -1;
temp = 0;
printf("\noutput final Random Updates\n");
while (temp<nowrites && ((error=PF_GetNextPage(fd,&pagenum,&buf))== PFE_OK || error == PFE_EOF) )
//for(i=0; i<nowrites; i++)
{
if(error == PFE_EOF)
{
printf("output EOF reached\n");
pagenum = -1;
continue;
}
i = pagenum;
if ((error=PF_UnfixPage(fd,pagenum,FALSE))!= PFE_OK){
printf("outputerror - %d\n", error);
PF_PrintError("unfix");
exit(1);
}
int j = rand();
if(j%2 == 0) continue;
temp++;
//printf("output i : %d", i);
//int j = rand() % filesize;
updatefile(FILE2,stale_file2,i);
if(temp==nowrites/2)
{
printf("output Before garbage collection1:-\n");
printf("output reads - %d\n", reads);
printf("output erases(in units of blocks) - %d\n", erases/PAGEINBLOCK);
printf("output writes - %d\n", writes);
numpages(FILE2,stale_file2);
printf("output total number of pages - %d, stale pages is - %d\n", numtotal,numstale);
garbagecollect(FILE2,stale_file2);
printf("output After garbage collection1:-\n");
printf("output reads - %d\n", reads);
printf("output erases(in units of blocks) - %d\n", erases/PAGEINBLOCK);
printf("output writes - %d\n", writes);
numpages(FILE2,stale_file2);
printf("output total number of pages - %d, stale pages is - %d\n", numtotal,numstale);
}
}
if(error != PFE_OK)
{
PF_PrintError("output");
}
printf("output Before garbage collection2\n");
printf("output reads - %d\n", reads);
printf("output erases(in units of blocks) - %d\n", erases/PAGEINBLOCK);
printf("output writes - %d\n", writes);
numpages(FILE2,stale_file2);
printf("output total number of pages - %d, stale pages is - %d\n", numtotal,numstale);
garbagecollect(FILE2,stale_file2);
printf("output file size - %d\n", filesize);
printf("output Number of updates - %d, %d\n", temp, nowrites);
printf("output Number of pages in block - %d\n", PAGEINBLOCK);
printf("output After garbage collection2\n");
printf("output final reads - %d\n", reads);
printf("output final erases(in units of blocks) - %d\n", erases/PAGEINBLOCK);
printf("output final writes - %d\n", writes);
numpages(FILE2,stale_file2);
printf("output total number of pages - %d, stale pages is - %d\n", numtotal,numstale);
if ((error=PF_CloseFile(fd))!= PFE_OK){
PF_PrintError("close file");
exit(1);
}
readfile(FILE2, stale_file2);
//readfile(FILE1,stale_file1);
//garbagecollect(FILE1,stale_file1);
//readfile(FILE1,stale_file1);
}
external_sort() {
int i, j, k, error;
int * buf;
int run_size = 1, run_count = B_R;
int MEM = PF_MAX_BUFS - 1, PREVMEM = MEM;
int fd_inpfile;
if ((fd_inpfile=PF_OpenFile(TMPFILE))<0){
PF_PrintError("Error in opening file");
exit(1);
}
int fd_finalfile;
if ((fd_finalfile=PF_OpenFile(FILE))<0){
PF_PrintError("Error in opening file");
exit(1);
}
// We assume that inpfile contains all the runs and then we merge them and put them in finalfile
// when 1 merge pass is over, we again write back everything to the tmpfile
// repeat
while(run_count > 1) {
int pagenum_finalfile = -1;
//printf("Run count: %d\n", run_count);
//printf("Run Size: %d\n", run_size);
MEM = PREVMEM;
int number_of_merges = run_count / MEM;
if(run_count % MEM != 0)
number_of_merges++;
//printf("Number of merges: %d\n", number_of_merges);
for(i = 0 ; i < number_of_merges ; i++) {
//MEM is the number of runs we merge at once. If say MEM = 4 and run_count = 11, then in the last merge step, we merge just 3 runs. So we set MEM accordingly to 3
if(run_count % MEM != 0 && i == number_of_merges - 1)
MEM = run_count % MEM;
int be[MEM], en[MEM]; // be[j] stores the beginning of j-th run and en[j] = 1 + ending of j-th run. Idea is that when be[j] == en[j], then that run is finished
for(j = 1; j <= MEM; j++) {
be[j - 1] = i * PREVMEM * run_size + (j - 1) * run_size; //simple arithmetic
en[j - 1] = i * PREVMEM * run_size + j * run_size;
}
if(i == number_of_merges - 1) //Since the extremely last run can be shorter than other runs, we need to take care of it. This run will come into picture only in the very last merging iteration
en[MEM - 1] = en[MEM - 1] < B_R ? en[MEM - 1] : B_R; //the ending of the very last run cannot be more than B_R
//for(j = 0; j < MEM; j++)
//printf("be: %d, en: %d\n", be[j], en[j]);
int blocks_written = run_size * (MEM - 1) + (en[MEM - 1] - be[MEM - 1]); //in 1 merge of iteration, we count the number of blocks we are going to write. Again, take care of the last run separately
//printf("blocks written: %d\n", blocks_written);
int *arr[MEM]; // the buffers for the runs that will be merged
for(j = 0 ; j < MEM ; j++) {
//printf("Getting page: %d\n", be[j]);
error = PF_GetThisPage(fd_inpfile, be[j], &arr[j]); // get the first page of each run in the buffer
if(error != PFE_OK)
printf("Error in getting page\n");
//printf("Page value: %d\n", *arr[j]);
}
//start writing blocks one by one
for(k = 0 ; k < blocks_written ; k++) { // here goes the actual merge step
int least = (1 << 30), least_index = -1 ; // least_index -> the run whose head is the smallest. least -> the value of the corresponding head
// Note -> be[j] also acts as a pointer to the starting of each run. during the merging process, the value of be[j] will increment
// if be[j] == en[j], then that particular run is finished
//printf("Yo\n");
for(j = 0 ; j < MEM ; j++) { // this for loop finds the run whose head is the smallest and also the smallest value of the head
if(en[j] != be[j] && *arr[j] < least) {
least = *arr[j];
least_index = j;
}
//if(en[j] == be[j])
//printf("Ignoring\n");
}
//printf("Least: %d\n", least);
//get the next page of the finalfile where the least block is to be written
//note that pagenum_finalfile will automatically be incremented by 1 after the line below
//printf("Getting page number %d of finalfile\n", pagenum_finalfile + 1);
if((error=PF_GetNextPage(fd_finalfile,&pagenum_finalfile,&buf))!= PFE_OK) {
PF_PrintError("Can't get page");
exit(1);
}
//printf("Value of pagenum_finalfile now: %d\n", pagenum_finalfile);
//printf("Writing it by changing buffer\n");
*((int *)buf) = least; // assign the buffer the least value
//printf("unfixing the written page\n");
//unfix the written page
if ((error=PF_UnfixPage(fd_finalfile,pagenum_finalfile,TRUE))!= PFE_OK){
PF_PrintError("unfix buffer\n");
exit(1);
}
//printf("Unfixing the least page\n");
//unfix the least page as well
if ((error=PF_UnfixPage(fd_inpfile,be[least_index],TRUE))!= PFE_OK){
PF_PrintError("unfix buffer\n");
exit(1);
}
be[least_index]++;
if(be[least_index] < en[least_index]) { // if the run which had the *least* head isn't finished, get its next page
be[least_index]--;
if((error=PF_GetNextPage(fd_inpfile , &be[least_index] , &arr[least_index]))!= PFE_OK) {
PF_PrintError("Can't get page");
exit(1);
}
//printf("Fetching next page: %d\n", *arr[least_index]);
}
}
//printf("End of 1 merge iteration\n");
}
copyfiles();
//printf("\n");
run_count = number_of_merges;
run_size = run_size * PREVMEM;
//printf("End of iteration\n");
//readfile(FILE);
}
//swap files here
if ((error=PF_CloseFile(fd_inpfile))!= PFE_OK){
PF_PrintError("Error in closing input file\n");
exit(1);
}
if ((error=PF_CloseFile(fd_finalfile))!= PFE_OK){
PF_PrintError("Error in closing temporary file\n");
exit(1);
}
}
