Bug 1542106 - Cache the last (line number, offset => column) mapping returned and use it to optimize a subsequent lookup that's further along in the same line. r=arai

Differential Revision: https://phabricator.services.mozilla.com/D26270

--HG--
extra : moz-landing-system : lando

js/src/frontend/TokenStream.cpp

@@ -690,11 +690,26 @@ uint32_t GeneralTokenStreamChars<Unit, AnyCharsAccess>::computeColumn(
 
   const TokenStreamAnyChars& anyChars = anyCharsAccess();
 
-  const Unit* begin =
-      this->sourceUnits.codeUnitPtrAt(anyChars.lineStart(lineToken));
+  uint32_t lineNumber = anyChars.srcCoords.lineNumber(lineToken);
+
+  uint32_t beginOffset;
+  uint32_t partialCols;
+  if (lineNumber == lastLineForColumn_ && lastOffsetForColumn_ <= offset) {
+    beginOffset = lastOffsetForColumn_;
+    partialCols = lastColumn_;
+  } else {
+    beginOffset = anyChars.lineStart(lineToken);
+    partialCols = 0;
+  }
+
+  const Unit* begin = this->sourceUnits.codeUnitPtrAt(beginOffset);
   const Unit* end = this->sourceUnits.codeUnitPtrAt(offset);
 
-  auto partialCols = AssertedCast<uint32_t>(ComputeColumn(begin, end));
+  partialCols += AssertedCast<uint32_t>(ComputeColumn(begin, end));
+
+  lastLineForColumn_ = lineNumber;
+  lastOffsetForColumn_ = offset;
+  lastColumn_ = partialCols;
   return (lineToken.isFirstLine() ? anyChars.options_.column : 0) + partialCols;
 }
 

js/src/frontend/TokenStream.h

@@ -1941,6 +1941,22 @@ class GeneralTokenStreamChars : public SpecializedTokenStreamCharsBase<Unit> {
     return static_cast<TokenStreamSpecific*>(this);
   }
 
+ private:
+  // Computing accurate column numbers requires linearly iterating through all
+  // source units in the line to account for multi-unit code points; on long
+  // lines requiring many column computations, this becomes quadratic.
+  //
+  // However, because usually we need columns for advancing offsets through
+  // scripts, caching the last ((line number, offset) => relative column)
+  // mapping -- in similar manner to how |SourceUnits::lastIndex_| is used to
+  // cache (offset => line number) mappings -- lets us avoid re-iterating
+  // through the line prefix in most cases.
+
+  mutable uint32_t lastLineForColumn_ = UINT32_MAX;
+  mutable uint32_t lastOffsetForColumn_ = UINT32_MAX;
+  mutable uint32_t lastColumn_ = 0;
+
+ protected:
   uint32_t computeColumn(LineToken lineToken, uint32_t offset) const;
   void computeLineAndColumn(uint32_t offset, uint32_t* line,
                             uint32_t* column) const;