mail ckpt

src/index/Makefile

@@ -1,11 +1,12 @@
 include ../mk/sysconf
 
 BIGLIB = ../lib/librcl.a
+MIMELIB = ../bincimapmime/libmime.a
 
 PROGS = recollindex csguess
 all: $(PROGS)
 
-RECOLLINDEX_OBJS= recollindex.o $(BIGLIB)
+RECOLLINDEX_OBJS= recollindex.o $(BIGLIB) $(MIMELIB)
 recollindex : $(RECOLLINDEX_OBJS)
 	$(CXX) $(CXXFLAGS) -o recollindex $(RECOLLINDEX_OBJS) \
 		$(LIBXAPIAN) $(LIBICONV) $(LIBSYS)

src/index/indexer.cpp

@@ -1,5 +1,5 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: indexer.cpp,v 1.8 2005-03-17 15:35:49 dockes Exp $ (C) 2004 J.F.Dockes";
+static char rcsid[] = "@(#$Id: indexer.cpp,v 1.9 2005-03-25 09:40:27 dockes Exp $ (C) 2004 J.F.Dockes";
 #endif
 #include <stdio.h>
 #include <sys/stat.h>
@@ -146,18 +146,25 @@ DbIndexer::processone(const std::string &fn, const struct stat *stp,
 	return FsTreeWalker::FtwOk;
     }
 
-    Rcl::Doc doc;
-    if (!internfile(fn, config, doc, tmpdir))
-	return FsTreeWalker::FtwOk;
+    FileInterner interner(fn, config, tmpdir);
+    FileInterner::Status fis = FileInterner::FIAgain;
+    while (fis == FileInterner::FIAgain) {
+	Rcl::Doc doc;
+	string ipath;
+	fis = interner.internfile(doc, ipath);
+	if (fis == FileInterner::FIError)
+	    break;
 
-    // Set up common fields:
-    char ascdate[20];
-    sprintf(ascdate, "%ld", long(stp->st_ctime));
-    doc.mtime = ascdate;
+	// Set up common fields:
+	char ascdate[20];
+	sprintf(ascdate, "%ld", long(stp->st_ctime));
+	doc.mtime = ascdate;
+	doc.ipath = ipath;
 
-    // Do database-specific work to update document data
-    if (!db.add(fn, doc)) 
-	return FsTreeWalker::FtwError;
+	// Do database-specific work to update document data
+	if (!db.add(fn, doc)) 
+	    return FsTreeWalker::FtwError;
+    }
 
     return FsTreeWalker::FtwOk;
 }

src/index/mimetype.cpp

@@ -1,14 +1,47 @@
 #ifndef lint
-static char rcsid[] = "@(#$Id: mimetype.cpp,v 1.5 2005-02-09 13:34:08 dockes Exp $ (C) 2004 J.F.Dockes";
+static char rcsid[] = "@(#$Id: mimetype.cpp,v 1.6 2005-03-25 09:40:27 dockes Exp $ (C) 2004 J.F.Dockes";
 #endif
 
 #include <ctype.h>
 
 #include <string>
 using std::string;
+#include <list>
+using std::list;
 
 #include "mimetype.h"
 #include "debuglog.h"
+#include "execmd.h"
+#include "conftree.h"
+
+static string mimetypefromdata(const string &fn)
+{
+    list<string> args;
+
+    args.push_back("-i");
+    args.push_back(fn);
+    ExecCmd ex;
+    string result;
+    string cmd = "file";
+    int status = ex.doexec(cmd, args, 0, &result);
+    if (status) {
+	LOGERR(("mimetypefromdata: doexec: status 0x%x\n", status));
+	return "";
+    }
+    // LOGDEB(("mimetypefromdata: %s [%s]\n", result.c_str(), fn.c_str()));
+    list<string> res;
+    ConfTree::stringToStrings(result, res);
+    if (res.size() <= 1) 
+	return "";
+    list<string>::iterator it = res.begin();
+    it++;
+    string mime = *it;
+
+    if (mime.length() > 0 && !isalpha(mime[mime.length() - 1]))
+	mime.erase(mime.length() -1);
+
+    return mime;
+}
 
 string mimetype(const string &fn, ConfTree *mtypes)
 {
@@ -38,8 +71,9 @@ string mimetype(const string &fn, ConfTree *mtypes)
 
     // If the file name has a suffix and we find it in the map, we're done
     string::size_type dot = fn.find_last_of(".");
+    string suff;
     if (dot != string::npos) {
-	string suff = fn.substr(dot);
+        suff = fn.substr(dot);
 	for (unsigned int i = 0; i < suff.length(); i++)
 	    suff[i] = tolower(suff[i]);
 
@@ -48,7 +82,9 @@ string mimetype(const string &fn, ConfTree *mtypes)
 	    return mtype;
     }
 
-    // Look at file data ? One day maybe
+    // Look at file data ? Only when no suffix
+    if (suff.empty())
+	return mimetypefromdata(fn);
     return "";
 }