moved code around to make smallut and pathut less recoll-specific and reusable. No actual changes

src/rcldb/rcldb.cpp

@@ -42,6 +42,7 @@ using namespace std;
 #include "unacpp.h"
 #include "conftree.h"
 #include "pathut.h"
+#include "rclutil.h"
 #include "smallut.h"
 #include "chrono.h"
 #include "utf8iter.h"
@@ -126,6 +127,21 @@ static inline string make_parentterm(const string& udi)
     return pterm;
 }
 
+static void utf8truncate(string& s, int maxlen)
+{
+    if (s.size() <= string::size_type(maxlen)) {
+        return;
+    }
+    Utf8Iter iter(s);
+    string::size_type pos = 0;
+    while (iter++ != string::npos)
+        if (iter.getBpos() < string::size_type(maxlen)) {
+            pos = iter.getBpos();
+        }
+
+    s.erase(pos);
+}
+
 Db::Native::Native(Db *db) 
     : m_rcldb(db), m_isopen(false), m_iswritable(false),
       m_noversionwrite(false)

src/rcldb/rcldoc.cpp

@@ -14,9 +14,11 @@
  *   Free Software Foundation, Inc.,
  *   59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
  */
+#include "autoconfig.h"
 
 #include "rcldoc.h"
 #include "debuglog.h"
+#include "rclutil.h"
 
 namespace Rcl {
     const string Doc::keyabs("abstract");
@@ -70,5 +72,31 @@ namespace Rcl {
         if (dotext)
             LOGDEB(("Rcl::Doc::dump: text: \n[%s]\n", text.c_str()));
     }
+
+    // Copy ensuring no shared string data, for threading issues.
+    void Doc::copyto(Doc *d) const
+    {
+	d->url.assign(url.begin(), url.end());
+        d->idxurl.assign(idxurl.begin(), idxurl.end());
+        d->idxi = idxi;
+	d->ipath.assign(ipath.begin(), ipath.end());
+	d->mimetype.assign(mimetype.begin(), mimetype.end());
+	d->fmtime.assign(fmtime.begin(), fmtime.end());
+	d->dmtime.assign(dmtime.begin(), dmtime.end());
+	d->origcharset.assign(origcharset.begin(), origcharset.end());
+        map_ss_cp_noshr(meta, &d->meta);
+	d->syntabs = syntabs;
+	d->pcbytes.assign(pcbytes.begin(), pcbytes.end());
+	d->fbytes.assign(fbytes.begin(), fbytes.end());
+	d->dbytes.assign(dbytes.begin(), dbytes.end());
+	d->sig.assign(sig.begin(), sig.end());
+        d->text.assign(text.begin(), text.end());
+	d->pc = pc;
+	d->xdocid = xdocid;
+	d->idxi = idxi;
+	d->haspages = haspages;
+	d->haschildren = haschildren;
+	d->onlyxattr = onlyxattr;
+    }
 }
 

src/rcldb/rcldoc.h

@@ -163,33 +163,11 @@ class Doc {
 	onlyxattr = false;
     }
     // Copy ensuring no shared string data, for threading issues.
-    void copyto(Doc *d) const {
-	d->url.assign(url.begin(), url.end());
-        d->idxurl.assign(idxurl.begin(), idxurl.end());
-        d->idxi = idxi;
-	d->ipath.assign(ipath.begin(), ipath.end());
-	d->mimetype.assign(mimetype.begin(), mimetype.end());
-	d->fmtime.assign(fmtime.begin(), fmtime.end());
-	d->dmtime.assign(dmtime.begin(), dmtime.end());
-	d->origcharset.assign(origcharset.begin(), origcharset.end());
-        map_ss_cp_noshr(meta, &d->meta);
-	d->syntabs = syntabs;
-	d->pcbytes.assign(pcbytes.begin(), pcbytes.end());
-	d->fbytes.assign(fbytes.begin(), fbytes.end());
-	d->dbytes.assign(dbytes.begin(), dbytes.end());
-	d->sig.assign(sig.begin(), sig.end());
-        d->text.assign(text.begin(), text.end());
-	d->pc = pc;
-	d->xdocid = xdocid;
-	d->idxi = idxi;
-	d->haspages = haspages;
-	d->haschildren = haschildren;
-	d->onlyxattr = onlyxattr;
-    }
+    void copyto(Doc *d) const;
+
     Doc()
 	: idxi(0), syntabs(false), pc(0), xdocid(0),
-	  haspages(false), haschildren(false), onlyxattr(false)
-    {
+	  haspages(false), haschildren(false), onlyxattr(false) {
     }
     /** Get value for named field. If value pointer is 0, just test existence */
     bool getmeta(const string& nm, string *value = 0) const