diff options
author | Luke Shumaker <lukeshu@lukeshu.com> | 2023-01-30 23:00:11 -0700 |
---|---|---|
committer | Luke Shumaker <lukeshu@lukeshu.com> | 2023-01-30 23:00:11 -0700 |
commit | 5d5ac549dc1a25963418fda4b20ed181b7afd8d2 (patch) | |
tree | 8c5c2e9f5bd1c3a2208efe4d11965ecaa95472ec /internal | |
parent | 005dfe26e308b965c2f42c81d34a4b48757414a3 (diff) | |
parent | ccf8dc4b21bb1a547f118affab22bca3a02df270 (diff) |
Merge branch 'lukeshu/tune'
Diffstat (limited to 'internal')
-rw-r--r-- | internal/allwriter.go | 174 | ||||
-rw-r--r-- | internal/base64.go | 9 | ||||
-rw-r--r-- | internal/parse.go | 124 |
3 files changed, 303 insertions, 4 deletions
diff --git a/internal/allwriter.go b/internal/allwriter.go new file mode 100644 index 0000000..187aa8e --- /dev/null +++ b/internal/allwriter.go @@ -0,0 +1,174 @@ +// Copyright (C) 2023 Luke Shumaker <lukeshu@lukeshu.com> +// +// SPDX-License-Identifier: GPL-2.0-or-later + +package internal + +import ( + "io" + "unicode/utf8" +) + +// interfaces ///////////////////////////////////////////////////////////////// + +type RuneWriter interface { + WriteRune(rune) (int, error) +} + +// An AllWriter is the union of several common writer interfaces. +type AllWriter interface { + io.Writer + io.ByteWriter + RuneWriter + io.StringWriter +} + +// implementations //////////////////////////////////////////////////////////// + +func WriteByte(w io.Writer, b byte) error { + var buf [1]byte + buf[0] = b + _, err := w.Write(buf[:]) + return err +} + +func WriteRune(w io.Writer, r rune) (int, error) { + var buf [utf8.UTFMax]byte + n := utf8.EncodeRune(buf[:], r) + return w.Write(buf[:n]) +} + +func WriteString(w io.Writer, s string) (int, error) { + return w.Write([]byte(s)) +} + +// wrappers /////////////////////////////////////////////////////////////////// + +// NNN + +type ( + writerNNN interface{ io.Writer } + writerNNNWrapper struct{ writerNNN } +) + +func (w writerNNNWrapper) WriteByte(b byte) error { return WriteByte(w, b) } +func (w writerNNNWrapper) WriteRune(r rune) (int, error) { return WriteRune(w, r) } +func (w writerNNNWrapper) WriteString(s string) (int, error) { return WriteString(w, s) } + +// NNY + +type ( + writerNNY interface { + io.Writer + io.StringWriter + } + writerNNYWrapper struct{ writerNNY } +) + +func (w writerNNYWrapper) WriteByte(b byte) error { return WriteByte(w, b) } +func (w writerNNYWrapper) WriteRune(r rune) (int, error) { return WriteRune(w, r) } + +// NYN + +type ( + writerNYN interface { + io.Writer + RuneWriter + } + writerNYNWrapper struct{ writerNYN } +) + +func (w writerNYNWrapper) WriteByte(b byte) error { return WriteByte(w, b) } +func (w writerNYNWrapper) WriteString(s string) (int, error) { return WriteString(w, s) } + +// NYY + +type ( + writerNYY interface { + io.Writer + RuneWriter + io.StringWriter + } + writerNYYWrapper struct{ writerNYY } +) + +func (w writerNYYWrapper) WriteByte(b byte) error { return WriteByte(w, b) } + +// YNN + +type ( + writerYNN interface { + io.Writer + io.ByteWriter + } + writerYNNWrapper struct{ writerYNN } +) + +func (w writerYNNWrapper) WriteRune(r rune) (int, error) { return WriteRune(w, r) } +func (w writerYNNWrapper) WriteString(s string) (int, error) { return WriteString(w, s) } + +// YNY + +type ( + writerYNY interface { + io.Writer + io.ByteWriter + io.StringWriter + } + writerYNYWrapper struct{ writerYNY } +) + +func (w writerYNYWrapper) WriteRune(r rune) (int, error) { return WriteRune(w, r) } + +// YYN + +type ( + writerYYN interface { + io.Writer + io.ByteWriter + RuneWriter + } + writerYYNWrapper struct{ writerYYN } +) + +func (w writerYYNWrapper) WriteString(s string) (int, error) { return WriteString(w, s) } + +// NewAllWriter wraps an io.Writer turning it in to an AllWriter. If +// the io.Writer already has any of the other write methods, then its +// native version of those methods are used. +func NewAllWriter(inner io.Writer) AllWriter { + switch inner := inner.(type) { + // 3 Y bits + case AllWriter: // YYY: + return inner + // 2 Y bits + case writerNYY: + return writerNYYWrapper{writerNYY: inner} + case writerYNY: + return writerYNYWrapper{writerYNY: inner} + case writerYYN: + return writerYYNWrapper{writerYYN: inner} + // 1 Y bit + case writerNNY: + return writerNNYWrapper{writerNNY: inner} + case writerNYN: + return writerNYNWrapper{writerNYN: inner} + case writerYNN: + return writerYNNWrapper{writerYNN: inner} + // 0 Y bits + default: // NNN: + return writerNNNWrapper{writerNNN: inner} + } +} + +// discard ///////////////////////////////////////////////////////////////////// + +// Discard is like io.Discard, but implements AllWriter. +var Discard = discard{} + +type discard struct{} + +func (discard) Write(p []byte) (int, error) { return len(p), nil } +func (discard) WriteByte(b byte) error { return nil } +func (discard) WriteRune(r rune) (int, error) { return 0, nil } +func (discard) WriteString(s string) (int, error) { return len(s), nil } diff --git a/internal/base64.go b/internal/base64.go index 15adbf4..291a229 100644 --- a/internal/base64.go +++ b/internal/base64.go @@ -19,7 +19,10 @@ type base64Decoder struct { bufLen int } -func NewBase64Decoder(w io.Writer) io.WriteCloser { +func NewBase64Decoder(w io.Writer) interface { + io.WriteCloser + RuneWriter +} { return &base64Decoder{ dst: w, } @@ -112,6 +115,10 @@ func (dec *base64Decoder) Write(dat []byte) (int, error) { return len(dat), nil } +func (dec *base64Decoder) WriteRune(r rune) (int, error) { + return WriteRune(dec, r) +} + func (dec *base64Decoder) Close() error { if dec.bufLen == 0 { return nil diff --git a/internal/parse.go b/internal/parse.go index b11aae6..36db4a9 100644 --- a/internal/parse.go +++ b/internal/parse.go @@ -313,6 +313,13 @@ type Parser struct { // a ["x","y" // ["x","y"] stack []RuneType + + barriers []barrier +} + +type barrier struct { + closed bool + stack []RuneType } func (par *Parser) init() { @@ -345,8 +352,26 @@ func (par *Parser) stackString() string { return buf.String() } +func (par *Parser) depth() int { + n := len(par.stack) + for _, barrier := range par.barriers { + n += len(barrier.stack) + } + return n +} + func (par *Parser) StackIsEmpty() bool { - return len(par.stack) == 0 || (len(par.stack) == 1 && par.stack[0] == runeTypeAny) + if len(par.barriers) > 0 { + return false + } + if len(par.stack) == 0 { + return true + } + return len(par.stack) == 1 && par.stack[0] == runeTypeAny +} + +func (par *Parser) StackSize() int { + return len(par.stack) } // Reset all Parser state. @@ -356,6 +381,99 @@ func (par *Parser) Reset() { } } +// PushReadBarrier causes the parser to expect EOF once the end of the +// element that is started by the current top-of-stack is reached, +// until this is un-done with PopBarrier. It essentially turns the +// parser in to a sub-parser. +// +// PushReadBarrier may only be called at the beginning of an element, +// whether that be +// +// - runeTypeAny +// - RuneTypeObjectBeg +// - RuneTypeArrayBeg +// - RuneTypeStringBeg +// - RuneTypeNumberIntNeg, RuneTypeNumberIntZero, RuneTypeNumberIntDig +// - RuneTypeTrueT +// - RuneTypeFalseF +// - RuneTypeNullN +func (par *Parser) PushReadBarrier() { + // Sanity checking. + par.init() + if len(par.stack) == 0 { + panic(errors.New("illegal PushReadBarrier call: empty stack")) + } + curState := par.stack[len(par.stack)-1] + switch curState { + case runeTypeAny, + RuneTypeObjectBeg, + RuneTypeArrayBeg, + RuneTypeStringBeg, + RuneTypeNumberIntNeg, RuneTypeNumberIntZero, RuneTypeNumberIntDig, + RuneTypeTrueT, + RuneTypeFalseF, + RuneTypeNullN: + // OK + default: + panic(fmt.Errorf("illegal PushReadBarrier call: %q", curState)) + } + // Actually push. + par.barriers = append(par.barriers, barrier{ + closed: par.closed, + stack: par.stack[:len(par.stack)-1], + }) + par.stack = []RuneType{curState} +} + +// PushWriteBarrier causes the parser to expect EOF once the end of +// the about-to-start element is reached, until this is un-done with +// PopBarrier. It essentially turns the parser in to a sub-parser. +// +// PushWriteBarrier may only be called at the places where an element +// of any type may start: +// +// - runeTypeAny for top-level and object-value elements +// - RuneTypeArrayBeg for array-item elements +// +// PushWriteBarrier signals intent to write an element; if it is +// called in a place where an element is optional (at the beginning of +// an array), it becomes a syntax error to not write the element. +func (par *Parser) PushWriteBarrier() { + par.init() + if len(par.stack) == 0 { + panic(errors.New("illegal PushWriteBarrier call: empty stack")) + } + switch par.stack[len(par.stack)-1] { + case runeTypeAny: + par.popState() + par.barriers = append(par.barriers, barrier{ + closed: par.closed, + stack: par.stack, + }) + par.stack = []RuneType{runeTypeAny} + case RuneTypeArrayBeg: + par.replaceState(RuneTypeArrayComma) + par.barriers = append(par.barriers, barrier{ + closed: par.closed, + stack: par.stack, + }) + par.stack = []RuneType{runeTypeAny} + default: + panic(fmt.Errorf("illegal PushWriteBarrier call: %q", par.stack[len(par.stack)-1])) + } +} + +// PopBarrier reverses a call to PushReadBarrier or PushWriteBarrier. +func (par *Parser) PopBarrier() { + if len(par.barriers) == 0 { + panic(errors.New("illegal PopBarrier call: empty barrier stack")) + } + barrier := par.barriers[len(par.barriers)-1] + par.barriers = par.barriers[:len(par.barriers)-1] + par.closed = barrier.closed + par.stack = append(barrier.stack, par.stack...) +} + // HandleEOF feeds EOF to the Parser. The returned RuneType is either // RuneTypeEOF or RuneTypeError. // @@ -435,12 +553,12 @@ func (par *Parser) HandleRune(c rune) (RuneType, error) { case 0x0020, 0x000A, 0x000D, 0x0009: return RuneTypeSpace, nil case '{': - if par.MaxDepth > 0 && len(par.stack) > par.MaxDepth { + if par.MaxDepth > 0 && par.depth() > par.MaxDepth { return RuneTypeError, ErrParserExceededMaxDepth } return par.replaceState(RuneTypeObjectBeg), nil case '[': - if par.MaxDepth > 0 && len(par.stack) > par.MaxDepth { + if par.MaxDepth > 0 && par.depth() > par.MaxDepth { return RuneTypeError, ErrParserExceededMaxDepth } return par.replaceState(RuneTypeArrayBeg), nil |