Disable filters with missing helpers for the whole indexing pass

src/internfile/mh_exec.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: mh_exec.cpp,v 1.12 2008-10-04 14:26:59 dockes Exp $ (C) 2005 J.F.Dockes";
+static char rcsid[] = "@(#$Id: mh_exec.cpp,v 1.13 2008-10-06 06:22:46 dockes Exp $ (C) 2005 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -23,6 +23,10 @@ static char rcsid[] = "@(#$Id: mh_exec.cpp,v 1.12 2008-10-04 14:26:59 dockes Exp
 #include "mh_html.h"
 #include "debuglog.h"
 #include "cancelcheck.h"
+#include "smallut.h"
+
+#include <sys/types.h>
+#include <sys/wait.h>
 
 #ifndef NO_NAMESPACES
 using namespace std;
@@ -42,6 +46,10 @@ bool MimeHandlerExec::next_document()
     if (m_havedoc == false)
 	return false;
     m_havedoc = false;
+    if (missingHelper) {
+	LOGDEB(("MimeHandlerExec::next_document(): helper known missing\n"));
+	return false;
+    }
     if (params.empty()) {
 	// Hu ho
 	LOGERR(("MimeHandlerExec::mkDoc: empty params\n"));
@@ -59,7 +67,7 @@ bool MimeHandlerExec::next_document()
     if (!m_ipath.empty())
 	myparams.push_back(m_ipath);
 
-    // Execute command and store the result text
+    // Execute command, store the output
     string& output = m_metaData["content"];
     output.erase();
     ExecCmd mexec;
@@ -68,15 +76,37 @@ bool MimeHandlerExec::next_document()
     mexec.putenv(m_forPreview ? "RECOLL_FILTER_FORPREVIEW=yes" :
 		"RECOLL_FILTER_FORPREVIEW=no");
     int status = mexec.doexec(cmd, myparams, 0, &output);
+
     if (status) {
-	LOGERR(("MimeHandlerExec: command status 0x%x: %s\n", 
+	LOGERR(("MimeHandlerExec: command status 0x%x for %s\n", 
 		status, cmd.c_str()));
-	// If the output string begins with RECFILTERROR, then it's 
+	if (WIFEXITED(status) && WEXITSTATUS(status) == 127) {
-	// interpretable error information
+	    // That's how execmd signals a failed exec (most probably
-	if (output.find("RECFILTERROR") == 0)
+	    // a missing command). Let'hope no filter uses the same value as
+	    // an exit status... Disable myself permanently and signal the 
+	    // missing cmd.
+	    missingHelper = true;
+	    m_reason = string("RECFILTERROR HELPERNOTFOUND ") + cmd;
+	} else if (output.find("RECFILTERROR") == 0) {
+	    // If the output string begins with RECFILTERROR, then it's 
+	    // interpretable error information out from a recoll script
 	    m_reason = output;
+	    list<string> lerr;
+	    stringToStrings(output, lerr);
+	    if (lerr.size() > 2) {
+		list<string>::iterator it = lerr.begin();
+		it++;
+		if (*it == "HELPERNOTFOUND") {
+		    // No use trying again and again to execute this filter, 
+		    // it won't work.
+		    missingHelper = true;
+		}
+	    }		    
+	}
 	return false;
     }
+
+    // Success. Store some external metadata
     m_metaData["origcharset"] = m_defcharset;
     // Default charset: all recoll filters output utf-8, but this
     // could still be overridden by the content-type meta tag.

src/internfile/mh_exec.h

@@ -16,7 +16,7 @@
  */
 #ifndef _MH_EXEC_H_INCLUDED_
 #define _MH_EXEC_H_INCLUDED_
-/* @(#$Id: mh_exec.h,v 1.7 2008-10-04 14:26:59 dockes Exp $  (C) 2004 J.F.Dockes */
+/* @(#$Id: mh_exec.h,v 1.8 2008-10-06 06:22:46 dockes Exp $  (C) 2004 J.F.Dockes */
 
 #include <string>
 #include <list>
@@ -33,8 +33,10 @@ using std::string;
  */
 class MimeHandlerExec : public RecollFilter {
  public:
-    // params, cfgMtype and chgCharset do not get reset by
+    // Members not reset by clear().  params, cfgMtype and chgCharset
-    // clear(). They define what I am
+    // actually define what I am.  missingHelper is a permanent error
+    // (no use to try and execute over and over something that's not
+    // here).
     list<string> params;
     // The defaults for external filters is to output html except if defined 
     // otherwise in the config.
@@ -42,8 +44,10 @@ class MimeHandlerExec : public RecollFilter {
     // For ext programs which don't output html, the output charset
     // has to be known: ie they have a --charset utf-8 like option.
     string cfgCharset; 
+    bool missingHelper;
 
-    MimeHandlerExec(const string& mt) : RecollFilter(mt) {}
+    MimeHandlerExec(const string& mt) : RecollFilter(mt), missingHelper(false) 
+    {}
     virtual ~MimeHandlerExec() {}
     virtual bool set_document_file(const string &file_path) {
 	m_fn = file_path;

src/internfile/mimehandler.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: mimehandler.cpp,v 1.23 2008-10-04 14:26:59 dockes Exp $ (C) 2004 J.F.Dockes";
+static char rcsid[] = "@(#$Id: mimehandler.cpp,v 1.24 2008-10-06 06:22:46 dockes Exp $ (C) 2004 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -35,8 +35,9 @@ using namespace std;
 #include "mh_text.h"
 #include "mh_unknown.h"
 
-// Pool of already known and created handlers
+// Pool of already known and created handlers. There can be several instance
-static map<string, Dijon::Filter*>  o_handlers;
+// for a given mime type (think email attachment in email message)
+static multimap<string, Dijon::Filter*>  o_handlers;
 
 /** Create internal handler object appropriate for given mime type */
 static Dijon::Filter *mhFactory(const string &mime)
@@ -127,9 +128,10 @@ MimeHandlerExec *mhExecFactory(RclConfig *cfg, const string& mtype, string& hs)
 /* Return mime handler to pool */
 void returnMimeHandler(Dijon::Filter *handler)
 {
+    typedef multimap<string, Dijon::Filter*>::value_type value_type;
     if (handler) {
 	handler->clear();
-	o_handlers[handler->get_mime_type()] = handler;
+	o_handlers.insert(value_type(handler->get_mime_type(), handler));
     }
 }
 

src/utils/execmd.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: execmd.cpp,v 1.26 2007-11-08 09:34:40 dockes Exp $ (C) 2004 J.F.Dockes";
+static char rcsid[] = "@(#$Id: execmd.cpp,v 1.27 2008-10-06 06:22:47 dockes Exp $ (C) 2004 J.F.Dockes";
 #endif
 /*
  *   This program is free software; you can redistribute it and/or modify
@@ -374,7 +374,7 @@ int ExecCmd::doexec(const string &cmd, const list<string>& args,
 	// Hu ho
 	LOGERR(("ExecCmd::doexec: execvp(%s) failed. errno %d\n", cmd.c_str(),
 		errno));
-	_exit(128);
+	_exit(127);
     }
     /* This cant be reached: to make cc happy */
     return -1;