support lookbehind

CJex#23 CJex#41 CJex#48 CJex#54
liudongmiao · Jun 28, 2022 · 790dabe · 790dabe
1 parent a5fa510
commit 790dabe
Show file tree

Hide file tree

Showing 5 changed files with 102 additions and 8 deletions.
diff --git a/dist/regulex.js b/dist/regulex.js
diff --git a/src/RegExp.js b/src/RegExp.js
@@ -284,6 +284,12 @@ var NFABuilders=(function _() {
       case AssertNegativeLookahead:
         f=_negativeLookahead(node);
         break;
+      case AssertLookbehind:
+        f=_lookbehind(node);
+        break;
+      case AssertNegativeLookbehind:
+        f=_negativeLookbehind(node);
+        break;
     }
     return _newAssert(node,from,f);
 
@@ -304,10 +310,21 @@ var NFABuilders=(function _() {
         return ret.acceptable;
       };
     }
+    function _lookbehind(node) {
+      var m=NFA(tree2NFA(node.sub,['start']));
+      return function _Lookbehind(stack,c,i,state,s) {
+        var ret=m.input(s,i,null,stack);
+        return ret.acceptable;
+      };
+    }
     function _negativeLookahead(node) {
       var f=_lookahead(node);
       return function _NLookahead() {return !f.apply(this,arguments)};
     }
+    function _negativeLookbehind(node) {
+      var f=_lookbehind(node);
+      return function _NLookbehind() {return !f.apply(this,arguments)};
+    }
 
     function _isBoundary(i,s) {return !!(_isWordChar(i-1,s) ^ _isWordChar(i,s))}
     function _isWordChar(i,s) {return i!==-1 && i!==s.length && /\w/.test(s[i])}

diff --git a/src/parse.js b/src/parse.js
@@ -20,6 +20,8 @@ var Constants={
   //Assertion Type Constants
   AssertLookahead:"AssertLookahead",
   AssertNegativeLookahead:"AssertNegativeLookahead",
+  AssertLookbehind:"AssertLookbehind",
+  AssertNegativeLookbehind:"AssertNegativeLookbehind",
   AssertNonWordBoundary:"AssertNonWordBoundary",
   AssertWordBoundary:"AssertWordBoundary",
   AssertEnd:"AssertEnd",
@@ -350,8 +352,14 @@ function _checkRepeat(node) {
   if (node.repeat) {
     var astype = node.assertionType;
     var msg = 'Nothing to repeat! Repeat after assertion doesn\'t make sense!';
-    if (astype === 'AssertLookahead'  || astype === 'AssertNegativeLookahead' ) {
-      var assertifier = astype === 'AssertLookahead' ? '?=' : '?!';
+    var assertifiers = {
+      'AssertLookahead': '?=',
+      'AssertNegativeLookahead': '?!',
+      'AssertLookbehind': '?<=',
+      'AssertNegativeLookbehind': '?<!',
+    }
+    var assertifier = assertifiers[astype];
+    if (assertifier !== undefined) {
       var pattern = '('+assertifier+'b)';
       msg += '\n/a'+pattern+'+/、/a'+pattern+'{1,n}/ are the same as /a'+pattern+'/。\n' +
               '/a'+pattern+'*/、/a'+pattern+'{0,n}/、/a'+pattern+'?/ are the same as /a/。';
@@ -567,10 +575,13 @@ var actions=(function _() {
     group.num=undefined;
     stack.groupCounter.i--;
   }
-  function groupToAssertion(stack,c,i) { // Convert /(?!)/,/(?=)/ to AssertNode
+  function groupToAssertion(stack,c,i,state) { // Convert /(?!)/,/(?=)/ to AssertNode
     var group=stack._parentGroup;
     group.type=ASSERT_NODE;
     group.assertionType= c=='=' ? AssertLookahead : AssertNegativeLookahead ;
+    if (state === 'groupNameStart') {
+      group.assertionType= c==='=' ? AssertLookbehind : AssertNegativeLookbehind ;
+    }
     // Caveat!!! Assertion group no need to capture
     group.num=undefined;
     stack.groupCounter.i--;
@@ -957,6 +968,8 @@ var config={
     ['groupQualify>groupQualifiedStart',':',actions.groupNonCapture],//group non-capturing
     ['groupQualify>groupQualifiedStart','=',actions.groupToAssertion],//group positive lookahead
     ['groupQualify>groupQualifiedStart','!',actions.groupToAssertion],//group negative lookahead
+    ['groupNameStart>groupQualifiedStart','=',actions.groupToAssertion],//group positive lookbehind
+    ['groupNameStart>groupQualifiedStart','!',actions.groupToAssertion],//group negative lookbehind
     ['groupQualify>groupQualifiedStart','>',actions.groupAtomicGroup],//group atomic-group
     ['groupQualify>groupNameStart','<'],
     ['groupNameStart>groupName','a-zA-Z', actions.groupName],//group name

diff --git a/src/visualize.js b/src/visualize.js
@@ -643,6 +643,14 @@ var plotNode={
       fg="Purple";
       //txt="Negative\nLookahead!"; // break line
       txt="Not followed by:";
+    } else if (nat === AssertLookbehind) {
+      lineColor="LightSeaGreen";
+      fg="Brown";
+      txt="Preceded by:";
+    } else if (nat === AssertNegativeLookbehind) {
+      lineColor="#F69";
+      fg="MediumVioletRed";
+      txt="Not preceded by:";
     }
 
     var sub=plotNode.group(node,x,y);
@@ -709,6 +717,8 @@ var hlColorMap={
   ')':'blue',
   '?=':'darkgreen',
   '?!':'red',
+  '?<!':'brown',
+  '?<=':'darkcyan',
   '?:':'grey',
   '[':'navy',
   ']':'navy',
@@ -740,6 +750,10 @@ function highlight(tree) {
       if (node.type===ASSERT_NODE) {
         if (node.assertionType===AssertLookahead) {
           texts.push(text('?='));
+        } else if (node.assertionType === AssertLookbehind) {
+          texts.push(text('?<='));
+        } else if (node.assertionType === AssertNegativeLookbehind) {
+          texts.push(text('?<!'));
         } else  {
           texts.push(text('?!'));
         }

diff --git a/tests/testData.js b/tests/testData.js
@@ -2009,6 +2009,56 @@ var re2ast =[{ raw: '\\\\{3}',
     chars: 'b',
     raw: 'b'
   }],
+}, {
+  raw: '(?<!foo)bar.*(?<=foo)bar',
+  groupCount: 0,
+  tree: [{
+    type: 'assert',
+    num: undefined,
+    sub: [{
+      type: 'exact',
+      indices: [4, 7],
+      chars: 'foo',
+      raw: 'foo'
+    }],
+    indices: [0, 8],
+    assertionType: 'AssertNegativeLookbehind',
+    endParenIndex: 7,
+    raw: '(?<!foo)'
+  }, {
+    type: 'exact',
+    indices: [8, 11],
+    chars: 'bar',
+    raw: 'bar',
+  }, {
+    type: 'dot',
+    indices: [11, 13],
+    repeat: {
+      min: 0,
+      max: Infinity,
+      nonGreedy: false,
+      possessive: false
+    },
+    raw: '.*'
+  }, {
+    type: 'assert',
+    num: undefined,
+    sub: [{
+      type: 'exact',
+      indices: [17, 20],
+      chars: 'foo',
+      raw: 'foo'
+    }],
+    indices: [13, 21],
+    assertionType: 'AssertLookbehind',
+    endParenIndex: 20,
+    raw: '(?<=foo)'
+  }, {
+    type: 'exact',
+    indices: [21, 24],
+    chars: 'bar',
+    raw: 'bar'
+  }]
 }];