mh_mail now uses mimetype() to try and better identify application/octet-stream

src/index/mimetype.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: mimetype.cpp,v 1.20 2006-12-20 09:54:18 dockes Exp $ (C) 2004 J.F.Dockes";
+static char rcsid[] = "@(#$Id: mimetype.cpp,v 1.21 2008-07-01 10:29:45 dockes Exp $ (C) 2004 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -57,13 +57,11 @@ using namespace std;
 
 static string mimetypefromdata(const string &fn, bool usfc)
 {
-    string mime;
-
     // In any case first try the internal identifier
-    mime = idFile(fn.c_str());
+    string mime = idFile(fn.c_str());
 
 #ifdef USE_SYSTEM_FILE_COMMAND
-    if (usfc && mime == "") {
+    if (usfc && mime.empty()) {
 	// Last resort: use "file -i"
 	list<string> args;
 
@@ -75,7 +73,7 @@ static string mimetypefromdata(const string &fn, bool usfc)
 	int status = ex.doexec(cmd, args, 0, &result);
 	if (status) {
 	    LOGERR(("mimetypefromdata: doexec: status 0x%x\n", status));
-	    return "";
+	    return string();
 	}
 	// LOGDEB(("mimetypefromdata: %s [%s]\n", result.c_str(), fn.c_str()));
 
@@ -88,7 +86,7 @@ static string mimetypefromdata(const string &fn, bool usfc)
 	list<string> res;
 	stringToStrings(result, res);
 	if (res.size() <= 1)
-	    return "";
+	    return string();
 	list<string>::iterator it = res.begin();
 	mime = *++it;
 	// Remove possible punctuation at the end
@@ -96,7 +94,7 @@ static string mimetypefromdata(const string &fn, bool usfc)
 	    mime.erase(mime.length() -1);
 	// File -i will sometimes return strange stuff (ie: "very small file")
 	if(mime.find("/") == string::npos) 
-	    mime = "";
+	    mime.clear();
     }
 #endif
 
@@ -109,17 +107,18 @@ static string mimetypefromdata(const string &fn, bool usfc)
 string mimetype(const string &fn, const struct stat *stp,
 		RclConfig *cfg, bool usfc)
 {
-    if (S_ISDIR(stp->st_mode))
-	return "application/x-fsdirectory";
-    if (!S_ISREG(stp->st_mode))
-	return "application/x-fsspecial";
-
+    if (stp) {
+	if (S_ISDIR(stp->st_mode))
+	    return "application/x-fsdirectory";
+	if (!S_ISREG(stp->st_mode))
+	    return "application/x-fsspecial";
+    }
     if (cfg == 0)
-	return "";
+	return string();
 
     if (cfg->inStopSuffixes(fn)) {
 	LOGDEB(("mimetype: fn [%s] in stopsuffixes\n", fn.c_str()));
-	return "";
+	return string();
     }
 
     // First look for suffix in mimetype map
@@ -135,7 +134,9 @@ string mimetype(const string &fn, const struct stat *stp,
 	    return mtype;
     }
 
-    // Then examine data
+    // Finally examine data
+    if (!stp)
+	return string();
     return mimetypefromdata(fn, usfc);
 }
 

src/index/mimetype.h

@@ -16,16 +16,22 @@
  */
 #ifndef _MIMETYPE_H_INCLUDED_
 #define _MIMETYPE_H_INCLUDED_
-/* @(#$Id: mimetype.h,v 1.6 2006-12-19 08:40:50 dockes Exp $  (C) 2004 J.F.Dockes */
+/* @(#$Id: mimetype.h,v 1.7 2008-07-01 10:29:45 dockes Exp $  (C) 2004 J.F.Dockes */
 
 #include <string>
 
 class RclConfig;
 struct stat;
+
 /**
- * Try to determine a mime type for filename. 
- * This may imply more than matching the suffix, the name must be usable
- * to actually access file data.
+ * Try to determine a mime type for file. 
+ *
+ * If stp is not null, this may imply more than matching the suffix,
+ * the name must be usable to actually access file data.
+ * @param filename file/path name to use
+ * @param stp if not null use st_mode bits for directories etc.
+ * @param cfg recoll config
+ * @param usfc Use system's 'file' command as last resort (or not)
  */
 string mimetype(const std::string &filename, const struct stat *stp,
 		RclConfig *cfg, bool usfc);

src/internfile/mh_mail.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: mh_mail.cpp,v 1.31 2007-12-13 06:58:21 dockes Exp $ (C) 2005 J.F.Dockes";
+static char rcsid[] = "@(#$Id: mh_mail.cpp,v 1.32 2008-07-01 10:29:45 dockes Exp $ (C) 2005 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -36,6 +36,8 @@ static char rcsid[] = "@(#$Id: mh_mail.cpp,v 1.31 2007-12-13 06:58:21 dockes Exp
 #include "debuglog.h"
 #include "smallut.h"
 #include "mh_html.h"
+#include "rclconfig.h"
+#include "mimetype.h"
 
 // binc imap mime definitions
 #include "mime.h"
@@ -103,7 +105,7 @@ bool MimeHandlerMail::skip_to_document(const string& ipath)
     LOGDEB(("MimeHandlerMail::skip_to_document(%s)\n", ipath.c_str()));
     if (m_idx == -1) {
 	// No decoding done yet. If ipath is null need do nothing
-	if (ipath == "" || ipath == "-1")
+	if (ipath.empty() || ipath == "-1")
 	    return true;
 	// ipath points to attachment: need to decode message
 	if (!next_document()) {
@@ -189,7 +191,7 @@ bool MimeHandlerMail::processAttach()
 	    att->m_charset.c_str(),
 	    att->m_filename.c_str()));
 
-    m_metaData["content"] = "";
+    m_metaData["content"] = string();
     string& body = m_metaData["content"];
     att->m_part->getBody(body, 0, att->m_part->bodylength);
     string decoded;
@@ -215,6 +217,16 @@ bool MimeHandlerMail::processAttach()
 	}
     }
 
+    // Special case for application/octet-stream: try to better
+    // identify content, using file name if set
+    if (m_metaData["mimetype"] == "application/octet-stream" &&
+	!m_metaData["filename"].empty()) {
+	string mt = mimetype(m_metaData["filename"], 0,	
+			     RclConfig::getMainConfig(), false);
+	if (!mt.empty()) 
+	    m_metaData["mimetype"] = mt;
+    }
+
     // Ipath
     char nbuf[10];
     sprintf(nbuf, "%d", m_idx);