From b511fe8adfa17635d3a0e781bb319647e28819a5 Mon Sep 17 00:00:00 2001
From: Art Cancro <ajc@citadel.org>
Date: Thu, 9 Feb 2006 04:00:10 +0000
Subject: [PATCH] Allow the user to specify a default character set for email
 headers.

---
 webcit/messages.c    | 57 +++++++++++++++++++++++++++++++++++---------
 webcit/preferences.c | 24 ++++++++++++-------
 2 files changed, 62 insertions(+), 19 deletions(-)

diff --git a/webcit/messages.c b/webcit/messages.c
index 724dab8c5..861b2a60e 100644
--- a/webcit/messages.c
+++ b/webcit/messages.c
@@ -67,14 +67,53 @@ void utf8ify_rfc822_string(char *buf) {
 	char encoding[16];
 	char istr[1024];
 	iconv_t ic = (iconv_t)(-1) ;
-	char *ibuf;		   /**< Buffer of characters to be converted */
-	char *obuf;		   /**< Buffer for converted characters      */
-	size_t ibuflen;	   /**< Length of input buffer	       */
-	size_t obuflen;	   /**< Length of output buffer	      */
-	char *isav;		   /**< Saved pointer to input buffer	*/
-	char *osav;		   /**< Saved pointer to output buffer       */
+	char *ibuf;			/**< Buffer of characters to be converted */
+	char *obuf;			/**< Buffer for converted characters */
+	size_t ibuflen;			/**< Length of input buffer */
+	size_t obuflen;			/**< Length of output buffer */
+	char *isav;			/**< Saved pointer to input buffer */
+	char *osav;			/**< Saved pointer to output buffer */
 	int passes = 0;
+	int i;
+	int illegal_non_rfc2047_encoding = 0;
 
+	/** Sometimes, badly formed messages contain strings which were simply
+	 *  written out directly in some foreign character set instead of
+	 *  using RFC2047 encoding.  This is illegal but we will attempt to
+	 *  handle it anyway by converting from a user-specified default
+	 *  charset to UTF-8 if we see any nonprintable characters.
+	 */
+	for (i=0; i<strlen(buf); ++i) {
+		if ((buf[i] < 32) || (buf[i] > 126)) {
+			illegal_non_rfc2047_encoding = 1;
+		}
+	}
+	if (illegal_non_rfc2047_encoding) {
+		char default_header_charset[128];
+		get_preference("default_header_charset", default_header_charset, sizeof default_header_charset);
+		if ( (strcasecmp(default_header_charset, "UTF-8")) && (strcasecmp(default_header_charset, "us-ascii")) ) {
+			ic = ctdl_iconv_open("UTF-8", default_header_charset);
+			if (ic != (iconv_t)(-1) ) {
+				ibuf = malloc(1024);
+				isav = ibuf;
+				safestrncpy(ibuf, buf, 1024);
+				ibuflen = strlen(ibuf);
+				obuflen = 1024;
+				obuf = (char *) malloc(obuflen);
+				osav = obuf;
+				iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
+				osav[1024-obuflen] = 0;
+				strcpy(buf, osav);
+				free(osav);
+				iconv_close(ic);
+				free(isav);
+			}
+		}
+	}
+
+	/** Now we handle foreign character sets properly encoded
+	 *  in RFC2047 format.
+	 */
 	while (start=strstr(buf, "=?"), end=strstr(buf, "?="),
 		((start != NULL) && (end != NULL) && (end > start)) )
 	{
@@ -82,10 +121,6 @@ void utf8ify_rfc822_string(char *buf) {
 		extract_token(encoding, start, 2, '?', sizeof encoding);
 		extract_token(istr, start, 3, '?', sizeof istr);
 
-		/*strcpy(start, "");
-		++end;
-		++end;*/
-
 		ibuf = malloc(1024);
 		isav = ibuf;
 		if (!strcasecmp(encoding, "B")) {	/**< base64 */
@@ -95,7 +130,7 @@ void utf8ify_rfc822_string(char *buf) {
 			ibuflen = CtdlDecodeQuotedPrintable(ibuf, istr, strlen(istr));
 		}
 		else {
-			strcpy(ibuf, istr);		/**< huh? */
+			strcpy(ibuf, istr);		/**< unknown encoding */
 			ibuflen = strlen(istr);
 		}
 
diff --git a/webcit/preferences.c b/webcit/preferences.c
index a7362134e..a21468c63 100644
--- a/webcit/preferences.c
+++ b/webcit/preferences.c
@@ -145,10 +145,11 @@ void get_preference(char *key, char *value, size_t value_len) {
 }
 
 /**
- * \brief Write a key into the citadel settings database
- * \param key key whichs value is to be modified
- * \param value value to set
- * \param save_to_server really write it????
+ * \brief	Write a key into the webcit preferences database for this user
+ *
+ * \params	key		key whichs value is to be modified
+ * \param	value		value to set
+ * \param	save_to_server	1 = flush all data to the server, 0 = cache it for now
  */
 void set_preference(char *key, char *value, int save_to_server) {
 	int num_prefs;
@@ -370,7 +371,16 @@ void display_preferences(void)
 		"	</script>				"
 	);
 
+	/** Character set to assume is in use for improperly encoded headers */
+	get_preference("default_header_charset", buf, sizeof buf);
+	if (buf[0] == 0) strcpy(buf, "UTF-8");
+	wprintf("<tr><td>");
+	wprintf(_("Default character set for email headers:"));
+	wprintf("</td><td>");
+	wprintf("<input type=\"text\" NAME=\"default_header_charset\" MAXLENGTH=\"32\" VALUE=\"%s\">", buf);
+	wprintf("</td></tr>");
 
+	/** submit buttons */
 	wprintf("</table>\n"
 		"<input type=\"submit\" name=\"change_button\" value=\"%s\">"
 		"&nbsp;"
@@ -379,11 +389,8 @@ void display_preferences(void)
 		_("Cancel")
 	);
 
-	wprintf("</form></center>\n");
-
 	/** end form */
-
-
+	wprintf("</form></center>\n");
 	wprintf("</td></tr></table></div>\n");
 	wDumpContent(1);
 }
@@ -412,6 +419,7 @@ void set_preferences(void)
 	set_preference("use_sig", bstr("use_sig"), 0);
 	set_preference("daystart", bstr("daystart"), 0);
 	set_preference("dayend", bstr("dayend"), 0);
+	set_preference("default_header_charset", bstr("default_header_charset"), 0);
 
 	euid_escapize(ebuf, bstr("signature"));
 	set_preference("signature", ebuf, 1);
-- 
2.39.2