FINAL cregex update for now: optimize/change callback mfun API. Also, cstr_printf() cannot take self as print argument.

4 files changed, 26 insertions, 29 deletions

diff --git a/examples/regex_replace.c b/examples/regex_replace.c
index 0b6c9a1e..2bb9517f 100644
--- a/examples/regex_replace.c
+++ b/examples/regex_replace.c
@@ -2,13 +2,14 @@
 #include <stc/cstr.h>
 #include <stc/cregex.h>
 
-cstr sub_20y(int i, csview m) {
-    if (i == 1) { // year
+bool add_10_years(int i, csview m, cstr* mstr) {
+    if (i == 1) { // group 1 matches year
         int year;
         sscanf(m.str, "%4d", &year);
-        return cstr_from_fmt("%04d", year - 20);
+        cstr_printf(mstr, "%04d", year + 10);
+        return true;
     }
-    return cstr_from_sv(m);
+    return false;
 }
 
 int main()
@@ -18,31 +19,31 @@ int main()
 
     c_auto (cstr, str)
     {
-        printf("input: %s\n\n", input);
+        printf("INPUT: %s\n", input);
 
         /* replace with a fixed string, extended all-in-one call: */
         cstr_take(&str, cregex_replace_p(input, pattern, "YYYY-MM-DD"));
         printf("fixed: %s\n", cstr_str(&str));
 
-        /* US date format, and subtract 20 years: */
-        cstr_take(&str, cregex_replace_pe(input, pattern, "\\1/\\3/\\2", sub_20y, 0, 0));
-        printf("us-20: %s\n", cstr_str(&str));
+        /* US date format, and add 10 years to dates: */
+        cstr_take(&str, cregex_replace_pe(input, pattern, "\\1/\\3/\\2", add_10_years, 0, 0));
+        printf("us+10: %s\n", cstr_str(&str));
 
         /* Wrap first date inside []: */
         cstr_take(&str, cregex_replace_pe(input, pattern, "[\\0]", NULL, 1, 0));
         printf("brack: %s\n", cstr_str(&str));
 
-        /* Wrap all words in {} */
-        cstr_take(&str, cregex_replace_p("[52] apples and [31] mangoes", "[a-z]+", "{\\0}"));
-        printf("curly: %s\n", cstr_str(&str));
-
-        /* European date format, compile RE separately */
+        /* European date format. Show how to compile RE separately */
         cregex re = cregex_from(pattern, 0);
         if (cregex_captures(&re) == 0)
               continue;
         cstr_take(&str, cregex_replace(input, &re, "\\3.\\2.\\1", NULL, 0));
         cregex_drop(&re);
         printf("euros: %s\n", cstr_str(&str));
+
+        /* Wrap all words in {} */
+        cstr_take(&str, cregex_replace_p("[52] apples and [31] mangoes", "[a-z]+", "{\\0}"));
+        printf("curly: %s\n", cstr_str(&str));
     }
 }
 
diff --git a/include/stc/cregex.h b/include/stc/cregex.h
index f93a0c03..fff8ccb1 100644
--- a/include/stc/cregex.h
+++ b/include/stc/cregex.h
@@ -98,11 +98,11 @@ int cregex_match_p(const char* input, const char* pattern,
 
 /* replace regular expression */ 
 cstr cregex_replace(const char* input, const cregex* re, const char* replace,
-                    cstr (*mfun)(int i, csview match), unsigned count);
+                    bool (*mfun)(int i, csview match, cstr* mstr), unsigned count);
 
-/* replace + compile RE pattern and extra arguments */
+/* replace + compile RE pattern, and extra arguments */
 cstr cregex_replace_pe(const char* input, const char* pattern, const char* replace,
-                       cstr (*mfun)(int i, csview match), unsigned count, int cflags);
+                       bool (*mfun)(int i, csview match, cstr* mstr), unsigned count, int cflags);
 static inline
 cstr cregex_replace_p(const char* input, const char* pattern, const char* replace)
     { return cregex_replace_pe(input, pattern, replace, NULL, 0, 0); }
diff --git a/include/stc/cstr.h b/include/stc/cstr.h
index 8395f127..6db5ef49 100644
--- a/include/stc/cstr.h
+++ b/include/stc/cstr.h
@@ -511,8 +511,7 @@ STC_DEF int cstr_vfmt(cstr* self, const char* fmt, va_list args) {
     va_list args2;
     va_copy(args2, args);
     const int n = vsnprintf(NULL, (size_t)0, fmt, args);
-    cstr_reserve(self, n);
-    vsprintf(cstr_data(self), fmt, args2);
+    vsprintf(cstr_reserve(self, n), fmt, args2);
     va_end(args2);
     _cstr_set_size(self, n);
     return n;
@@ -532,13 +531,12 @@ STC_DEF cstr cstr_from_fmt(const char* fmt, ...) {
     return s;
 }
 
+/* NB! self-data in args is UB */
 STC_DEF int cstr_printf(cstr* self, const char* fmt, ...) {
-    cstr s = cstr_null;
     va_list args;
     va_start(args, fmt);
-    const int n = cstr_vfmt(&s, fmt, args);
+    const int n = cstr_vfmt(self, fmt, args);
     va_end(args);
-    cstr_drop(self); *self = s;
     return n;
 }
 
diff --git a/src/cregex.c b/src/cregex.c
index 8705b335..02e3c3d8 100644
--- a/src/cregex.c
+++ b/src/cregex.c
@@ -1157,10 +1157,11 @@ regexec(const Reprog *progp,    /* program to run */
 
 static void
 build_subst_string(const char* replace, unsigned nmatch, const csview match[],
-                   cstr (*mfun)(int i, csview match), cstr* subst) {
+                   bool (*mfun)(int i, csview match, cstr* mstr), cstr* subst) {
     cstr_clear(subst);
     unsigned len = 0, cap = cstr_capacity(*subst);
     char* dst = cstr_data(subst);
+    cstr mstr = cstr_null;
 
     while (*replace != '\0') {
         if (*replace == '\\') {
@@ -1171,15 +1172,11 @@ build_subst_string(const char* replace, unsigned nmatch, const csview match[],
             case '5': case '6': case '7': case '8': case '9':
                 i = num - '0';
                 if (i < nmatch) {
-                    csview m;
-                    cstr mstr = cstr_null;
-                    if (mfun) { mstr = mfun(i, match[i]); m = cstr_sv(&mstr); }
-                    else m = match[i];
+                    csview m = mfun && mfun(i, match[i], &mstr) ? cstr_sv(&mstr) : match[i];
                     if (len + m.size >= cap)
                         dst = cstr_reserve(subst, cap = cap*3/2 + m.size);
                     for (const char* rp = m.str; rp != (m.str + m.size); ++rp)
                         dst[len++] = *rp;
-                    cstr_drop(&mstr);
                 }
                 ++replace;
             case '\0':
@@ -1190,6 +1187,7 @@ build_subst_string(const char* replace, unsigned nmatch, const csview match[],
             dst = cstr_reserve(subst, cap = cap*3/2 + 4);
         dst[len++] = *replace++;
     }
+    cstr_drop(&mstr);
     _cstr_set_size(subst, len);
 }
 
@@ -1233,7 +1231,7 @@ int cregex_match_p(const char* input, const char* pattern,
 
 cstr
 cregex_replace(const char* input, const cregex* re, const char* replace,
-               cstr (*mfun)(int i, csview match), unsigned count) {
+               bool (*mfun)(int i, csview match, cstr* mstr), unsigned count) {
     cstr out = cstr_null;
     cstr subst = cstr_null;
     size_t from = 0;
@@ -1255,7 +1253,7 @@ cregex_replace(const char* input, const cregex* re, const char* replace,
 
 cstr
 cregex_replace_pe(const char* input, const char* pattern, const char* replace,
-                  cstr (*mfun)(int i, csview match), unsigned count, int cflags) {
+                  bool (*mfun)(int i, csview match, cstr* mstr), unsigned count, int cflags) {
     cregex re = cregex_init();
     int res = cregex_compile(&re, pattern, cflags);
     if (res < 0)