|
|
|
@ -3,14 +3,42 @@
|
|
|
|
|
#include <string.h>
|
|
|
|
|
#include <stdint.h>
|
|
|
|
|
|
|
|
|
|
#if TIME_SECTIONS
|
|
|
|
|
#include <time.h>
|
|
|
|
|
|
|
|
|
|
clock_t start, end;
|
|
|
|
|
double section_time, total_time;
|
|
|
|
|
char * curr_section;
|
|
|
|
|
|
|
|
|
|
#define START_TIME(section_name) do {\
|
|
|
|
|
curr_section = section_name;\
|
|
|
|
|
printf("BEGIN %s\n", curr_section);\
|
|
|
|
|
start = clock();\
|
|
|
|
|
} while (0);
|
|
|
|
|
|
|
|
|
|
#define END_TIME do {\
|
|
|
|
|
end = clock();\
|
|
|
|
|
section_time = ((double) (end - start)) / CLOCKS_PER_SEC;\
|
|
|
|
|
printf("END %s: %fs\n\n", curr_section, section_time);\
|
|
|
|
|
total_time += section_time;\
|
|
|
|
|
} while (0);
|
|
|
|
|
|
|
|
|
|
#else
|
|
|
|
|
// TIME_SECTIONS not specified, define to be nothing
|
|
|
|
|
#define START_TIME(section_name) 0;
|
|
|
|
|
#define END_TIME 0;
|
|
|
|
|
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#define WORDS 12971
|
|
|
|
|
#define WORDLEN 5
|
|
|
|
|
#define CHARMASK ~(0b11u << 5u)
|
|
|
|
|
|
|
|
|
|
typedef struct {
|
|
|
|
|
typedef struct signature {
|
|
|
|
|
char * word;
|
|
|
|
|
uint32_t sign;
|
|
|
|
|
} signature;
|
|
|
|
|
|
|
|
|
|
uint32_t
|
|
|
|
|
rawWC = 0,
|
|
|
|
|
fiveUniqWC = 0,
|
|
|
|
@ -18,10 +46,10 @@ uint32_t
|
|
|
|
|
;
|
|
|
|
|
|
|
|
|
|
int compare(const void * a, const void * b) {
|
|
|
|
|
return (((signature *)a)->sign) - (((signature *)b)->sign);
|
|
|
|
|
return ((signature *)a)->sign - ((signature *)b)->sign;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void * rmdups(signature * arr) {
|
|
|
|
|
signature* rmdups(signature * arr) {
|
|
|
|
|
// list is already sorted
|
|
|
|
|
uint32_t nitems = fiveUniqWC;
|
|
|
|
|
for (uint32_t i = 0; i < nitems; /* handled in loop */) {
|
|
|
|
@ -35,7 +63,6 @@ void * rmdups(signature * arr) {
|
|
|
|
|
}
|
|
|
|
|
cookedWC = nitems;
|
|
|
|
|
return arr;
|
|
|
|
|
// return realloc(arr, cookedWC * sizeof(signature));
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
uint32_t getsig(char * word) {
|
|
|
|
@ -51,52 +78,71 @@ uint32_t getsig(char * word) {
|
|
|
|
|
return result;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void * makeArray(FILE * f) {
|
|
|
|
|
signature* makeArray(FILE * f) {
|
|
|
|
|
|
|
|
|
|
signature* sigs = (signature*) calloc(WORDS, sizeof(signature));
|
|
|
|
|
|
|
|
|
|
char word[WORDLEN + sizeof(char)]; // 5 letters + \n
|
|
|
|
|
uint32_t charsRead;
|
|
|
|
|
|
|
|
|
|
while (fread(word, sizeof(char), WORDLEN + sizeof(char), f) != 0) {
|
|
|
|
|
rawWC++;
|
|
|
|
|
// only assign sign so far, more efficient
|
|
|
|
|
uint32_t s = getsig(word);
|
|
|
|
|
if (s != 0) {
|
|
|
|
|
// no duplicate word letters in word
|
|
|
|
|
sigs[fiveUniqWC].word = strndup(word, 5);
|
|
|
|
|
sigs[fiveUniqWC].sign = s;
|
|
|
|
|
fiveUniqWC++;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// return realloc(sigs, fiveUniqWC * sizeof(signature));
|
|
|
|
|
return sigs;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int main(int argc, char * argv[]) {
|
|
|
|
|
#if TIME_SECTIONS
|
|
|
|
|
clock_t prog_start = clock();
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#if ! ASSUME_SAFE
|
|
|
|
|
if (argc != 2) {
|
|
|
|
|
fprintf(stderr, "usage: %s file\n", argv[0]);
|
|
|
|
|
return 2;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
FILE * f = fopen(argv[1], "r");
|
|
|
|
|
|
|
|
|
|
#if ! ASSUME_SAFE
|
|
|
|
|
if (f == NULL) {
|
|
|
|
|
fprintf(stderr, "%s: %s: no such file or directory\n", argv[0], argv[1]);
|
|
|
|
|
return 2;
|
|
|
|
|
}
|
|
|
|
|
signature* sigs = (signature*) makeArray(f);
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
START_TIME("making array")
|
|
|
|
|
signature* sigs = makeArray(f);
|
|
|
|
|
END_TIME
|
|
|
|
|
|
|
|
|
|
fclose(f);
|
|
|
|
|
|
|
|
|
|
#if PRINT_INTERMEDIATES
|
|
|
|
|
printf("read %u words\n", rawWC);
|
|
|
|
|
printf("kept %u words with 5 unique letters\n", fiveUniqWC);
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
START_TIME("sorting")
|
|
|
|
|
qsort(sigs, fiveUniqWC, sizeof(signature), compare);
|
|
|
|
|
sigs = (signature*) rmdups(sigs);
|
|
|
|
|
END_TIME
|
|
|
|
|
|
|
|
|
|
START_TIME("removing duplicates")
|
|
|
|
|
sigs = rmdups(sigs);
|
|
|
|
|
END_TIME
|
|
|
|
|
|
|
|
|
|
#if PRINT_INTERMEDIATES
|
|
|
|
|
printf("ended with %u totally unique words\n", cookedWC);
|
|
|
|
|
#if 0
|
|
|
|
|
for (uint32_t i = 0; i < cookedWC; i++) {
|
|
|
|
|
printf("%u\n", sigs[i].sign);
|
|
|
|
|
}
|
|
|
|
|
#else
|
|
|
|
|
#endif
|
|
|
|
|
START_TIME("finding words")
|
|
|
|
|
for (uint32_t i1 = 0; i1 < cookedWC; i1++) {
|
|
|
|
|
uint32_t a = sigs[i1].sign;
|
|
|
|
|
|
|
|
|
@ -106,7 +152,7 @@ int main(int argc, char * argv[]) {
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
uint32_t ab = a | b;
|
|
|
|
|
#if print
|
|
|
|
|
#if PRINT_INTERMEDIATES
|
|
|
|
|
fprintf(stderr, "%s:%s\n", sigs[i1].word, sigs[i2].word);
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
@ -116,7 +162,7 @@ int main(int argc, char * argv[]) {
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
uint32_t abc = ab | c;
|
|
|
|
|
#if print
|
|
|
|
|
#if PRINT_INTERMEDIATES
|
|
|
|
|
fprintf(stderr, "%s:%s:%s\n", sigs[i1].word, sigs[i2].word, sigs[i3].word);
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
@ -126,7 +172,7 @@ int main(int argc, char * argv[]) {
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
uint32_t abcd = abc | d;
|
|
|
|
|
#if print
|
|
|
|
|
#if PRINT_INTERMEDIATES
|
|
|
|
|
fprintf(stderr, "%s:%s:%s:%s\n", sigs[i1].word, sigs[i2].word, sigs[i3].word, sigs[i4].word);
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
@ -136,13 +182,22 @@ int main(int argc, char * argv[]) {
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
fprintf(stderr, "%s, %s, %s, %s, %s\n", sigs[i1].word, sigs[i2].word, sigs[i3].word, sigs[i4].word, sigs[i5].word);
|
|
|
|
|
// goto loopend;
|
|
|
|
|
#if FIND_ONE
|
|
|
|
|
goto loopend;
|
|
|
|
|
#endif
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
// loopend:
|
|
|
|
|
#if FIND_ONE
|
|
|
|
|
loopend:
|
|
|
|
|
#endif
|
|
|
|
|
END_TIME
|
|
|
|
|
|
|
|
|
|
#if TIME_SECTIONS
|
|
|
|
|
clock_t prog_end = clock();
|
|
|
|
|
printf("%s took %fs total\n", argv[0], ((double)(prog_end - prog_start)) / CLOCKS_PER_SEC);
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|