summaryrefslogtreecommitdiff
path: root/src/modules/filters/utf8utf16.cpp
diff options
context:
space:
mode:
authorRoberto C. Sanchez <roberto@connexer.com>2014-03-29 10:53:59 -0400
committerRoberto C. Sanchez <roberto@connexer.com>2014-03-29 10:53:59 -0400
commit03134fa5f6f25d92724ce4c183f9bbe12a9e37dc (patch)
tree847326a4de82f0241ac87cbbc427a1b92a696a02 /src/modules/filters/utf8utf16.cpp
parentd7469385b05b9510338407fa123e9ad090f80af6 (diff)
Imported Upstream version 1.5.11
Diffstat (limited to 'src/modules/filters/utf8utf16.cpp')
-rw-r--r--src/modules/filters/utf8utf16.cpp78
1 files changed, 78 insertions, 0 deletions
diff --git a/src/modules/filters/utf8utf16.cpp b/src/modules/filters/utf8utf16.cpp
new file mode 100644
index 0000000..5c1614c
--- /dev/null
+++ b/src/modules/filters/utf8utf16.cpp
@@ -0,0 +1,78 @@
+/******************************************************************************
+ *
+ * UTF8UTF16 - SWFilter descendant to convert UTF-8 to UTF-16
+ *
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+
+#include <utf8utf16.h>
+#include <swbuf.h>
+
+SWORD_NAMESPACE_START
+
+UTF8UTF16::UTF8UTF16() {
+}
+
+char UTF8UTF16::processText(SWBuf &text, const SWKey *key, const SWModule *module) {
+ const unsigned char *from;
+ unsigned long ch;
+ signed short utf16;
+ unsigned char from2[7];
+
+ SWBuf orig = text;
+
+ from = (const unsigned char *)orig.c_str();
+
+ // -------------------------------
+ for (text = ""; *from; from++) {
+ ch = 0;
+ //case: ANSI
+ if ((*from & 128) != 128) {
+ text.setSize(text.size()+2);
+ *((unsigned short *)(text.getRawData()+(text.size()-2))) = (unsigned short)*from;
+ continue;
+ }
+ //case: Invalid UTF-8 (illegal continuing byte in initial position)
+ if ((*from & 128) && ((*from & 64) != 64)) {
+ continue;
+ }
+ //case: 2+ byte codepoint
+ from2[0] = *from;
+ from2[0] <<= 1;
+ int subsequent;
+ for (subsequent = 1; (from2[0] & 128) && (subsequent < 7); subsequent++) {
+ from2[0] <<= 1;
+ from2[subsequent] = from[subsequent];
+ from2[subsequent] &= 63;
+ ch <<= 6;
+ ch |= from2[subsequent];
+ }
+ subsequent--;
+ from2[0] <<= 1;
+ char significantFirstBits = 8 - (2+subsequent);
+
+ ch |= (((short)from2[0]) << (((6*subsequent)+significantFirstBits)-8));
+ from += subsequent;
+ if (ch < 0x10000) {
+ text.setSize(text.size()+2);
+ *((unsigned short *)(text.getRawData()+(text.size()-2))) = (unsigned short)ch;
+ }
+ else {
+ utf16 = (signed short)((ch - 0x10000) / 0x400 + 0xD800);
+ text.setSize(text.size()+2);
+ *((unsigned short *)(text.getRawData()+(text.size()-2))) = (unsigned short)utf16;
+ utf16 = (signed short)((ch - 0x10000) % 0x400 + 0xDC00);
+ text.setSize(text.size()+2);
+ *((unsigned short *)(text.getRawData()+(text.size()-2))) = (unsigned short)utf16;
+ }
+ }
+ text.setSize(text.size()+2);
+ *((unsigned short *)(text.getRawData()+(text.size()-2))) = (unsigned short)0;
+
+ return 0;
+
+}
+
+SWORD_NAMESPACE_END