From 368fb5f02bf26a1d4b2a15a204cacd3a15bfaa88 Mon Sep 17 00:00:00 2001
From: ilor <kailoran@gmail.com>
Date: Tue, 24 May 2011 12:24:30 +0200
Subject: [PATCH] simple swig icu UnicodeString wrapper with __str__,
__unicode__ and length()
---
swig/corpus2.i | 1 +
swig/unicodestring.i | 42 ++++++++++++++++++++++++++++++++++++++++++
2 files changed, 43 insertions(+)
create mode 100644 swig/unicodestring.i
diff --git a/swig/corpus2.i b/swig/corpus2.i
index a805a32..087727c 100644
--- a/swig/corpus2.i
+++ b/swig/corpus2.i
@@ -4,6 +4,7 @@
%module corpus2
%include "boost_shared_ptr.i"
+%include "unicodestring.i"
%include "libcorpus2exception.i"
%include "libcorpusannotatedsentence.i"
%include "libcorpusannotationchannel.i"
diff --git a/swig/unicodestring.i b/swig/unicodestring.i
new file mode 100644
index 0000000..30997d7
--- /dev/null
+++ b/swig/unicodestring.i
@@ -0,0 +1,42 @@
+#ifndef SWIG_UNICODESTRING_I
+#define SWIG_UNICODESTRING_I
+
+%module unicodestring
+%{
+ #include <unicode/unistr.h>
+%}
+
+%include "std_string.i"
+%include "std_except.i"
+
+
+class UnicodeString {
+ public:
+ int length();
+ UChar* getBuffer();
+ UChar* getTerminatedBuffer();
+
+%pythoncode %{
+ def __unicode__(self):
+ return self.as_utf16().decode('utf16')
+
+ def __str__(self):
+ return self.as_utf8()
+%}
+
+};
+
+%extend UnicodeString {
+ std::string as_utf16() {
+ return std::string((char*)self->getTerminatedBuffer(), self->length()*2);
+ }
+
+ std::string as_utf8() {
+ std::string r;
+ self->toUTF8String(r);
+ return r;
+ }
+}
+
+
+#endif /* SWIG_UNICODESTRING_I */
--
GitLab