1
0
forked from aniani/vim

patch 8.2.3139: functions for string manipulation are spread out

Problem:    Functions for string manipulation are spread out.
Solution:   Move string related functions to a new source file. (Yegappan
            Lakshmanan, closes #8470)
This commit is contained in:
Yegappan Lakshmanan
2021-07-10 21:29:18 +02:00
committed by Bram Moolenaar
parent 31e21766d6
commit a2438132a6
21 changed files with 1673 additions and 1627 deletions

View File

@@ -29,14 +29,10 @@ static void f_balloon_split(typval_T *argvars, typval_T *rettv);
# endif
#endif
static void f_byte2line(typval_T *argvars, typval_T *rettv);
static void byteidx(typval_T *argvars, typval_T *rettv, int comp);
static void f_byteidx(typval_T *argvars, typval_T *rettv);
static void f_byteidxcomp(typval_T *argvars, typval_T *rettv);
static void f_call(typval_T *argvars, typval_T *rettv);
static void f_changenr(typval_T *argvars, typval_T *rettv);
static void f_char2nr(typval_T *argvars, typval_T *rettv);
static void f_charcol(typval_T *argvars, typval_T *rettv);
static void f_charidx(typval_T *argvars, typval_T *rettv);
static void f_col(typval_T *argvars, typval_T *rettv);
static void f_confirm(typval_T *argvars, typval_T *rettv);
static void f_copy(typval_T *argvars, typval_T *rettv);
@@ -83,7 +79,6 @@ static void f_hasmapto(typval_T *argvars, typval_T *rettv);
static void f_hlID(typval_T *argvars, typval_T *rettv);
static void f_hlexists(typval_T *argvars, typval_T *rettv);
static void f_hostname(typval_T *argvars, typval_T *rettv);
static void f_iconv(typval_T *argvars, typval_T *rettv);
static void f_index(typval_T *argvars, typval_T *rettv);
static void f_input(typval_T *argvars, typval_T *rettv);
static void f_inputdialog(typval_T *argvars, typval_T *rettv);
@@ -173,19 +168,6 @@ static void f_spellbadword(typval_T *argvars, typval_T *rettv);
static void f_spellsuggest(typval_T *argvars, typval_T *rettv);
static void f_split(typval_T *argvars, typval_T *rettv);
static void f_srand(typval_T *argvars, typval_T *rettv);
static void f_str2list(typval_T *argvars, typval_T *rettv);
static void f_str2nr(typval_T *argvars, typval_T *rettv);
static void f_strcharlen(typval_T *argvars, typval_T *rettv);
static void f_strchars(typval_T *argvars, typval_T *rettv);
static void f_strgetchar(typval_T *argvars, typval_T *rettv);
static void f_stridx(typval_T *argvars, typval_T *rettv);
static void f_strlen(typval_T *argvars, typval_T *rettv);
static void f_strcharpart(typval_T *argvars, typval_T *rettv);
static void f_strpart(typval_T *argvars, typval_T *rettv);
static void f_strridx(typval_T *argvars, typval_T *rettv);
static void f_strtrans(typval_T *argvars, typval_T *rettv);
static void f_strdisplaywidth(typval_T *argvars, typval_T *rettv);
static void f_strwidth(typval_T *argvars, typval_T *rettv);
static void f_submatch(typval_T *argvars, typval_T *rettv);
static void f_substitute(typval_T *argvars, typval_T *rettv);
static void f_swapinfo(typval_T *argvars, typval_T *rettv);
@@ -198,10 +180,6 @@ static void f_synconcealed(typval_T *argvars, typval_T *rettv);
static void f_tabpagebuflist(typval_T *argvars, typval_T *rettv);
static void f_taglist(typval_T *argvars, typval_T *rettv);
static void f_tagfiles(typval_T *argvars, typval_T *rettv);
static void f_tolower(typval_T *argvars, typval_T *rettv);
static void f_toupper(typval_T *argvars, typval_T *rettv);
static void f_tr(typval_T *argvars, typval_T *rettv);
static void f_trim(typval_T *argvars, typval_T *rettv);
static void f_type(typval_T *argvars, typval_T *rettv);
static void f_virtcol(typval_T *argvars, typval_T *rettv);
static void f_visualmode(typval_T *argvars, typval_T *rettv);
@@ -2377,50 +2355,6 @@ f_byte2line(typval_T *argvars UNUSED, typval_T *rettv)
#endif
}
static void
byteidx(typval_T *argvars, typval_T *rettv, int comp UNUSED)
{
char_u *t;
char_u *str;
varnumber_T idx;
str = tv_get_string_chk(&argvars[0]);
idx = tv_get_number_chk(&argvars[1], NULL);
rettv->vval.v_number = -1;
if (str == NULL || idx < 0)
return;
t = str;
for ( ; idx > 0; idx--)
{
if (*t == NUL) // EOL reached
return;
if (enc_utf8 && comp)
t += utf_ptr2len(t);
else
t += (*mb_ptr2len)(t);
}
rettv->vval.v_number = (varnumber_T)(t - str);
}
/*
* "byteidx()" function
*/
static void
f_byteidx(typval_T *argvars, typval_T *rettv)
{
byteidx(argvars, rettv, FALSE);
}
/*
* "byteidxcomp()" function
*/
static void
f_byteidxcomp(typval_T *argvars, typval_T *rettv)
{
byteidx(argvars, rettv, TRUE);
}
/*
* "call(func, arglist [, dict])" function
*/
@@ -2552,58 +2486,6 @@ f_charcol(typval_T *argvars, typval_T *rettv)
get_col(argvars, rettv, TRUE);
}
/*
* "charidx()" function
*/
static void
f_charidx(typval_T *argvars, typval_T *rettv)
{
char_u *str;
varnumber_T idx;
varnumber_T countcc = FALSE;
char_u *p;
int len;
int (*ptr2len)(char_u *);
rettv->vval.v_number = -1;
if (argvars[0].v_type != VAR_STRING || argvars[1].v_type != VAR_NUMBER
|| (argvars[2].v_type != VAR_UNKNOWN
&& argvars[2].v_type != VAR_NUMBER
&& argvars[2].v_type != VAR_BOOL))
{
emsg(_(e_invarg));
return;
}
str = tv_get_string_chk(&argvars[0]);
idx = tv_get_number_chk(&argvars[1], NULL);
if (str == NULL || idx < 0)
return;
if (argvars[2].v_type != VAR_UNKNOWN)
countcc = tv_get_bool(&argvars[2]);
if (countcc < 0 || countcc > 1)
{
semsg(_(e_using_number_as_bool_nr), countcc);
return;
}
if (enc_utf8 && countcc)
ptr2len = utf_ptr2len;
else
ptr2len = mb_ptr2len;
for (p = str, len = 0; p <= str + idx; len++)
{
if (*p == NUL)
return;
p += ptr2len(p);
}
rettv->vval.v_number = len > 0 ? len - 1 : 0;
}
win_T *
get_optional_window(typval_T *argvars, int idx)
{
@@ -5780,37 +5662,6 @@ f_hostname(typval_T *argvars UNUSED, typval_T *rettv)
rettv->vval.v_string = vim_strsave(hostname);
}
/*
* iconv() function
*/
static void
f_iconv(typval_T *argvars UNUSED, typval_T *rettv)
{
char_u buf1[NUMBUFLEN];
char_u buf2[NUMBUFLEN];
char_u *from, *to, *str;
vimconv_T vimconv;
rettv->v_type = VAR_STRING;
rettv->vval.v_string = NULL;
str = tv_get_string(&argvars[0]);
from = enc_canonize(enc_skip(tv_get_string_buf(&argvars[1], buf1)));
to = enc_canonize(enc_skip(tv_get_string_buf(&argvars[2], buf2)));
vimconv.vc_type = CONV_NONE;
convert_setup(&vimconv, from, to);
// If the encodings are equal, no conversion needed.
if (vimconv.vc_type == CONV_NONE)
rettv->vval.v_string = vim_strsave(str);
else
rettv->vval.v_string = string_convert(&vimconv, str, NULL);
convert_setup(&vimconv, NULL, NULL);
vim_free(from);
vim_free(to);
}
/*
* "index()" function
*/
@@ -8786,452 +8637,6 @@ theend:
p_cpo = save_cpo;
}
/*
* "str2list()" function
*/
static void
f_str2list(typval_T *argvars, typval_T *rettv)
{
char_u *p;
int utf8 = FALSE;
if (rettv_list_alloc(rettv) == FAIL)
return;
if (argvars[1].v_type != VAR_UNKNOWN)
utf8 = (int)tv_get_bool_chk(&argvars[1], NULL);
p = tv_get_string(&argvars[0]);
if (has_mbyte || utf8)
{
int (*ptr2len)(char_u *);
int (*ptr2char)(char_u *);
if (utf8 || enc_utf8)
{
ptr2len = utf_ptr2len;
ptr2char = utf_ptr2char;
}
else
{
ptr2len = mb_ptr2len;
ptr2char = mb_ptr2char;
}
for ( ; *p != NUL; p += (*ptr2len)(p))
list_append_number(rettv->vval.v_list, (*ptr2char)(p));
}
else
for ( ; *p != NUL; ++p)
list_append_number(rettv->vval.v_list, *p);
}
/*
* "str2nr()" function
*/
static void
f_str2nr(typval_T *argvars, typval_T *rettv)
{
int base = 10;
char_u *p;
varnumber_T n;
int what = 0;
int isneg;
if (argvars[1].v_type != VAR_UNKNOWN)
{
base = (int)tv_get_number(&argvars[1]);
if (base != 2 && base != 8 && base != 10 && base != 16)
{
emsg(_(e_invarg));
return;
}
if (argvars[2].v_type != VAR_UNKNOWN && tv_get_bool(&argvars[2]))
what |= STR2NR_QUOTE;
}
p = skipwhite(tv_get_string_strict(&argvars[0]));
isneg = (*p == '-');
if (*p == '+' || *p == '-')
p = skipwhite(p + 1);
switch (base)
{
case 2: what |= STR2NR_BIN + STR2NR_FORCE; break;
case 8: what |= STR2NR_OCT + STR2NR_OOCT + STR2NR_FORCE; break;
case 16: what |= STR2NR_HEX + STR2NR_FORCE; break;
}
vim_str2nr(p, NULL, NULL, what, &n, NULL, 0, FALSE);
// Text after the number is silently ignored.
if (isneg)
rettv->vval.v_number = -n;
else
rettv->vval.v_number = n;
}
/*
* "strgetchar()" function
*/
static void
f_strgetchar(typval_T *argvars, typval_T *rettv)
{
char_u *str;
int len;
int error = FALSE;
int charidx;
int byteidx = 0;
rettv->vval.v_number = -1;
str = tv_get_string_chk(&argvars[0]);
if (str == NULL)
return;
len = (int)STRLEN(str);
charidx = (int)tv_get_number_chk(&argvars[1], &error);
if (error)
return;
while (charidx >= 0 && byteidx < len)
{
if (charidx == 0)
{
rettv->vval.v_number = mb_ptr2char(str + byteidx);
break;
}
--charidx;
byteidx += MB_CPTR2LEN(str + byteidx);
}
}
/*
* "stridx()" function
*/
static void
f_stridx(typval_T *argvars, typval_T *rettv)
{
char_u buf[NUMBUFLEN];
char_u *needle;
char_u *haystack;
char_u *save_haystack;
char_u *pos;
int start_idx;
needle = tv_get_string_chk(&argvars[1]);
save_haystack = haystack = tv_get_string_buf_chk(&argvars[0], buf);
rettv->vval.v_number = -1;
if (needle == NULL || haystack == NULL)
return; // type error; errmsg already given
if (argvars[2].v_type != VAR_UNKNOWN)
{
int error = FALSE;
start_idx = (int)tv_get_number_chk(&argvars[2], &error);
if (error || start_idx >= (int)STRLEN(haystack))
return;
if (start_idx >= 0)
haystack += start_idx;
}
pos = (char_u *)strstr((char *)haystack, (char *)needle);
if (pos != NULL)
rettv->vval.v_number = (varnumber_T)(pos - save_haystack);
}
/*
* "string()" function
*/
void
f_string(typval_T *argvars, typval_T *rettv)
{
char_u *tofree;
char_u numbuf[NUMBUFLEN];
rettv->v_type = VAR_STRING;
rettv->vval.v_string = tv2string(&argvars[0], &tofree, numbuf,
get_copyID());
// Make a copy if we have a value but it's not in allocated memory.
if (rettv->vval.v_string != NULL && tofree == NULL)
rettv->vval.v_string = vim_strsave(rettv->vval.v_string);
}
/*
* "strlen()" function
*/
static void
f_strlen(typval_T *argvars, typval_T *rettv)
{
rettv->vval.v_number = (varnumber_T)(STRLEN(
tv_get_string(&argvars[0])));
}
static void
strchar_common(typval_T *argvars, typval_T *rettv, int skipcc)
{
char_u *s = tv_get_string(&argvars[0]);
varnumber_T len = 0;
int (*func_mb_ptr2char_adv)(char_u **pp);
func_mb_ptr2char_adv = skipcc ? mb_ptr2char_adv : mb_cptr2char_adv;
while (*s != NUL)
{
func_mb_ptr2char_adv(&s);
++len;
}
rettv->vval.v_number = len;
}
/*
* "strcharlen()" function
*/
static void
f_strcharlen(typval_T *argvars, typval_T *rettv)
{
strchar_common(argvars, rettv, TRUE);
}
/*
* "strchars()" function
*/
static void
f_strchars(typval_T *argvars, typval_T *rettv)
{
varnumber_T skipcc = FALSE;
if (argvars[1].v_type != VAR_UNKNOWN)
skipcc = tv_get_bool(&argvars[1]);
if (skipcc < 0 || skipcc > 1)
semsg(_(e_using_number_as_bool_nr), skipcc);
else
strchar_common(argvars, rettv, skipcc);
}
/*
* "strdisplaywidth()" function
*/
static void
f_strdisplaywidth(typval_T *argvars, typval_T *rettv)
{
char_u *s = tv_get_string(&argvars[0]);
int col = 0;
if (argvars[1].v_type != VAR_UNKNOWN)
col = (int)tv_get_number(&argvars[1]);
rettv->vval.v_number = (varnumber_T)(linetabsize_col(col, s) - col);
}
/*
* "strwidth()" function
*/
static void
f_strwidth(typval_T *argvars, typval_T *rettv)
{
char_u *s = tv_get_string_strict(&argvars[0]);
rettv->vval.v_number = (varnumber_T)(mb_string2cells(s, -1));
}
/*
* "strcharpart()" function
*/
static void
f_strcharpart(typval_T *argvars, typval_T *rettv)
{
char_u *p;
int nchar;
int nbyte = 0;
int charlen;
int skipcc = FALSE;
int len = 0;
int slen;
int error = FALSE;
p = tv_get_string(&argvars[0]);
slen = (int)STRLEN(p);
nchar = (int)tv_get_number_chk(&argvars[1], &error);
if (!error)
{
if (argvars[2].v_type != VAR_UNKNOWN
&& argvars[3].v_type != VAR_UNKNOWN)
{
skipcc = tv_get_bool(&argvars[3]);
if (skipcc < 0 || skipcc > 1)
{
semsg(_(e_using_number_as_bool_nr), skipcc);
return;
}
}
if (nchar > 0)
while (nchar > 0 && nbyte < slen)
{
if (skipcc)
nbyte += mb_ptr2len(p + nbyte);
else
nbyte += MB_CPTR2LEN(p + nbyte);
--nchar;
}
else
nbyte = nchar;
if (argvars[2].v_type != VAR_UNKNOWN)
{
charlen = (int)tv_get_number(&argvars[2]);
while (charlen > 0 && nbyte + len < slen)
{
int off = nbyte + len;
if (off < 0)
len += 1;
else
{
if (skipcc)
len += mb_ptr2len(p + off);
else
len += MB_CPTR2LEN(p + off);
}
--charlen;
}
}
else
len = slen - nbyte; // default: all bytes that are available.
}
/*
* Only return the overlap between the specified part and the actual
* string.
*/
if (nbyte < 0)
{
len += nbyte;
nbyte = 0;
}
else if (nbyte > slen)
nbyte = slen;
if (len < 0)
len = 0;
else if (nbyte + len > slen)
len = slen - nbyte;
rettv->v_type = VAR_STRING;
rettv->vval.v_string = vim_strnsave(p + nbyte, len);
}
/*
* "strpart()" function
*/
static void
f_strpart(typval_T *argvars, typval_T *rettv)
{
char_u *p;
int n;
int len;
int slen;
int error = FALSE;
p = tv_get_string(&argvars[0]);
slen = (int)STRLEN(p);
n = (int)tv_get_number_chk(&argvars[1], &error);
if (error)
len = 0;
else if (argvars[2].v_type != VAR_UNKNOWN)
len = (int)tv_get_number(&argvars[2]);
else
len = slen - n; // default len: all bytes that are available.
// Only return the overlap between the specified part and the actual
// string.
if (n < 0)
{
len += n;
n = 0;
}
else if (n > slen)
n = slen;
if (len < 0)
len = 0;
else if (n + len > slen)
len = slen - n;
if (argvars[2].v_type != VAR_UNKNOWN && argvars[3].v_type != VAR_UNKNOWN)
{
int off;
// length in characters
for (off = n; off < slen && len > 0; --len)
off += mb_ptr2len(p + off);
len = off - n;
}
rettv->v_type = VAR_STRING;
rettv->vval.v_string = vim_strnsave(p + n, len);
}
/*
* "strridx()" function
*/
static void
f_strridx(typval_T *argvars, typval_T *rettv)
{
char_u buf[NUMBUFLEN];
char_u *needle;
char_u *haystack;
char_u *rest;
char_u *lastmatch = NULL;
int haystack_len, end_idx;
needle = tv_get_string_chk(&argvars[1]);
haystack = tv_get_string_buf_chk(&argvars[0], buf);
rettv->vval.v_number = -1;
if (needle == NULL || haystack == NULL)
return; // type error; errmsg already given
haystack_len = (int)STRLEN(haystack);
if (argvars[2].v_type != VAR_UNKNOWN)
{
// Third argument: upper limit for index
end_idx = (int)tv_get_number_chk(&argvars[2], NULL);
if (end_idx < 0)
return; // can never find a match
}
else
end_idx = haystack_len;
if (*needle == NUL)
{
// Empty string matches past the end.
lastmatch = haystack + end_idx;
}
else
{
for (rest = haystack; *rest != '\0'; ++rest)
{
rest = (char_u *)strstr((char *)rest, (char *)needle);
if (rest == NULL || rest > haystack + end_idx)
break;
lastmatch = rest;
}
}
if (lastmatch == NULL)
rettv->vval.v_number = -1;
else
rettv->vval.v_number = (varnumber_T)(lastmatch - haystack);
}
/*
* "strtrans()" function
*/
static void
f_strtrans(typval_T *argvars, typval_T *rettv)
{
rettv->v_type = VAR_STRING;
rettv->vval.v_string = transstr(tv_get_string(&argvars[0]));
}
/*
* "submatch()" function
*/
@@ -9627,236 +9032,6 @@ f_taglist(typval_T *argvars, typval_T *rettv)
(void)get_tags(rettv->vval.v_list, tag_pattern, fname);
}
/*
* "tolower(string)" function
*/
static void
f_tolower(typval_T *argvars, typval_T *rettv)
{
rettv->v_type = VAR_STRING;
rettv->vval.v_string = strlow_save(tv_get_string(&argvars[0]));
}
/*
* "toupper(string)" function
*/
static void
f_toupper(typval_T *argvars, typval_T *rettv)
{
rettv->v_type = VAR_STRING;
rettv->vval.v_string = strup_save(tv_get_string(&argvars[0]));
}
/*
* "tr(string, fromstr, tostr)" function
*/
static void
f_tr(typval_T *argvars, typval_T *rettv)
{
char_u *in_str;
char_u *fromstr;
char_u *tostr;
char_u *p;
int inlen;
int fromlen;
int tolen;
int idx;
char_u *cpstr;
int cplen;
int first = TRUE;
char_u buf[NUMBUFLEN];
char_u buf2[NUMBUFLEN];
garray_T ga;
in_str = tv_get_string(&argvars[0]);
fromstr = tv_get_string_buf_chk(&argvars[1], buf);
tostr = tv_get_string_buf_chk(&argvars[2], buf2);
// Default return value: empty string.
rettv->v_type = VAR_STRING;
rettv->vval.v_string = NULL;
if (fromstr == NULL || tostr == NULL)
return; // type error; errmsg already given
ga_init2(&ga, (int)sizeof(char), 80);
if (!has_mbyte)
// not multi-byte: fromstr and tostr must be the same length
if (STRLEN(fromstr) != STRLEN(tostr))
{
error:
semsg(_(e_invarg2), fromstr);
ga_clear(&ga);
return;
}
// fromstr and tostr have to contain the same number of chars
while (*in_str != NUL)
{
if (has_mbyte)
{
inlen = (*mb_ptr2len)(in_str);
cpstr = in_str;
cplen = inlen;
idx = 0;
for (p = fromstr; *p != NUL; p += fromlen)
{
fromlen = (*mb_ptr2len)(p);
if (fromlen == inlen && STRNCMP(in_str, p, inlen) == 0)
{
for (p = tostr; *p != NUL; p += tolen)
{
tolen = (*mb_ptr2len)(p);
if (idx-- == 0)
{
cplen = tolen;
cpstr = p;
break;
}
}
if (*p == NUL) // tostr is shorter than fromstr
goto error;
break;
}
++idx;
}
if (first && cpstr == in_str)
{
// Check that fromstr and tostr have the same number of
// (multi-byte) characters. Done only once when a character
// of in_str doesn't appear in fromstr.
first = FALSE;
for (p = tostr; *p != NUL; p += tolen)
{
tolen = (*mb_ptr2len)(p);
--idx;
}
if (idx != 0)
goto error;
}
(void)ga_grow(&ga, cplen);
mch_memmove((char *)ga.ga_data + ga.ga_len, cpstr, (size_t)cplen);
ga.ga_len += cplen;
in_str += inlen;
}
else
{
// When not using multi-byte chars we can do it faster.
p = vim_strchr(fromstr, *in_str);
if (p != NULL)
ga_append(&ga, tostr[p - fromstr]);
else
ga_append(&ga, *in_str);
++in_str;
}
}
// add a terminating NUL
(void)ga_grow(&ga, 1);
ga_append(&ga, NUL);
rettv->vval.v_string = ga.ga_data;
}
/*
* "trim({expr})" function
*/
static void
f_trim(typval_T *argvars, typval_T *rettv)
{
char_u buf1[NUMBUFLEN];
char_u buf2[NUMBUFLEN];
char_u *head = tv_get_string_buf_chk(&argvars[0], buf1);
char_u *mask = NULL;
char_u *tail;
char_u *prev;
char_u *p;
int c1;
int dir = 0;
rettv->v_type = VAR_STRING;
rettv->vval.v_string = NULL;
if (head == NULL)
return;
if (argvars[1].v_type != VAR_UNKNOWN && argvars[1].v_type != VAR_STRING)
{
semsg(_(e_invarg2), tv_get_string(&argvars[1]));
return;
}
if (argvars[1].v_type == VAR_STRING)
{
mask = tv_get_string_buf_chk(&argvars[1], buf2);
if (argvars[2].v_type != VAR_UNKNOWN)
{
int error = 0;
// leading or trailing characters to trim
dir = (int)tv_get_number_chk(&argvars[2], &error);
if (error)
return;
if (dir < 0 || dir > 2)
{
semsg(_(e_invarg2), tv_get_string(&argvars[2]));
return;
}
}
}
if (dir == 0 || dir == 1)
{
// Trim leading characters
while (*head != NUL)
{
c1 = PTR2CHAR(head);
if (mask == NULL)
{
if (c1 > ' ' && c1 != 0xa0)
break;
}
else
{
for (p = mask; *p != NUL; MB_PTR_ADV(p))
if (c1 == PTR2CHAR(p))
break;
if (*p == NUL)
break;
}
MB_PTR_ADV(head);
}
}
tail = head + STRLEN(head);
if (dir == 0 || dir == 2)
{
// Trim trailing characters
for (; tail > head; tail = prev)
{
prev = tail;
MB_PTR_BACK(head, prev);
c1 = PTR2CHAR(prev);
if (mask == NULL)
{
if (c1 > ' ' && c1 != 0xa0)
break;
}
else
{
for (p = mask; *p != NUL; MB_PTR_ADV(p))
if (c1 == PTR2CHAR(p))
break;
if (*p == NUL)
break;
}
}
}
rettv->vval.v_string = vim_strnsave(head, tail - head);
}
/*
* "type(expr)" function
*/