uncompression+linux port

src/internfile/internfile.cpp

@@ -0,0 +1,165 @@
+#ifndef lint
+static char rcsid[] = "@(#$Id: internfile.cpp,v 1.1 2005-02-04 14:21:17 dockes Exp $ (C) 2004 J.F.Dockes";
+#endif
+#include <unistd.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <errno.h>
+
+#include <string>
+#include <iostream>
+using namespace std;
+
+#include "internfile.h"
+#include "mimetype.h"
+#include "debuglog.h"
+#include "mimehandler.h"
+#include "execmd.h"
+#include "pathut.h"
+
+static bool uncompressfile(RclConfig *conf, const string& ifn, 
+			   const list<string>& cmdv, string& tdir, 
+			   string& tfile)
+{
+    const char *tmpdir = getenv("RECOLL_TMPDIR");
+    if (!tmpdir)
+	tmpdir = getenv("TMPDIR");
+    if (!tmpdir)
+	tmpdir = "/tmp";
+    tdir = tmpdir;
+    path_cat(tdir, "rcltmpXXXXXX");
+    {
+	char *cp = strdup(tdir.c_str());
+	if (!cp) {
+	    LOGERR(("uncompressfile: out of memory (for file name !)\n"));
+	    return false;
+	}	
+	if (!mktemp(cp)) {
+	    free(cp);
+	    LOGERR(("uncompressfile: mktemp failed\n"));
+	    return false;
+	}	
+	tdir = cp;
+	free(cp);
+    }
+
+    if (mkdir(tdir.c_str(), 0700) < 0) {
+	LOGERR(("uncompressfile: mkdir %s failed\n", tdir.c_str()));
+	return false;
+    }
+
+    string cmd = find_filter(conf, cmdv.front());
+
+    // Substitute file name and temp dir in command elements
+    list<string>::const_iterator it = cmdv.begin();
+    ++it;
+    list<string> args;
+    for (; it != cmdv.end(); it++) {
+	string s = *it;
+	string ns;
+	string::const_iterator it1;
+	for (it1 = s.begin(); it1 != s.end();it1++) {
+	    if (*it1 == '%') {
+		if (++it1 == s.end()) {
+		    ns += '%';
+		    break;
+		}
+		if (*it1 == '%')
+		    ns += '%';
+		if (*it1 == 'f')
+		    ns += ifn;
+		if (*it1 == 't')
+		    ns += tdir;
+	    } else {
+		ns += *it1;
+	    }
+	}
+	args.push_back(ns);
+    }
+
+    // Execute command and retrieve output file name, check that it exists
+    ExecCmd ex;
+    int status = ex.doexec(cmd, args, 0, &tfile);
+    if (status) {
+	LOGERR(("uncompressfile: doexec: status 0x%x\n", status));
+	rmdir(tdir.c_str());
+	return false;
+    }
+    if (!tfile.empty() && tfile[tfile.length() - 1] == '\n')
+	tfile.erase(tfile.length() - 1, 1);
+    return true;
+}
+
+static void tmpcleanup(const string& tdir, const string& tfile)
+{
+    if (tdir.empty())
+	return;
+    if (!tfile.empty()) {
+	if (unlink(tfile.c_str()) < 0) {
+	    LOGERR(("tmpcleanup: unlink(%s) errno %d\n", tfile.c_str(), 
+		    errno));
+	    return;
+	}
+    }
+    if (rmdir(tdir.c_str()) < 0) {
+	LOGERR(("tmpcleanup: rmdir(%s) errno %d\n", tdir.c_str(), errno));
+	return;
+    }
+}
+
+bool internfile(const std::string &ifn, RclConfig *config, Rcl::Doc& doc)
+{
+    string fn = ifn;
+    string tdir;
+    string tfile;
+    MimeHandler *handler = 0;
+    bool ret = false;
+
+    string mime = mimetype(fn, config->getMimeMap());
+    if (mime.empty()) {
+	// No mime type ?? pass on.
+	LOGDEB(("internfile: (no mime) [%s]\n", fn.c_str()));
+	return false;
+    }
+
+    // First check for a compressed file
+    list<string>ucmd;
+    if (getUncompressor(mime, config->getMimeConf(), ucmd)) {
+	if (!uncompressfile(config, fn, ucmd, tdir, tfile)) 
+	    return false;
+	LOGDEB(("internfile: after ucomp: tdir %s, tfile %s\n", 
+		tdir.c_str(), tfile.c_str()));
+	fn = tfile;
+	mime = mimetype(fn, config->getMimeMap());
+	if (mime.empty()) {
+	    // No mime type ?? pass on.
+	    LOGDEB(("internfile: (no mime) [%s]\n", fn.c_str()));
+	    goto out;
+	}
+
+    }
+    
+    // Look for appropriate handler
+    handler = getMimeHandler(mime, config->getMimeConf());
+    if (!handler) {
+	// No handler for this type, for now :(
+	LOGDEB(("internfile: %s : no handler\n", mime.c_str()));
+	goto out;
+    }
+
+    LOGDEB(("internfile: %s [%s]\n", mime.c_str(), fn.c_str()));
+
+    // Turn file into a document. The document has fields for title, body 
+    // etc.,  all text converted to utf8
+    if (!handler->worker(config, fn,  mime, doc)) {
+	goto out;
+    }
+    doc.mimetype = mime;
+
+    // Clean up. We delete the temp file and its father directory
+    ret = true;
+ out:
+    delete handler;
+    tmpcleanup(tdir, tfile);
+    return ret;
+}

src/internfile/internfile.h

@@ -0,0 +1,14 @@
+#ifndef _INTERNFILE_H_INCLUDED_
+#define _INTERNFILE_H_INCLUDED_
+/* @(#$Id: internfile.h,v 1.1 2005-02-04 14:21:17 dockes Exp $  (C) 2004 J.F.Dockes */
+
+#include <string>
+
+#include "rclconfig.h"
+#include "rcldb.h"
+
+/// Turn external file into internal representation, according to mime type etc
+extern bool internfile(const std::string &fn, RclConfig *config, 
+		       Rcl::Doc& doc);
+
+#endif /* _INTERNFILE_H_INCLUDED_ */

src/internfile/mimehandler.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: mimehandler.cpp,v 1.7 2005-02-04 09:39:44 dockes Exp $ (C) 2004 J.F.Dockes";
+static char rcsid[] = "@(#$Id: mimehandler.cpp,v 1.8 2005-02-04 14:21:17 dockes Exp $ (C) 2004 J.F.Dockes";
 #endif
 
 #include <iostream>
@@ -71,6 +71,12 @@ class MimeHandlerExec : public MimeHandler {
 bool MimeHandlerExec::worker(RclConfig *conf, const string &fn, 
 			     const string &mtype, Rcl::Doc &docout)
 {
+    if (params.empty()) {
+	// Hu ho
+	LOGERR(("MimeHandlerExec::worker: empty params for mime %s\n",
+		mtype.c_str()));
+	return false;
+    }
     // Command name
     string cmd = find_filter(conf, params.front());
     
@@ -163,8 +169,14 @@ bool getUncompressor(const std::string &mtype, ConfTree *mhandlers,
     string hs;
 
     mhandlers->get(mtype, hs, "");
+    if (hs.empty())
+	return false;
     list<string> tokens;
     ConfTree::stringToStrings(hs, tokens);
+    if (tokens.empty()) {
+	LOGERR(("getUncompressor: empty spec for mtype %s\n", mtype.c_str()));
+	return false;
+    }
     if (stringlowercmp("uncompress", tokens.front())) 
 	return false;
     list<string>::iterator it = tokens.begin();