added optional extended file attributes support

src/internfile/internfile.cpp

@@ -19,6 +19,7 @@ static char rcsid[] = "@(#$Id: internfile.cpp,v 1.46 2008-10-10 08:04:54 dockes
  */
 
 #ifndef TEST_INTERNFILE
+#include "autoconfig.h"
 
 #include <unistd.h>
 #include <fcntl.h>
@@ -132,7 +133,7 @@ void FileInterner::tmpcleanup()
 //
 // Empty handler on return says that we're in error, this will be
 // processed by the first call to internfile().
-FileInterner::FileInterner(const std::string &f, const struct stat *stp,
+FileInterner::FileInterner(const string &f, const struct stat *stp,
 			   RclConfig *cnf, 
 			   const string& td, const string *imime)
     : m_cfg(cnf), m_fn(f), m_forPreview(imime?true:false), m_tdir(td)
@@ -337,7 +338,12 @@ static const string keytt("title");
 bool FileInterner::dijontorcl(Rcl::Doc& doc)
 {
     Dijon::Filter *df = m_handlers.back();
-    const std::map<std::string, std::string>& docdata = df->get_meta_data();
+    if (df == 0) {
+	//??
+	LOGERR(("FileInterner::dijontorcl: null top handler ??\n"));
+	return false;
+    }
+    const map<string, string>& docdata = df->get_meta_data();
 
     for (map<string,string>::const_iterator it = docdata.begin(); 
 	 it != docdata.end(); it++) {
@@ -357,6 +363,18 @@ bool FileInterner::dijontorcl(Rcl::Doc& doc)
 	doc.meta[Rcl::Doc::keyabs] = doc.meta[keyds];
 	doc.meta.erase(keyds);
     }
+#ifdef RCL_USE_XATTR
+    // Finally set any data possibly coming out of the extended file attributes
+    // these override any values from inside the file.
+    RecollFilter *rf = dynamic_cast<RecollFilter*>(df);
+    if (rf != 0) {
+	const map<string, string>& ffa = rf->getFieldsFromAttrs();
+	for (map<string,string>::const_iterator it = ffa.begin(); 
+	     it != ffa.end(); it++) {
+	    doc.meta[it->first] = it->second;
+	}
+    }
+#endif //RCL_USE_XATTR
     return true;
 }
 
@@ -425,7 +443,7 @@ enum addResols {ADD_OK, ADD_CONTINUE, ADD_BREAK, ADD_ERROR};
 // and possibly add a filter/handler to the stack
 int FileInterner::addHandler()
 {
-    const std::map<std::string, std::string>& docdata = 
+    const map<string, string>& docdata = 
 	m_handlers.back()->get_meta_data();
     string charset, mimetype;
     getKeyValue(docdata, keycs, charset);

src/internfile/mh_exec.h

@@ -50,6 +50,7 @@ class MimeHandlerExec : public RecollFilter {
     {}
     virtual ~MimeHandlerExec() {}
     virtual bool set_document_file(const string &file_path) {
+	RecollFilter::set_document_file(file_path);
 	m_fn = file_path;
 	m_havedoc = true;
 	return true;

src/internfile/mh_html.cpp

@@ -41,6 +41,7 @@ using namespace std;
 bool MimeHandlerHtml::set_document_file(const string &fn)
 {
     LOGDEB0(("textHtmlToDoc: %s\n", fn.c_str()));
+    RecollFilter::set_document_file(fn);
     string otext;
     if (!file_to_string(fn, otext)) {
 	LOGINFO(("textHtmlToDoc: cant read: %s\n", fn.c_str()));

src/internfile/mh_mail.cpp

@@ -78,6 +78,7 @@ void MimeHandlerMail::clear()
 bool MimeHandlerMail::set_document_file(const string &fn)
 {
     LOGDEB(("MimeHandlerMail::set_document_file(%s)\n", fn.c_str()));
+    RecollFilter::set_document_file(fn);
     if (m_fd >= 0) {
 	close(m_fd);
 	m_fd = -1;

src/internfile/mh_mbox.cpp

@@ -57,6 +57,7 @@ void MimeHandlerMbox::clear()
 bool MimeHandlerMbox::set_document_file(const string &fn)
 {
     LOGDEB(("MimeHandlerMbox::set_document_file(%s)\n", fn.c_str()));
+    RecollFilter::set_document_file(fn);
     m_fn = fn;
     if (m_vfp) {
 	fclose((FILE *)m_vfp);

src/internfile/mh_text.cpp

@@ -34,6 +34,7 @@ using namespace std;
 // Process a plain text file
 bool MimeHandlerText::set_document_file(const string &fn)
 {
+    RecollFilter::set_document_file(fn);
     string otext;
     string reason;
     if (!file_to_string(fn, otext, &reason)) {

src/internfile/mh_unknown.h

@@ -30,7 +30,8 @@ class MimeHandlerUnknown : public RecollFilter {
  public:
     MimeHandlerUnknown(const string& mt) : RecollFilter(mt) {}
     virtual ~MimeHandlerUnknown() {}
-    virtual bool set_document_string(const string&) {
+    virtual bool set_document_string(const string& fn) {
+	RecollFilter::set_document_file(fn);
 	return m_havedoc = true;
     }
     virtual bool set_document_file(const string&) {

src/internfile/mimehandler.cpp

@@ -17,16 +17,21 @@ static char rcsid[] = "@(#$Id: mimehandler.cpp,v 1.25 2008-10-09 09:19:37 dockes
  *   Free Software Foundation, Inc.,
  *   59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
  */
+#include "autoconfig.h"
 
+#include <errno.h>
 #include <iostream>
 #include <string>
-
+#include <vector>
 using namespace std;
 
 #include "mimehandler.h"
 #include "debuglog.h"
 #include "rclconfig.h"
 #include "smallut.h"
+#ifdef RCL_USE_XATTR
+#include "pxattr.h"
+#endif // RCL_USE_XATTR
 
 #include "mh_exec.h"
 #include "mh_html.h"
@@ -35,6 +40,40 @@ using namespace std;
 #include "mh_text.h"
 #include "mh_unknown.h"
 
+// Common code for all docs that are a file (not subdocs). If extended
+// attributes support is enabled, fetch the data.
+bool RecollFilter::set_document_file(const string& path)
+{
+#ifdef RCL_USE_XATTR
+    RclConfig* rclconfig = RclConfig::getMainConfig();
+    if (rclconfig == 0) {
+	LOGERR(("RecollFilter::set_document_file: no config\n"));
+	return false;
+    }
+    vector<string> xnames;
+    if (!pxattr::list(path, &xnames)) {
+	LOGERR(("xattrToMeta: pxattr::list failed, errno %d\n", errno));
+	return false;
+    }
+    const map<string, string>& xtof = rclconfig->getXattrToField();
+    for (vector<string>::const_iterator it = xnames.begin();
+	 it != xnames.end(); it++) {
+	map<string, string>::const_iterator mit;
+	if ((mit = xtof.find(*it)) != xtof.end()) {
+	    string value;
+	    if (!pxattr::get(path, *it, &value, pxattr::PXATTR_NOFOLLOW)) {
+		LOGERR(("xattrToMeta: pxattr::get failed for %s, errno %d\n", 
+			(*it).c_str(), errno));
+		continue;
+	    }
+	    // Encode should we ?
+	    m_fieldsFromAttrs[mit->second] = value;
+	}
+    }
+#endif // RCL_USE_XATTR
+    return true;
+}
+
 // Pool of already known and created handlers. There can be several instance
 // for a given mime type (think email attachment in email message)
 static multimap<string, Dijon::Filter*>  o_handlers;

src/internfile/mimehandler.h

@@ -17,11 +17,16 @@
 #ifndef _MIMEHANDLER_H_INCLUDED_
 #define _MIMEHANDLER_H_INCLUDED_
 /* @(#$Id: mimehandler.h,v 1.16 2008-10-04 14:26:59 dockes Exp $  (C) 2004 J.F.Dockes */
+#include "autoconfig.h"
 
 #include <string>
 #include <list>
 using std::string;
 using std::list;
+#ifdef RCL_USE_XATTR
+#include <map>
+using std::map;
+#endif // RCL_USE_XATTR
 
 #include <Filter.h>
 
@@ -51,6 +56,10 @@ public:
     // We don't use this for now
     virtual bool set_document_uri(const std::string &) {return false;}
 
+    /// This does the extended attributes thing if enabled and should
+    /// be called from subclasses.
+    virtual bool set_document_file(const string &file_path);
+
     // Default implementations
     virtual bool set_document_string(const std::string &) {return false;}
     virtual bool set_document_data(const char *cp, unsigned int sz) {
@@ -77,15 +86,26 @@ public:
     }
 
     virtual void clear() {
-	m_forPreview = m_havedoc = false;
 	Dijon::Filter::clear();
+	m_forPreview = m_havedoc = false;
+	m_defcharset.clear();
+	m_reason.clear();
+#ifdef RCL_USE_XATTR
+	m_fieldsFromAttrs.clear();
+#endif // RCL_USE_XATTR
     }
+#ifdef RCL_USE_XATTR
+    const map<string, string>& getFieldsFromAttrs() {return m_fieldsFromAttrs;}
+#endif // RCL_USE_XATTR
 
 protected:
     bool   m_forPreview;
     string m_defcharset;
     string m_reason;
     bool   m_havedoc;
+#ifdef RCL_USE_XATTR
+    map<string, string> m_fieldsFromAttrs;
+#endif // RCL_USE_XATTR
 };
 
 /**
@@ -97,11 +117,12 @@ protected:
  *     indexedmimetypes (if this is set at all).
  */
 extern Dijon::Filter *getMimeHandler(const std::string &mtyp, RclConfig *cfg,
-
 				     bool filtertypes=false);
+
 /// Free up filter for reuse (you can also delete it)
 extern void returnMimeHandler(Dijon::Filter *);
 
 /// Can this mime type be interned ?
 extern bool canIntern(const std::string mimetype, RclConfig *cfg);
+
 #endif /* _MIMEHANDLER_H_INCLUDED_ */