From 7fb480447736bd7e7371c53e5def624724028240 Mon Sep 17 00:00:00 2001
From: hiro <hiro@ee746299-78ed-0310-b773-934348b2243d>
Date: Thu, 16 Nov 2017 07:52:54 +0000
Subject: supported UTF-16 text attachments on sending.

git-svn-id: svn://sylpheed.sraoss.jp/sylpheed/trunk@3573 ee746299-78ed-0310-b773-934348b2243d
---
 libsylph/codeconv.c | 73 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 libsylph/codeconv.h |  8 +++++-
 2 files changed, 80 insertions(+), 1 deletion(-)

(limited to 'libsylph')

diff --git a/libsylph/codeconv.c b/libsylph/codeconv.c
index 8781f446..1ba52ac3 100644
--- a/libsylph/codeconv.c
+++ b/libsylph/codeconv.c
@@ -1681,6 +1681,9 @@ static const struct {
 	{C_GEORGIAN_PS,		CS_GEORGIAN_PS},
 	{C_TCVN5712_1,		CS_TCVN5712_1},
 	{C_ISO_8859_16,		CS_ISO_8859_16},
+	{C_UTF_16,		CS_UTF_16},
+	{C_UTF_16BE,		CS_UTF_16BE},
+	{C_UTF_16LE,		CS_UTF_16LE},
 };
 
 static const struct {
@@ -2656,6 +2659,7 @@ CharSet conv_check_file_encoding(const gchar *file)
 	CharSet enc;
 	const gchar *enc_str;
 	gboolean is_locale = TRUE, is_utf8 = TRUE;
+	size_t size;
 
 	g_return_val_if_fail(file != NULL, C_AUTO);
 
@@ -2669,6 +2673,75 @@ CharSet conv_check_file_encoding(const gchar *file)
 		return C_AUTO;
 	}
 
+	/* UTF-16 check */
+	if ((size = fread(buf, 2, BUFFSIZE / 2, fp)) > 0) {
+		CharSet guess_enc = C_AUTO;
+
+		debug_print("conv_check_file_encoding: check first %d bytes of file %s\n", size * 2, file);
+
+		/* BOM check */
+		if ((buf[0] & 0xff) == 0xfe && (buf[1] & 0xff) == 0xff) {
+			debug_print("conv_check_file_encoding: UTF-16 BOM (BE) found\n");
+			guess_enc = C_UTF_16; /* UTF-16BE */
+		} else if ((buf[0] & 0xff) == 0xff && (buf[1] & 0xff) == 0xfe) {
+			debug_print("conv_check_file_encoding: UTF-16 BOM (LE) found\n");
+			guess_enc = C_UTF_16; /* UTF-16LE */
+		}
+		if (guess_enc != C_AUTO) {
+			fclose(fp);
+			return guess_enc;
+		}
+
+		/* search UTF-16 CR/LF */
+		if (memchr(buf, 0x00, size * 2) != NULL) {
+			gint i;
+			guchar c1, c2;
+
+			for (i = 0; i < size; i++) {
+				c1 = buf[i * 2] & 0xff;
+				c2 = buf[i * 2 + 1] & 0xff;
+				if (c1 == 0x00 && c2 == 0x0d) { /* UTF-16BE CR */
+					i++;
+					if (i >= size) {
+						break;
+					}
+					c1 = buf[i * 2] & 0xff;
+					c2 = buf[i * 2 + 1] & 0xff;
+					if (c1 == 0x00 && c2 == 0x0a) { /* UTF-16BE LF */
+						guess_enc = C_UTF_16BE;
+						break;
+					}
+				} else if (c1 == 0x0d && c2 == 0x00) { /* UTF-16LE CR */
+					i++;
+					if (i >= size) {
+						break;
+					}
+					c1 = buf[i * 2] & 0xff;
+					c2 = buf[i * 2 + 1] & 0xff;
+					if (c1 == 0x0a && c2 == 0x00) { /* UTF-16LE LF */
+						guess_enc = C_UTF_16LE;
+						break;
+					}
+				} else if (c1 == 0x00 && c2 == 0x0a) { /* UTF-16BE LF */
+					guess_enc = C_UTF_16BE;
+					break;
+				} else if (c1 == 0x0a && c2 == 0x00) { /* UTF-16LE LF */
+					guess_enc = C_UTF_16LE;
+					break;
+				}
+			}
+
+			if (guess_enc != C_AUTO) {
+				debug_print("conv_check_file_encoding: %s detected\n",
+					    conv_get_charset_str(guess_enc));
+				fclose(fp);
+				return guess_enc;
+			}
+		}
+	}
+
+	rewind(fp);
+
 	while (fgets(buf, sizeof(buf), fp) != NULL) {
 		gchar *str;
 		gint error = 0;
diff --git a/libsylph/codeconv.h b/libsylph/codeconv.h
index 121de0da..6f6fc365 100644
--- a/libsylph/codeconv.h
+++ b/libsylph/codeconv.h
@@ -92,7 +92,10 @@ typedef enum
 	C_WINDOWS_874,
 	C_GEORGIAN_PS,
 	C_TCVN5712_1,
-	C_ISO_8859_16
+	C_ISO_8859_16,
+	C_UTF_16,
+	C_UTF_16BE,
+	C_UTF_16LE
 } CharSet;
 
 typedef enum
@@ -180,6 +183,9 @@ struct _CodeConverter
 #define CS_GEORGIAN_PS		"GEORGIAN-PS"
 #define CS_TCVN5712_1		"TCVN5712-1"
 #define CS_ISO_8859_16		"ISO-8859-16"
+#define CS_UTF_16		"UTF-16"
+#define CS_UTF_16BE		"UTF-16BE"
+#define CS_UTF_16LE		"UTF-16LE"
 
 #define C_INTERNAL		C_UTF_8
 #define CS_INTERNAL		CS_UTF_8
-- 
cgit v1.2.3