summaryrefslogtreecommitdiff
path: root/do/split
diff options
context:
space:
mode:
Diffstat (limited to 'do/split')
-rw-r--r--do/split235
1 files changed, 235 insertions, 0 deletions
diff --git a/do/split b/do/split
new file mode 100644
index 0000000000..904d29ae72
--- /dev/null
+++ b/do/split
@@ -0,0 +1,235 @@
+int
+do_split(TARG,spat,limit,gimme,arglast)
+STR *TARG;
+register SPAT *spat;
+register int limit;
+int gimme;
+int *arglast;
+{
+ register ARRAY *ary = stack;
+ STR **st = ary->ary_array;
+ register int sp = arglast[0] + 1;
+ register char *s = str_get(st[sp]);
+ char *strend = s + st[sp--]->str_cur;
+ register STR *dstr;
+ register char *m;
+ int iters = 0;
+ int maxiters = (strend - s) + 10;
+ int i;
+ char *orig;
+ int origlimit = limit;
+ int realarray = 0;
+
+ if (!spat || !s)
+ fatal("panic: do_split");
+ else if (spat->spat_runtime) {
+ nointrp = "|)";
+ sp = eval(spat->spat_runtime,G_SCALAR,sp);
+ st = stack->ary_array;
+ m = str_get(dstr = st[sp--]);
+ nointrp = "";
+ if (*m == ' ' && dstr->str_cur == 1) {
+ str_set(dstr,"\\s+");
+ m = dstr->str_ptr;
+ spat->spat_flags |= SPAT_SKIPWHITE;
+ }
+ if (spat->spat_regexp) {
+ regfree(spat->spat_regexp);
+ spat->spat_regexp = Null(REGEXP*); /* avoid possible double free */
+ }
+ spat->spat_regexp = regcomp(m,m+dstr->str_cur,
+ spat->spat_flags & SPAT_FOLD);
+ if (spat->spat_flags & SPAT_KEEP ||
+ (spat->spat_runtime->arg_type == O_ITEM &&
+ (spat->spat_runtime[1].arg_type & A_MASK) == A_SINGLE) ) {
+ arg_free(spat->spat_runtime); /* it won't change, so */
+ spat->spat_runtime = Nullarg; /* no point compiling again */
+ }
+ }
+#ifdef DEBUGGING
+ if (debug & 8) {
+ deb("2.SPAT /%s/\n",spat->spat_regexp->precomp);
+ }
+#endif
+ ary = stab_xarray(spat->spat_repl[1].arg_ptr.arg_stab);
+ if (ary && (gimme != G_ARRAY || (spat->spat_flags & SPAT_ONCE))) {
+ realarray = 1;
+ if (!(ary->ary_flags & ARF_REAL)) {
+ ary->ary_flags |= ARF_REAL;
+ for (i = ary->ary_fill; i >= 0; i--)
+ ary->ary_array[i] = Nullstr; /* don't free mere refs */
+ }
+ ary->ary_fill = -1;
+ sp = -1; /* temporarily switch stacks */
+ }
+ else
+ ary = stack;
+ orig = s;
+ if (spat->spat_flags & SPAT_SKIPWHITE) {
+ while (isSPACE(*s))
+ s++;
+ }
+ if (!limit)
+ limit = maxiters + 2;
+ if (strEQ("\\s+",spat->spat_regexp->precomp)) {
+ while (--limit) {
+ /*SUPPRESS 530*/
+ for (m = s; m < strend && !isSPACE(*m); m++) ;
+ if (m >= strend)
+ break;
+ dstr = Str_new(30,m-s);
+ str_nset(dstr,s,m-s);
+ if (!realarray)
+ str_2mortal(dstr);
+ (void)astore(ary, ++sp, dstr);
+ /*SUPPRESS 530*/
+ for (s = m + 1; s < strend && isSPACE(*s); s++) ;
+ }
+ }
+ else if (strEQ("^",spat->spat_regexp->precomp)) {
+ while (--limit) {
+ /*SUPPRESS 530*/
+ for (m = s; m < strend && *m != '\n'; m++) ;
+ m++;
+ if (m >= strend)
+ break;
+ dstr = Str_new(30,m-s);
+ str_nset(dstr,s,m-s);
+ if (!realarray)
+ str_2mortal(dstr);
+ (void)astore(ary, ++sp, dstr);
+ s = m;
+ }
+ }
+ else if (spat->spat_short) {
+ i = spat->spat_short->str_cur;
+ if (i == 1) {
+ int fold = (spat->spat_flags & SPAT_FOLD);
+
+ i = *spat->spat_short->str_ptr;
+ if (fold && isUPPER(i))
+ i = tolower(i);
+ while (--limit) {
+ if (fold) {
+ for ( m = s;
+ m < strend && *m != i &&
+ (!isUPPER(*m) || tolower(*m) != i);
+ m++) /*SUPPRESS 530*/
+ ;
+ }
+ else /*SUPPRESS 530*/
+ for (m = s; m < strend && *m != i; m++) ;
+ if (m >= strend)
+ break;
+ dstr = Str_new(30,m-s);
+ str_nset(dstr,s,m-s);
+ if (!realarray)
+ str_2mortal(dstr);
+ (void)astore(ary, ++sp, dstr);
+ s = m + 1;
+ }
+ }
+ else {
+#ifndef lint
+ while (s < strend && --limit &&
+ (m=fbminstr((unsigned char*)s, (unsigned char*)strend,
+ spat->spat_short)) )
+#endif
+ {
+ dstr = Str_new(31,m-s);
+ str_nset(dstr,s,m-s);
+ if (!realarray)
+ str_2mortal(dstr);
+ (void)astore(ary, ++sp, dstr);
+ s = m + i;
+ }
+ }
+ }
+ else {
+ maxiters += (strend - s) * spat->spat_regexp->nparens;
+ while (s < strend && --limit &&
+ regexec(spat->spat_regexp, s, strend, orig, 1, Nullstr, TRUE) ) {
+ if (spat->spat_regexp->subbase
+ && spat->spat_regexp->subbase != orig) {
+ m = s;
+ s = orig;
+ orig = spat->spat_regexp->subbase;
+ s = orig + (m - s);
+ strend = s + (strend - m);
+ }
+ m = spat->spat_regexp->startp[0];
+ dstr = Str_new(32,m-s);
+ str_nset(dstr,s,m-s);
+ if (!realarray)
+ str_2mortal(dstr);
+ (void)astore(ary, ++sp, dstr);
+ if (spat->spat_regexp->nparens) {
+ for (i = 1; i <= spat->spat_regexp->nparens; i++) {
+ s = spat->spat_regexp->startp[i];
+ m = spat->spat_regexp->endp[i];
+ dstr = Str_new(33,m-s);
+ str_nset(dstr,s,m-s);
+ if (!realarray)
+ str_2mortal(dstr);
+ (void)astore(ary, ++sp, dstr);
+ }
+ }
+ s = spat->spat_regexp->endp[0];
+ }
+ }
+ if (realarray)
+ iters = sp + 1;
+ else
+ iters = sp - arglast[0];
+ if (iters > maxiters)
+ fatal("Split loop");
+ if (s < strend || origlimit) { /* keep field after final delim? */
+ dstr = Str_new(34,strend-s);
+ str_nset(dstr,s,strend-s);
+ if (!realarray)
+ str_2mortal(dstr);
+ (void)astore(ary, ++sp, dstr);
+ iters++;
+ }
+ else {
+#ifndef I286x
+ while (iters > 0 && ary->ary_array[sp]->str_cur == 0)
+ iters--,sp--;
+#else
+ char *zaps;
+ int zapb;
+
+ if (iters > 0) {
+ zaps = str_get(afetch(ary,sp,FALSE));
+ zapb = (int) *zaps;
+ }
+
+ while (iters > 0 && (!zapb)) {
+ iters--,sp--;
+ if (iters > 0) {
+ zaps = str_get(afetch(ary,iters-1,FALSE));
+ zapb = (int) *zaps;
+ }
+ }
+#endif
+ }
+ if (realarray) {
+ ary->ary_fill = sp;
+ if (gimme == G_ARRAY) {
+ sp++;
+ astore(stack, arglast[0] + 1 + sp, Nullstr);
+ Copy(ary->ary_array, stack->ary_array + arglast[0] + 1, sp, STR*);
+ return arglast[0] + sp;
+ }
+ }
+ else {
+ if (gimme == G_ARRAY)
+ return sp;
+ }
+ sp = arglast[0] + 1;
+ str_numset(TARG,(double)iters);
+ STABSET(TARG);
+ st[sp] = TARG;
+ return sp;
+}
+