commit 4105e73

Backport from xs.

  1@@ -59,6 +59,7 @@ xs_val *xs_insert_m(xs_val *data, int offset, const char *mem, int size);
  2 #define xs_append_m(data, mem, size) xs_insert_m(data, xs_size(data) - 1, mem, size)
  3 
  4 xs_str *xs_str_new(const char *str);
  5+xs_str *xs_str_new_sz(const char *mem, int sz);
  6 xs_str *xs_str_wrap_i(const char *prefix, xs_str *str, const char *suffix);
  7 #define xs_str_prepend_i(str, prefix) xs_str_wrap_i(prefix, str, NULL)
  8 #define xs_str_cat(str, suffix) xs_str_wrap_i(NULL, str, suffix)
  9@@ -72,6 +73,8 @@ int xs_starts_and_ends(const char *prefix, const char *str, const char *suffix);
 10 #define xs_startswith(str, prefix) xs_starts_and_ends(prefix, str, NULL)
 11 #define xs_endswith(str, suffix) xs_starts_and_ends(NULL, str, suffix)
 12 xs_str *xs_crop_i(xs_str *str, int start, int end);
 13+xs_str *xs_lstrip_chars_i(xs_str *str, const char *chars);
 14+xs_str *xs_rstrip_chars_i(xs_str *str, const char *chars);
 15 xs_str *xs_strip_chars_i(xs_str *str, const char *chars);
 16 #define xs_strip_i(str) xs_strip_chars_i(str, " \r\n\t\v\f")
 17 xs_str *xs_tolower_i(xs_str *str);
 18@@ -424,6 +427,17 @@ xs_str *xs_str_new(const char *str)
 19 }
 20 
 21 
 22+xs_str *xs_str_new_sz(const char *mem, int sz)
 23+/* creates a new string from a memory block, adding an asciiz */
 24+{
 25+    xs_str *s = xs_realloc(NULL, _xs_blk_size(sz + 1));
 26+    memcpy(s, mem, sz);
 27+    s[sz] = '\0';
 28+
 29+    return s;
 30+}
 31+
 32+
 33 xs_str *xs_str_wrap_i(const char *prefix, xs_str *str, const char *suffix)
 34 /* wraps str with prefix and suffix */
 35 {
 36@@ -546,26 +560,36 @@ xs_str *xs_crop_i(xs_str *str, int start, int end)
 37 }
 38 
 39 
 40-xs_str *xs_strip_chars_i(xs_str *str, const char *chars)
 41-/* strips the string of chars from the start and the end */
 42+xs_str *xs_lstrip_chars_i(xs_str *str, const char *chars)
 43+/* strips all chars from the start of str */
 44 {
 45-    XS_ASSERT_TYPE(str, XSTYPE_STRING);
 46+    int n;
 47 
 48+    for (n = 0; str[n] && strchr(chars, str[n]); n++);
 49+
 50+    if (n)
 51+        str = xs_collapse(str, 0, n);
 52+
 53+    return str;
 54+}
 55+
 56+
 57+xs_str *xs_rstrip_chars_i(xs_str *str, const char *chars)
 58+/* strips all chars from the end of str */
 59+{
 60     int n;
 61 
 62-    /* strip first from the end */
 63     for (n = strlen(str); n > 0 && strchr(chars, str[n - 1]); n--);
 64     str[n] = '\0';
 65 
 66-    if (str[0]) {
 67-        /* now strip from the beginning */
 68-        for (n = 0; str[n] && strchr(chars, str[n]); n++);
 69+    return str;
 70+}
 71 
 72-        if (n)
 73-            str = xs_collapse(str, 0, n);
 74-    }
 75 
 76-    return str;
 77+xs_str *xs_strip_chars_i(xs_str *str, const char *chars)
 78+/* strips the string of chars from the start and the end */
 79+{
 80+    return xs_lstrip_chars_i(xs_rstrip_chars_i(str, chars), chars);
 81 }
 82 
 83 
 84@@ -859,11 +883,9 @@ xs_list *xs_split_n(const char *str, const char *sep, int times)
 85     list = xs_list_new();
 86 
 87     while (times > 0 && (ss = strstr(str, sep)) != NULL) {
 88-        /* add the first part (without the asciiz) */
 89-        list = xs_list_append_m(list, str, ss - str);
 90-
 91-        /* add the asciiz */
 92-        list = xs_insert_m(list, xs_size(list) - 1, "", 1);
 93+        /* create a new string with this slice and add it to the list */
 94+        xs *s = xs_str_new_sz(str, ss - str);
 95+        list = xs_list_append(list, s);
 96 
 97         /* skip past the separator */
 98         str = ss + sz;
 99@@ -1131,8 +1153,7 @@ int xs_data_size(const xs_data *value)
100 void xs_data_get(const xs_data *value, void *data)
101 /* copies the raw data stored inside value into data */
102 {
103-    int size = _xs_get_24b(value + 1) - 4;
104-    memcpy(data, &value[4], size);
105+    memcpy(data, &value[4], xs_data_size(value));
106 }
107 
108 

 1@@ -33,12 +33,12 @@ xs_list *xs_regex_split_n(const char *str, const char *rx, int count)
 2 
 3     while (count > 0 && !regexec(&re, (p = str + offset), 1, &rm, offset > 0 ? REG_NOTBOL : 0)) {
 4         /* add first the leading part of the string */
 5-        list = xs_list_append_m(list, p, rm.rm_so);
 6-        list = xs_insert_m(list, xs_size(list) - 1, "", 1);
 7+        xs *s1 = xs_str_new_sz(p, rm.rm_so);
 8+        list = xs_list_append(list, s1);
 9 
10         /* add now the matched text as the separator */
11-        list = xs_list_append_m(list, p + rm.rm_so, rm.rm_eo - rm.rm_so);
12-        list = xs_insert_m(list, xs_size(list) - 1, "", 1);
13+        xs *s2 = xs_str_new_sz(p + rm.rm_so, rm.rm_eo - rm.rm_so);
14+        list = xs_list_append(list, s2);
15 
16         /* move forward */
17         offset += rm.rm_eo;

 1@@ -16,6 +16,7 @@
 2  unsigned int xs_unicode_to_lower(unsigned int cpoint);
 3  int xs_unicode_nfd(unsigned int cpoint, unsigned int *base, unsigned int *diac);
 4  int xs_unicode_nfc(unsigned int base, unsigned int diac, unsigned int *cpoint);
 5+ int xs_unicode_is_alpha(unsigned int cpoint);
 6 
 7 #ifdef XS_IMPLEMENTATION
 8 
 9@@ -101,6 +102,15 @@ unsigned int xs_utf8_dec(char **str)
10 }
11 
12 
13+static int int_range_cmp(const void *p1, const void *p2)
14+{
15+    const unsigned int *a = p1;
16+    const unsigned int *b = p2;
17+
18+    return *a < b[0] ? -1 : *a > b[1] ? 1 : 0;
19+}
20+
21+
22 /* intentionally dead simple */
23 
24 static unsigned int xs_unicode_width_table[] = {
25@@ -119,20 +129,12 @@ static unsigned int xs_unicode_width_table[] = {
26 int xs_unicode_width(unsigned int cpoint)
27 /* returns the width in columns of a Unicode codepoint (somewhat simplified) */
28 {
29-    unsigned int *p = xs_unicode_width_table;
30-    unsigned int *e = p + sizeof(xs_unicode_width_table) / sizeof(unsigned int);
31-
32-    while (p < e) {
33-        if (cpoint < p[0])
34-            return 1;
35-
36-        if (cpoint >= p[0] && cpoint <= p[1])
37-            return p[2];
38-
39-        p += 3;
40-    }
41+    unsigned int *r = bsearch(&cpoint, xs_unicode_width_table,
42+                        sizeof(xs_unicode_width_table) / (sizeof(unsigned int) * 3),
43+                        sizeof(unsigned int) * 3,
44+                        int_range_cmp);
45 
46-    return 0;
47+    return r ? r[2] : 1;
48 }
49 
50 
51@@ -232,6 +234,18 @@ int xs_unicode_nfc(unsigned int base, unsigned int diac, unsigned int *cpoint)
52 }
53 
54 
55+int xs_unicode_is_alpha(unsigned int cpoint)
56+/* checks if a codepoint is an alpha (i.e. a letter) */
57+{
58+    unsigned int *r = bsearch(&cpoint, xs_unicode_alpha_table,
59+                        sizeof(xs_unicode_alpha_table) / (sizeof(unsigned int) * 2),
60+                        sizeof(unsigned int) * 2,
61+                        int_range_cmp);
62+
63+    return !!r;
64+}
65+
66+
67 #endif /* _XS_UNICODE_TBL_H */
68 
69 #endif /* XS_IMPLEMENTATION */

1@@ -1 +1 @@
2-/* fdd04f1862e0d8bdebb7b438798914643895d43f */
3+/* 5bf06243b37eec60e48e53d87d4d147d01ad9924 */