Transcode mh_execm text/plain output like we do for mh_exec. Adjust handling of transcoding errors. These changes should fix most cases of non-utf8 text making it to unac/index

src/internfile/mh_exec.cpp

@@ -153,17 +153,24 @@ void MimeHandlerExec::finaldetails()
     // cfgFilterOutputCharset comes from the mimeconf filter definition line
     string charset = cfgFilterOutputCharset.empty() ? "utf-8" : 
 	cfgFilterOutputCharset;
+    bool trustcharset = true;
     if (!stringlowercmp("default", charset)) {
 	charset = m_dfltInputCharset;
+	trustcharset = false;
     }
     string mt = cfgFilterOutputMtype.empty() ? "text/html" : 
 	cfgFilterOutputMtype;
-    if (!mt.compare(cstr_textplain) && stringlowercmp("utf-8", charset)) {
+
+    // If this is text/plain and not utf-8 or untrusted, transcode to utf-8.
+    if (!mt.compare(cstr_textplain) && 
+	(!trustcharset || stringlowercmp("utf-8", charset))) {
 	string transcoded;
 	int ecnt;
 	if (!transcode(output, transcoded, charset, "UTF-8", &ecnt)) {
 	    LOGERR(("mh_exec: transcode failed from [%s] to UTF-8\n",
 		    charset.c_str()));
+	    // Erase text in this case: it's garbage
+	    output.clear();
 	} else {
 	    if (ecnt) {
 		LOGDEB(("mh_exec: %d transcoding errors  from [%s] to UTF-8\n",

src/internfile/mh_execm.cpp

@@ -284,15 +284,37 @@ bool MimeHandlerExecMultiple::next_document()
 
     // Charset. For many document types it doesn't matter. For text
     // and html it does. We supply a default from the
-    // configuration. We should do the text transcoding to utf-8 here
-    // like exec::finaldetails does.
+    // configuration. 
+    bool trustcharset = true;
     if (charset.empty()) {
 	charset = cfgFilterOutputCharset.empty() ? "utf-8" : 
 	    cfgFilterOutputCharset;
 	if (!stringlowercmp("default", charset)) {
+	    trustcharset = false;
 	    charset = m_dfltInputCharset;
 	}
     }
+
+    string& output = m_metaData[cstr_content];
+    if (!m_metaData[cstr_mimetype].compare(cstr_textplain) && 
+	(!trustcharset || stringlowercmp("utf-8", charset))) {
+	string transcoded;
+	int ecnt;
+	if (!transcode(output, transcoded, charset, "UTF-8", &ecnt)) {
+	    LOGERR(("mh_execm: transcode failed from [%s] to UTF-8\n",
+		    charset.c_str()));
+	    // Erase text in this case: it's garbage
+	    output.clear();
+	} else {
+	    if (ecnt) {
+		LOGDEB(("mh_exec: %d transcoding errors  from [%s] to UTF-8\n",
+			ecnt, charset.c_str()));
+	    }
+	    output = transcoded;
+	    charset = "utf-8";
+	}
+    }
+    
     m_metaData[cstr_charset] = charset;
     
     if (eofnext_received)

src/internfile/mh_mail.cpp

@@ -259,13 +259,13 @@ bool MimeHandlerMail::processAttach()
     // Special case for text/plain content. Internfile should deal
     // with this but it expects text/plain to be utf-8 already, so we
     // handle the transcoding if needed
-    if (m_metaData[cstr_mimetype] == cstr_textplain && 
-	stringicmp(m_metaData[cstr_charset], "UTF-8")) {
+    if (m_metaData[cstr_mimetype] == cstr_textplain) {
 	string utf8;
 	if (!transcode(body, utf8, m_metaData[cstr_charset], "UTF-8")) {
 	    LOGERR(("  processAttach: transcode to utf-8 failed "
 		    "for charset [%s]\n", m_metaData[cstr_charset].c_str()));
-	    // Just let it through and hope for the best...
+ 	    // can't transcode at all -> data is garbage just erase it
+ 	    body.clear();
 	} else {
 	    body = utf8;
 	}