Return absolute minimum capture data from OnigRegExp

Atom/src/OnigRegexpExtension.mm

@@ -37,37 +37,27 @@ public:
     return resultArray;
   }
 
-  CefRefPtr<CefV8Value> GetCaptureTree(CefRefPtr<CefV8Value> string, CefRefPtr<CefV8Value> index) {
+  CefRefPtr<CefV8Value> GetCaptureIndices(CefRefPtr<CefV8Value> string, CefRefPtr<CefV8Value> index) {
     OnigResult *result = [m_regex search:stringFromCefV8Value(string) start:index->GetIntValue()];
     if ([result count] == 0) return CefV8Value::CreateNull();
-    return BuildCaptureTree(result);
+    return BuildCaptureIndices(result);
   }
 
-  CefRefPtr<CefV8Value> BuildCaptureTree(OnigResult *result) {
-    int index = 0;
-    return BuildCaptureTree(result, index);
-  }
+  CefRefPtr<CefV8Value> BuildCaptureIndices(OnigResult *result) {
+    CefRefPtr<CefV8Value> array = CefV8Value::CreateArray();
+    int i = 0;
 
-  CefRefPtr<CefV8Value> BuildCaptureTree(OnigResult *result, int &index) {
-    int currentIndex = index++;
-    int startPosition = [result locationAt:currentIndex];
-    int endPosition = startPosition + [result lengthAt:currentIndex];
-    
-    CefRefPtr<CefV8Value> tree = CefV8Value::CreateArray();    
-    int i = 0;    
-    tree->SetValue(i++, CefV8Value::CreateInt(currentIndex));
-    tree->SetValue(i++, CefV8Value::CreateInt(startPosition));
-    tree->SetValue(i++, CefV8Value::CreateInt(endPosition));
-
-    while (index < [result count] && [result locationAt:index] < endPosition) {
-      if ([result lengthAt:index] == 0) {
-        index++;
-      } else {
-        tree->SetValue(i++, BuildCaptureTree(result, index));
-      }
+    int resultCount = [result count];
+    for (int index = 0; index < resultCount; index++) {
+      int captureLength = [result lengthAt:index];
+      if (captureLength == 0) continue;
+      int captureStart = [result locationAt:index];
+      array->SetValue(i++, CefV8Value::CreateInt(index));
+      array->SetValue(i++, CefV8Value::CreateInt(captureStart));
+      array->SetValue(i++, CefV8Value::CreateInt(captureStart + captureLength));
     }
     
-    return tree;
+    return array;
   }
 
   CefRefPtr<CefV8Value> CaptureCount() {
@@ -91,11 +81,11 @@ bool OnigRegexpExtension::Execute(const CefString& name,
                             CefRefPtr<CefV8Value>& retval,
                             CefString& exception) {
 
-  if (name == "getCaptureTree") {
+  if (name == "getCaptureIndices") {
     CefRefPtr<CefV8Value> string = arguments[0];
     CefRefPtr<CefV8Value> index = arguments.size() > 1 ? arguments[1] : CefV8Value::CreateInt(0);
     OnigRegexpUserData *userData = (OnigRegexpUserData *)object->GetUserData().get();
-    retval = userData->GetCaptureTree(string, index);
+    retval = userData->GetCaptureIndices(string, index);
     return true;
   }
   else if (name == "buildOnigRegExp") {

src/app/text-mate-grammar.coffee

@@ -92,22 +92,17 @@ class Rule
     regexPatternPairs
 
   getNextTokens: (stack, line, position) ->
-    captureTree = @regex.getCaptureTree(line, position)
-    return {} unless captureTree?[2] > 0 # ignore zero-length matches
+    captureIndices = @regex.getCaptureIndices(line, position)
 
-    firstCapture = captureTree[3]
-    [firstCaptureIndex, firstCaptureStart, firstCaptureEnd]  = firstCapture
+    return {} unless captureIndices?[2] > 0 # ignore zero-length matches
+
+    shiftCapture(captureIndices)
+
+    [firstCaptureIndex, firstCaptureStart, firstCaptureEnd] = captureIndices
     pattern = @patternsByCaptureIndex[firstCaptureIndex]
-
-    @adjustCaptureTreeIndices(firstCapture, firstCaptureIndex)
-    nextTokens = pattern.handleMatch(stack, line, firstCapture)
+    nextTokens = pattern.handleMatch(stack, line, captureIndices)
     { nextTokens, tokensStartPosition: firstCaptureStart, tokensEndPosition: firstCaptureEnd }
 
-  adjustCaptureTreeIndices: (tree, startIndex) ->
-    tree[0] -= startIndex
-    for capture in tree[3..]
-      @adjustCaptureTreeIndices(capture, startIndex)
-
   getNextMatch: (line, position) ->
     nextMatch = null
     matchedPattern = null
@@ -156,16 +151,16 @@ class Pattern
       rule = @grammar.ruleForInclude(@include)
       rule.getNextMatch(line, position)
     else
-      { match: @regex.getCaptureTree(line, position), pattern: this }
+      { match: @regex.getCaptureIndices(line, position), pattern: this }
 
-  handleMatch: (stack, line, captureTree) ->
+  handleMatch: (stack, line, captureIndices) ->
     scopes = _.pluck(stack, "scopeName")
     scopes.push(@scopeName) unless @popRule
 
     if @captures
-      tokens = @getTokensForCaptureTree(line, captureTree, scopes)
+      tokens = @getTokensForCaptureIndices(line, captureIndices, scopes)
     else
-      [start, end] = captureTree[1..2]
+      [start, end] = captureIndices[1..2]
       tokens = [{ value: line[start...end], scopes: scopes }]
 
     if @pushRule
@@ -175,22 +170,24 @@ class Pattern
 
     tokens
 
-  getTokensForCaptureTree: (line, parentCapture, scopes) ->
-    [parentCaptureIndex, parentCaptureStart, parentCaptureEnd, childCaptures...] = parentCapture
+  getTokensForCaptureIndices: (line, captureIndices, scopes, indexOffset=captureIndices[0]) ->
+    [parentCaptureIndex, parentCaptureStart, parentCaptureEnd] = shiftCapture(captureIndices)
+    relativeParentCaptureIndex = parentCaptureIndex - indexOffset
 
     tokens = []
-    if scope = @captures[parentCaptureIndex]?.name
+    if scope = @captures[relativeParentCaptureIndex]?.name
       scopes = scopes.concat(scope)
 
     previousChildCaptureEnd = parentCaptureStart
-    for childCapture in childCaptures
-      [childCaptureIndex, childCaptureStart, childCaptureEnd] = childCapture
+    while captureIndices.length and captureIndices[1] < parentCaptureEnd
+      [childCaptureIndex, childCaptureStart, childCaptureEnd] = captureIndices
+
       if childCaptureStart > previousChildCaptureEnd
         tokens.push
           value: line[previousChildCaptureEnd...childCaptureStart]
           scopes: scopes
 
-      captureTokens = @getTokensForCaptureTree(line, childCapture, scopes)
+      captureTokens = @getTokensForCaptureIndices(line, captureIndices, scopes, indexOffset)
       tokens.push(captureTokens...)
       previousChildCaptureEnd = childCaptureEnd
 
@@ -200,3 +197,7 @@ class Pattern
         scopes: scopes
 
     tokens
+
+shiftCapture = (captureIndices) ->
+  [captureIndices.shift(), captureIndices.shift(), captureIndices.shift()]
+

src/stdlib/onig-reg-exp-extension.js

@@ -1,7 +1,7 @@
 (function() {
   native function buildOnigRegExp(source);
   native function search(string, index);
-  native function getCaptureTree(source, index);
+  native function getCaptureIndices(source, index);
   native function getCaptureCount();
 
   function OnigRegExp(source) {
@@ -13,7 +13,7 @@
   }
 
   OnigRegExp.prototype.search = search;
-  OnigRegExp.prototype.getCaptureTree = getCaptureTree;
+  OnigRegExp.prototype.getCaptureIndices = getCaptureIndices;
   OnigRegExp.prototype.getCaptureCount = getCaptureCount;
 
   this.OnigRegExp = OnigRegExp;