Commit 031561e8 authored by Jeff Walden's avatar Jeff Walden
Browse files

Bug 1476866 - Replace SourceUnits<CharT>::isRawEOLChar (where CharT had rather... 

Bug 1476866 - Replace SourceUnits<CharT>::isRawEOLChar (where CharT had rather uncertain meaning) with standalone IsLineTerminator of clearer meaning.  r=arai
parent 563f25bd

js/src/frontend/TokenStream.cpp

+9 −12
Original line number Diff line number Diff line
@@ -583,7 +583,7 @@ TokenStreamChars<char16_t, AnyCharsAccess>::getNonAsciiCodePoint(int32_t lead, i 


            *codePoint = '\n';

        } else {

            MOZ_ASSERT(!SourceUnits::isRawEOLChar(*codePoint));

            MOZ_ASSERT(!IsLineTerminator(AssertedCast<char32_t>(*codePoint)));

        }



        return true;
@@ -593,13 +593,13 @@ TokenStreamChars<char16_t, AnyCharsAccess>::getNonAsciiCodePoint(int32_t lead, i 
    if (MOZ_UNLIKELY(this->sourceUnits.atEnd() ||

                     !unicode::IsTrailSurrogate(this->sourceUnits.peekCodeUnit())))

    {

        MOZ_ASSERT(!SourceUnits::isRawEOLChar(*codePoint));

        MOZ_ASSERT(!IsLineTerminator(AssertedCast<char32_t>(*codePoint)));

        return true;

    }



    // Otherwise we have a multi-unit code point.

    *codePoint = unicode::UTF16Decode(lead, this->sourceUnits.getCodeUnit());

    MOZ_ASSERT(!SourceUnits::isRawEOLChar(*codePoint));

    MOZ_ASSERT(!IsLineTerminator(AssertedCast<char32_t>(*codePoint)));

    return true;

}
@@ -644,7 +644,7 @@ SourceUnits<char16_t>::findWindowStart(size_t offset) 
        // This stops at U+2028 LINE SEPARATOR or U+2029 PARAGRAPH SEPARATOR in

        // string and template literals.  These code points do affect line and

        // column coordinates, even as they encode their literal values.

        if (isRawEOLChar(c))

        if (IsLineTerminator(c))

            break;



        // Don't allow invalid UTF-16 in pre-context.  (Current users don't
@@ -697,7 +697,7 @@ SourceUnits<char16_t>::findWindowEnd(size_t offset) 
        // This stops at U+2028 LINE SEPARATOR or U+2029 PARAGRAPH SEPARATOR in

        // string and template literals.  These code points do affect line and

        // column coordinates, even as they encode their literal values.

        if (isRawEOLChar(c))

        if (IsLineTerminator(c))

            break;



        // Don't allow invalid UTF-16 in post-context.  (Current users don't
@@ -1601,7 +1601,7 @@ SpecializedTokenStreamCharsBase<char16_t>::infallibleConsumeRestOfSingleLineComm 
{

    while (MOZ_LIKELY(!this->sourceUnits.atEnd())) {

        char16_t unit = this->sourceUnits.peekCodeUnit();

        if (SourceUnits::isRawEOLChar(unit))

        if (IsLineTerminator(unit))

            return;



        this->sourceUnits.consumeKnownCodeUnit(unit);
@@ -1777,17 +1777,14 @@ TokenStreamSpecific<CharT, AnyCharsAccess>::regexpLiteral(TokenStart start, Toke 
            break;

        }



        // NOTE: Non-ASCII LineTerminators were handled by

        //       ProcessNonAsciiCodePoint calls above.

        if (unit == '\r' || unit == '\n') {

            ReportUnterminatedRegExp(unit);

            return badToken();

        }



        // We're accumulating regular expression *source* text here: source

        // text matching a line break will appear as U+005C REVERSE SOLIDUS

        // U+006E LATIN SMALL LETTER N, and |unit| here would be the latter

        // code point.

        MOZ_ASSERT(!SourceUnits::isRawEOLChar(unit));



        MOZ_ASSERT(!IsLineTerminator(AssertedCast<char32_t>(unit)));

        if (!this->charBuffer.append(unit))

            return badToken();

    } while (true);

js/src/frontend/TokenStream.h

+20 −7
Original line number Diff line number Diff line
@@ -949,6 +949,26 @@ CodeUnitValue(mozilla::Utf8Unit unit) 
template<typename CharT>

class TokenStreamCharsBase;



template<typename T>

inline bool

IsLineTerminator(T) = delete;



inline bool

IsLineTerminator(char32_t codePoint)

{

    return codePoint == '\n' ||

           codePoint == '\r' ||

           codePoint == unicode::LINE_SEPARATOR ||

           codePoint == unicode::PARA_SEPARATOR;

}



inline bool

IsLineTerminator(char16_t unit)

{

    // Every LineTerminator fits in char16_t, so this is exact.

    return IsLineTerminator(static_cast<char32_t>(unit));

}



// This is the low-level interface to the JS source code buffer.  It just gets

// raw Unicode code units -- 16-bit char16_t units of source text that are not

// (always) full code points, and 8-bit units of UTF-8 source text soon.
@@ -1132,13 +1152,6 @@ class SourceUnits 
#endif

    }



    static bool isRawEOLChar(int32_t c) {

        return c == '\n' ||

               c == '\r' ||

               c == unicode::LINE_SEPARATOR ||

               c == unicode::PARA_SEPARATOR;

    }



    /**

     * The maximum radius of code around the location of an error that should

     * be included in a syntax error message -- this many code units to either