From 3fe3f8410626706688f66e06f4f52e57c1c96d72 Mon Sep 17 00:00:00 2001 From: Nigel Tao Date: Thu, 2 Apr 2020 21:56:53 +1100 Subject: [PATCH] Add double-curly blocks --- doc/changelog.md | 1 + doc/wuffs-the-language.md | 4 + lang/parse/parse.go | 63 +- lang/render/render.go | 10 +- lang/token/list.go | 38 +- std/gif/decode_config.wuffs | 185 ++-- std/gif/decode_gif.wuffs | 185 ++-- std/json/decode_json.wuffs | 1949 ++++++++++++++++++----------------- 8 files changed, 1243 insertions(+), 1192 deletions(-) diff --git a/doc/changelog.md b/doc/changelog.md index b6a30615..f5639cd9 100644 --- a/doc/changelog.md +++ b/doc/changelog.md @@ -14,6 +14,7 @@ - Added `std/wbmp`. - Added `WUFFS_BASE__PIXEL_BLEND__SRC_OVER`. - Added `WUFFS_BASE__PIXEL_FORMAT__BGR_565`. +- Added double-curly blocks. - Added interfaces. - Added preprocessor. - Added single-quoted strings. diff --git a/doc/wuffs-the-language.md b/doc/wuffs-the-language.md index 1c081b49..f93beb6e 100644 --- a/doc/wuffs-the-language.md +++ b/doc/wuffs-the-language.md @@ -30,6 +30,10 @@ Minor features or restrictions include: - Labeled jumps look like `break.loopname` and `continue.loopname`, for a matching `while.loopname`. The `while.loopname`'s closing curly brace must be followed by `endwhile.loopname`. +- A `while true` loop's block may use double-curly braces: `{{` and `}}`. The + formatter will not add an indent to the code inside the block. This is useful + when using `while true {{ etc; break; etc; break }}` to simulate what would + be a (forwards) `goto` in other languages' straight-line code. Wuffs code is formatted by the [`wuffsfmt`](https://godoc.org/github.com/google/wuffs/cmd/wuffsfmt) program. diff --git a/lang/parse/parse.go b/lang/parse/parse.go index 3a931cbf..a3053945 100644 --- a/lang/parse/parse.go +++ b/lang/parse/parse.go @@ -205,7 +205,7 @@ func (p *parser) parseTopLevelDecl() (*a.Node, error) { } p.allowVar = true - body, err := p.parseBlock() + body, err := p.parseBlock(false) if err != nil { return nil, err } @@ -341,11 +341,13 @@ func (p *parser) parseList(stop t.ID, parseElem func(*parser) (*a.Node, error)) ret := []*a.Node(nil) for len(p.src) > 0 { - if p.src[0].ID == stop { + if id := p.src[0].ID; id == stop { if stop == t.IDCloseParen || stop == t.IDCloseBracket { p.src = p.src[1:] } return ret, nil + } else if (stop == t.IDOpenDoubleCurly) && (id == t.IDOpenCurly) { + return ret, nil } elem, err := parseElem(p) @@ -531,18 +533,34 @@ func (p *parser) parseBracket(sep t.ID) (op t.ID, ei *a.Expr, ej *a.Expr, err er return sep, ei, ej, nil } -func (p *parser) parseBlock() ([]*a.Node, error) { - if x := p.peek1(); x != t.IDOpenCurly { - got := p.tm.ByID(x) - return nil, fmt.Errorf(`parse: expected "{", got %q at %s:%d`, got, p.filename, p.line()) +func (p *parser) parseBlock(doubleCurly bool) ([]*a.Node, error) { + if doubleCurly { + if x := p.peek1(); x != t.IDOpenDoubleCurly { + got := p.tm.ByID(x) + return nil, fmt.Errorf(`parse: expected "{{", got %q at %s:%d`, got, p.filename, p.line()) + } + } else { + if x := p.peek1(); x != t.IDOpenCurly { + got := p.tm.ByID(x) + return nil, fmt.Errorf(`parse: expected "{", got %q at %s:%d`, got, p.filename, p.line()) + } } p.src = p.src[1:] block := []*a.Node(nil) - for len(p.src) > 0 { - if p.src[0].ID == t.IDCloseCurly { - p.src = p.src[1:] - return block, nil + for { + if len(p.src) == 0 { + return nil, fmt.Errorf(`parse: expected "}" or "}}" at %s:%d`, p.filename, p.line()) + } + + if doubleCurly { + if p.src[0].ID == t.IDCloseDoubleCurly { + break + } + } else { + if p.src[0].ID == t.IDCloseCurly { + break + } } s, err := p.parseStatement() @@ -557,7 +575,9 @@ func (p *parser) parseBlock() ([]*a.Node, error) { } p.src = p.src[1:] } - return nil, fmt.Errorf(`parse: expected "}" at %s:%d`, p.filename, p.line()) + + p.src = p.src[1:] + return block, nil } func (p *parser) assertsSorted(asserts []*a.Node) error { @@ -735,7 +755,12 @@ func (p *parser) parseStatement1() (*a.Node, error) { if err != nil { return nil, err } - body, err := p.parseBlock() + doubleCurly := p.peek1() == t.IDOpenDoubleCurly + if doubleCurly && ((condition.Operator() != 0) || (condition.Ident() != t.IDTrue)) { + return nil, fmt.Errorf(`parse: double {{ }} while loop condition isn't "true" at %s:%d`, + p.filename, p.line()) + } + body, err := p.parseBlock(doubleCurly) if err != nil { return nil, err } @@ -747,6 +772,10 @@ func (p *parser) parseStatement1() (*a.Node, error) { return nil, fmt.Errorf(`parse: expected endwhile%s at %s:%d`, dotLabel, p.filename, p.line()) } + if doubleCurly && !a.Terminates(body) { + return nil, fmt.Errorf(`parse: double {{ }} while loop doesn't terminate at %s:%d`, + p.filename, p.line()) + } return a.NewWhile(label, condition, asserts, body).AsNode(), nil } return p.parseAssignNode() @@ -838,7 +867,7 @@ func (p *parser) parseAsserts() ([]*a.Node, error) { if p.peek1() == t.IDComma { p.src = p.src[1:] var err error - if asserts, err = p.parseList(t.IDOpenCurly, (*parser).parseAssertNode); err != nil { + if asserts, err = p.parseList(t.IDOpenDoubleCurly, (*parser).parseAssertNode); err != nil { return nil, err } if err := p.assertsSorted(asserts); err != nil { @@ -916,7 +945,7 @@ func (p *parser) parseIOBindNode() (*a.Node, error) { } p.src = p.src[1:] - body, err := p.parseBlock() + body, err := p.parseBlock(false) if err != nil { return nil, err } @@ -938,7 +967,7 @@ func (p *parser) parseIf() (*a.If, error) { return nil, fmt.Errorf(`parse: if-condition %q is not effect-free at %s:%d`, condition.Str(p.tm), p.filename, p.line()) } - bodyIfTrue, err := p.parseBlock() + bodyIfTrue, err := p.parseBlock(false) if err != nil { return nil, err } @@ -951,7 +980,7 @@ func (p *parser) parseIf() (*a.If, error) { return nil, err } } else { - bodyIfFalse, err = p.parseBlock() + bodyIfFalse, err = p.parseBlock(false) if err != nil { return nil, err } @@ -1042,7 +1071,7 @@ func (p *parser) parseIterateBlock(label t.ID, assigns []*a.Node) (*a.Iterate, e if err != nil { return nil, err } - body, err := p.parseBlock() + body, err := p.parseBlock(false) if err != nil { return nil, err } diff --git a/lang/render/render.go b/lang/render/render.go index 2efa21b1..64c30cff 100644 --- a/lang/render/render.go +++ b/lang/render/render.go @@ -110,9 +110,11 @@ func Render(w io.Writer, tm *t.Map, src []t.Token, comments []string) (err error // line containing the matching open token. buf = buf[:0] indentAdjustment := 0 - if lineTokens[0].ID.IsClose() { + if id := lineTokens[0].ID; id == t.IDCloseDoubleCurly { + // No-op. + } else if id.IsClose() { indentAdjustment-- - } else if hanging && lineTokens[0].ID != t.IDOpenCurly { + } else if hanging && ((id != t.IDOpenCurly) && (id != t.IDOpenDoubleCurly)) { indentAdjustment++ } buf = appendTabs(buf, indent+indentAdjustment) @@ -189,7 +191,9 @@ func Render(w io.Writer, tm *t.Map, src []t.Token, comments []string) (err error } commentLine = line + 1 prevLine = line - prevLineHanging = prevLineHanging && lineTokens[len(lineTokens)-1].ID != t.IDOpenCurly + lastID := lineTokens[len(lineTokens)-1].ID + prevLineHanging = prevLineHanging && + (lastID != t.IDOpenCurly) && (lastID != t.IDOpenDoubleCurly) } // Print any trailing comments. diff --git a/lang/token/list.go b/lang/token/list.go index 867d334b..c38c9631 100644 --- a/lang/token/list.go +++ b/lang/token/list.go @@ -242,16 +242,18 @@ const ( minOpen = 0x10 maxOpen = 0x17 - IDOpenParen = ID(0x10) - IDOpenBracket = ID(0x11) - IDOpenCurly = ID(0x12) + IDOpenParen = ID(0x10) + IDOpenBracket = ID(0x11) + IDOpenCurly = ID(0x12) + IDOpenDoubleCurly = ID(0x13) minClose = 0x18 maxClose = 0x1F - IDCloseParen = ID(0x18) - IDCloseBracket = ID(0x19) - IDCloseCurly = ID(0x1A) + IDCloseParen = ID(0x18) + IDCloseBracket = ID(0x19) + IDCloseCurly = ID(0x1A) + IDCloseDoubleCurly = ID(0x1B) ) const ( @@ -670,13 +672,15 @@ var builtInsByID = [nBuiltInIDs]string{ IDQuestion: "?", IDColon: ":", - IDOpenParen: "(", - IDCloseParen: ")", - IDOpenBracket: "[", + IDOpenParen: "(", + IDOpenBracket: "[", + IDOpenCurly: "{", + IDOpenDoubleCurly: "{{", - IDCloseBracket: "]", - IDOpenCurly: "{", - IDCloseCurly: "}", + IDCloseParen: ")", + IDCloseBracket: "]", + IDCloseCurly: "}", + IDCloseDoubleCurly: "}}", IDPlusEq: "+=", IDMinusEq: "-=", @@ -1029,8 +1033,6 @@ var squiggles = [256]ID{ ')': IDCloseParen, '[': IDOpenBracket, ']': IDCloseBracket, - '{': IDOpenCurly, - '}': IDCloseCurly, ',': IDComma, '!': IDExclam, @@ -1105,6 +1107,14 @@ var lexers = [256][]suffixLexer{ {"=", IDGreaterEq}, {"", IDGreaterThan}, }, + '{': { + {"{", IDOpenDoubleCurly}, + {"", IDOpenCurly}, + }, + '}': { + {"}", IDCloseDoubleCurly}, + {"", IDCloseCurly}, + }, '~': { {"mod<<=", IDTildeModShiftLEq}, {"mod<<", IDTildeModShiftL}, diff --git a/std/gif/decode_config.wuffs b/std/gif/decode_config.wuffs index f86ec9c5..d84935a7 100644 --- a/std/gif/decode_config.wuffs +++ b/std/gif/decode_config.wuffs @@ -544,105 +544,106 @@ pri func config_decoder.decode_ae?(src: base.io_reader) { var is_iccp : base.bool var is_xmp : base.bool - while.goto_done true { - block_size = args.src.read_u8?() - if block_size == 0 { - return ok - } + while.goto_done true {{ + block_size = args.src.read_u8?() + if block_size == 0 { + return ok + } - // Look only for an 11 byte "ANIMEXTS1.0", "NETSCAPE2.0" or other - // extension, as per: - // - http://www.vurdalakov.net/misc/gif/animexts-looping-application-extension - // - http://www.vurdalakov.net/misc/gif/netscape-looping-application-extension - // - // Other extensions include XMP metadata. - if block_size <> 11 { + // Look only for an 11 byte "ANIMEXTS1.0", "NETSCAPE2.0" or other + // extension, as per: + // - http://www.vurdalakov.net/misc/gif/animexts-looping-application-extension + // - http://www.vurdalakov.net/misc/gif/netscape-looping-application-extension + // + // Other extensions include XMP metadata. + if block_size <> 11 { + args.src.skip32?(n: block_size as base.u32) + break.goto_done + } + is_animexts = true + is_netscape = true + is_iccp = true + is_xmp = true + block_size = 0 // Re-purpose the block_size variable as a counter. + while block_size < 11 { + c = args.src.read_u8?() + is_animexts = is_animexts and (c == animexts1dot0[block_size]) + is_netscape = is_netscape and (c == netscape2dot0[block_size]) + is_iccp = is_iccp and (c == iccrgbg1012[block_size]) + is_xmp = is_xmp and (c == xmpdataxmp[block_size]) + block_size += 1 + } endwhile + + if is_animexts or is_netscape { + // Those 11 bytes should be followed by 0x03, 0x01 and then the loop + // count. + block_size = args.src.read_u8?() + if block_size <> 3 { args.src.skip32?(n: block_size as base.u32) break.goto_done } - is_animexts = true - is_netscape = true - is_iccp = true - is_xmp = true - block_size = 0 // Re-purpose the block_size variable as a counter. - while block_size < 11 { - c = args.src.read_u8?() - is_animexts = is_animexts and (c == animexts1dot0[block_size]) - is_netscape = is_netscape and (c == netscape2dot0[block_size]) - is_iccp = is_iccp and (c == iccrgbg1012[block_size]) - is_xmp = is_xmp and (c == xmpdataxmp[block_size]) - block_size += 1 - } endwhile + c = args.src.read_u8?() + if c <> 0x01 { + args.src.skip32?(n: 2) + break.goto_done + } + this.num_loops = args.src.read_u16le_as_u32?() + this.seen_num_loops = true - if is_animexts or is_netscape { - // Those 11 bytes should be followed by 0x03, 0x01 and then the loop - // count. - block_size = args.src.read_u8?() - if block_size <> 3 { - args.src.skip32?(n: block_size as base.u32) - break.goto_done - } - c = args.src.read_u8?() - if c <> 0x01 { - args.src.skip32?(n: 2) - break.goto_done - } - this.num_loops = args.src.read_u16le_as_u32?() - this.seen_num_loops = true - - // A loop count of N, in the wire format, actually means "repeat N - // times after the first play", if N is positive. A zero N means to - // loop forever. Playing the frames exactly once is denoted by the - // *absence* of this NETSCAPE2.0 application extension. - // - // For example, if there are four frames: A, B, C, D, and N is 2, then - // each frame is actually played N+1 or 3 times: ABCDABCDABCD. - // - // Thus, we increment N if it is positive. The comparison against - // 0xFFFF will never fail, but is necessary for the overflow checker. - if (0 < this.num_loops) and (this.num_loops <= 0xFFFF) { - this.num_loops += 1 - } - - } else if this.ignore_metadata { - // No-op. - - } else if is_iccp and this.report_metadata_iccp { - while args.src.available() <= 0, - post args.src.available() > 0, - { - yield? base."$short read" - } endwhile - this.metadata_chunk_length_value = args.src.peek_u8_as_u64() - args.src.skip32_fast!(actual: 1, worst_case: 1) - this.metadata_fourcc_value = 'ICCP'be - this.metadata_io_position = args.src.position() ~sat+ this.metadata_chunk_length_value - this.call_sequence = 1 - return base."@metadata reported" - - } else if is_xmp and this.report_metadata_xmp { - while args.src.available() <= 0, - post args.src.available() > 0, - { - yield? base."$short read" - } endwhile - this.metadata_chunk_length_value = args.src.peek_u8_as_u64() - if this.metadata_chunk_length_value > 0 { - // The +1 is because XMP metadata's encoding includes each - // block's leading byte (the block size) as part of the - // metadata passed to the caller. - this.metadata_chunk_length_value += 1 - } else { - args.src.skip32_fast!(actual: 1, worst_case: 1) - } - this.metadata_fourcc_value = 'XMP 'be - this.metadata_io_position = args.src.position() ~sat+ this.metadata_chunk_length_value - this.call_sequence = 1 - return base."@metadata reported" + // A loop count of N, in the wire format, actually means "repeat N + // times after the first play", if N is positive. A zero N means to + // loop forever. Playing the frames exactly once is denoted by the + // *absence* of this NETSCAPE2.0 application extension. + // + // For example, if there are four frames: A, B, C, D, and N is 2, then + // each frame is actually played N+1 or 3 times: ABCDABCDABCD. + // + // Thus, we increment N if it is positive. The comparison against + // 0xFFFF will never fail, but is necessary for the overflow checker. + if (0 < this.num_loops) and (this.num_loops <= 0xFFFF) { + this.num_loops += 1 } - break.goto_done - } endwhile.goto_done + } else if this.ignore_metadata { + // No-op. + + } else if is_iccp and this.report_metadata_iccp { + while args.src.available() <= 0, + post args.src.available() > 0, + { + yield? base."$short read" + } endwhile + this.metadata_chunk_length_value = args.src.peek_u8_as_u64() + args.src.skip32_fast!(actual: 1, worst_case: 1) + this.metadata_fourcc_value = 'ICCP'be + this.metadata_io_position = args.src.position() ~sat+ this.metadata_chunk_length_value + this.call_sequence = 1 + return base."@metadata reported" + + } else if is_xmp and this.report_metadata_xmp { + while args.src.available() <= 0, + post args.src.available() > 0, + { + yield? base."$short read" + } endwhile + this.metadata_chunk_length_value = args.src.peek_u8_as_u64() + if this.metadata_chunk_length_value > 0 { + // The +1 is because XMP metadata's encoding includes each + // block's leading byte (the block size) as part of the + // metadata passed to the caller. + this.metadata_chunk_length_value += 1 + } else { + args.src.skip32_fast!(actual: 1, worst_case: 1) + } + this.metadata_fourcc_value = 'XMP 'be + this.metadata_io_position = args.src.position() ~sat+ this.metadata_chunk_length_value + this.call_sequence = 1 + return base."@metadata reported" + } + + break.goto_done + }} endwhile.goto_done + this.skip_blocks?(src: args.src) } diff --git a/std/gif/decode_gif.wuffs b/std/gif/decode_gif.wuffs index 0d56bda3..232ea29f 100644 --- a/std/gif/decode_gif.wuffs +++ b/std/gif/decode_gif.wuffs @@ -619,105 +619,106 @@ pri func decoder.decode_ae?(src: base.io_reader) { var is_iccp : base.bool var is_xmp : base.bool - while.goto_done true { - block_size = args.src.read_u8?() - if block_size == 0 { - return ok - } + while.goto_done true {{ + block_size = args.src.read_u8?() + if block_size == 0 { + return ok + } - // Look only for an 11 byte "ANIMEXTS1.0", "NETSCAPE2.0" or other - // extension, as per: - // - http://www.vurdalakov.net/misc/gif/animexts-looping-application-extension - // - http://www.vurdalakov.net/misc/gif/netscape-looping-application-extension - // - // Other extensions include XMP metadata. - if block_size <> 11 { + // Look only for an 11 byte "ANIMEXTS1.0", "NETSCAPE2.0" or other + // extension, as per: + // - http://www.vurdalakov.net/misc/gif/animexts-looping-application-extension + // - http://www.vurdalakov.net/misc/gif/netscape-looping-application-extension + // + // Other extensions include XMP metadata. + if block_size <> 11 { + args.src.skip32?(n: block_size as base.u32) + break.goto_done + } + is_animexts = true + is_netscape = true + is_iccp = true + is_xmp = true + block_size = 0 // Re-purpose the block_size variable as a counter. + while block_size < 11 { + c = args.src.read_u8?() + is_animexts = is_animexts and (c == animexts1dot0[block_size]) + is_netscape = is_netscape and (c == netscape2dot0[block_size]) + is_iccp = is_iccp and (c == iccrgbg1012[block_size]) + is_xmp = is_xmp and (c == xmpdataxmp[block_size]) + block_size += 1 + } endwhile + + if is_animexts or is_netscape { + // Those 11 bytes should be followed by 0x03, 0x01 and then the loop + // count. + block_size = args.src.read_u8?() + if block_size <> 3 { args.src.skip32?(n: block_size as base.u32) break.goto_done } - is_animexts = true - is_netscape = true - is_iccp = true - is_xmp = true - block_size = 0 // Re-purpose the block_size variable as a counter. - while block_size < 11 { - c = args.src.read_u8?() - is_animexts = is_animexts and (c == animexts1dot0[block_size]) - is_netscape = is_netscape and (c == netscape2dot0[block_size]) - is_iccp = is_iccp and (c == iccrgbg1012[block_size]) - is_xmp = is_xmp and (c == xmpdataxmp[block_size]) - block_size += 1 - } endwhile + c = args.src.read_u8?() + if c <> 0x01 { + args.src.skip32?(n: 2) + break.goto_done + } + this.num_loops = args.src.read_u16le_as_u32?() + this.seen_num_loops = true - if is_animexts or is_netscape { - // Those 11 bytes should be followed by 0x03, 0x01 and then the loop - // count. - block_size = args.src.read_u8?() - if block_size <> 3 { - args.src.skip32?(n: block_size as base.u32) - break.goto_done - } - c = args.src.read_u8?() - if c <> 0x01 { - args.src.skip32?(n: 2) - break.goto_done - } - this.num_loops = args.src.read_u16le_as_u32?() - this.seen_num_loops = true - - // A loop count of N, in the wire format, actually means "repeat N - // times after the first play", if N is positive. A zero N means to - // loop forever. Playing the frames exactly once is denoted by the - // *absence* of this NETSCAPE2.0 application extension. - // - // For example, if there are four frames: A, B, C, D, and N is 2, then - // each frame is actually played N+1 or 3 times: ABCDABCDABCD. - // - // Thus, we increment N if it is positive. The comparison against - // 0xFFFF will never fail, but is necessary for the overflow checker. - if (0 < this.num_loops) and (this.num_loops <= 0xFFFF) { - this.num_loops += 1 - } - - } else if this.ignore_metadata { - // No-op. - - } else if is_iccp and this.report_metadata_iccp { - while args.src.available() <= 0, - post args.src.available() > 0, - { - yield? base."$short read" - } endwhile - this.metadata_chunk_length_value = args.src.peek_u8_as_u64() - args.src.skip32_fast!(actual: 1, worst_case: 1) - this.metadata_fourcc_value = 'ICCP'be - this.metadata_io_position = args.src.position() ~sat+ this.metadata_chunk_length_value - this.call_sequence = 1 - return base."@metadata reported" - - } else if is_xmp and this.report_metadata_xmp { - while args.src.available() <= 0, - post args.src.available() > 0, - { - yield? base."$short read" - } endwhile - this.metadata_chunk_length_value = args.src.peek_u8_as_u64() - if this.metadata_chunk_length_value > 0 { - // The +1 is because XMP metadata's encoding includes each - // block's leading byte (the block size) as part of the - // metadata passed to the caller. - this.metadata_chunk_length_value += 1 - } else { - args.src.skip32_fast!(actual: 1, worst_case: 1) - } - this.metadata_fourcc_value = 'XMP 'be - this.metadata_io_position = args.src.position() ~sat+ this.metadata_chunk_length_value - this.call_sequence = 1 - return base."@metadata reported" + // A loop count of N, in the wire format, actually means "repeat N + // times after the first play", if N is positive. A zero N means to + // loop forever. Playing the frames exactly once is denoted by the + // *absence* of this NETSCAPE2.0 application extension. + // + // For example, if there are four frames: A, B, C, D, and N is 2, then + // each frame is actually played N+1 or 3 times: ABCDABCDABCD. + // + // Thus, we increment N if it is positive. The comparison against + // 0xFFFF will never fail, but is necessary for the overflow checker. + if (0 < this.num_loops) and (this.num_loops <= 0xFFFF) { + this.num_loops += 1 } - break.goto_done - } endwhile.goto_done + } else if this.ignore_metadata { + // No-op. + + } else if is_iccp and this.report_metadata_iccp { + while args.src.available() <= 0, + post args.src.available() > 0, + { + yield? base."$short read" + } endwhile + this.metadata_chunk_length_value = args.src.peek_u8_as_u64() + args.src.skip32_fast!(actual: 1, worst_case: 1) + this.metadata_fourcc_value = 'ICCP'be + this.metadata_io_position = args.src.position() ~sat+ this.metadata_chunk_length_value + this.call_sequence = 1 + return base."@metadata reported" + + } else if is_xmp and this.report_metadata_xmp { + while args.src.available() <= 0, + post args.src.available() > 0, + { + yield? base."$short read" + } endwhile + this.metadata_chunk_length_value = args.src.peek_u8_as_u64() + if this.metadata_chunk_length_value > 0 { + // The +1 is because XMP metadata's encoding includes each + // block's leading byte (the block size) as part of the + // metadata passed to the caller. + this.metadata_chunk_length_value += 1 + } else { + args.src.skip32_fast!(actual: 1, worst_case: 1) + } + this.metadata_fourcc_value = 'XMP 'be + this.metadata_io_position = args.src.position() ~sat+ this.metadata_chunk_length_value + this.call_sequence = 1 + return base."@metadata reported" + } + + break.goto_done + }} endwhile.goto_done + this.skip_blocks?(src: args.src) } diff --git a/std/json/decode_json.wuffs b/std/json/decode_json.wuffs index ebe3ca18..367bfacb 100644 --- a/std/json/decode_json.wuffs +++ b/std/json/decode_json.wuffs @@ -157,617 +157,173 @@ pub func decoder.decode_tokens?(dst: base.token_writer, src: base.io_reader, wor expect = 0x1EB2 // EXPECT_VALUE while.outer true { - while.goto_parsed_a_leaf_value true { - if args.dst.available() <= 0 { - yield? base."$short write" - continue.outer - } + while.goto_parsed_a_leaf_value true {{ + if args.dst.available() <= 0 { + yield? base."$short write" + continue.outer + } - // Consume whitespace. - whitespace_length = 0 - c = 0 - class = 0 - while.ws true, - inv args.dst.available() > 0, - post args.src.available() > 0, - { - if args.src.available() <= 0 { - if whitespace_length > 0 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0, - link: 0x0, - length: whitespace_length) - whitespace_length = 0 - } - if args.src.is_closed() { - return "#bad input" - } - yield? base."$short read" - whitespace_length = 0 - continue.outer - } - - c = args.src.peek_u8() - class = lut_classes[c] - if class <> 0x00 { // 0x00 is CLASS_WHITESPACE. - break.ws - } - args.src.skip32_fast!(actual: 1, worst_case: 1) - - if whitespace_length >= 0xFFFE { + // Consume whitespace. + whitespace_length = 0 + c = 0 + class = 0 + while.ws true, + inv args.dst.available() > 0, + post args.src.available() > 0, + { + if args.src.available() <= 0 { + if whitespace_length > 0 { args.dst.write_simple_token_fast!( value_major: 0, value_minor: 0, link: 0x0, - length: 0xFFFF) + length: whitespace_length) whitespace_length = 0 - continue.outer } - whitespace_length += 1 - } endwhile.ws + if args.src.is_closed() { + return "#bad input" + } + yield? base."$short read" + whitespace_length = 0 + continue.outer + } - // Emit whitespace. - if whitespace_length > 0 { + c = args.src.peek_u8() + class = lut_classes[c] + if class <> 0x00 { // 0x00 is CLASS_WHITESPACE. + break.ws + } + args.src.skip32_fast!(actual: 1, worst_case: 1) + + if whitespace_length >= 0xFFFE { args.dst.write_simple_token_fast!( value_major: 0, value_minor: 0, link: 0x0, - length: whitespace_length) + length: 0xFFFF) whitespace_length = 0 + continue.outer + } + whitespace_length += 1 + } endwhile.ws + + // Emit whitespace. + if whitespace_length > 0 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0, + link: 0x0, + length: whitespace_length) + whitespace_length = 0 + if args.dst.available() <= 0 { + continue.outer + } + } + + // Check expected character classes. + if 0 == (expect & ((1 as base.u32) << class)) { + return "#bad input" + } + + // These assertions are redundant (the Wuffs compiler should already + // know these facts; deleting these assertions should still compile) + // but are listed explicitly to guard against future edits to the code + // above inadvertently invalidating these assertions. + assert args.dst.available() > 0 + assert args.src.available() > 0 + + if class == 0x01 { // 0x01 is CLASS_STRING. + // -------- BEGIN parse strings. + // Emit the leading '"'. + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0013, + link: 0x1, + length: 1) + args.src.skip32_fast!(actual: 1, worst_case: 1) + + while.string_loop_outer true { if args.dst.available() <= 0 { - continue.outer + yield? base."$short write" + continue.string_loop_outer } - } - // Check expected character classes. - if 0 == (expect & ((1 as base.u32) << class)) { - return "#bad input" - } - - // These assertions are redundant (the Wuffs compiler should - // already know these facts; deleting these assertions should still - // compile) but are listed explicitly to guard against future edits - // to the code above inadvertently invalidating these assertions. - assert args.dst.available() > 0 - assert args.src.available() > 0 - - if class == 0x01 { // 0x01 is CLASS_STRING. - // -------- BEGIN parse strings. - // Emit the leading '"'. - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0013, - link: 0x1, - length: 1) - args.src.skip32_fast!(actual: 1, worst_case: 1) - - while.string_loop_outer true { - if args.dst.available() <= 0 { - yield? base."$short write" + string_length = 0 + while.string_loop_inner true, + pre args.dst.available() > 0, + { + if args.src.available() <= 0 { + if string_length > 0 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0021, + link: 0x3, + length: string_length) + string_length = 0 + } + if args.src.is_closed() { + return "#bad input" + } + yield? base."$short read" + string_length = 0 continue.string_loop_outer } - string_length = 0 - while.string_loop_inner true, - pre args.dst.available() > 0, + // As an optimization, consume non-special ASCII 4 bytes at + // a time. + while args.src.available() > 4, + inv args.dst.available() > 0, + inv args.src.available() > 0, { - if args.src.available() <= 0 { - if string_length > 0 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: string_length) - string_length = 0 - } - if args.src.is_closed() { - return "#bad input" - } - yield? base."$short read" + c4 = args.src.peek_u32le() + if 0x00 <> (lut_chars[0xFF & (c4 >> 0)] | + lut_chars[0xFF & (c4 >> 8)] | + lut_chars[0xFF & (c4 >> 16)] | + lut_chars[0xFF & (c4 >> 24)]) { + break + } + args.src.skip32_fast!(actual: 4, worst_case: 4) + if string_length > (0xFFFB - 4) { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0021, + link: 0x3, + length: string_length + 4) string_length = 0 continue.string_loop_outer } + string_length += 4 + } endwhile - // As an optimization, consume non-special ASCII 4 bytes at a time. - while args.src.available() > 4, - inv args.dst.available() > 0, - inv args.src.available() > 0, - { - c4 = args.src.peek_u32le() - if 0x00 <> (lut_chars[0xFF & (c4 >> 0)] | - lut_chars[0xFF & (c4 >> 8)] | - lut_chars[0xFF & (c4 >> 16)] | - lut_chars[0xFF & (c4 >> 24)]) { - break - } - args.src.skip32_fast!(actual: 4, worst_case: 4) - if string_length > (0xFFFB - 4) { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: string_length + 4) - string_length = 0 - continue.string_loop_outer - } - string_length += 4 - } endwhile + c = args.src.peek_u8() + char = lut_chars[c] - c = args.src.peek_u8() - char = lut_chars[c] - - if char == 0x00 { // Non-special ASCII. - args.src.skip32_fast!(actual: 1, worst_case: 1) - if string_length >= 0xFFFB { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: 0xFFFC) - string_length = 0 - continue.string_loop_outer - } - string_length += 1 - continue.string_loop_inner - - } else if char == 0x01 { // '"' - if string_length <> 0 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: string_length) - string_length = 0 - } - break.string_loop_outer - - } else if char == 0x02 { // '\\'. - if string_length > 0 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: string_length) - string_length = 0 - if args.dst.available() <= 0 { - continue.string_loop_outer - } - } - assert args.dst.available() > 0 - - if args.src.available() < 2 { - if args.src.is_closed() { - return "#bad backslash-escape" - } - yield? base."$short read" - string_length = 0 - char = 0 - continue.string_loop_outer - } - c = (args.src.peek_u16le() >> 8) as base.u8 - backslash = lut_backslashes[c] - if (backslash & 0x80) <> 0 { - args.src.skip32_fast!(actual: 2, worst_case: 2) - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_0000 | ((backslash & 0x7F) as base.u32), - link: 0x3, - length: 2) - continue.string_loop_outer - - } else if backslash <> 0 { - if this.quirk_enabled_allow_backslash_etc[backslash & 7] { - args.src.skip32_fast!(actual: 2, worst_case: 2) - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_0000 | (lut_quirky_backslashes[backslash & 7] as base.u32), - link: 0x3, - length: 2) - continue.string_loop_outer - } - - } else if c == 'u' { - // -------- BEGIN backslash-u. - if args.src.available() < 6 { - if args.src.is_closed() { - return "#bad backslash-escape" - } - yield? base."$short read" - string_length = 0 - char = 0 - continue.string_loop_outer - } - - uni4_string = args.src.peek_u48le_as_u64() >> 16 - uni4_value = 0 - uni4_ok = 0x80 - - c = lut_hexadecimal_digits[0xFF & (uni4_string >> 0)] - uni4_ok &= c - uni4_value |= ((c & 0x0F) as base.u32) << 12 - c = lut_hexadecimal_digits[0xFF & (uni4_string >> 8)] - uni4_ok &= c - uni4_value |= ((c & 0x0F) as base.u32) << 8 - c = lut_hexadecimal_digits[0xFF & (uni4_string >> 16)] - uni4_ok &= c - uni4_value |= ((c & 0x0F) as base.u32) << 4 - c = lut_hexadecimal_digits[0xFF & (uni4_string >> 24)] - uni4_ok &= c - uni4_value |= ((c & 0x0F) as base.u32) << 0 - - if uni4_ok == 0 { - // It wasn't 4 hexadecimal digits. No-op - // (and fall through to "#bad - // backslash-escape"). - - } else if (uni4_value < 0xD800) or (0xDFFF < uni4_value) { - // Not a Unicode surrogate. We're good. - args.src.skip32_fast!(actual: 6, worst_case: 6) - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_0000 | uni4_value, - link: 0x3, - length: 6) - continue.string_loop_outer - - } else if uni4_value >= 0xDC00 { - // Low surrogate. No-op (and fall through - // to "#bad backslash-escape"). - - } else { - // High surrogate, which needs to be - // followed by a "\\u1234" low surrogate. - // We've already peeked 6 bytes for the - // high surrogate. We need 12 in total: - // another 8 bytes at an offset of 4. - if args.src.available() < 12 { - if args.src.is_closed() { - if this.quirk_enabled_replace_invalid_unicode { - args.src.skip32_fast!(actual: 6, worst_case: 6) - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_FFFD, - link: 0x3, - length: 6) - continue.string_loop_outer - } - return "#bad backslash-escape" - } - yield? base."$short read" - string_length = 0 - uni4_value = 0 - char = 0 - continue.string_loop_outer - } - uni4_string = args.src.peek_u64le_at(offset: 4) >> 16 - - // Look for the low surrogate's "\\u". - if ((0xFF & (uni4_string >> 0)) <> '\\') or - ((0xFF & (uni4_string >> 8)) <> 'u') { - uni4_high_surrogate = 0 - uni4_value = 0 - uni4_ok = 0 - } else { - uni4_high_surrogate = - 0x1_0000 + ((uni4_value - 0xD800) << 10) - uni4_value = 0 - uni4_ok = 0x80 - uni4_string >>= 16 - - c = lut_hexadecimal_digits[0xFF & (uni4_string >> 0)] - uni4_ok &= c - uni4_value |= ((c & 0x0F) as base.u32) << 12 - c = lut_hexadecimal_digits[0xFF & (uni4_string >> 8)] - uni4_ok &= c - uni4_value |= ((c & 0x0F) as base.u32) << 8 - c = lut_hexadecimal_digits[0xFF & (uni4_string >> 16)] - uni4_ok &= c - uni4_value |= ((c & 0x0F) as base.u32) << 4 - c = lut_hexadecimal_digits[0xFF & (uni4_string >> 24)] - uni4_ok &= c - uni4_value |= ((c & 0x0F) as base.u32) << 0 - } - - if (uni4_ok <> 0) and - (0xDC00 <= uni4_value) and (uni4_value <= 0xDFFF) { - - // Emit a single token for the surrogate pair. - uni4_value -= 0xDC00 - args.src.skip32_fast!(actual: 12, worst_case: 12) - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_0000 | uni4_high_surrogate | uni4_value, - link: 0x3, - length: 12) - continue.string_loop_outer - } - } - - if this.quirk_enabled_replace_invalid_unicode { - if args.src.available() < 6 { - return "#internal error: inconsistent I/O" - } - args.src.skip32_fast!(actual: 6, worst_case: 6) - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_FFFD, - link: 0x3, - length: 6) - continue.string_loop_outer - } - // -------- END backslash-u. - - } else if (c == 'U') and - this.quirk_enabled_allow_backslash_capital_u { - // -------- BEGIN backslash-capital-u. - if args.src.available() < 10 { - if args.src.is_closed() { - return "#bad backslash-escape" - } - yield? base."$short read" - string_length = 0 - char = 0 - continue.string_loop_outer - } - uni8_string = args.src.peek_u64le_at(offset: 2) - uni8_value = 0 - uni8_ok = 0x80 - - c = lut_hexadecimal_digits[0xFF & (uni8_string >> 0)] - uni8_ok &= c - uni8_value |= ((c & 0x0F) as base.u32) << 28 - c = lut_hexadecimal_digits[0xFF & (uni8_string >> 8)] - uni8_ok &= c - uni8_value |= ((c & 0x0F) as base.u32) << 24 - c = lut_hexadecimal_digits[0xFF & (uni8_string >> 16)] - uni8_ok &= c - uni8_value |= ((c & 0x0F) as base.u32) << 20 - c = lut_hexadecimal_digits[0xFF & (uni8_string >> 24)] - uni8_ok &= c - uni8_value |= ((c & 0x0F) as base.u32) << 16 - c = lut_hexadecimal_digits[0xFF & (uni8_string >> 32)] - uni8_ok &= c - uni8_value |= ((c & 0x0F) as base.u32) << 12 - c = lut_hexadecimal_digits[0xFF & (uni8_string >> 40)] - uni8_ok &= c - uni8_value |= ((c & 0x0F) as base.u32) << 8 - c = lut_hexadecimal_digits[0xFF & (uni8_string >> 48)] - uni8_ok &= c - uni8_value |= ((c & 0x0F) as base.u32) << 4 - c = lut_hexadecimal_digits[0xFF & (uni8_string >> 56)] - uni8_ok &= c - uni8_value |= ((c & 0x0F) as base.u32) << 0 - - if uni8_ok == 0 { - // It wasn't 8 hexadecimal digits. No-op - // (and fall through to "#bad - // backslash-escape"). - - } else if (uni8_value < 0xD800) or ( - (0xDFFF < uni8_value) and (uni8_value <= 0x10_FFFF)) { - // Not a Unicode surrogate. We're good. - args.src.skip32_fast!(actual: 10, worst_case: 10) - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_0000 | (uni8_value & 0x1F_FFFF), - link: 0x3, - length: 10) - continue.string_loop_outer - } else if this.quirk_enabled_replace_invalid_unicode { - args.src.skip32_fast!(actual: 10, worst_case: 10) - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_FFFD, - link: 0x3, - length: 10) - continue.string_loop_outer - } - // -------- END backslash-capital-u. - - } else if (c == 'x') and - this.quirk_enabled_allow_backslash_x { - // -------- BEGIN backslash-x - if args.src.available() < 4 { - if args.src.is_closed() { - return "#bad backslash-escape" - } - yield? base."$short read" - string_length = 0 - char = 0 - continue.string_loop_outer - } - - backslash_x_length = 0 - while (backslash_x_length <= 0xFFFB) and (args.src.available() >= 4), - inv args.dst.available() > 0, - { - backslash_x_string = args.src.peek_u32le() - backslash_x_ok = 0x80 - - c = lut_hexadecimal_digits[0xFF & (backslash_x_string >> 16)] - backslash_x_ok &= c - c = lut_hexadecimal_digits[0xFF & (backslash_x_string >> 24)] - backslash_x_ok &= c - - if (backslash_x_ok == 0) or - ((backslash_x_string & 0xFFFF) <> 0x785C) { - // It wasn't "\\x34", for some - // hexadecimal digits "34". - break - } - args.src.skip32_fast!(actual: 4, worst_case: 4) - backslash_x_length += 4 - } endwhile - - if backslash_x_length == 0 { - return "#bad backslash-escape" - } - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0080, - link: 0x3, - length: backslash_x_length) - continue.string_loop_outer - // -------- END backslash-x - } - - return "#bad backslash-escape" - - } else if char == 0x03 { // 2-byte UTF-8. - if args.src.available() < 2 { - if string_length > 0 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: string_length) - string_length = 0 - if args.dst.available() <= 0 { - continue.string_loop_outer - } - } - if args.src.is_closed() { - if this.quirk_enabled_replace_invalid_unicode { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_FFFD, - link: 0x3, - length: 1) - args.src.skip32_fast!(actual: 1, worst_case: 1) - continue.string_loop_outer - } - return "#bad UTF-8" - } - yield? base."$short read" - string_length = 0 - char = 0 - continue.string_loop_outer - } - multi_byte_utf8 = args.src.peek_u16le_as_u32() - if (multi_byte_utf8 & 0xC000) == 0x8000 { - multi_byte_utf8 = (0x00_07C0 & (multi_byte_utf8 ~mod<< 6)) | - (0x00_003F & (multi_byte_utf8 >> 8)) - args.src.skip32_fast!(actual: 2, worst_case: 2) - if string_length >= 0xFFF8 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: string_length + 2) - string_length = 0 - continue.string_loop_outer - } - string_length += 2 - continue.string_loop_inner - } - - } else if char == 0x04 { // 3-byte UTF-8. - if args.src.available() < 3 { - if string_length > 0 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: string_length) - string_length = 0 - if args.dst.available() <= 0 { - continue.string_loop_outer - } - } - if args.src.is_closed() { - if this.quirk_enabled_replace_invalid_unicode { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_FFFD, - link: 0x3, - length: 1) - args.src.skip32_fast!(actual: 1, worst_case: 1) - continue.string_loop_outer - } - return "#bad UTF-8" - } - yield? base."$short read" - string_length = 0 - char = 0 - continue.string_loop_outer - } - multi_byte_utf8 = args.src.peek_u24le_as_u32() - if (multi_byte_utf8 & 0xC0_C000) == 0x80_8000 { - multi_byte_utf8 = (0x00_F000 & (multi_byte_utf8 ~mod<< 12)) | - (0x00_0FC0 & (multi_byte_utf8 >> 2)) | - (0x00_003F & (multi_byte_utf8 >> 16)) - if (0x07FF < multi_byte_utf8) and - ((multi_byte_utf8 < 0xD800) or (0xDFFF < multi_byte_utf8)) { - - args.src.skip32_fast!(actual: 3, worst_case: 3) - if string_length >= 0xFFF8 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: string_length + 3) - string_length = 0 - continue.string_loop_outer - } - string_length += 3 - continue.string_loop_inner - } - } - - } else if char == 0x05 { // 4-byte UTF-8. - if args.src.available() < 4 { - if string_length > 0 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: string_length) - string_length = 0 - if args.dst.available() <= 0 { - continue.string_loop_outer - } - } - if args.src.is_closed() { - if this.quirk_enabled_replace_invalid_unicode { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_FFFD, - link: 0x3, - length: 1) - args.src.skip32_fast!(actual: 1, worst_case: 1) - continue.string_loop_outer - } - return "#bad UTF-8" - } - yield? base."$short read" - string_length = 0 - char = 0 - continue.string_loop_outer - } - multi_byte_utf8 = args.src.peek_u32le() - if (multi_byte_utf8 & 0xC0C0_C000) == 0x8080_8000 { - multi_byte_utf8 = (0x1C_0000 & (multi_byte_utf8 ~mod<< 18)) | - (0x03_F000 & (multi_byte_utf8 ~mod<< 4)) | - (0x00_0FC0 & (multi_byte_utf8 >> 10)) | - (0x00_003F & (multi_byte_utf8 >> 24)) - if (0xFFFF < multi_byte_utf8) and (multi_byte_utf8 <= 0x10_FFFF) { - args.src.skip32_fast!(actual: 4, worst_case: 4) - if string_length >= 0xFFF8 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0021, - link: 0x3, - length: string_length + 4) - string_length = 0 - continue.string_loop_outer - } - string_length += 4 - continue.string_loop_inner - } - } + if char == 0x00 { // Non-special ASCII. + args.src.skip32_fast!(actual: 1, worst_case: 1) + if string_length >= 0xFFFB { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0021, + link: 0x3, + length: 0xFFFC) + string_length = 0 + continue.string_loop_outer } + string_length += 1 + continue.string_loop_inner + } else if char == 0x01 { // '"' + if string_length <> 0 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0021, + link: 0x3, + length: string_length) + string_length = 0 + } + break.string_loop_outer + + } else if char == 0x02 { // '\\'. if string_length > 0 { args.dst.write_simple_token_fast!( value_major: 0, @@ -779,330 +335,774 @@ pub func decoder.decode_tokens?(dst: base.token_writer, src: base.io_reader, wor continue.string_loop_outer } } - if char == 0x80 { - return "#bad C0 control code" - } - if this.quirk_enabled_replace_invalid_unicode { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x60_FFFD, - link: 0x3, - length: 1) - args.src.skip32_fast!(actual: 1, worst_case: 1) + assert args.dst.available() > 0 + + if args.src.available() < 2 { + if args.src.is_closed() { + return "#bad backslash-escape" + } + yield? base."$short read" + string_length = 0 + char = 0 continue.string_loop_outer } - return "#bad UTF-8" - } endwhile.string_loop_inner - } endwhile.string_loop_outer + c = (args.src.peek_u16le() >> 8) as base.u8 + backslash = lut_backslashes[c] + if (backslash & 0x80) <> 0 { + args.src.skip32_fast!(actual: 2, worst_case: 2) + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_0000 | ((backslash & 0x7F) as base.u32), + link: 0x3, + length: 2) + continue.string_loop_outer - // Emit the trailing '"'. - while true { - if args.src.available() <= 0 { - if args.src.is_closed() { - return "#bad input" + } else if backslash <> 0 { + if this.quirk_enabled_allow_backslash_etc[backslash & 7] { + args.src.skip32_fast!(actual: 2, worst_case: 2) + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_0000 | (lut_quirky_backslashes[backslash & 7] as base.u32), + link: 0x3, + length: 2) + continue.string_loop_outer + } + + } else if c == 'u' { + // -------- BEGIN backslash-u. + if args.src.available() < 6 { + if args.src.is_closed() { + return "#bad backslash-escape" + } + yield? base."$short read" + string_length = 0 + char = 0 + continue.string_loop_outer + } + + uni4_string = args.src.peek_u48le_as_u64() >> 16 + uni4_value = 0 + uni4_ok = 0x80 + + c = lut_hexadecimal_digits[0xFF & (uni4_string >> 0)] + uni4_ok &= c + uni4_value |= ((c & 0x0F) as base.u32) << 12 + c = lut_hexadecimal_digits[0xFF & (uni4_string >> 8)] + uni4_ok &= c + uni4_value |= ((c & 0x0F) as base.u32) << 8 + c = lut_hexadecimal_digits[0xFF & (uni4_string >> 16)] + uni4_ok &= c + uni4_value |= ((c & 0x0F) as base.u32) << 4 + c = lut_hexadecimal_digits[0xFF & (uni4_string >> 24)] + uni4_ok &= c + uni4_value |= ((c & 0x0F) as base.u32) << 0 + + if uni4_ok == 0 { + // It wasn't 4 hexadecimal digits. No-op (and + // fall through to "#bad backslash-escape"). + + } else if (uni4_value < 0xD800) or (0xDFFF < uni4_value) { + // Not a Unicode surrogate. We're good. + args.src.skip32_fast!(actual: 6, worst_case: 6) + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_0000 | uni4_value, + link: 0x3, + length: 6) + continue.string_loop_outer + + } else if uni4_value >= 0xDC00 { + // Low surrogate. No-op (and fall through to + // "#bad backslash-escape"). + + } else { + // High surrogate, which needs to be followed + // by a "\\u1234" low surrogate. We've already + // peeked 6 bytes for the high surrogate. We + // need 12 in total: another 8 bytes at an + // offset of 4. + if args.src.available() < 12 { + if args.src.is_closed() { + if this.quirk_enabled_replace_invalid_unicode { + args.src.skip32_fast!(actual: 6, worst_case: 6) + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_FFFD, + link: 0x3, + length: 6) + continue.string_loop_outer + } + return "#bad backslash-escape" + } + yield? base."$short read" + string_length = 0 + uni4_value = 0 + char = 0 + continue.string_loop_outer + } + uni4_string = args.src.peek_u64le_at(offset: 4) >> 16 + + // Look for the low surrogate's "\\u". + if ((0xFF & (uni4_string >> 0)) <> '\\') or + ((0xFF & (uni4_string >> 8)) <> 'u') { + uni4_high_surrogate = 0 + uni4_value = 0 + uni4_ok = 0 + } else { + uni4_high_surrogate = + 0x1_0000 + ((uni4_value - 0xD800) << 10) + uni4_value = 0 + uni4_ok = 0x80 + uni4_string >>= 16 + + c = lut_hexadecimal_digits[0xFF & (uni4_string >> 0)] + uni4_ok &= c + uni4_value |= ((c & 0x0F) as base.u32) << 12 + c = lut_hexadecimal_digits[0xFF & (uni4_string >> 8)] + uni4_ok &= c + uni4_value |= ((c & 0x0F) as base.u32) << 8 + c = lut_hexadecimal_digits[0xFF & (uni4_string >> 16)] + uni4_ok &= c + uni4_value |= ((c & 0x0F) as base.u32) << 4 + c = lut_hexadecimal_digits[0xFF & (uni4_string >> 24)] + uni4_ok &= c + uni4_value |= ((c & 0x0F) as base.u32) << 0 + } + + if (uni4_ok <> 0) and + (0xDC00 <= uni4_value) and (uni4_value <= 0xDFFF) { + + // Emit a single token for the surrogate + // pair. + uni4_value -= 0xDC00 + args.src.skip32_fast!(actual: 12, worst_case: 12) + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_0000 | uni4_high_surrogate | uni4_value, + link: 0x3, + length: 12) + continue.string_loop_outer + } + } + + if this.quirk_enabled_replace_invalid_unicode { + if args.src.available() < 6 { + return "#internal error: inconsistent I/O" + } + args.src.skip32_fast!(actual: 6, worst_case: 6) + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_FFFD, + link: 0x3, + length: 6) + continue.string_loop_outer + } + // -------- END backslash-u. + + } else if (c == 'U') and + this.quirk_enabled_allow_backslash_capital_u { + // -------- BEGIN backslash-capital-u. + if args.src.available() < 10 { + if args.src.is_closed() { + return "#bad backslash-escape" + } + yield? base."$short read" + string_length = 0 + char = 0 + continue.string_loop_outer + } + uni8_string = args.src.peek_u64le_at(offset: 2) + uni8_value = 0 + uni8_ok = 0x80 + + c = lut_hexadecimal_digits[0xFF & (uni8_string >> 0)] + uni8_ok &= c + uni8_value |= ((c & 0x0F) as base.u32) << 28 + c = lut_hexadecimal_digits[0xFF & (uni8_string >> 8)] + uni8_ok &= c + uni8_value |= ((c & 0x0F) as base.u32) << 24 + c = lut_hexadecimal_digits[0xFF & (uni8_string >> 16)] + uni8_ok &= c + uni8_value |= ((c & 0x0F) as base.u32) << 20 + c = lut_hexadecimal_digits[0xFF & (uni8_string >> 24)] + uni8_ok &= c + uni8_value |= ((c & 0x0F) as base.u32) << 16 + c = lut_hexadecimal_digits[0xFF & (uni8_string >> 32)] + uni8_ok &= c + uni8_value |= ((c & 0x0F) as base.u32) << 12 + c = lut_hexadecimal_digits[0xFF & (uni8_string >> 40)] + uni8_ok &= c + uni8_value |= ((c & 0x0F) as base.u32) << 8 + c = lut_hexadecimal_digits[0xFF & (uni8_string >> 48)] + uni8_ok &= c + uni8_value |= ((c & 0x0F) as base.u32) << 4 + c = lut_hexadecimal_digits[0xFF & (uni8_string >> 56)] + uni8_ok &= c + uni8_value |= ((c & 0x0F) as base.u32) << 0 + + if uni8_ok == 0 { + // It wasn't 8 hexadecimal digits. No-op (and + // fall through to "#bad backslash-escape"). + + } else if (uni8_value < 0xD800) or ( + (0xDFFF < uni8_value) and (uni8_value <= 0x10_FFFF)) { + // Not a Unicode surrogate. We're good. + args.src.skip32_fast!(actual: 10, worst_case: 10) + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_0000 | (uni8_value & 0x1F_FFFF), + link: 0x3, + length: 10) + continue.string_loop_outer + } else if this.quirk_enabled_replace_invalid_unicode { + args.src.skip32_fast!(actual: 10, worst_case: 10) + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_FFFD, + link: 0x3, + length: 10) + continue.string_loop_outer + } + // -------- END backslash-capital-u. + + } else if (c == 'x') and + this.quirk_enabled_allow_backslash_x { + // -------- BEGIN backslash-x + if args.src.available() < 4 { + if args.src.is_closed() { + return "#bad backslash-escape" + } + yield? base."$short read" + string_length = 0 + char = 0 + continue.string_loop_outer + } + + backslash_x_length = 0 + while (backslash_x_length <= 0xFFFB) and (args.src.available() >= 4), + inv args.dst.available() > 0, + { + backslash_x_string = args.src.peek_u32le() + backslash_x_ok = 0x80 + + c = lut_hexadecimal_digits[0xFF & (backslash_x_string >> 16)] + backslash_x_ok &= c + c = lut_hexadecimal_digits[0xFF & (backslash_x_string >> 24)] + backslash_x_ok &= c + + if (backslash_x_ok == 0) or + ((backslash_x_string & 0xFFFF) <> 0x785C) { + // It wasn't "\\x34", for some hexadecimal + // digits "34". + break + } + args.src.skip32_fast!(actual: 4, worst_case: 4) + backslash_x_length += 4 + } endwhile + + if backslash_x_length == 0 { + return "#bad backslash-escape" + } + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0080, + link: 0x3, + length: backslash_x_length) + continue.string_loop_outer + // -------- END backslash-x } - yield? base."$short read" - continue - } - if args.dst.available() <= 0 { - yield? base."$short write" - continue - } - args.src.skip32_fast!(actual: 1, worst_case: 1) - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x40_0013, - link: 0x2, - length: 1) - break - } endwhile - // As above, expect must have contained EXPECT_STRING. If it - // didn't also contain EXPECT_NUMBER then we were parsing an - // object key and the next token should be ':'. - if 0 == (expect & 0x0010) { // 0x0010 is (1 << CLASS_NUMBER). - expect = 0x1008 // 0x1008 is EXPECT_COLON. - continue.outer - } - break.goto_parsed_a_leaf_value - // -------- END parse strings. + return "#bad backslash-escape" - } else if class == 0x02 { // 0x02 is CLASS_COMMA. - args.src.skip32_fast!(actual: 1, worst_case: 1) - // The ',' is filler. - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0, - link: 0x0, - length: 1) - // What's valid after a comma depends on whether or not we're - // in an array or an object. - if 0 == (expect & 0x0100) { // 0x0100 is (1 << CLASS_CLOSE_SQUARE_BRACKET). - if this.quirk_enabled_allow_extra_comma { - expect = 0x1042 // 0x1042 is EXPECT_STRING | EXPECT_CLOSE_CURLY_BRACE. - } else { - expect = 0x1002 // 0x1002 is EXPECT_STRING. - } - } else { - if this.quirk_enabled_allow_extra_comma { - expect = 0x1FB2 // 0x0FB2 is EXPECT_VALUE | EXPECT_CLOSE_SQUARE_BRACKET. - } else { - expect = 0x1EB2 // 0x0EB2 is EXPECT_VALUE. - } - } - continue.outer + } else if char == 0x03 { // 2-byte UTF-8. + if args.src.available() < 2 { + if string_length > 0 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0021, + link: 0x3, + length: string_length) + string_length = 0 + if args.dst.available() <= 0 { + continue.string_loop_outer + } + } + if args.src.is_closed() { + if this.quirk_enabled_replace_invalid_unicode { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_FFFD, + link: 0x3, + length: 1) + args.src.skip32_fast!(actual: 1, worst_case: 1) + continue.string_loop_outer + } + return "#bad UTF-8" + } + yield? base."$short read" + string_length = 0 + char = 0 + continue.string_loop_outer + } + multi_byte_utf8 = args.src.peek_u16le_as_u32() + if (multi_byte_utf8 & 0xC000) == 0x8000 { + multi_byte_utf8 = (0x00_07C0 & (multi_byte_utf8 ~mod<< 6)) | + (0x00_003F & (multi_byte_utf8 >> 8)) + args.src.skip32_fast!(actual: 2, worst_case: 2) + if string_length >= 0xFFF8 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0021, + link: 0x3, + length: string_length + 2) + string_length = 0 + continue.string_loop_outer + } + string_length += 2 + continue.string_loop_inner + } - } else if class == 0x03 { // 0x03 is CLASS_COLON. - args.src.skip32_fast!(actual: 1, worst_case: 1) - // The ':' is filler. - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0, - link: 0x0, - length: 1) - expect = 0x1EB2 // 0x1EB2 is EXPECT_VALUE. - continue.outer + } else if char == 0x04 { // 3-byte UTF-8. + if args.src.available() < 3 { + if string_length > 0 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0021, + link: 0x3, + length: string_length) + string_length = 0 + if args.dst.available() <= 0 { + continue.string_loop_outer + } + } + if args.src.is_closed() { + if this.quirk_enabled_replace_invalid_unicode { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_FFFD, + link: 0x3, + length: 1) + args.src.skip32_fast!(actual: 1, worst_case: 1) + continue.string_loop_outer + } + return "#bad UTF-8" + } + yield? base."$short read" + string_length = 0 + char = 0 + continue.string_loop_outer + } + multi_byte_utf8 = args.src.peek_u24le_as_u32() + if (multi_byte_utf8 & 0xC0_C000) == 0x80_8000 { + multi_byte_utf8 = (0x00_F000 & (multi_byte_utf8 ~mod<< 12)) | + (0x00_0FC0 & (multi_byte_utf8 >> 2)) | + (0x00_003F & (multi_byte_utf8 >> 16)) + if (0x07FF < multi_byte_utf8) and + ((multi_byte_utf8 < 0xD800) or (0xDFFF < multi_byte_utf8)) { - } else if class == 0x04 { // 0x04 is CLASS_NUMBER. - // -------- BEGIN parse numbers. - while true, - pre args.dst.available() > 0, - { - number_length = this.decode_number!(src: args.src) - number_status = number_length >> 8 - vminor = 0xA0_0403 - if (number_length & 0x80) <> 0 { - vminor = 0xA0_0401 + args.src.skip32_fast!(actual: 3, worst_case: 3) + if string_length >= 0xFFF8 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0021, + link: 0x3, + length: string_length + 3) + string_length = 0 + continue.string_loop_outer + } + string_length += 3 + continue.string_loop_inner + } + } + + } else if char == 0x05 { // 4-byte UTF-8. + if args.src.available() < 4 { + if string_length > 0 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0021, + link: 0x3, + length: string_length) + string_length = 0 + if args.dst.available() <= 0 { + continue.string_loop_outer + } + } + if args.src.is_closed() { + if this.quirk_enabled_replace_invalid_unicode { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_FFFD, + link: 0x3, + length: 1) + args.src.skip32_fast!(actual: 1, worst_case: 1) + continue.string_loop_outer + } + return "#bad UTF-8" + } + yield? base."$short read" + string_length = 0 + char = 0 + continue.string_loop_outer + } + multi_byte_utf8 = args.src.peek_u32le() + if (multi_byte_utf8 & 0xC0C0_C000) == 0x8080_8000 { + multi_byte_utf8 = (0x1C_0000 & (multi_byte_utf8 ~mod<< 18)) | + (0x03_F000 & (multi_byte_utf8 ~mod<< 4)) | + (0x00_0FC0 & (multi_byte_utf8 >> 10)) | + (0x00_003F & (multi_byte_utf8 >> 24)) + if (0xFFFF < multi_byte_utf8) and (multi_byte_utf8 <= 0x10_FFFF) { + args.src.skip32_fast!(actual: 4, worst_case: 4) + if string_length >= 0xFFF8 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x40_0021, + link: 0x3, + length: string_length + 4) + string_length = 0 + continue.string_loop_outer + } + string_length += 4 + continue.string_loop_inner + } + } } - number_length = number_length & 0x7F - if number_status == 0 { + + if string_length > 0 { args.dst.write_simple_token_fast!( value_major: 0, - value_minor: vminor, - link: 0x0, - length: number_length) - break + value_minor: 0x40_0021, + link: 0x3, + length: string_length) + string_length = 0 + if args.dst.available() <= 0 { + continue.string_loop_outer + } } + if char == 0x80 { + return "#bad C0 control code" + } + if this.quirk_enabled_replace_invalid_unicode { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x60_FFFD, + link: 0x3, + length: 1) + args.src.skip32_fast!(actual: 1, worst_case: 1) + continue.string_loop_outer + } + return "#bad UTF-8" + } endwhile.string_loop_inner + } endwhile.string_loop_outer - while number_length > 0 { - number_length -= 1 - if args.src.can_undo_byte() { - args.src.undo_byte!() - } else { - return "#internal error: inconsistent I/O" - } - } endwhile - - if number_status == 1 { - if this.quirk_enabled_allow_inf_nan_numbers { - this.decode_inf_nan?(dst: args.dst, src: args.src) - break - } + // Emit the trailing '"'. + while true { + if args.src.available() <= 0 { + if args.src.is_closed() { return "#bad input" - } else if number_status == 2 { - return "#unsupported number length" - } else { - yield? base."$short read" - while args.dst.available() <= 0, - post args.dst.available() > 0, - { - yield? base."$short write" - } endwhile } - } endwhile - break.goto_parsed_a_leaf_value - // -------- END parse numbers. - - } else if class == 0x05 { // 0x05 is CLASS_OPEN_CURLY_BRACE. - vminor = 0x20_4011 - if depth == 0 { - // No-op. - } else if 0 <> (expect_after_value & 0x0040) { // 0x0040 is (1 << CLASS_CLOSE_CURLY_BRACE). - vminor = 0x20_4041 - } else { - vminor = 0x20_4021 + yield? base."$short read" + continue } - if depth >= 1024 { - return "#unsupported recursion depth" + if args.dst.available() <= 0 { + yield? base."$short write" + continue } - stack_byte = depth / 32 - stack_bit = depth & 31 - this.stack[stack_byte] |= (1 as base.u32) << stack_bit - depth += 1 - args.src.skip32_fast!(actual: 1, worst_case: 1) args.dst.write_simple_token_fast!( value_major: 0, - value_minor: vminor, - link: 0x0, + value_minor: 0x40_0013, + link: 0x2, length: 1) - expect = 0x1042 // 0x1042 is (EXPECT_CLOSE_CURLY_BRACE | EXPECT_STRING). - expect_after_value = 0x1044 // 0x1044 is (EXPECT_CURLY_CLOSE_BRACE | EXPECT_COMMA). - continue.outer + break + } endwhile - } else if class == 0x06 { // 0x06 is CLASS_CLOSE_CURLY_BRACE. - args.src.skip32_fast!(actual: 1, worst_case: 1) - if depth <= 1 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x20_1042, - link: 0x0, - length: 1) - break.outer - } - depth -= 1 - stack_byte = (depth - 1) / 32 - stack_bit = (depth - 1) & 31 - if 0 == (this.stack[stack_byte] & ((1 as base.u32) << stack_bit)) { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x20_2042, - link: 0x0, - length: 1) - // 0x1104 is (EXPECT_SQUARE_CLOSE_BRACKET | EXPECT_COMMA). - expect = 0x1104 - expect_after_value = 0x1104 + // As above, expect must have contained EXPECT_STRING. If it didn't + // also contain EXPECT_NUMBER then we were parsing an object key + // and the next token should be ':'. + if 0 == (expect & 0x0010) { // 0x0010 is (1 << CLASS_NUMBER). + expect = 0x1008 // 0x1008 is EXPECT_COLON. + continue.outer + } + break.goto_parsed_a_leaf_value + // -------- END parse strings. + + } else if class == 0x02 { // 0x02 is CLASS_COMMA. + args.src.skip32_fast!(actual: 1, worst_case: 1) + // The ',' is filler. + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0, + link: 0x0, + length: 1) + // What's valid after a comma depends on whether or not we're in an + // array or an object. + if 0 == (expect & 0x0100) { // 0x0100 is (1 << CLASS_CLOSE_SQUARE_BRACKET). + if this.quirk_enabled_allow_extra_comma { + expect = 0x1042 // 0x1042 is EXPECT_STRING | EXPECT_CLOSE_CURLY_BRACE. } else { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x20_4042, - link: 0x0, - length: 1) - // 0x1044 is (EXPECT_CURLY_CLOSE_BRACE | EXPECT_COMMA). - expect = 0x1044 - expect_after_value = 0x1044 + expect = 0x1002 // 0x1002 is EXPECT_STRING. } - continue.outer - - } else if class == 0x07 { // 0x07 is CLASS_OPEN_SQUARE_BRACKET. - vminor = 0x20_2011 - if depth == 0 { - // No-op. - } else if 0 <> (expect_after_value & 0x0040) { // 0x0040 is (1 << CLASS_CLOSE_CURLY_BRACE). - vminor = 0x20_2041 + } else { + if this.quirk_enabled_allow_extra_comma { + expect = 0x1FB2 // 0x0FB2 is EXPECT_VALUE | EXPECT_CLOSE_SQUARE_BRACKET. } else { - vminor = 0x20_2021 - } - if depth >= 1024 { - return "#unsupported recursion depth" - } - stack_byte = depth / 32 - stack_bit = depth & 31 - this.stack[stack_byte] &= 0xFFFF_FFFF ^ ((1 as base.u32) << stack_bit) - depth += 1 - - args.src.skip32_fast!(actual: 1, worst_case: 1) - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: vminor, - link: 0x0, - length: 1) - expect = 0x1FB2 // 0x1FB2 is (EXPECT_CLOSE_SQUARE_BRACKET | EXPECT_VALUE). - expect_after_value = 0x1104 // 0x1104 is (EXPECT_CLOSE_SQUARE_BRACKET | EXPECT_COMMA). - continue.outer - - } else if class == 0x08 { // 0x08 is CLASS_CLOSE_SQUARE_BRACKET. - args.src.skip32_fast!(actual: 1, worst_case: 1) - if depth <= 1 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x20_1022, - link: 0x0, - length: 1) - break.outer - } - depth -= 1 - stack_byte = (depth - 1) / 32 - stack_bit = (depth - 1) & 31 - if 0 == (this.stack[stack_byte] & ((1 as base.u32) << stack_bit)) { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x20_2022, - link: 0x0, - length: 1) - // 0x1104 is (EXPECT_CLOSE_SQUARE_BRACKET | EXPECT_COMMA). - expect = 0x1104 - expect_after_value = 0x1104 - } else { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x20_4022, - link: 0x0, - length: 1) - // 0x1044 is (EXPECT_CLOSE_CURLY_BRACE | EXPECT_COMMA). - expect = 0x1044 - expect_after_value = 0x1044 - } - continue.outer - - } else if class == 0x09 { // 0x09 is CLASS_FALSE. - match = args.src.match7(a: '\x05false'le) - if match == 0 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x80_0004, - link: 0x0, - length: 5) - if args.src.available() < 5 { - return "#internal error: inconsistent I/O" - } - args.src.skip32_fast!(actual: 5, worst_case: 5) - break.goto_parsed_a_leaf_value - } else if match == 1 { - yield? base."$short read" - continue.outer - } - - } else if class == 0x0A { // 0x0A is CLASS_TRUE. - match = args.src.match7(a: '\x04true'le) - if match == 0 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x80_0008, - link: 0x0, - length: 4) - if args.src.available() < 4 { - return "#internal error: inconsistent I/O" - } - args.src.skip32_fast!(actual: 4, worst_case: 4) - break.goto_parsed_a_leaf_value - } else if match == 1 { - yield? base."$short read" - continue.outer - } - - } else if class == 0x0B { // 0x0B is CLASS_NULL_NAN_INF. - match = args.src.match7(a: '\x04null'le) - if match == 0 { - args.dst.write_simple_token_fast!( - value_major: 0, - value_minor: 0x80_0002, - link: 0x0, - length: 4) - if args.src.available() < 4 { - return "#internal error: inconsistent I/O" - } - args.src.skip32_fast!(actual: 4, worst_case: 4) - break.goto_parsed_a_leaf_value - } else if match == 1 { - yield? base."$short read" - continue.outer - } - - if this.quirk_enabled_allow_inf_nan_numbers { - this.decode_inf_nan?(dst: args.dst, src: args.src) - break.goto_parsed_a_leaf_value - } - - } else if class == 0x0C { // 0x0C is CLASS_COMMENT. - if this.quirk_enabled_allow_comment_block or this.quirk_enabled_allow_comment_line { - this.decode_comment?(dst: args.dst, src: args.src) - continue.outer + expect = 0x1EB2 // 0x0EB2 is EXPECT_VALUE. } } + continue.outer - return "#bad input" - } endwhile.goto_parsed_a_leaf_value + } else if class == 0x03 { // 0x03 is CLASS_COLON. + args.src.skip32_fast!(actual: 1, worst_case: 1) + // The ':' is filler. + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0, + link: 0x0, + length: 1) + expect = 0x1EB2 // 0x1EB2 is EXPECT_VALUE. + continue.outer + + } else if class == 0x04 { // 0x04 is CLASS_NUMBER. + // -------- BEGIN parse numbers. + while true, + pre args.dst.available() > 0, + { + number_length = this.decode_number!(src: args.src) + number_status = number_length >> 8 + vminor = 0xA0_0403 + if (number_length & 0x80) <> 0 { + vminor = 0xA0_0401 + } + number_length = number_length & 0x7F + if number_status == 0 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: vminor, + link: 0x0, + length: number_length) + break + } + + while number_length > 0 { + number_length -= 1 + if args.src.can_undo_byte() { + args.src.undo_byte!() + } else { + return "#internal error: inconsistent I/O" + } + } endwhile + + if number_status == 1 { + if this.quirk_enabled_allow_inf_nan_numbers { + this.decode_inf_nan?(dst: args.dst, src: args.src) + break + } + return "#bad input" + } else if number_status == 2 { + return "#unsupported number length" + } else { + yield? base."$short read" + while args.dst.available() <= 0, + post args.dst.available() > 0, + { + yield? base."$short write" + } endwhile + } + } endwhile + break.goto_parsed_a_leaf_value + // -------- END parse numbers. + + } else if class == 0x05 { // 0x05 is CLASS_OPEN_CURLY_BRACE. + vminor = 0x20_4011 + if depth == 0 { + // No-op. + } else if 0 <> (expect_after_value & 0x0040) { // 0x0040 is (1 << CLASS_CLOSE_CURLY_BRACE). + vminor = 0x20_4041 + } else { + vminor = 0x20_4021 + } + if depth >= 1024 { + return "#unsupported recursion depth" + } + stack_byte = depth / 32 + stack_bit = depth & 31 + this.stack[stack_byte] |= (1 as base.u32) << stack_bit + depth += 1 + + args.src.skip32_fast!(actual: 1, worst_case: 1) + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: vminor, + link: 0x0, + length: 1) + expect = 0x1042 // 0x1042 is (EXPECT_CLOSE_CURLY_BRACE | EXPECT_STRING). + expect_after_value = 0x1044 // 0x1044 is (EXPECT_CURLY_CLOSE_BRACE | EXPECT_COMMA). + continue.outer + + } else if class == 0x06 { // 0x06 is CLASS_CLOSE_CURLY_BRACE. + args.src.skip32_fast!(actual: 1, worst_case: 1) + if depth <= 1 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x20_1042, + link: 0x0, + length: 1) + break.outer + } + depth -= 1 + stack_byte = (depth - 1) / 32 + stack_bit = (depth - 1) & 31 + if 0 == (this.stack[stack_byte] & ((1 as base.u32) << stack_bit)) { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x20_2042, + link: 0x0, + length: 1) + // 0x1104 is (EXPECT_SQUARE_CLOSE_BRACKET | EXPECT_COMMA). + expect = 0x1104 + expect_after_value = 0x1104 + } else { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x20_4042, + link: 0x0, + length: 1) + // 0x1044 is (EXPECT_CURLY_CLOSE_BRACE | EXPECT_COMMA). + expect = 0x1044 + expect_after_value = 0x1044 + } + continue.outer + + } else if class == 0x07 { // 0x07 is CLASS_OPEN_SQUARE_BRACKET. + vminor = 0x20_2011 + if depth == 0 { + // No-op. + } else if 0 <> (expect_after_value & 0x0040) { // 0x0040 is (1 << CLASS_CLOSE_CURLY_BRACE). + vminor = 0x20_2041 + } else { + vminor = 0x20_2021 + } + if depth >= 1024 { + return "#unsupported recursion depth" + } + stack_byte = depth / 32 + stack_bit = depth & 31 + this.stack[stack_byte] &= 0xFFFF_FFFF ^ ((1 as base.u32) << stack_bit) + depth += 1 + + args.src.skip32_fast!(actual: 1, worst_case: 1) + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: vminor, + link: 0x0, + length: 1) + expect = 0x1FB2 // 0x1FB2 is (EXPECT_CLOSE_SQUARE_BRACKET | EXPECT_VALUE). + expect_after_value = 0x1104 // 0x1104 is (EXPECT_CLOSE_SQUARE_BRACKET | EXPECT_COMMA). + continue.outer + + } else if class == 0x08 { // 0x08 is CLASS_CLOSE_SQUARE_BRACKET. + args.src.skip32_fast!(actual: 1, worst_case: 1) + if depth <= 1 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x20_1022, + link: 0x0, + length: 1) + break.outer + } + depth -= 1 + stack_byte = (depth - 1) / 32 + stack_bit = (depth - 1) & 31 + if 0 == (this.stack[stack_byte] & ((1 as base.u32) << stack_bit)) { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x20_2022, + link: 0x0, + length: 1) + // 0x1104 is (EXPECT_CLOSE_SQUARE_BRACKET | EXPECT_COMMA). + expect = 0x1104 + expect_after_value = 0x1104 + } else { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x20_4022, + link: 0x0, + length: 1) + // 0x1044 is (EXPECT_CLOSE_CURLY_BRACE | EXPECT_COMMA). + expect = 0x1044 + expect_after_value = 0x1044 + } + continue.outer + + } else if class == 0x09 { // 0x09 is CLASS_FALSE. + match = args.src.match7(a: '\x05false'le) + if match == 0 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x80_0004, + link: 0x0, + length: 5) + if args.src.available() < 5 { + return "#internal error: inconsistent I/O" + } + args.src.skip32_fast!(actual: 5, worst_case: 5) + break.goto_parsed_a_leaf_value + } else if match == 1 { + yield? base."$short read" + continue.outer + } + + } else if class == 0x0A { // 0x0A is CLASS_TRUE. + match = args.src.match7(a: '\x04true'le) + if match == 0 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x80_0008, + link: 0x0, + length: 4) + if args.src.available() < 4 { + return "#internal error: inconsistent I/O" + } + args.src.skip32_fast!(actual: 4, worst_case: 4) + break.goto_parsed_a_leaf_value + } else if match == 1 { + yield? base."$short read" + continue.outer + } + + } else if class == 0x0B { // 0x0B is CLASS_NULL_NAN_INF. + match = args.src.match7(a: '\x04null'le) + if match == 0 { + args.dst.write_simple_token_fast!( + value_major: 0, + value_minor: 0x80_0002, + link: 0x0, + length: 4) + if args.src.available() < 4 { + return "#internal error: inconsistent I/O" + } + args.src.skip32_fast!(actual: 4, worst_case: 4) + break.goto_parsed_a_leaf_value + } else if match == 1 { + yield? base."$short read" + continue.outer + } + + if this.quirk_enabled_allow_inf_nan_numbers { + this.decode_inf_nan?(dst: args.dst, src: args.src) + break.goto_parsed_a_leaf_value + } + + } else if class == 0x0C { // 0x0C is CLASS_COMMENT. + if this.quirk_enabled_allow_comment_block or this.quirk_enabled_allow_comment_line { + this.decode_comment?(dst: args.dst, src: args.src) + continue.outer + } + } + + return "#bad input" + }} endwhile.goto_parsed_a_leaf_value // We've just parsed a leaf (non-container) value: literal (null, // false, true), number or string. @@ -1124,97 +1124,67 @@ pri func decoder.decode_number!(src: base.io_reader) base.u32[..= 0x3FF] { var n : base.u32[..= 0x3FF] var floating_point : base.u32[..= 0x80] - while.goto_done true { - n = 0 + while.goto_done true {{ + n = 0 + + // Peek. + if args.src.available() <= 0 { + if not args.src.is_closed() { + n |= 0x300 + } + break.goto_done + } + c = args.src.peek_u8() + + // Scan the optional minus sign. + if c <> '-' { + assert args.src.available() > 0 + assert n <= 1 + } else { + n += 1 + args.src.skip32_fast!(actual: 1, worst_case: 1) // Peek. if args.src.available() <= 0 { if not args.src.is_closed() { n |= 0x300 } + n |= 0x100 // A '-' without digits is invalid. break.goto_done } c = args.src.peek_u8() - // Scan the optional minus sign. - if c <> '-' { - assert args.src.available() > 0 - assert n <= 1 - } else { - n += 1 - args.src.skip32_fast!(actual: 1, worst_case: 1) + assert args.src.available() > 0 + assert n <= 1 + } - // Peek. - if args.src.available() <= 0 { - if not args.src.is_closed() { - n |= 0x300 - } - n |= 0x100 // A '-' without digits is invalid. - break.goto_done - } - c = args.src.peek_u8() - - assert args.src.available() > 0 - assert n <= 1 - } - - // Scan the opening digits. - if c == '0' { - n += 1 - args.src.skip32_fast!(actual: 1, worst_case: 1) - assert n <= 99 - } else { - n = this.decode_digits!(src: args.src, n: n) - if n > 99 { - break.goto_done - } - assert n <= 99 - } - - // Peek. - if args.src.available() <= 0 { - if not args.src.is_closed() { - n |= 0x300 - } + // Scan the opening digits. + if c == '0' { + n += 1 + args.src.skip32_fast!(actual: 1, worst_case: 1) + assert n <= 99 + } else { + n = this.decode_digits!(src: args.src, n: n) + if n > 99 { break.goto_done } - c = args.src.peek_u8() + assert n <= 99 + } - // Scan the optional fraction. - if c <> '.' { - assert args.src.available() > 0 - assert n <= 99 - } else { - if n >= 99 { - n |= 0x200 - break.goto_done - } - n += 1 - args.src.skip32_fast!(actual: 1, worst_case: 1) - floating_point = 0x80 - - n = this.decode_digits!(src: args.src, n: n) - if n > 99 { - break.goto_done - } - - // Peek. - if args.src.available() <= 0 { - if not args.src.is_closed() { - n |= 0x300 - } - break.goto_done - } - c = args.src.peek_u8() - - assert args.src.available() > 0 - assert n <= 99 + // Peek. + if args.src.available() <= 0 { + if not args.src.is_closed() { + n |= 0x300 } + break.goto_done + } + c = args.src.peek_u8() - // Scan the optional 'E' or 'e'. - if (c <> 'E') and (c <> 'e') { - break.goto_done - } + // Scan the optional fraction. + if c <> '.' { + assert args.src.available() > 0 + assert n <= 99 + } else { if n >= 99 { n |= 0x200 break.goto_done @@ -1222,36 +1192,67 @@ pri func decoder.decode_number!(src: base.io_reader) base.u32[..= 0x3FF] { n += 1 args.src.skip32_fast!(actual: 1, worst_case: 1) floating_point = 0x80 - assert n <= 99 + + n = this.decode_digits!(src: args.src, n: n) + if n > 99 { + break.goto_done + } // Peek. if args.src.available() <= 0 { if not args.src.is_closed() { n |= 0x300 } - n |= 0x100 // An 'E' or 'e' without digits is invalid. break.goto_done } c = args.src.peek_u8() - // Scan the optional '+' or '-'. - if (c <> '+') and (c <> '-') { - assert n <= 99 - } else { - if n >= 99 { - n |= 0x200 - break.goto_done - } - n += 1 - args.src.skip32_fast!(actual: 1, worst_case: 1) - assert n <= 99 - } - - // Scan the exponent digits. - n = this.decode_digits!(src: args.src, n: n) + assert args.src.available() > 0 + assert n <= 99 + } + // Scan the optional 'E' or 'e'. + if (c <> 'E') and (c <> 'e') { break.goto_done - } endwhile.goto_done + } + if n >= 99 { + n |= 0x200 + break.goto_done + } + n += 1 + args.src.skip32_fast!(actual: 1, worst_case: 1) + floating_point = 0x80 + assert n <= 99 + + // Peek. + if args.src.available() <= 0 { + if not args.src.is_closed() { + n |= 0x300 + } + n |= 0x100 // An 'E' or 'e' without digits is invalid. + break.goto_done + } + c = args.src.peek_u8() + + // Scan the optional '+' or '-'. + if (c <> '+') and (c <> '-') { + assert n <= 99 + } else { + if n >= 99 { + n |= 0x200 + break.goto_done + } + n += 1 + args.src.skip32_fast!(actual: 1, worst_case: 1) + assert n <= 99 + } + + // Scan the exponent digits. + n = this.decode_digits!(src: args.src, n: n) + + break.goto_done + }} endwhile.goto_done + return n | floating_point } @@ -1299,6 +1300,7 @@ pri func decoder.decode_leading?(dst: base.token_writer, src: base.io_reader) { this.allow_leading_ars = this.quirk_enabled_allow_leading_ascii_record_separator this.allow_leading_ubom = this.quirk_enabled_allow_leading_unicode_byte_order_mark + while this.allow_leading_ars or this.allow_leading_ubom { if args.dst.available() <= 0 { yield? base."$short write" @@ -1634,5 +1636,4 @@ pri func decoder.decode_trailing_new_line?(dst: base.token_writer, src: base.io_ whitespace_length += 1 } endwhile.inner } endwhile.outer - }