Changeset 3322

Timestamp:

2002-08-05T15:15:18+12:00 (22 years ago)

Author:

sjboddie

Message:

Attempted to improve search term highlighting. Simple phrase searches will
now be highlighted correctly rather than simply highlighting all terms that
appear in the phrase. There are still many cases where highlighting won't
work quite as expected though, e.g. when stemming is turned on and there
are stemmed terms within the phrase and when there is more than one phrase
in the query string.

Location:

trunk/gsdl/src/recpt

Files:

• Makefile.in (modified) (3 diffs)
• documentaction.cpp (modified) (9 diffs)
• documentaction.h (modified) (2 diffs)
• highlighttext.cpp (added)
• highlighttext.h (added)
• win32.mak (modified) (3 diffs)

trunk/gsdl/src/recpt/Makefile.in

@@ -138 +138 @@
         datelistbrowserclass.h invbrowserclass.h pagedbrowserclass.h \
         htmlbrowserclass.h phindbrowserclass.h historydb.h collectoraction.h \
-        phindaction.h summarise.h
+        phindaction.h summarise.h highlighttext.h
 
 RECEPTHEADERS = 
@@ -160 +160 @@
         invbrowserclass.cpp pagedbrowserclass.cpp htmlbrowserclass.cpp \
         phindbrowserclass.cpp  historydb.cpp collectoraction.cpp argdb.cpp \
-        browseactiontools.h formattools.cpp phindaction.cpp summarise.cpp
+        browseactiontools.h formattools.cpp phindaction.cpp summarise.cpp \
+        highlighttext.cpp
 
 
@@ -192 +193 @@
         pagedbrowserclass.o htmlbrowserclass.o phindbrowserclass.o \
         historydb.o $(Z_COMOBJS) collectoraction.o phindaction.o \
-        summarise.o
+        summarise.o highlighttext.o
 
 RECPTOBJECTS = recptmain.o

trunk/gsdl/src/recpt/documentaction.cpp

@@ -31 +31 @@
 #include "unitool.h"
 #include "gsdltools.h"
+#include "highlighttext.h"
 
 documentaction::documentaction () {
@@ -106 +107 @@
 
   // in this action "hl" is the "highlighting on/
-  // highlighting off control
+  // highlighting off control ("hl" == "2" is a
+  // special case that forces phrase highlighting
+  // to be used even if the query string doesn't
+  // appear to be a phrase)
   arg_ainfo.shortname = "hl";
   arg_ainfo.longname = "highlighting on/off";
@@ -181 +185 @@
   // check hl argument
   int arg_hl = args.getintarg("hl");
-  if (arg_hl != 0 && arg_hl != 1) {
+  if (arg_hl < 0 || arg_hl > 2) {
     logout << "Warning: \"hl\" argument out of range (" << arg_hl << ")\n";
     cgiarginfo *hlinfo = argsinfo.getarginfo ("hl");
@@ -782 +786 @@
                   text_t &collection, recptproto *collectproto, 
                   displayclass &disp, outconvertclass &outconvert, 
-                  ostream &textout, ostream &logout) {
+                  ostream &textout, ostream &logout,
+                  cgiargsclass &args) {
   
   DocumentRequest_t docrequest;
@@ -798 +803 @@
       // (wanttext will equal 2 if we want text and other stuff too)
       if (wanttext == 1)
-    if (highlight)
-      highlighttext(docresponse.doc, terminfo, disp, outconvert, textout, logout);
-    else
+    if (highlight) {
+      highlighttext(docresponse.doc, args, terminfo, disp, outconvert, textout);
+    } else {
       textout << outconvert << disp << docresponse.doc;
+    }
     }
   
     if (wanttext != 1) {
       text_t doctext 
-      = get_formatted_string (collection, collectproto,
-                  docinfo, disp, formatlistptr, docresponse.doc,
-                  logout); 
+    = get_formatted_string (collection, collectproto,
+                docinfo, disp, formatlistptr, docresponse.doc,
+                logout); 
       
-      if (highlight)
-    highlighttext(doctext, terminfo, disp, outconvert, textout, logout);
-      else
+      if (highlight) {
+    highlighttext(doctext, args, terminfo, disp, outconvert, textout);
+      } else {
     textout << outconvert << disp << doctext;
+      }
     }
   }
@@ -876 +883 @@
       output_text (inforesponse.docInfo[0], formatlistptr, queryresponse.termInfo, 
            OID, highlight, hastxt, wanttext, collection, collectproto, 
-           disp, outconvert, textout, logout);
+           disp, outconvert, textout, logout, args);
 
 
@@ -919 +926 @@
       output_text (thisdocinfo, formatlistptr, queryresponse.termInfo, 
                OID, highlight, hastxt, wanttext, collection, 
-               collectproto, disp, outconvert, textout, logout);
+               collectproto, disp, outconvert, textout, logout, args);
       
     }
@@ -936 +943 @@
       output_text (*sechere, formatlistptr, queryresponse.termInfo, 
                (*sechere).OID, highlight, shastxt, wanttext, collection, 
-               collectproto, disp, outconvert, textout, logout);
+               collectproto, disp, outconvert, textout, logout, args);
       count ++;
       sechere ++;
@@ -945 +952 @@
   delete formatlistptr;
 }
-
-// highlighttext highlights query terms in text string and outputs the resulting text string
-void documentaction::highlighttext(text_t &text, const TermInfo_tarray &terms, 
-                   displayclass &disp, outconvertclass &outconvert, 
-                   ostream &textout, ostream &/*logout*/) {
-
-  text_tmap allterms;
-  text_tmap::const_iterator it;
-
-  // first load all the term variations into a map
-  TermInfo_tarray::const_iterator this_term = terms.begin();
-  TermInfo_tarray::const_iterator last_term = terms.end();
-  while (this_term != last_term) {
-    text_tarray::const_iterator this_var = (*this_term).matchTerms.begin();
-    text_tarray::const_iterator last_var = (*this_term).matchTerms.end();
-    while (this_var != last_var) {
-      allterms[*this_var] = 1;
-      this_var ++;
-    }
-    this_term ++;
-  }
-
-  // get the text to start and end a hightlight
-  text_t starthighlight = "<b><u>";
-  text_t endhighlight = "</u></b>";
-  if (disp.isdefaultmacro("Global", "starthighlight"))
-    disp.expandstring("Global", "_starthighlight_", starthighlight);
-  if (disp.isdefaultmacro("Global", "endhighlight"))
-    disp.expandstring("Global", "_endhighlight_", endhighlight);
-
-
-  text_t::iterator here = text.begin(); 
-  text_t::iterator end = text.end();
-  text_t word, buffer;
-  while (here != end) {
-    if (is_unicode_letdig(*here)) {
-      // not word boundary
-      word.push_back(*here);
-      here++;
-
-    } else {
-      // found word boundary
-      // add last word if there was one
-      if (!word.empty()) {
-    it = allterms.find(word);
-    if (it != allterms.end()) {
-      word = starthighlight + word + endhighlight;
-    }
-    buffer += word;
-        word.clear();
-      }
-
-      if (*here == '<') {
-        // skip over rest of html tag
-    while ((here != end) && (*here != '>')) {
-      buffer.push_back(*here);
-      here++;
-    }
-      }
-
-      buffer.push_back(*here);
-      here++;
-
-      if (buffer.size() > 1024) {
-    textout << outconvert << disp << buffer;
-    buffer.clear();
-      }
-    }
-  }
-  textout << outconvert << disp << buffer;
-}

trunk/gsdl/src/recpt/documentaction.h

@@ -54 +54 @@
                 text_t &collection, recptproto *collectproto,
                 displayclass &disp, outconvertclass &outconvert,
-                ostream &textout, ostream &logout);
+                ostream &textout, ostream &logout,
+                cgiargsclass &args);
 
   virtual void output_document (const text_t &OID, cgiargsclass &args, 
@@ -60 +61 @@
                 outconvertclass &outconvert, ostream &textout, 
                 ostream &logout);
-
-  void highlighttext(text_t &text, const TermInfo_tarray &terms, 
-             displayclass &disp, outconvertclass &outconvert, 
-             ostream &textout, ostream &logout);
 
   formatinfo_t formatinfo;

trunk/gsdl/src/recpt/win32.mak

@@ -80 +80 @@
         htmlbrowserclass.h historydb.h phindbrowserclass.h collectoraction.h \
         nullproto.h argdb.h browseaction.h browseactiontools.h phindaction.h \
-        summarise.h
+        summarise.h highlighttext.h
 
 CGIHEADERS = cgiwrapper.h
@@ -97 +97 @@
         historydb.cpp phindbrowserclass.cpp collectoraction.cpp nullproto.cpp \
         argdb.cpp browseaction.cpp browseactiontools.cpp phindaction.cpp \
-        summarise.cpp
+        summarise.cpp highlighttext.cpp
 
 CGISOURCES = librarymain.cpp cgiwrapper.cpp
@@ -114 +114 @@
         historydb.obj phindbrowserclass.obj collectoraction.obj nullproto.obj \
         argdb.obj browseaction.obj browseactiontools.obj phindaction.obj \
-        summarise.obj
+        summarise.obj highlighttext.obj
 
 CGIOBJECTS = librarymain.obj cgiwrapper.obj