Backport from xs.

3 files changed, 134 insertions, 13 deletions

diff --git a/xs_json.h b/xs_json.h
index 388cfe0..d702a85 100644
--- a/xs_json.h
+++ b/xs_json.h
@@ -5,6 +5,7 @@
 #define _XS_JSON_H
 
 xs_str *xs_json_dumps_pp(const xs_val *data, int indent);
+int xs_json_dump_pp(const xs_val *data, int indent, FILE *f);
 #define xs_json_dumps(data) xs_json_dumps_pp(data, 0)
 xs_val *xs_json_loads(const xs_str *json);
 
@@ -165,6 +166,19 @@ xs_str *xs_json_dumps_pp(const xs_val *data, int indent)
 }
 
 
+int xs_json_dump_pp(const xs_val *data, int indent, FILE *f)
+/* dumps data into a file as JSON */
+{
+    xs *j = xs_json_dumps_pp(data, indent);
+
+    if (j == NULL)
+        return 0;
+
+    fwrite(j, strlen(j), 1, f);
+    return 1;
+}
+
+
 /** JSON loads **/
 
 /* this code comes mostly from the Minimum Profit Text Editor (MPDM) */
diff --git a/xs_unicode.h b/xs_unicode.h
index d45b52e..48cd660 100644
--- a/xs_unicode.h
+++ b/xs_unicode.h
@@ -5,8 +5,15 @@
 #define _XS_UNICODE_H
 
  xs_str *xs_utf8_enc(xs_str *str, unsigned int cpoint);
- char *xs_utf8_dec(const char *str, unsigned int *cpoint);
-
+ unsigned int xs_utf8_dec(char **str);
+ unsigned int *_xs_unicode_upper_search(unsigned int cpoint);
+ unsigned int *_xs_unicode_lower_search(unsigned int cpoint);
+ #define xs_unicode_is_upper(cpoint) (!!_xs_unicode_upper_search(cpoint))
+ #define xs_unicode_is_lower(cpoint) (!!_xs_unicode_lower_search(cpoint))
+ unsigned int xs_unicode_to_upper(unsigned int cpoint);
+ unsigned int xs_unicode_to_lower(unsigned int cpoint);
+ int xs_unicode_nfd(unsigned int cpoint, unsigned int *base, unsigned int *diac);
+ int xs_unicode_nfc(unsigned int base, unsigned int diac, unsigned int *cpoint);
 
 #ifdef XS_IMPLEMENTATION
 
@@ -50,46 +57,146 @@ xs_str *xs_utf8_enc(xs_str *str, unsigned int cpoint)
 }
 
 
-char *xs_utf8_dec(const char *str, unsigned int *cpoint)
-/* decodes an utf-8 char inside str into cpoint and returns the next position */
+unsigned int xs_utf8_dec(char **str)
+/* decodes an utf-8 char inside str and updates the pointer */
 {
-    unsigned char *p = (unsigned char *)str;
+    unsigned char *p = (unsigned char *)*str;
+    unsigned int cpoint = 0;
     int c = *p++;
     int cb = 0;
 
     if ((c & 0x80) == 0) { /* 1 byte char */
-        *cpoint = c;
+        cpoint = c;
     }
     else
     if ((c & 0xe0) == 0xc0) { /* 2 byte char */
-        *cpoint = (c & 0x1f) << 6;
+        cpoint = (c & 0x1f) << 6;
         cb = 1;
     }
     else
     if ((c & 0xf0) == 0xe0) { /* 3 byte char */
-        *cpoint = (c & 0x0f) << 12;
+        cpoint = (c & 0x0f) << 12;
         cb = 2;
     }
     else
     if ((c & 0xf8) == 0xf0) { /* 4 byte char */
-        *cpoint = (c & 0x07) << 18;
+        cpoint = (c & 0x07) << 18;
         cb = 3;
     }
 
     /* process the continuation bytes */
     while (cb--) {
         if ((*p & 0xc0) == 0x80)
-            *cpoint |= (*p++ & 0x3f) << (cb * 6);
+            cpoint |= (*p++ & 0x3f) << (cb * 6);
         else {
-            *cpoint = 0xfffd;
+            cpoint = 0xfffd;
             break;
         }
     }
 
-    return (char *)p;
+    *str = (char *)p;
+    return cpoint;
+}
+
+
+#ifdef _XS_UNICODE_TBL_H
+
+/* include xs_unicode_tbl.h before to use these functions */
+
+static int int_cmp(const void *p1, const void *p2)
+{
+    const unsigned int *a = p1;
+    const unsigned int *b = p2;
+
+    return *a < *b ? -1 : *a > *b ? 1 : 0;
+}
+
+
+unsigned int *_xs_unicode_upper_search(unsigned int cpoint)
+/* searches for an uppercase codepoint in the case fold table */
+{
+    return bsearch(&cpoint, xs_unicode_case_fold_table,
+        sizeof(xs_unicode_case_fold_table) / (sizeof(unsigned int) * 2),
+        sizeof(unsigned int) * 2,
+        int_cmp);
+}
+
+
+unsigned int *_xs_unicode_lower_search(unsigned int cpoint)
+/* searches for a lowercase codepoint in the case fold table */
+{
+    unsigned int *p = xs_unicode_case_fold_table + 1;
+    unsigned int *e = xs_unicode_case_fold_table +
+            sizeof(xs_unicode_case_fold_table) / sizeof(unsigned int);
+
+    while (p < e) {
+        if (cpoint == *p)
+            return p;
+
+        p += 2;
+    }
+
+    return NULL;
+}
+
+
+unsigned int xs_unicode_to_upper(unsigned int cpoint)
+/* returns the cpoint to uppercase */
+{
+    unsigned int *p = _xs_unicode_lower_search(cpoint);
+
+    return p == NULL ? cpoint : p[-1];
+}
+
+
+unsigned int xs_unicode_to_lower(unsigned int cpoint)
+/* returns the cpoint to lowercase */
+{
+    unsigned int *p = _xs_unicode_upper_search(cpoint);
+
+    return p == NULL ? cpoint : p[1];
 }
 
 
+int xs_unicode_nfd(unsigned int cpoint, unsigned int *base, unsigned int *diac)
+/* applies unicode Normalization Form D */
+{
+    unsigned int *r = bsearch(&cpoint, xs_unicode_nfd_table,
+                        sizeof(xs_unicode_nfd_table) / (sizeof(unsigned int) * 3),
+                        sizeof(unsigned int) * 3,
+                        int_cmp);
+
+    if (r != NULL) {
+        *base = r[1];
+        *diac = r[2];
+    }
+
+    return !!r;
+}
+
+
+int xs_unicode_nfc(unsigned int base, unsigned int diac, unsigned int *cpoint)
+/* applies unicode Normalization Form C */
+{
+    unsigned int *p = xs_unicode_nfd_table;
+    unsigned int *e = xs_unicode_nfd_table +
+        sizeof(xs_unicode_nfd_table) / sizeof(unsigned int);
+
+    while (p < e) {
+        if (p[1] == base && p[2] == diac) {
+            *cpoint = p[0];
+            return 1;
+        }
+
+        p += 3;
+    }
+
+    return 0;
+}
+
+
+#endif /* _XS_UNICODE_TBL_H */
+
 #endif /* XS_IMPLEMENTATION */
 
 #endif /* _XS_UNICODE_H */
diff --git a/xs_version.h b/xs_version.h
index bc84175..6f137ea 100644
--- a/xs_version.h
+++ b/xs_version.h
@@ -1 +1 @@
-/* 4c151c4cc5b7c6980c1f0dd733a3fab0a30f9695 */
+/* 5ad148b1c1dbbf7b4550c9fcd13d96ac6def2d21 */