Changeset 28760 for main/trunk/greenstone2/runtime-src/src/recpt/formattools.cpp

Timestamp:

2013-12-13T13:22:44+13:00 (10 years ago)

Author:

ak19

Message:

Adding in the first working version of the formatconverter program which uses formattools to convert GS2 statements to GS3. Not all the GS2 terms have GS3 equivalents yet and the current ones still need to be run by Dr Bainbridge, but nested IFs and ORs seem to work alright in general. Kathy made the important changes to Makefile.in to get the new formatconverter.cpp to compile. formatconverter.cpp uses the new GS2-to-GS3 specific functions added to formattools.cpp

File:

• main/trunk/greenstone2/runtime-src/src/recpt/formattools.cpp (modified) (6 diffs)

main/trunk/greenstone2/runtime-src/src/recpt/formattools.cpp

@@ -49 +49 @@
                   ResultDocInfo_t &docinfo, displayclass &disp, 
                   text_tmap &options, ostream& logout);
+
 static text_t format_text (const text_t& collection, recptproto* collectproto,
                   ResultDocInfo_t &docinfo, displayclass &disp, 
@@ -58 +59 @@
                   ostream &logout);
 
+static text_t transform_to_GS3_format (format_t *formatlistptr);
 
 void metadata_t::clear() {
@@ -127 +129 @@
         (*(here+2) == 'd' || *(here+2) == 'D') &&
         (*(here+3) == '>' || *(here+3) == ' '))
+      //|| *(here+3) == '\t' || *(here+3) == '\n'))
       return true;
       } else return false;
@@ -1552 +1555 @@
 
 
-
-
 static bool uses_expression(const text_t& collection, recptproto* collectproto,
                 ResultDocInfo_t &docinfo,
@@ -1740 +1741 @@
     // This is a tad tricky.  When we expand a string like _cgiargmode_, that is
     // a cgi argument macro that has not been set, it evaluates to itself.
-    // Therefore, were have to say that a piece of text evalautes true if
+    // Therefore, we have to say that a piece of text evaluates true if
     // it is non-empty and if it is a cgi argument evaulating to itself.
 
@@ -2057 +2058 @@
   return summary;
 }
+
+//-------------- GS3 related functions --------------
+// copy of the other uses_expression function, but without using the extra GS2-runtime-specific parameters
+static bool uses_expression(const text_t& outstring, text_t& lhs_expr,
+                text_t& op_expr, text_t& rhs_expr)
+{
+  // Note: the string may not be of the form: str1 op str2, however
+  // to deterine this we have to process it on the assumption it is,
+  // and if at any point an 'erroneous' value is encountered, return
+  // false and let something else have a go at evaluating it
+
+  // Starting at the end of the string and working backwards ..
+
+  const int outstring_len = outstring.size();
+
+  // skip over white space
+  int rhs_end = rscan_over_whitespace(outstring,outstring_len-1);
+
+  if (rhs_end<=0) { 
+    // no meaningful text or (rhs_end==0) no room for operator
+    return false;
+  }
+
+  // check for ' or " and then scan over token
+  const char potential_quote = outstring[rhs_end];
+  int rhs_start=rhs_end;
+  bool quoted = false;
+
+  if ((potential_quote == '\'') || (potential_quote == '\"')) {
+    --rhs_end;
+    rhs_start = rscan_for(outstring,rhs_end-1,potential_quote) +1;
+    quoted = true;
+  }
+  else {
+    rhs_start = rscan_for_whitespace(outstring,rhs_end-1) +1;
+  }
+
+  if ((rhs_end-rhs_start)<0) {
+    // no meaningful rhs expression
+    return false;
+  }
+
+  // form rhs_expr
+  rhs_expr = extract_substr(outstring,rhs_start,rhs_end);
+
+  // skip over white space
+  const int to_whitespace = (quoted) ? 2 : 1;
+
+  int op_end = rscan_over_whitespace(outstring,rhs_start-to_whitespace);
+  int op_start = rscan_for_whitespace(outstring,op_end-1)+1;
+
+  if ((op_end<0) && (op_start<0)) {
+    // no meaningful expression operator
+    return false;
+  }
+
+  if (op_end-op_start<0) {
+    // no meaningful expression operator
+    return false;
+  }
+
+  op_expr = extract_substr(outstring,op_start,op_end);
+
+
+  // check for operator
+  if ((op_expr != "eq") && (op_expr != "ne") && (op_expr != "gt") &&
+     (op_expr != "ge") && (op_expr != "lt") && (op_expr != "le") && (op_expr != "==") && (op_expr != "!=") &&  (op_expr != ">") && (op_expr != ">=") && (op_expr != "<") && (op_expr != "<=") && (op_expr != "sw") && (op_expr != "ew")) {
+
+    // not a valid operator
+    return false;
+  }
+
+  int lhs_end = rscan_over_whitespace(outstring,op_start-1);
+  if (lhs_end<0) {
+    // no meaningful lhs expression
+    return false;
+  }
+
+  int lhs_start = scan_over_whitespace(outstring,0);
+
+  // form lhs_expr from remainder of string
+  lhs_expr = extract_substr(outstring,lhs_start,lhs_end);
+
+  return true;
+}
+
+// [ex.Title] -> ex.Title
+static text_t remove_bracket_bookends(const text_t &str) {
+  
+  if(str[0] == '[' && str[str.size()-1] == ']') {
+    return substr (str.begin()+1, str.end()-1);
+  } else {
+    return str;
+  }
+}
+
+static text_t get_gs3_if (const decision_t &decision, format_t *ifptr, format_t *elseptr)
+{
+  text_t ifstmt ="<gsf:switch>";
+
+
+  if (decision.command == dMeta) {
+    ifstmt += "<gsf:metadata name=\"";
+    ifstmt += remove_bracket_bookends(decision.meta.metaname);
+    ifstmt += "\"/>"; 
+  }
+
+  else { //if(decision.command == dText) 
+
+    text_t outstring = decision.text;
+
+    // Check for if expression in form: str1 op str2 
+    // (such as [x] eq "y")
+    text_t lhs_expr, op_expr, rhs_expr;
+    if (uses_expression(outstring,lhs_expr,op_expr,rhs_expr)) {
+
+      text_t if_operator = op_expr;
+      if (op_expr == "eq" || op_expr == "==") {
+    if_operator = "equals";
+      } else if (op_expr == "ne" || op_expr == "!=") {
+    if_operator = "notEquals";
+      } else if (op_expr == "gt" || op_expr == ">") {
+    if_operator = "greaterThan";
+      } else if (op_expr == "lt" || op_expr == "<") {
+    if_operator = "lessThan";
+      } else if (op_expr == "ge" || op_expr == ">=") {
+    if_operator = "greaterThanOrEquals";
+      } else if (op_expr == "le" || op_expr == "<=") {
+    if_operator = "lessThanOrEquals";
+      } else if (op_expr == "sw") {
+    if_operator = "startsWith";
+      } else if (op_expr == "ew") {
+    if_operator = "endsWith";
+      }
+
+      ifstmt += "<gsf:metadata name=\"";
+      ifstmt += remove_bracket_bookends(lhs_expr);
+      ifstmt += "\"/>"; 
+
+      ifstmt += "<gsf:when test=\"";
+      ifstmt += if_operator; // the test operator
+      ifstmt += "\" test-value=\"";
+      ifstmt += remove_bracket_bookends(rhs_expr); // the test-value
+      ifstmt += "\">";
+    } 
+    else {
+      ifstmt += "<gsf:metadata name=\"";
+      ifstmt += remove_bracket_bookends(decision.text);
+      ifstmt += "\"/>";
+      ifstmt += "<gsf:when test=\"exists\">";
+    }
+  }
+   
+  // if portion
+  text_t if_body = "";
+  while(ifptr != NULL) { // body of if can contain a list of items to be transformed into GS3 format stmts
+    if_body += transform_to_GS3_format (ifptr);
+    ifptr = ifptr->nextptr;
+  }
+  ifstmt += if_body;
+  ifstmt += "</gsf:when>";
+
+  // else portion
+  if(elseptr != NULL) {
+
+    ifstmt += "<gsf:otherwise>";
+    text_t else_body = ""; // body of else can contain a list of items to be transformed into GS3 format stmts
+    while(elseptr != NULL) {
+      else_body += transform_to_GS3_format (elseptr);
+      elseptr = elseptr->nextptr;
+    }
+    ifstmt += else_body;
+    ifstmt += "</gsf:otherwise>";
+  }  
+
+  ifstmt += "</gsf:switch>";
+  return ifstmt;
+}
+
+
+static text_t get_gs3_or (format_t *orptr) {
+  text_t result = "<gsf:choose-metadata>";
+
+  while(orptr != NULL) {
+    text_t or_body = transform_to_GS3_format (orptr);
+    if (!or_body.empty()) {
+      result += or_body;
+    }
+
+    orptr = orptr->nextptr;    
+  }
+  result += "</gsf:choose-metadata>";
+  return result;
+}
+
+// what about all the <td>? Does that get stored in formatlistptr, such as under the ->text field?
+text_t get_GS3_formatstring (format_t *formatlistptr) {
+  text_t result;
+
+  while (formatlistptr != NULL) {
+    result += transform_to_GS3_format(formatlistptr);    
+    formatlistptr = formatlistptr->nextptr;
+  }
+  
+  return result;
+}
+
+text_t transform_to_GS3_format (format_t *formatlistptr) {
+
+  if (formatlistptr == NULL) return "";
+  
+  switch (formatlistptr->command) {
+  case comOID:
+    return "<gsf:OID/>";
+  case comTopOID:
+    return "<gsf:metadata name='OID' select='root' />"; // for now try this
+  case comRank:
+    return "<gsf:rank/>";
+  case comText:
+    return formatlistptr->text; // [text]? or any string that is not a command or reserved
+  case comLink:
+    return "<gsf:link type='document'>"; // type?
+  case comEndLink:
+    return "</gsf:link>";
+  case comHref:
+    return "<gsf:lib name=\"href\"/>"; // ??? in gslib xsl, output comment marking not sure what this maps to
+  case comIcon:
+    return "<gsf:icon type='document'/>";
+  case comNum:
+    return "<gsf:lib name=\"num\"/>"; // ??? in gslib xsl, output comment marking not sure what this maps to
+  case comRel: //if [RelatedDocuments] appears in format string, collect relation data
+    return "<gsf:lib name=\"RelatedDocuments\"/>"; // output comment marking this as deprecated or to be implemented for GS3 in gslib xslt
+  case comSummary:
+    return "<gsf:lib name=\"Summary\"/>"; // in gslib xslt output comment marking this as to be implemented for GS3 
+    // need to invent this for GS3 based on what GS2 does
+  case comAssocLink:
+    return "<gsf:link type='source'>";
+  case comEndAssocLink:
+    return "</gsf:link>";
+  case comMeta:
+    return "<gsf:metadata name=\"" + formatlistptr->meta.metaname + "\" />";//?
+  case comDoc:
+    return "<gsf:text/>";
+  case comImage: // the cover img seems to be handled by some magic code in GS3
+    return "<gsf:lib name=\"image\"/>"; // in gslib xslt output a comment marking this as not working in the same way in GS3 as in GS2. TOC currently not relocatable
+  case comTOC:
+    return "<gsf:lib name=\"TOC\"/>"; // in gslib xslt output a comment marking this as not working in the same way in GS3 as in GS2. TOC currently not relocatable
+    // need to think about whether an equivalent actually exists
+    // return "<gsf:option name=\"TOC\" value=\"true\"/>"; // this is wrong
+  case comDocumentButtonDetach:
+    return "<gsf:lib name=\"DocumentButtonDetach\"/>"; // output comment marking this as deprecated in gslib xslt
+  case comDocumentButtonHighlight:
+    return "<gsf:lib name=\"DocumentButtonHighlight\"/>"; // output comment marking this as deprecated in gslib xslt
+  case comDocumentButtonExpandContents:
+    return "<gsf:lib name=\"DocumentButtonExpandContents\"/>"; // output comment marking this as deprecated in gslib xslt
+  case comDocumentButtonExpandText:
+    return "<gsf:lib name=\"DocumentButtonExpandText\"/>"; // output comment marking this as deprecated in gslib xslt
+  case comHighlight:
+    return "<span class=\"highlight\">";
+    break;
+  case comEndHighlight:
+    return "</span>";
+    break;
+  case comMetadataSpanWrap:
+    metadata_wrap=true;  metadata_wrap_type="span"; return "";
+    break;
+  case comEndMetadataSpanWrap:
+    metadata_wrap=false; metadata_wrap_type="";     return "";
+    break;
+  case comMetadataDivWrap:
+    metadata_wrap=true;  metadata_wrap_type="div";  return "";
+    break;
+  case comEndMetadataDivWrap:
+    metadata_wrap=false; metadata_wrap_type="";     return "";
+    break;
+  case comIf:
+    return get_gs3_if (formatlistptr->decision, formatlistptr->ifptr, formatlistptr->elseptr);
+  case comOr:
+    return get_gs3_or (formatlistptr->orptr);
+    //return "<gsf:choose-metadata>"+get_gs3_or (formatlistptr->orptr)+"</gsf:choose-metadata>";
+  case comDocTermsFreqTotal:
+    return "<gsf:lib name=\"DocTermsFreqTotal\"/>";
+  case comCollection: // trying to get all the metadata for a collection. How is this done in GS3???
+    return "<gsf:lib name=\"collection\"/>";
+  }
+  return "";
+}