split.c
上传用户:tsgydb
上传日期:2007-04-14
资源大小:10674k
文件大小:7k
- #include <stdio.h>
- #include <string.h>
- /*
- - split - divide a string into fields, like awk split()
- = int split(char *string, char *fields[], int nfields, char *sep);
- */
- int /* number of fields, including overflow */
- split(string, fields, nfields, sep)
- char *string;
- char *fields[]; /* list is not NULL-terminated */
- int nfields; /* number of entries available in fields[] */
- char *sep; /* "" white, "c" single char, "ab" [ab]+ */
- {
- register char *p = string;
- register char c; /* latest character */
- register char sepc = sep[0];
- register char sepc2;
- register int fn;
- register char **fp = fields;
- register char *sepp;
- register int trimtrail;
- /* white space */
- if (sepc == ' ') {
- while ((c = *p++) == ' ' || c == 't')
- continue;
- p--;
- trimtrail = 1;
- sep = " t"; /* note, code below knows this is 2 long */
- sepc = ' ';
- } else
- trimtrail = 0;
- sepc2 = sep[1]; /* now we can safely pick this up */
- /* catch empties */
- if (*p == ' ')
- return(0);
- /* single separator */
- if (sepc2 == ' ') {
- fn = nfields;
- for (;;) {
- *fp++ = p;
- fn--;
- if (fn == 0)
- break;
- while ((c = *p++) != sepc)
- if (c == ' ')
- return(nfields - fn);
- *(p-1) = ' ';
- }
- /* we have overflowed the fields vector -- just count them */
- fn = nfields;
- for (;;) {
- while ((c = *p++) != sepc)
- if (c == ' ')
- return(fn);
- fn++;
- }
- /* not reached */
- }
- /* two separators */
- if (sep[2] == ' ') {
- fn = nfields;
- for (;;) {
- *fp++ = p;
- fn--;
- while ((c = *p++) != sepc && c != sepc2)
- if (c == ' ') {
- if (trimtrail && **(fp-1) == ' ')
- fn++;
- return(nfields - fn);
- }
- if (fn == 0)
- break;
- *(p-1) = ' ';
- while ((c = *p++) == sepc || c == sepc2)
- continue;
- p--;
- }
- /* we have overflowed the fields vector -- just count them */
- fn = nfields;
- while (c != ' ') {
- while ((c = *p++) == sepc || c == sepc2)
- continue;
- p--;
- fn++;
- while ((c = *p++) != ' ' && c != sepc && c != sepc2)
- continue;
- }
- /* might have to trim trailing white space */
- if (trimtrail) {
- p--;
- while ((c = *--p) == sepc || c == sepc2)
- continue;
- p++;
- if (*p != ' ') {
- if (fn == nfields+1)
- *p = ' ';
- fn--;
- }
- }
- return(fn);
- }
- /* n separators */
- fn = 0;
- for (;;) {
- if (fn < nfields)
- *fp++ = p;
- fn++;
- for (;;) {
- c = *p++;
- if (c == ' ')
- return(fn);
- sepp = sep;
- while ((sepc = *sepp++) != ' ' && sepc != c)
- continue;
- if (sepc != ' ') /* it was a separator */
- break;
- }
- if (fn < nfields)
- *(p-1) = ' ';
- for (;;) {
- c = *p++;
- sepp = sep;
- while ((sepc = *sepp++) != ' ' && sepc != c)
- continue;
- if (sepc == ' ') /* it wasn't a separator */
- break;
- }
- p--;
- }
- /* not reached */
- }
- #ifdef TEST_SPLIT
- /*
- * test program
- * pgm runs regression
- * pgm sep splits stdin lines by sep
- * pgm str sep splits str by sep
- * pgm str sep n splits str by sep n times
- */
- int
- main(argc, argv)
- int argc;
- char *argv[];
- {
- char buf[512];
- register int n;
- # define MNF 10
- char *fields[MNF];
- if (argc > 4)
- for (n = atoi(argv[3]); n > 0; n--) {
- (void) strcpy(buf, argv[1]);
- }
- else if (argc > 3)
- for (n = atoi(argv[3]); n > 0; n--) {
- (void) strcpy(buf, argv[1]);
- (void) split(buf, fields, MNF, argv[2]);
- }
- else if (argc > 2)
- dosplit(argv[1], argv[2]);
- else if (argc > 1)
- while (fgets(buf, sizeof(buf), stdin) != NULL) {
- buf[strlen(buf)-1] = ' '; /* stomp newline */
- dosplit(buf, argv[1]);
- }
- else
- regress();
- exit(0);
- }
- dosplit(string, seps)
- char *string;
- char *seps;
- {
- # define NF 5
- char *fields[NF];
- register int nf;
- nf = split(string, fields, NF, seps);
- print(nf, NF, fields);
- }
- print(nf, nfp, fields)
- int nf;
- int nfp;
- char *fields[];
- {
- register int fn;
- register int bound;
- bound = (nf > nfp) ? nfp : nf;
- printf("%d:t", nf);
- for (fn = 0; fn < bound; fn++)
- printf(""%s"%s", fields[fn], (fn+1 < nf) ? ", " : "n");
- }
- #define RNF 5 /* some table entries know this */
- struct {
- char *str;
- char *seps;
- int nf;
- char *fi[RNF];
- } tests[] = {
- "", " ", 0, { "" },
- " ", " ", 2, { "", "" },
- "x", " ", 1, { "x" },
- "xy", " ", 1, { "xy" },
- "x y", " ", 2, { "x", "y" },
- "abc def g ", " ", 5, { "abc", "def", "", "g", "" },
- " a bcd", " ", 4, { "", "", "a", "bcd" },
- "a b c d e f", " ", 6, { "a", "b", "c", "d", "e f" },
- " a b c d ", " ", 6, { "", "a", "b", "c", "d " },
- "", " _", 0, { "" },
- " ", " _", 2, { "", "" },
- "x", " _", 1, { "x" },
- "x y", " _", 2, { "x", "y" },
- "ab _ cd", " _", 2, { "ab", "cd" },
- " a_b c ", " _", 5, { "", "a", "b", "c", "" },
- "a b c_d e f", " _", 6, { "a", "b", "c", "d", "e f" },
- " a b c d ", " _", 6, { "", "a", "b", "c", "d " },
- "", " _~", 0, { "" },
- " ", " _~", 2, { "", "" },
- "x", " _~", 1, { "x" },
- "x y", " _~", 2, { "x", "y" },
- "ab _~ cd", " _~", 2, { "ab", "cd" },
- " a_b c~", " _~", 5, { "", "a", "b", "c", "" },
- "a b_c d~e f", " _~", 6, { "a", "b", "c", "d", "e f" },
- "~a b c d ", " _~", 6, { "", "a", "b", "c", "d " },
- "", " _~-", 0, { "" },
- " ", " _~-", 2, { "", "" },
- "x", " _~-", 1, { "x" },
- "x y", " _~-", 2, { "x", "y" },
- "ab _~- cd", " _~-", 2, { "ab", "cd" },
- " a_b c~", " _~-", 5, { "", "a", "b", "c", "" },
- "a b_c-d~e f", " _~-", 6, { "a", "b", "c", "d", "e f" },
- "~a-b c d ", " _~-", 6, { "", "a", "b", "c", "d " },
- "", " ", 0, { "" },
- " ", " ", 2, { "", "" },
- "x", " ", 1, { "x" },
- "xy", " ", 1, { "xy" },
- "x y", " ", 2, { "x", "y" },
- "abc def g ", " ", 4, { "abc", "def", "g", "" },
- " a bcd", " ", 3, { "", "a", "bcd" },
- "a b c d e f", " ", 6, { "a", "b", "c", "d", "e f" },
- " a b c d ", " ", 6, { "", "a", "b", "c", "d " },
- "", "", 0, { "" },
- " ", "", 0, { "" },
- "x", "", 1, { "x" },
- "xy", "", 1, { "xy" },
- "x y", "", 2, { "x", "y" },
- "abc def g ", "", 3, { "abc", "def", "g" },
- "t a bcd", "", 2, { "a", "bcd" },
- " a tbt c ", "", 3, { "a", "b", "c" },
- "a b c d e ", "", 5, { "a", "b", "c", "d", "e" },
- "a btc d e f", "", 6, { "a", "b", "c", "d", "e f" },
- " a b c d e f ", "", 6, { "a", "b", "c", "d", "e f " },
- NULL, NULL, 0, { NULL },
- };
- regress()
- {
- char buf[512];
- register int n;
- char *fields[RNF+1];
- register int nf;
- register int i;
- register int printit;
- register char *f;
- for (n = 0; tests[n].str != NULL; n++) {
- (void) strcpy(buf, tests[n].str);
- fields[RNF] = NULL;
- nf = split(buf, fields, RNF, tests[n].seps);
- printit = 0;
- if (nf != tests[n].nf) {
- printf("split `%s' by `%s' gave %d fields, not %dn",
- tests[n].str, tests[n].seps, nf, tests[n].nf);
- printit = 1;
- } else if (fields[RNF] != NULL) {
- printf("split() went beyond array endn");
- printit = 1;
- } else {
- for (i = 0; i < nf && i < RNF; i++) {
- f = fields[i];
- if (f == NULL)
- f = "(NULL)";
- if (strcmp(f, tests[n].fi[i]) != 0) {
- printf("split `%s' by `%s', field %d is `%s', not `%s'n",
- tests[n].str, tests[n].seps,
- i, fields[i], tests[n].fi[i]);
- printit = 1;
- }
- }
- }
- if (printit)
- print(nf, RNF, fields);
- }
- }
- #endif