Merge pull request #3 from cdibbs/master

Modified ebnf-transform to allow symbol aliases.
zaach · Aug 10, 2013 · ea638f6 · ea638f6
2 parents 706a5c4 + 982d9b6
commit ea638f6
Show file tree

Hide file tree

Showing 7 changed files with 506 additions and 122 deletions.
diff --git a/Makefile b/Makefile
@@ -8,6 +8,9 @@ build:
 	node ./node_modules/.bin/jison bnf.y bnf.l
 	mv bnf.js parser.js
 
+	node ./node_modules/.bin/jison ebnf.y
+	mv ebnf.js transform-parser.js
+
 test:
 	node tests/all-tests.js
 

diff --git a/bnf.l b/bnf.l
@@ -1,3 +1,5 @@
+id                        [a-zA-Z][a-zA-Z0-9_-]*
+
 %x action code
 %s bnf ebnf
 
@@ -14,7 +16,8 @@
 \s+                     /* skip whitespace */
 "//".*                  /* skip comment */
 "/*"(.|\n|\r)*?"*/"     /* skip comment */
-[a-zA-Z][a-zA-Z0-9_-]*  return 'ID';
+"["{id}"]"              yytext = yytext.substr(1, yyleng-2); return 'ALIAS';
+{id}                    return 'ID';
 '"'[^"]+'"'             yytext = yytext.substr(1, yyleng-2); return 'STRING';
 "'"[^']+"'"             yytext = yytext.substr(1, yyleng-2); return 'STRING';
 ":"                     return ':';

diff --git a/bnf.y b/bnf.y
@@ -122,7 +122,9 @@ handle_sublist
     ;
 
 expression_suffix
-    : expression suffix
+    : expression suffix ALIAS
+        {$$ = $expression + $suffix + "[" + $ALIAS + "]"; }
+    | expression suffix
         {$$ = $expression + $suffix; }
     ;
 

diff --git a/ebnf-transform.js b/ebnf-transform.js
@@ -2,14 +2,31 @@ var EBNF = (function(){
     var parser = require('./transform-parser.js');
 
     var transformExpression = function(e, opts, emit) {
-        var type = e[0], value = e[1], name;
+        var type = e[0], value = e[1], name = false;
+
+        if (type === 'xalias') {
+            type = e[1];
+            value = e[2]
+            name = e[3];
+            if (type) {
+                e = e.slice(1,2);
+            } else {
+                e = value;
+                type = e[0];
+                value = e[1];
+            }
+        }
 
         if (type === 'symbol') {
-            if (e[1][0] === '\\') emit (e[1][1]);
-            else if (e[1][0] === '\'') emit (e[1].substring(1, e[1].length-1));
-            else emit (e[1]);
+            var n;
+            if (e[1][0] === '\\') n = e[1][1];
+            else if (e[1][0] === '\'') n = e[1].substring(1, e[1].length-1);
+            else n = e[1];
+            emit(n + (name ? "["+name+"]" : ""));
         } else if (type === "+") {
-            name = opts.production + "_repetition_plus" + opts.repid++;
+            if (!name) {
+                name = opts.production + "_repetition_plus" + opts.repid++;
+            }
             emit(name);
 
             opts = optsForProduction(name, opts.grammar);
@@ -22,7 +39,9 @@ var EBNF = (function(){
                 ]
             ];
         } else if (type === "*") {
-            name = opts.production + "_repetition" + opts.repid++;
+            if (!name) {
+                name = opts.production + "_repetition" + opts.repid++;
+            }
             emit(name);
 
             opts = optsForProduction(name, opts.grammar);
@@ -34,7 +53,9 @@ var EBNF = (function(){
                 ]
             ];
         } else if (type ==="?") {
-            name = opts.production + "_option" + opts.optid++;
+            if (!name) {
+                name = opts.production + "_option" + opts.optid++;
+            }
             emit(name);
 
             opts = optsForProduction(name, opts.grammar);
@@ -45,7 +66,9 @@ var EBNF = (function(){
             if (value.length == 1) {
                 emit(transformExpressionList(value[0], opts));
             } else {
-                name = opts.production + "_group" + opts.groupid++;
+                if (!name) {
+                    name = opts.production + "_group" + opts.groupid++;
+                }
                 emit(name);
 
                 opts = optsForProduction(name, opts.grammar);

diff --git a/ebnf.y b/ebnf.y
@@ -0,0 +1,66 @@
+/* EBNF grammar spec */
+
+%lex
+
+id                        [a-zA-Z][a-zA-Z0-9_-]*
+
+%%
+\s+             /* skip whitespace */
+{id}           return 'symbol';
+"["{id}"]"     yytext = yytext.substr(1, yyleng-2); return 'ALIAS';
+"'"[^']*"'"    return 'symbol';
+"."            return 'symbol';
+
+bar            return 'bar';
+"("            return '(';
+")"            return ')';
+"*"            return '*';
+"?"            return '?';
+"|"            return '|';
+"+"            return '+';
+<<EOF>>        return 'EOF';
+/lex
+
+%start production
+
+%%
+
+production
+  : handle EOF
+    { return $handle; }
+  ;
+
+handle_list
+  : handle
+    { $$ = [$handle]; }
+  | handle_list '|' handle
+    { $handle_list.push($handle); }
+  ;
+
+handle
+  :
+    { $$ = []; }
+  | handle expression_suffix
+    { $handle.push($expression_suffix); }
+  ;
+
+expression_suffix
+  : expression suffix ALIAS
+    { $$ = ['xalias', $suffix, $expression, $ALIAS]; }
+  | expression suffix
+    { if ($suffix) $$ = [$suffix, $expression]; else $$ = $expression; }
+  ;
+
+expression
+  : symbol
+    { $$ = ['symbol', $symbol]; }
+  | '(' handle_list ')'
+    { $$ = ['()', $handle_list]; }
+  ;
+
+suffix
+  : 
+  | '*'
+  | '?'
+  | '+'
+  ;
diff --git a/tests/ebnf.js b/tests/ebnf.js
@@ -44,6 +44,25 @@ function testBadParse(top, strings) {
     };
 }
 
+function testAlias(top, obj, str) {
+    return function() {
+        var grammar = {
+            "lex": {
+                "rules": [
+                    ["\\s+", ''],
+                    ["[A-Za-z]+", "return 'word';"],
+                    [",", "return ',';"],
+                    ["$", "return 'EOF';"]
+                ]
+            },
+            "start": "top",
+            "bnf": ebnf.transform({"top": [top]})
+        };
+        assert.deepEqual(grammar['bnf'], obj);
+        assert.ok(new Parser(grammar).parse(str));
+    };
+}
+
 var tests = {
     "test idempotent transform": function() {
         var first = {
@@ -63,10 +82,24 @@ var tests = {
     "test group () on simple phrase": testParse("(word word) EOF", "two words"),
     "test group () with multiple options on first option": testParse("((word word) | word) EOF", "hi there"),
     "test group () with multiple options on second option": testParse("((word word) | word) EOF", "hi"),
-    "test complex expression ( *, ?, () )": testParse("(word (',' word)*)? EOF ", ["", "hi", "hi, there"])
+    "test complex expression ( *, ?, () )": testParse("(word (',' word)*)? EOF ", ["", "hi", "hi, there"]),
+    "test named repeat (*)": testAlias("word*[bob] EOF",
+        { top: [ 'bob EOF' ],
+        bob: [ [ '', '$$ = [];' ], [ 'bob word', '$1.push($2);' ] ] }, "word"),
+    "test named repeat (+)": testAlias("word+[bob] EOF",
+        { top: [ 'bob EOF' ],
+        bob: [ [ 'word', '$$ = [$1];' ], [ 'bob word', '$1.push($2);' ] ] }, "wordy word"),
+    "test named group ()": testAlias("word[alice] (',' word)*[bob] EOF",
+        {"top":["word[alice] bob EOF"],"bob":[["","$$ = [];"],["bob , word","$1.push($2);"]]},
+        "one, two"),
+    "test named option (?)": testAlias("word[alex] word?[bob] EOF", { top: [ 'word[alex] bob EOF' ], bob: [ '', 'word' ] }, "oneor two"),
+    "test named complex expression (())": testAlias("word[alpha] (word[alex] (word[bob] word[carol] ',')+[david] word ',')*[enoch] EOF",
+        {"top":["word[alpha] enoch EOF"],"david":[["word[bob] word[carol] ,","$$ = [$1];"],["david word[bob] word[carol] ,","$1.push($2);"]],
+        "enoch":[["","$$ = [];"],["enoch word[alex] david word ,","$1.push($2);"]]},
+        "one two three four, five,"
+    )
 };
 
 for (var test in tests) {
     exports[test] = tests[test];
 }
-