commit 8586e44

Some optimizations.

 1@@ -208,6 +208,7 @@ static xs_val *_xs_json_load_lexer(FILE *f, js_type *t)
 2 {
 3     int c;
 4     xs_val *v = NULL;
 5+    int offset;
 6 
 7     *t = JS_ERROR;
 8 
 9@@ -236,6 +237,7 @@ static xs_val *_xs_json_load_lexer(FILE *f, js_type *t)
10         *t = JS_STRING;
11 
12         v = xs_str_new(NULL);
13+        offset = 0;
14 
15         while ((c = fgetc(f)) != '"' && c != EOF && *t != JS_ERROR) {
16             if (c == '\\') {
17@@ -274,11 +276,12 @@ static xs_val *_xs_json_load_lexer(FILE *f, js_type *t)
18                     break;
19                 }
20 
21-                v = xs_utf8_cat(v, cp);
22+                v = xs_utf8_insert(v, cp, &offset);
23             }
24             else {
25                 char cc = c;
26-                v = xs_append_m(v, &cc, 1);
27+                v = xs_insert_m(v, offset, &cc, 1);
28+                offset++;
29             }
30         }
31 

  1@@ -9,6 +9,7 @@
  2  unsigned int xs_utf8_dec(const char **str);
  3  int xs_unicode_width(unsigned int cpoint);
  4  int xs_is_surrogate(unsigned int cpoint);
  5+ int xs_is_diacritic(unsigned int cpoint);
  6  unsigned int xs_surrogate_dec(unsigned int p1, unsigned int p2);
  7  unsigned int xs_surrogate_enc(unsigned int cpoint);
  8  unsigned int *_xs_unicode_upper_search(unsigned int cpoint);
  9@@ -22,7 +23,12 @@
 10  int xs_unicode_is_alpha(unsigned int cpoint);
 11 
 12 #ifdef _XS_H
 13+ xs_str *xs_utf8_insert(xs_str *str, unsigned int cpoint, int *offset);
 14  xs_str *xs_utf8_cat(xs_str *str, unsigned int cpoint);
 15+ xs_str *xs_utf8_to_upper(const char *str);
 16+ xs_str *xs_utf8_to_lower(const char *str);
 17+ xs_str *xs_utf8_to_nfd(const char *str);
 18+ xs_str *xs_utf8_to_nfc(const char *str);
 19 #endif
 20 
 21 #ifdef XS_IMPLEMENTATION
 22@@ -144,6 +150,12 @@ int xs_unicode_width(unsigned int cpoint)
 23 }
 24 
 25 
 26+int xs_is_diacritic(unsigned int cpoint)
 27+{
 28+    return cpoint >= 0x300 && cpoint <= 0x36f;
 29+}
 30+
 31+
 32 /** surrogate pairs **/
 33 
 34 int xs_is_surrogate(unsigned int cpoint)
 35@@ -172,14 +184,27 @@ unsigned int xs_surrogate_enc(unsigned int cpoint)
 36 
 37 #ifdef _XS_H
 38 
 39-xs_str *xs_utf8_cat(xs_str *str, unsigned int cpoint)
 40+xs_str *xs_utf8_insert(xs_str *str, unsigned int cpoint, int *offset)
 41 /* encodes an Unicode codepoint to utf-8 into str */
 42 {
 43     char tmp[4];
 44 
 45     int c = xs_utf8_enc(tmp, cpoint);
 46 
 47-    return xs_append_m(str, tmp, c);
 48+    str = xs_insert_m(str, *offset, tmp, c);
 49+
 50+    *offset += c;
 51+
 52+    return str;
 53+}
 54+
 55+
 56+xs_str *xs_utf8_cat(xs_str *str, unsigned int cpoint)
 57+/* encodes an Unicode codepoint to utf-8 into str */
 58+{
 59+    int offset = strlen(str);
 60+
 61+    return xs_utf8_insert(str, cpoint, &offset);
 62 }
 63 
 64 #endif /* _XS_H */
 65@@ -232,6 +257,9 @@ unsigned int *_xs_unicode_lower_search(unsigned int cpoint)
 66 unsigned int xs_unicode_to_lower(unsigned int cpoint)
 67 /* returns the cpoint to lowercase */
 68 {
 69+    if (cpoint < 0x80)
 70+        return tolower(cpoint);
 71+
 72     unsigned int *p = _xs_unicode_upper_search(cpoint);
 73 
 74     return p == NULL ? cpoint : p[1];
 75@@ -241,6 +269,9 @@ unsigned int xs_unicode_to_lower(unsigned int cpoint)
 76 unsigned int xs_unicode_to_upper(unsigned int cpoint)
 77 /* returns the cpoint to uppercase */
 78 {
 79+    if (cpoint < 0x80)
 80+        return toupper(cpoint);
 81+
 82     unsigned int *p = _xs_unicode_lower_search(cpoint);
 83 
 84     return p == NULL ? cpoint : p[0];
 85@@ -317,6 +348,87 @@ int xs_unicode_is_alpha(unsigned int cpoint)
 86 }
 87 
 88 
 89+#ifdef _XS_H
 90+
 91+xs_str *xs_utf8_to_upper(const char *str)
 92+{
 93+    xs_str *s = xs_str_new(NULL);
 94+    unsigned int cpoint;
 95+    int offset = 0;
 96+
 97+    while ((cpoint = xs_utf8_dec(&str))) {
 98+        cpoint = xs_unicode_to_upper(cpoint);
 99+        s = xs_utf8_insert(s, cpoint, &offset);
100+    }
101+
102+    return s;
103+}
104+
105+
106+xs_str *xs_utf8_to_lower(const char *str)
107+{
108+    xs_str *s = xs_str_new(NULL);
109+    unsigned int cpoint;
110+    int offset = 0;
111+
112+    while ((cpoint = xs_utf8_dec(&str))) {
113+        cpoint = xs_unicode_to_lower(cpoint);
114+        s = xs_utf8_insert(s, cpoint, &offset);
115+    }
116+
117+    return s;
118+}
119+
120+
121+xs_str *xs_utf8_to_nfd(const char *str)
122+{
123+    xs_str *s = xs_str_new(NULL);
124+    unsigned int cpoint;
125+    int offset = 0;
126+
127+    while ((cpoint = xs_utf8_dec(&str))) {
128+        unsigned int base;
129+        unsigned int diac;
130+
131+        if (xs_unicode_nfd(cpoint, &base, &diac)) {
132+            s = xs_utf8_insert(s, base, &offset);
133+            s = xs_utf8_insert(s, diac, &offset);
134+        }
135+        else
136+            s = xs_utf8_insert(s, cpoint, &offset);
137+    }
138+
139+    return s;
140+}
141+
142+
143+xs_str *xs_utf8_to_nfc(const char *str)
144+{
145+    xs_str *s = xs_str_new(NULL);
146+    unsigned int cpoint;
147+    unsigned int base = 0;
148+    int offset = 0;
149+
150+    while ((cpoint = xs_utf8_dec(&str))) {
151+        if (xs_is_diacritic(cpoint)) {
152+            if (xs_unicode_nfc(base, cpoint, &base))
153+                continue;
154+        }
155+
156+        if (base)
157+            s = xs_utf8_insert(s, base, &offset);
158+
159+        base = cpoint;
160+    }
161+
162+    if (base)
163+        s = xs_utf8_insert(s, base, &offset);
164+
165+    return s;
166+}
167+
168+#endif /* _XS_H */
169+
170 #endif /* _XS_UNICODE_TBL_H */
171 
172 #endif /* XS_IMPLEMENTATION */

1@@ -1 +1 @@
2-/* c6eca9593f9b3d6791cba600e5950f682fdb36cf 2024-08-12T16:08:37+02:00 */
3+/* cc9ebd36ae640e4701277327fbba9996143076f6 2024-08-23T17:17:08+02:00 */