SDF -> BGF extractor is made much more advanced

git-svn-id: https://slps.svn.sourceforge.net/svnroot/slps@950 ab42f6e0-554d-0410-b580-99e487e6eeb2

shared/tools/sdf2bgf

@@ -3,7 +3,7 @@
 # Get our hands on basedir
 LOCAL1=${PWD}
 cd `dirname $0`
-cd ../../topics/extraction/sdf2bgf
+cd ../../topics/extraction/sdf
 XTR=${PWD}
 cd ${LOCAL1}
 
@@ -32,6 +32,23 @@ else
         else
             echo ")" >> _.def
             cat _.def | sglr -p ${XTR}/Main.tbl | asfe -e ${XTR}/Main.eqs | unparsePT > $i
+            # a fix for ->
+            perl -pi -w -e 's/strcon2chardata\(\"\-\>\"\)/\-\>/g;' $i
+            # a fix for <=
+            perl -pi -w -e 's/strcon2chardata\(\"\<\=\"\)/\&lt;\=/g;' $i
+            perl -pi -w -e "s/sqstrcon2chardata\(\'\<\=\'\)/\&lt;\=/g;" $i
+            # a fix for <<
+            perl -pi -w -e 's/strcon2chardata\(\"\<\<\"\)/\&lt;\&lt;/g;' $i
+            # a fix for <<=
+            perl -pi -w -e 's/strcon2chardata\(\"\<\<\=\"\)/\&lt;\&lt;\=/g;' $i
+            # a fix for >>
+            perl -pi -w -e 's/strcon2chardata\(\"\>\>\"\)/\&gt;\&gt;/g;' $i
+            # a fix for &=
+            perl -pi -w -e 's/strcon2chardata\(\"\&\=\"\)/\&amp;\=/g;' $i
+            # a fix for >=
+            perl -pi -w -e 's/\<terminal\>\>\=\<\/terminal\>/\<terminal\>\&gt;\=\<\/terminal\>/g;' $i
+            # a fix for >>=
+            perl -pi -w -e 's/\<terminal\>\>\>\=\<\/terminal\>/\<terminal\>\&gt;\&gt;\=\<\/terminal\>/g;' $i
         fi
     done
     rm -f _.def

topics/extraction/sdf2bgf/Main.asf → topics/extraction/sdf/Main.asf

@@ -93,14 +93,22 @@ trafoSymbol(&N1) =
     <nonterminal>&C1</nonterminal>
   </bgf:expression>
 
-[terminal]
+[terminal-1]
 &C1 := strcon2chardata(&Z1)
 ==================
 trafoSymbol(&Z1) =
   <bgf:expression>
     <terminal>&C1</terminal>
   </bgf:expression>
 
+[terminal-2]
+&C1 := sqstrcon2chardata(&Y1)
+==================
+trafoSymbol(&Y1) =
+  <bgf:expression>
+    <terminal>&C1</terminal>
+  </bgf:expression>
+
 equations
 
 [plus]
@@ -175,6 +183,16 @@ trafoSymbol((&S1 &S2 &S+1)) =
 	</sequence>
   </bgf:expression>
 
+[selector-1]
+&E1 := trafoSymbol(&N1)
+================
+trafoSymbol(&I1:&N1) =
+  <bgf:expression>
+	<selectable>
+		<selector>idcon2chardata(&I1)</selector>
+		&E1
+	</selectable>
+  </bgf:expression>
 
 [cons-to-label-only-cons]
 guessLabel({cons(&I1)}) = <label>idcon2chardata(&I1)</label>

topics/extraction/sdf2bgf/Main.sdf → topics/extraction/sdf/Main.sdf

@@ -49,4 +49,5 @@ module Main
 	"&S+" [0-9]+ -> Symbol+
 	"&Ss" [0-9]+ -> Symbols
 	"&Z"  [0-9]+ -> StrCon
+	"&Y"  [0-9]+ -> SingleQuotedStrCon
 	"&I"  [0-9]+ -> IdCon

topics/extraction/sdf2bgf/Tokens.asf → topics/extraction/sdf/Tokens.asf

@@ -14,10 +14,25 @@ sort2chardata(sort(&Head1 &Middle1 &Middles1 &Last1)) = chardata(&Char1 &Middle1
 
 equations
 
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%%%%% StrCon -> CharData
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
 %%[strcon-of-length-1]
 %% Not working! But needed for &, <, >, etc.
 %%strcon2chardata(strcon(\" strchar(\&) \")) = entityref(& name(letter(a) namechar(letter(m)) namechar(letter(p))) ;)
 
+[strcon-of-length-1-lt]
+strcon2chardata(strcon(\" strchar(\<) \")) = &lt;
+
+[strcon-of-length-1-gt]
+strcon2chardata(strcon(\" strchar(\>) \")) = &gt;
+
+[strcon-of-length-1-amp]
+strcon2chardata(strcon(\" strchar(\&) \")) = &amp;
+
+%%[strcon-of-length-1-minus]
+%%strcon2chardata(strcon(\" strchar(\-) \")) = "-"
 
 [default-strcon-of-length-1]
 strchar(&Chr1) := &StrChar1
@@ -30,6 +45,34 @@ chardata(&Char1 &Chars1) := strcon2chardata(strcon(\" &StrChar+2 \"))
 ===========================================================
 strcon2chardata(strcon(\" &StrChar1 &StrChar+2 \")) = chardata(&Chr1 &Char1 &Chars1)
 
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%%%%% SingleQuotedStrCon -> CharData
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+[sqstrcon-of-length-1-lt]
+sqstrcon2chardata(singlequotedstrcon(\' singlequotedstrchar(\<) \')) = &lt;
+
+[sqstrcon-of-length-1-gt]
+sqstrcon2chardata(singlequotedstrcon(\' singlequotedstrchar(\>) \')) = &gt;
+
+[sqstrcon-of-length-1-amp]
+sqstrcon2chardata(singlequotedstrcon(\' singlequotedstrchar(\&) \')) = &amp;
+
+[default-sqstrcon-of-length-1]
+singlequotedstrchar(&Ch1) := &SQStrChar1
+========================================
+sqstrcon2chardata(singlequotedstrcon(\' &SQStrChar1 \')) = chardata(&Ch1)
+
+[sqstrcon-of-length-greater-than-1]
+singlequotedstrchar(&Ch1) := &SQStrChar1,
+chardata(&Char1 &Chars1) := sqstrcon2chardata(singlequotedstrcon(\' &SQStrChar+2 \'))
+===========================================================
+sqstrcon2chardata(singlequotedstrcon(\' &SQStrChar1 &SQStrChar+2 \')) = chardata(&Ch1 &Char1 &Chars1)
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%%%%% IdCon -> CharData
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
 [idcon-of-length-1]
 idcon2chardata(idcon(&IdHead1)) = chardata(&IdHead1)
 

topics/extraction/sdf/Tokens.sdf

@@ -0,0 +1,56 @@
+module Tokens
+
+imports 
+	languages/sdf2/syntax/Literals
+	languages/sdf2/syntax/Sorts
+	languages/xml/syntax/XML
+	basic/StrCon
+	basic/IdentifierCon
+
+exports
+	sorts
+		CharData
+	context-free syntax
+		sort2chardata(Sort)			-> CharData
+		strcon2chardata(StrCon)			-> Content
+		sqstrcon2chardata(SingleQuotedStrCon)	-> Content
+		idcon2chardata(IdCon)			-> CharData
+		strchar2strcon(StrChar)  			  -> Content
+
+hiddens
+	lexical syntax
+    	%% Let's remind ourselves of the relevant lexical syntax.
+                  %%
+                  %%   1. languages/sdf2/syntax/Sorts
+                  %%     head:[A-Z] -> Sort {cons("one-char")}
+                  %%     head:[A-Z] middle:[A-Za-z0-9\-]* last:[A-Za-z0-9] -> Sort {cons("more-chars")}
+                  %%
+                  %%   2. languages/xml/syntax/XML
+                  %%     ~[\<\&]+ -> CharData {avoid}
+                  %%     ~[\<\&]* "]]>" ~[\<\&]* -> CharData {reject}
+                  %%
+                  %%   3. basic/StrCon
+                  %%     "\\n" -> StrChar {cons("newline")}
+                  %%     "\\t" -> StrChar {cons("tab")}
+                  %%     "\\\"" -> StrChar {cons("quote")}
+                  %%     "\\\\" -> StrChar {cons("backslash")}
+                  %%     "\\" a:[0-9]b:[0-9]c:[0-9] -> StrChar {cons("decimal")}
+                  %%     ~[\0-\31\n\t\"\\] -> StrChar {cons("normal")}
+                  %%     [\"] chars:StrChar* [\"] -> StrCon  {cons("default")}
+
+	lexical variables
+		"&Head" [0-9]*     -> [A-Z]                    %% Part of Sort
+		"&IdHead" [0-9]*   -> [A-Za-z]                 %% Part of IdCon
+		"&Middle" [0-9]*   -> [A-Za-z0-9\-]            %% Part of Sort
+		"&Middles" [0-9]*  -> [A-Za-z0-9\-]*           %% Part of Sort
+		"&Last" [0-9]*     -> [A-Za-z0-9]              %% Part of Sort
+		"&Char" [0-9]*     -> ~[\<\&]                  %% Part of CharData
+		"&Chars" [0-9]*    -> ~[\<\&]*                 %% Part of CharData
+		"&StrChar" [0-9]*  -> StrChar                  %% Part of StrCon
+		"&StrChar+" [0-9]* -> StrChar+                 %% Part of StrCon
+		"&Chr" [0-9]*      -> ~[\0-\31\n\t\"\\\<\&]
+		"&Ch" [0-9]*	 -> ~[\0-\31\n\t\'\\\<\&]
+		"&Both" -> ~[\0-\31\n\t\"\\\<\&] /\ ~[\<\&]
+		"&StrCon" [0-9]* -> StrCon
+		"&SQStrChar" [0-9]*  -> SingleQuotedStrChar    %% Part of SingleQuotedStrCon
+		"&SQStrChar+" [0-9]* -> SingleQuotedStrChar+    %% Part of SingleQuotedStrCon