Changeset 3182 for trunk/protemix

Timestamp:

2002-06-25T21:07:30+12:00 (22 years ago)

Author:

sjboddie

Message:

* empty log message *

Location:

trunk/protemix

Files:

• etc/collect.cfg (modified) (1 diff)
• macros/style.dm (modified) (1 diff)
• perllib/plugins/ProtemixPlug.pm (modified) (1 diff)
• rename.pl (modified) (1 diff)
• src/recpt/Makefile.in (modified) (1 diff)
• src/recpt/equivterms.cpp (added)
• src/recpt/ptmxdocaction.cpp (modified) (2 diffs)
• src/recpt/ptmxdocaction.h (modified) (1 diff)
• src/recpt/ptmxqueryaction.cpp (modified) (1 diff)

trunk/protemix/etc/collect.cfg

@@ -11 +11 @@
 
 plugin        GAPlug
-plugin        ProtemixPlug 
+plugin        ProtemixPlug
 plugin        ArcPlug
 plugin        RecPlug

trunk/protemix/macros/style.dm

@@ -126 +126 @@
 </tr>
 </table>
-<div style="margin-left:10; margin-right:10; margin-bottom:10">
+<center>
+<table border="0" cellpadding="0" cellspacing="0" width="638"><tr><td>
 }
 
 _footer_ {
 _pagefooterextra_
-</div>
+</td></tr></table>
 </body>
 </html>

trunk/protemix/perllib/plugins/ProtemixPlug.pm

@@ -176 +176 @@
 }
 
+# don't want to convert character entities for now as mgpp appears to be broken
+sub read_file {
+    my ($self, $filename, $encoding, $language, $textref) = @_;
+
+    &BasPlug::read_file($self, $filename, $encoding, $language, $textref);
+
+    # Convert entities to their UTF8 equivalents
+#    $$textref =~ s/&(lt|gt|amp|quot);/&z$1;/go;
+#    $$textref =~ s/&([^;]+);/&ghtml::getcharequiv($1,1)/gseo;
+#    $$textref =~ s/&z(lt|gt|amp|quot);/&$1;/go;
+}
+
+
 1;

trunk/protemix/rename.pl

@@ -91 +91 @@
     $noncss =~ s/\\\n/\n/sg;
 
+    # Occurances of 2HCl (lowercase el) were mistakenly changed to 2HC1
+    # (digit one) during OCR in some cases so we'll change them back
+    $noncss =~ s/2HC1/2HCl/g;
+
     # alter <img> tags to support renamed files
     $noncss =~ s/(<img src=\")([^\"]+)/$1 . &rename_file($2)/sige;

trunk/protemix/src/recpt/Makefile.in

@@ -67 +67 @@
 HEADERS = ptmxqueryaction.h ptmxdocaction.h
 
-SOURCES = ptmxmain.cpp ptmxqueryaction.cpp ptmxdocaction.cpp browsetools.cpp
+SOURCES = ptmxmain.cpp ptmxqueryaction.cpp ptmxdocaction.cpp browsetools.cpp equivterms.cpp
 
-OBJECTS = ptmxmain.o ptmxqueryaction.o ptmxdocaction.o browsetools.o
+OBJECTS = ptmxmain.o ptmxqueryaction.o ptmxdocaction.o browsetools.o equivterms.o
 
 EXEC    = ptmx

trunk/protemix/src/recpt/ptmxdocaction.cpp

@@ -27 +27 @@
 #include "ptmxdocaction.h"
 #include "OIDtools.h"
+#include "querytools.h"
 
 ptmxdocaction::ptmxdocaction () {
@@ -62 +63 @@
 }
 
+// overridden to format the querystring when highlighting text
+void ptmxdocaction::output_document (const text_t &OID, cgiargsclass &args, 
+                     recptproto *collectproto, displayclass &disp, 
+                     outconvertclass &outconvert, ostream &textout, 
+                     ostream &logout) {
+  FilterResponse_t inforesponse;
+  FilterResponse_t queryresponse;
+  text_tset metadata;
+  bool getParents = false;
+  bool highlight = false;
+  int wanttext = 0;
+  int arg_gt = args.getintarg("gt");
+  text_t &collection = args["c"];
+    
+  // if we have a query string and highlighting is turned on we need
+  // to redo the query to get the terms for highlighting
+  if (args.getintarg("hl")) {
+    FilterRequest_t request;
+    comerror_t err;
+    request.filterResultOptions = FRmatchTerms;
+    text_t formattedstring = args["q"];
+    format_querystring (formattedstring, args.getintarg("b"));
+
+    // ------ for ptmx collection
+
+    if (!args["c1"].empty() && args["c1"] != "Any") {
+      //    formattedstring = "[" + args["c1"] + "]:CL";
+      formattedstring = args["c1"];
+    }
+    if (!args["c2"].empty() && args["c2"] != "Any") {
+      if (!formattedstring.empty()) formattedstring += " & ";
+      //    formattedstring += "[" + args["c2"] + "]:CA";
+      formattedstring += args["c2"];
+    }
+    if (!args["c3"].empty() && args["c3"] != "Any") {
+      if (!formattedstring.empty()) formattedstring += " & ";
+      //    formattedstring += "[" + args["c3"] + "]:CS";
+      formattedstring += args["c3"];
+    }
+    if (!args["q"].empty()) {
+      if (!formattedstring.empty()) formattedstring += " & ";
+      formattedstring += args["q"];
+    }
+
+    void insert_equiv_terms(text_t&); // declaration
+    insert_equiv_terms(formattedstring);
+    
+    // --------------
+
+    if (!formattedstring.empty()) {
+
+      set_queryfilter_options (request, formattedstring, args);
+      collectproto->filter (args["c"], request, queryresponse, err, logout);
+      if (err != noError) {
+    outconvertclass text_t2ascii;
+    logout << text_t2ascii
+           << "documentaction::output_document: call to QueryFilter failed "
+           << "for " << args["c"] << " collection (" << get_comerror_string (err) << ")\n";
+    highlight = false;
+      } else {
+    highlight = true;
+      }
+    }
+  }
+
+
+  format_t *formatlistptr = new format_t();
+  parse_formatstring (formatinfo.DocumentText, formatlistptr, metadata, getParents);
+
+  metadata.insert ("hastxt");
+  metadata.insert ("haschildren");
+
+  if (formatinfo.DocumentText == "[Text]")
+    wanttext = 1;
+  else {
+    char *docformat = formatinfo.DocumentText.getcstr();
+    if (strstr (docformat, "[Text]") != NULL)
+      wanttext = 2;
+    delete docformat;
+  }
+    
+  if (get_info (OID, collection, metadata, getParents, collectproto, inforesponse, logout)) {
+    int hastxt = inforesponse.docInfo[0].metadata["hastxt"].values[0].getint();
+    int haschildren = inforesponse.docInfo[0].metadata["haschildren"].values[0].getint();
+    
+    if (arg_gt == 0) {
+      output_text (inforesponse.docInfo[0], formatlistptr, queryresponse.termInfo, 
+           OID, highlight, hastxt, wanttext, collection, collectproto, 
+           disp, outconvert, textout, logout);
+
+
+    } else {
+
+      ResultDocInfo_t thisdocinfo = inforesponse.docInfo[0];
+    
+      // text is to be expanded
+      text_t exOID = OID;
+      if (haschildren != 1) exOID = get_parent (OID);
+      if (exOID.empty()) exOID = OID;
+    
+      // if we're not in a document (i.e. we're in a top level classification)
+      // we need to pass "is_classify = true" to get_contents so that it 
+      // doesn't recurse all the way through each document in the classification
+      bool is_classify = false;
+      if (args["d"].empty()) is_classify = true;
+
+      get_contents (exOID, is_classify, metadata, collection, 
+            collectproto, inforesponse, logout);
+    
+      ResultDocInfo_tarray::iterator sechere = inforesponse.docInfo.begin();
+      ResultDocInfo_tarray::iterator secend = inforesponse.docInfo.end();
+
+      if (arg_gt == 1) {
+    // check if there are more than 10 sections containing text to be expanded - 
+    // if there are output warning message - this isn't a great way to do this
+    // since the sections may be very large or very small - one day I'll fix it 
+    // -- Stefan.
+    int seccount = 0;
+    while (sechere != secend) {
+      int shastxt = (*sechere).metadata["hastxt"].values[0].getint();
+      if (shastxt == 1) seccount ++;
+      if (seccount > 10) break;
+      sechere ++;
+    }
+    if (seccount > 10) {
+      // more than 10 sections so output warning message and text 
+      // for current section only
+      textout << outconvert << disp << "_document:textltwarning_";
+
+      output_text (thisdocinfo, formatlistptr, queryresponse.termInfo, 
+               OID, highlight, hastxt, wanttext, collection, 
+               collectproto, disp, outconvert, textout, logout);
+      
+    }
+    else arg_gt = 2;
+      }
+
+      if (arg_gt == 2) {
+    // get the text for each section
+    sechere = inforesponse.docInfo.begin();
+    int count = 0;
+    while (sechere != secend) {
+      textout << outconvert << disp << "\n<p><a name=" << (*sechere).OID << "></a>\n";
+
+      int shastxt = (*sechere).metadata["hastxt"].values[0].getint();
+
+      output_text (*sechere, formatlistptr, queryresponse.termInfo, 
+               (*sechere).OID, highlight, shastxt, wanttext, collection, 
+               collectproto, disp, outconvert, textout, logout);
+      count ++;
+      sechere ++;
+    }
+      }
+    }
+  }
+  delete formatlistptr;
+}

trunk/protemix/src/recpt/ptmxdocaction.h

@@ -41 +41 @@
   virtual void define_internal_macros (displayclass &disp, cgiargsclass &args, 
                        recptprotolistclass *protos, ostream &logout);
+
+  virtual void output_document (const text_t &OID, cgiargsclass &args, 
+                recptproto *collectproto, displayclass &disp, 
+                outconvertclass &outconvert, ostream &textout, 
+                ostream &logout);
+
 };
 

trunk/protemix/src/recpt/ptmxqueryaction.cpp

@@ -128 +128 @@
   }
 
-  logout << outconvert << "formattedstring: " << formattedstring << "\n";
+  // do synonym searching stuff
+  int insert_equiv_terms(text_t&); // forward declaration
+  insert_equiv_terms(formattedstring);
 
   if (!formattedstring.empty()) {