more safexx posix stuff

src/common/rclconfig.cpp

@@ -22,6 +22,7 @@
 #include <langinfo.h>
 #include <limits.h>
 #include "safesysstat.h"
+#include "safeunistd.h"
 #include <sys/param.h>
 #ifdef __FreeBSD__
 #include <osreldate.h>
@@ -179,7 +180,7 @@ RclConfig::RclConfig(const string *argcnf)
     // want to avoid the imperfect test in isDefaultConfig() if we actually know
     // this is the default conf
     if (!autoconfdir && !isDefaultConfig()) {
-	if (access(m_confdir.c_str(), 0) < 0) {
+	if (!path_exists(m_confdir)) {
 	    m_reason = "Explicitly specified configuration "
 		"directory must exist"
 		" (won't be automatically created). Use mkdir first";
@@ -187,7 +188,7 @@ RclConfig::RclConfig(const string *argcnf)
 	}
     }
 
-    if (access(m_confdir.c_str(), 0) < 0) {
+    if (!path_exists(m_confdir)) {
 	if (!initUserConfig()) 
 	    return;
     }
@@ -1401,7 +1402,7 @@ bool RclConfig::initUserConfig()
 
     // Use protective 700 mode to create the top configuration
     // directory: documents can be reconstructed from index data.
-    if (access(m_confdir.c_str(), 0) < 0 && 
+    if (!path_exists(m_confdir) && 
 	mkdir(m_confdir.c_str(), 0700) < 0) {
 	m_reason += string("mkdir(") + m_confdir + ") failed: " + 
 	    strerror(errno);
@@ -1410,7 +1411,7 @@ bool RclConfig::initUserConfig()
     string lang = localelang();
     for (int i = 0; i < ncffiles; i++) {
 	string dst = path_cat(m_confdir, string(configfiles[i])); 
-	if (access(dst.c_str(), 0) < 0) {
+	if (!path_exists(dst)) {
 	    FILE *fp = fopen(dst.c_str(), "w");
 	    if (fp) {
 		fprintf(fp, "%s\n", blurb);

src/common/syngroups.cpp

@@ -30,6 +30,16 @@
 
 using namespace std;
 
+// Note that we are storing each term twice. I don't think that the
+// size could possibly be a serious issue, but if it was, we could
+// reduce the storage a bit by storing (short hash)-> vector<int>
+// correspondances in the direct map, and then checking all the
+// resulting groups for the input word.
+//
+// As it is, a word can only index one group (the last it is found
+// in). It can be part of several groups though (appear in
+// expansions). I really don't know what we should do with multiple
+// groups anyway
 class SynGroups::Internal {
 public:
     Internal() : ok(false) {
@@ -40,10 +50,12 @@ public:
     // Group num to group
     STD_UNORDERED_MAP<int, vector<string> > groups;
 };
+
 bool SynGroups::ok() 
 {
     return m && m->ok;
 }
+
 SynGroups::~SynGroups()
 {
     delete m;
@@ -215,7 +227,7 @@ int main(int argc, char **argv)
     cout << group.size() << " terms in group\n";
     for (vector<string>::const_iterator it = group.begin();
 	 it != group.end(); it++) {
-	cout << *it << " ";
+	cout << "[" << *it << "] ";
     }
     cout << endl;
     return 0;

src/index/beaglequeue.cpp

@@ -19,6 +19,7 @@
 #include <string.h>
 #include <errno.h>
 #include "safesysstat.h"
+#include "safeunistd.h"
 
 #include "cstr.h"
 #include "pathut.h"

src/index/rclmonprc.cpp

@@ -27,6 +27,7 @@
 #include <pthread.h>
 #include <errno.h>
 #include <fnmatch.h>
+#include "safeunistd.h"
 
 #include <cstring>
 #include <cstdio>
@@ -439,7 +440,7 @@ bool RclMonEventQueue::pushEvent(const RclMonEvent &ev)
 static bool checkfileanddelete(const string& fname)
 {
     bool ret;
-    ret = access(fname.c_str(), 0) == 0;
+    ret = path_exists(fname);
     unlink(fname.c_str());
     return ret;
 }

src/index/rclmonrcv.cpp

@@ -22,6 +22,7 @@
 #include <cstdio>
 #include <cstring>
 #include "safesysstat.h"
+#include "safeunistd.h"
 
 #include "debuglog.h"
 #include "rclmon.h"

src/index/recollindex.cpp

@@ -24,6 +24,7 @@
 #include <sys/time.h>
 #include <sys/resource.h>
 #include <fcntl.h>
+#include "safeunistd.h"
 
 #include <iostream>
 #include <list>

src/internfile/mh_mbox.cpp

@@ -294,7 +294,7 @@ bool MimeHandlerMbox::set_document_file(const string& mt, const string &fn)
 
     // And double check for thunderbird 
     string tbirdmsf = fn + ".msf";
-    if ((m_quirks&MBOXQUIRK_TBIRD) == 0 && access(tbirdmsf.c_str(), 0) == 0) {
+    if ((m_quirks&MBOXQUIRK_TBIRD) == 0 && path_exists(tbirdmsf)) {
 	LOGDEB(("MimeHandlerMbox: detected unconfigured tbird mbox in %s\n",
 		fn.c_str()));
 	m_quirks |= MBOXQUIRK_TBIRD;

src/qtgui/fragbuts.cpp

@@ -148,7 +148,7 @@ FragButs::FragButs(QWidget* parent)
     m_fn = path_cat(theconfig->getConfDir(), "fragbuts.xml");
 
     string data, reason;
-    if (access(m_fn.c_str(), 0) != 0) {
+    if (!path_exists(m_fn)) {
         // config does not exist: try to create it from sample
         string src = path_cat(theconfig->getDatadir(), "examples");
         src = path_cat(src, "fragbuts.xml");

src/qtgui/multisave.cpp

@@ -123,7 +123,7 @@ void multiSave(QWidget *p, vector<Rcl::Doc>& docs)
     
     for (unsigned int i = 0; i != docs.size(); i++) {
 	string fn = path_cat(dir, filenames[i]);
-	if (access(fn.c_str(), 0) == 0) {
+	if (path_exists(fn)) {
 	    QMessageBox::warning(0, "Recoll",
 				 QWidget::tr("Unexpected file name collision, "
 				       "cancelling."));

src/qtgui/preview_w.cpp

@@ -14,6 +14,9 @@
  *   Free Software Foundation, Inc.,
  *   59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
  */
+#include "autoconfig.h"
+
+#include "safeunistd.h"
 
 #include <list>
 #include <utility>

src/qtgui/rclm_saveload.cpp

@@ -41,7 +41,7 @@ static QString prevDir()
         settings.value("/Recoll/prefs/lastQuerySaveDir").toString();
     string defpath = path_cat(theconfig->getConfDir(), "saved_queries");
     if (prevdir.isEmpty()) {
-        if (access(defpath.c_str(), 0) != 0) {
+        if (!path_exists(defpath)) {
             mkdir(defpath.c_str(), 0700);
         }
         return QString::fromLocal8Bit(defpath.c_str());

src/qtgui/rclm_view.cpp

@@ -16,6 +16,8 @@
  */
 #include "autoconfig.h"
 
+#include "safeunistd.h"
+
 #include <list>
 
 #include <QMessageBox>

src/qtgui/rclmain_w.cpp

@@ -17,6 +17,7 @@
 #include "autoconfig.h"
 
 #include <fcntl.h>
+#include "safeunistd.h"
 
 #include <utility>
 #include MEMORY_INCLUDE

src/qtgui/rtitool.cpp

@@ -18,6 +18,7 @@
 
 #include <stdio.h>
 #include "safesysstat.h"
+#include "safeunistd.h"
 
 #include <string>
 
@@ -60,7 +61,7 @@ void RTIToolW::init()
     connect(this->sesCB, SIGNAL(clicked(bool)), 
 	    this, SLOT(sesclicked(bool)));
     string autostartfile = path_cat(path_home(), rautostartfile);
-    if (access(autostartfile.c_str(), 0) == 0) {
+    if (path_exists(autostartfile)) {
 	sesCB->setChecked(true);
     }
 }
@@ -80,7 +81,7 @@ void RTIToolW::accept()
     if (sesCB->isChecked()) {
 	// Setting up daemon indexing autostart
 
-	if (::access(autostartfile.c_str(), 0) == 0) {
+	if (path_exists(autostartfile)) {
 	    QString msg = tr("Replacing: ") + 
 		QString::fromLocal8Bit(autostartfile.c_str());
 	
@@ -96,7 +97,7 @@ void RTIToolW::accept()
 	if (theconfig) {
 	    string sourcefile = path_cat(theconfig->getDatadir(), "examples");
 	    sourcefile = path_cat(sourcefile, "recollindex.desktop");
-	    if (::access(sourcefile.c_str(), 0) == 0) {
+	    if (path_exists(sourcefile)) {
 		file_to_string(sourcefile, text);
 	    }
 	}
@@ -138,7 +139,7 @@ void RTIToolW::accept()
 	exitdial = true;
     } else {
 	// Turning autostart off
-	if (::access(autostartfile.c_str(), 0) == 0) {
+	if (path_exists(autostartfile)) {
 	    QString msg = tr("Deleting: ") + 
 		QString::fromLocal8Bit(autostartfile.c_str());
 	

src/qtgui/uiprefs_w.cpp

@@ -565,7 +565,7 @@ void UIPrefsDialog::addExtraDbPB_clicked()
     if (input.isEmpty())
 	return;
     string dbdir = (const char *)input.toLocal8Bit();
-    if (access(path_cat(dbdir, "recoll.conf").c_str(), 0) == 0) {
+    if (path_exists(path_cat(dbdir, "recoll.conf"))) {
 	// Chosen dir is config dir.
 	RclConfig conf(&dbdir);
 	dbdir = conf.getDbDir();

src/utils/conftree.cpp

@@ -172,7 +172,7 @@ ConfSimple::ConfSimple(const char *fname, int readonly, bool tildexp)
 	// It seems that there is no separate 'create if not exists' 
 	// open flag. Have to truncate to create, but dont want to do 
 	// this to an existing file !
-	if (access(fname, 0) < 0) {
+	if (!path_exists(fname)) {
 	    mode |= ios::trunc;
 	}
 	input.open(fname, mode);

src/utils/pathut.cpp

@@ -433,6 +433,11 @@ long long path_filesize(const string& path)
     return (long long)st.st_size;
 }
 
+bool path_exists(const string& path)
+{
+    return access(path.c_str(), 0) == 0;
+}
+
 // Allowed punctuation in the path part of an URI according to RFC2396
 // -_.!~*'():@&=+$,
 /*

src/utils/pathut.h

@@ -18,8 +18,6 @@
 #define _PATHUT_H_INCLUDED_
 #include "autoconfig.h"
 
-#include <unistd.h>
-
 #include <string>
 #include <vector>
 #include <set>
@@ -75,6 +73,11 @@ extern bool path_isdir(const std::string& path);
 /// Retrieve file size
 extern long long path_filesize(const std::string& path);
 
+/// Check that path is traversable and last element exists
+/// Returns true if last elt could be checked to exist. False may mean that
+/// the file/dir does not exist or that an error occurred. 
+extern bool path_exists(const std::string& path);
+
 /// Dump directory
 extern bool readdir(const std::string& dir, std::string& reason, 
 		    std::set<std::string>& entries);