summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorPeng Wu <alexepico@gmail.com>2018-10-23 11:22:25 +0800
committerPeng Wu <alexepico@gmail.com>2018-10-23 11:22:25 +0800
commit9c08655cdf87e948dda61f452167b9281686334d (patch)
treed5f86dfea0aa95b1f1703e7fcccc8b7a7e0d0eaf
parent155ef10fcb6d2d69fe6d4e733693a23005473662 (diff)
downloadibus-libzhuyin-9c08655cdf87e948dda61f452167b9281686334d.zip
ibus-libzhuyin-9c08655cdf87e948dda61f452167b9281686334d.tar.gz
ibus-libzhuyin-9c08655cdf87e948dda61f452167b9281686334d.tar.xz
update class TradSimpConverter
-rw-r--r--src/ZYTradSimpConverter.cc99
1 files changed, 98 insertions, 1 deletions
diff --git a/src/ZYTradSimpConverter.cc b/src/ZYTradSimpConverter.cc
index 909f751..e7548e0 100644
--- a/src/ZYTradSimpConverter.cc
+++ b/src/ZYTradSimpConverter.cc
@@ -25,13 +25,20 @@
#include "ZYTradSimpConverter.h"
-#include <opencc.h>
+#ifdef HAVE_OPENCC
+# include <opencc.h>
+#else
+# include <cstring>
+# include <cstdlib>
+#endif
#include "ZYTypes.h"
#include "ZYString.h"
namespace ZY {
+#ifdef HAVE_OPENCC
+
class opencc {
public:
opencc (void)
@@ -64,4 +71,94 @@ TradSimpConverter::tradToSimp (const gchar *in, String &out)
cc.convert (in, out);
}
+#else
+
+static gint _xcmp (const gchar *p1, const gchar *p2, const gchar *str)
+{
+ for (;;) {
+ // both reach end
+ if (p1 == p2 && *str == '\0')
+ return 0;
+ // p1 reaches end
+ if (p1 == p2)
+ return -1;
+ // str reaches end
+ if (*str == '\0')
+ return 1;
+
+ if (*p1 < *str)
+ return -1;
+ if (*p1 > *str)
+ return 1;
+
+ p1 ++; str ++;
+ };
+}
+
+static gint _cmp (gconstpointer p1, gconstpointer p2)
+{
+ const gchar **pp = (const gchar **) p1;
+ const gchar **s2 = (const gchar **) p2;
+
+ return _xcmp (pp[0], pp[1], s2[0]);
+}
+
+#include "ZYTradSimpConverterTable.h"
+
+void
+TradSimpConverter::tradToSimp (const gchar *in, String &out)
+{
+ const gchar *pend;
+ const gchar *pp[2];
+ glong len;
+ glong begin;
+
+ if (!g_utf8_validate (in, -1 , NULL)) {
+ g_warning ("\%s\" is not an utf8 string!", in);
+ g_assert_not_reached ();
+ }
+
+ begin = 0;
+ pend = in + std::strlen (in);
+ len = g_utf8_strlen (in, -1); // length in charactoers
+ pp[0] = in;
+
+ while (pp[0] != pend) {
+ glong slen = std::min (len - begin, (glong) TRAD_TO_SIMP_MAX_LEN); // the length of sub string in character
+ pp[1] = g_utf8_offset_to_pointer (pp[0], slen); // the end of sub string
+
+ for (;;) {
+ const gchar **result;
+ result = (const gchar **) std::bsearch (pp, trad_to_simp,
+ G_N_ELEMENTS (trad_to_simp), sizeof (trad_to_simp[0]),
+ _cmp);
+
+ if (result != NULL) {
+ // found item in table,
+ // append the trad to out and adjust pointers
+ out << result[1];
+ pp[0] = pp[1];
+ begin += slen;
+ break;
+ }
+
+ if (slen == 1) {
+ // if only one character left,
+ // append origin character to out and adjust pointers
+ out.append (pp[0], pp[1] - pp[0]);
+ pp[0] = pp[1];
+ begin += 1;
+ break;
+ }
+
+ // if more than on characters left,
+ // adjust pp[1] to previous character
+ pp[1] = g_utf8_prev_char (pp[1]);
+ slen--;
+ }
+ }
+}
+
+#endif
+
};