tpl/internal: Sync go_templates

Closes #10411
This commit is contained in:
Bjørn Erik Pedersen
2022-11-14 19:13:09 +01:00
parent 58a98c7758
commit f6ab9553f4
34 changed files with 739 additions and 514 deletions

View File

@@ -111,31 +111,36 @@ type stateFn func(*lexer) stateFn
// lexer holds the state of the scanner.
type lexer struct {
name string // the name of the input; used only for error reports
input string // the string being scanned
leftDelim string // start of action
rightDelim string // end of action
emitComment bool // emit itemComment tokens.
pos Pos // current position in the input
start Pos // start position of this item
width Pos // width of last rune read from input
items chan item // channel of scanned items
parenDepth int // nesting depth of ( ) exprs
line int // 1+number of newlines seen
startLine int // start line of this item
breakOK bool // break keyword allowed
continueOK bool // continue keyword allowed
name string // the name of the input; used only for error reports
input string // the string being scanned
leftDelim string // start of action marker
rightDelim string // end of action marker
pos Pos // current position in the input
start Pos // start position of this item
atEOF bool // we have hit the end of input and returned eof
parenDepth int // nesting depth of ( ) exprs
line int // 1+number of newlines seen
startLine int // start line of this item
item item // item to return to parser
insideAction bool // are we inside an action?
options lexOptions
}
// lexOptions control behavior of the lexer. All default to false.
type lexOptions struct {
emitComment bool // emit itemComment tokens.
breakOK bool // break keyword allowed
continueOK bool // continue keyword allowed
}
// next returns the next rune in the input.
func (l *lexer) next() rune {
if int(l.pos) >= len(l.input) {
l.width = 0
l.atEOF = true
return eof
}
r, w := utf8.DecodeRuneInString(l.input[l.pos:])
l.width = Pos(w)
l.pos += l.width
l.pos += Pos(w)
if r == '\n' {
l.line++
}
@@ -149,23 +154,41 @@ func (l *lexer) peek() rune {
return r
}
// backup steps back one rune. Can only be called once per call of next.
// backup steps back one rune.
func (l *lexer) backup() {
l.pos -= l.width
// Correct newline count.
if l.width == 1 && l.input[l.pos] == '\n' {
l.line--
if !l.atEOF && l.pos > 0 {
r, w := utf8.DecodeLastRuneInString(l.input[:l.pos])
l.pos -= Pos(w)
// Correct newline count.
if r == '\n' {
l.line--
}
}
}
// emit passes an item back to the client.
func (l *lexer) emit(t itemType) {
l.items <- item{t, l.start, l.input[l.start:l.pos], l.startLine}
// thisItem returns the item at the current input point with the specified type
// and advances the input.
func (l *lexer) thisItem(t itemType) item {
i := item{t, l.start, l.input[l.start:l.pos], l.startLine}
l.start = l.pos
l.startLine = l.line
return i
}
// emit passes the trailing text as an item back to the parser.
func (l *lexer) emit(t itemType) stateFn {
return l.emitItem(l.thisItem(t))
}
// emitItem passes the specified item to the parser.
func (l *lexer) emitItem(i item) stateFn {
l.item = i
return nil
}
// ignore skips over the pending input before this point.
// It tracks newlines in the ignored text, so use it only
// for text that is skipped without calling l.next.
func (l *lexer) ignore() {
l.line += strings.Count(l.input[l.start:l.pos], "\n")
l.start = l.pos
@@ -191,25 +214,31 @@ func (l *lexer) acceptRun(valid string) {
// errorf returns an error token and terminates the scan by passing
// back a nil pointer that will be the next state, terminating l.nextItem.
func (l *lexer) errorf(format string, args ...any) stateFn {
l.items <- item{itemError, l.start, fmt.Sprintf(format, args...), l.startLine}
l.item = item{itemError, l.start, fmt.Sprintf(format, args...), l.startLine}
l.start = 0
l.pos = 0
l.input = l.input[:0]
return nil
}
// nextItem returns the next item from the input.
// Called by the parser, not in the lexing goroutine.
func (l *lexer) nextItem() item {
return <-l.items
}
// drain drains the output so the lexing goroutine will exit.
// Called by the parser, not in the lexing goroutine.
func (l *lexer) drain() {
for range l.items {
l.item = item{itemEOF, l.pos, "EOF", l.startLine}
state := lexText
if l.insideAction {
state = lexInsideAction
}
for {
state = state(l)
if state == nil {
return l.item
}
}
}
// lex creates a new scanner for the input string.
func lex(name, input, left, right string, emitComment bool) *lexer {
func lex(name, input, left, right string) *lexer {
if left == "" {
left = leftDelim
}
@@ -217,27 +246,17 @@ func lex(name, input, left, right string, emitComment bool) *lexer {
right = rightDelim
}
l := &lexer{
name: name,
input: input,
leftDelim: left,
rightDelim: right,
emitComment: emitComment,
items: make(chan item),
line: 1,
startLine: 1,
name: name,
input: input,
leftDelim: left,
rightDelim: right,
line: 1,
startLine: 1,
insideAction: false,
}
go l.run()
return l
}
// run runs the state machine for the lexer.
func (l *lexer) run() {
for state := lexText; state != nil; {
state = state(l)
}
close(l.items)
}
// state functions
const (
@@ -249,31 +268,33 @@ const (
// lexText scans until an opening action delimiter, "{{".
func lexText(l *lexer) stateFn {
l.width = 0
if x := strings.Index(l.input[l.pos:], l.leftDelim); x >= 0 {
ldn := Pos(len(l.leftDelim))
l.pos += Pos(x)
trimLength := Pos(0)
if hasLeftTrimMarker(l.input[l.pos+ldn:]) {
trimLength = rightTrimLength(l.input[l.start:l.pos])
}
l.pos -= trimLength
if l.pos > l.start {
if x > 0 {
l.pos += Pos(x)
// Do we trim any trailing space?
trimLength := Pos(0)
delimEnd := l.pos + Pos(len(l.leftDelim))
if hasLeftTrimMarker(l.input[delimEnd:]) {
trimLength = rightTrimLength(l.input[l.start:l.pos])
}
l.pos -= trimLength
l.line += strings.Count(l.input[l.start:l.pos], "\n")
l.emit(itemText)
i := l.thisItem(itemText)
l.pos += trimLength
l.ignore()
if len(i.val) > 0 {
return l.emitItem(i)
}
}
l.pos += trimLength
l.ignore()
return lexLeftDelim
}
l.pos = Pos(len(l.input))
// Correctly reached EOF.
if l.pos > l.start {
l.line += strings.Count(l.input[l.start:l.pos], "\n")
l.emit(itemText)
return l.emit(itemText)
}
l.emit(itemEOF)
return nil
return l.emit(itemEOF)
}
// rightTrimLength returns the length of the spaces at the end of the string.
@@ -298,6 +319,7 @@ func leftTrimLength(s string) Pos {
}
// lexLeftDelim scans the left delimiter, which is known to be present, possibly with a trim marker.
// (The text to be trimmed has already been emitted.)
func lexLeftDelim(l *lexer) stateFn {
l.pos += Pos(len(l.leftDelim))
trimSpace := hasLeftTrimMarker(l.input[l.pos:])
@@ -310,28 +332,27 @@ func lexLeftDelim(l *lexer) stateFn {
l.ignore()
return lexComment
}
l.emit(itemLeftDelim)
i := l.thisItem(itemLeftDelim)
l.insideAction = true
l.pos += afterMarker
l.ignore()
l.parenDepth = 0
return lexInsideAction
return l.emitItem(i)
}
// lexComment scans a comment. The left comment marker is known to be present.
func lexComment(l *lexer) stateFn {
l.pos += Pos(len(leftComment))
i := strings.Index(l.input[l.pos:], rightComment)
if i < 0 {
x := strings.Index(l.input[l.pos:], rightComment)
if x < 0 {
return l.errorf("unclosed comment")
}
l.pos += Pos(i + len(rightComment))
l.pos += Pos(x + len(rightComment))
delim, trimSpace := l.atRightDelim()
if !delim {
return l.errorf("comment ends before closing delimiter")
}
if l.emitComment {
l.emit(itemComment)
}
i := l.thisItem(itemComment)
if trimSpace {
l.pos += trimMarkerLen
}
@@ -340,23 +361,27 @@ func lexComment(l *lexer) stateFn {
l.pos += leftTrimLength(l.input[l.pos:])
}
l.ignore()
if l.options.emitComment {
return l.emitItem(i)
}
return lexText
}
// lexRightDelim scans the right delimiter, which is known to be present, possibly with a trim marker.
func lexRightDelim(l *lexer) stateFn {
trimSpace := hasRightTrimMarker(l.input[l.pos:])
_, trimSpace := l.atRightDelim()
if trimSpace {
l.pos += trimMarkerLen
l.ignore()
}
l.pos += Pos(len(l.rightDelim))
l.emit(itemRightDelim)
i := l.thisItem(itemRightDelim)
if trimSpace {
l.pos += leftTrimLength(l.input[l.pos:])
l.ignore()
}
return lexText
l.insideAction = false
return l.emitItem(i)
}
// lexInsideAction scans the elements inside action delimiters.
@@ -378,14 +403,14 @@ func lexInsideAction(l *lexer) stateFn {
l.backup() // Put space back in case we have " -}}".
return lexSpace
case r == '=':
l.emit(itemAssign)
return l.emit(itemAssign)
case r == ':':
if l.next() != '=' {
return l.errorf("expected :=")
}
l.emit(itemDeclare)
return l.emit(itemDeclare)
case r == '|':
l.emit(itemPipe)
return l.emit(itemPipe)
case r == '"':
return lexQuote
case r == '`':
@@ -410,20 +435,19 @@ func lexInsideAction(l *lexer) stateFn {
l.backup()
return lexIdentifier
case r == '(':
l.emit(itemLeftParen)
l.parenDepth++
return l.emit(itemLeftParen)
case r == ')':
l.emit(itemRightParen)
l.parenDepth--
if l.parenDepth < 0 {
return l.errorf("unexpected right paren %#U", r)
return l.errorf("unexpected right paren")
}
return l.emit(itemRightParen)
case r <= unicode.MaxASCII && unicode.IsPrint(r):
l.emit(itemChar)
return l.emit(itemChar)
default:
return l.errorf("unrecognized character in action: %#U", r)
}
return lexInsideAction
}
// lexSpace scans a run of space characters.
@@ -448,13 +472,11 @@ func lexSpace(l *lexer) stateFn {
return lexRightDelim // On the delim, so go right to that.
}
}
l.emit(itemSpace)
return lexInsideAction
return l.emit(itemSpace)
}
// lexIdentifier scans an alphanumeric.
func lexIdentifier(l *lexer) stateFn {
Loop:
for {
switch r := l.next(); {
case isAlphaNumeric(r):
@@ -468,22 +490,19 @@ Loop:
switch {
case key[word] > itemKeyword:
item := key[word]
if item == itemBreak && !l.breakOK || item == itemContinue && !l.continueOK {
l.emit(itemIdentifier)
} else {
l.emit(item)
if item == itemBreak && !l.options.breakOK || item == itemContinue && !l.options.continueOK {
return l.emit(itemIdentifier)
}
return l.emit(item)
case word[0] == '.':
l.emit(itemField)
return l.emit(itemField)
case word == "true", word == "false":
l.emit(itemBool)
return l.emit(itemBool)
default:
l.emit(itemIdentifier)
return l.emit(itemIdentifier)
}
break Loop
}
}
return lexInsideAction
}
// lexField scans a field: .Alphanumeric.
@@ -496,8 +515,7 @@ func lexField(l *lexer) stateFn {
// The $ has been scanned.
func lexVariable(l *lexer) stateFn {
if l.atTerminator() { // Nothing interesting follows -> "$".
l.emit(itemVariable)
return lexInsideAction
return l.emit(itemVariable)
}
return lexFieldOrVariable(l, itemVariable)
}
@@ -507,11 +525,9 @@ func lexVariable(l *lexer) stateFn {
func lexFieldOrVariable(l *lexer, typ itemType) stateFn {
if l.atTerminator() { // Nothing interesting follows -> "." or "$".
if typ == itemVariable {
l.emit(itemVariable)
} else {
l.emit(itemDot)
return l.emit(itemVariable)
}
return lexInsideAction
return l.emit(itemDot)
}
var r rune
for {
@@ -524,8 +540,7 @@ func lexFieldOrVariable(l *lexer, typ itemType) stateFn {
if !l.atTerminator() {
return l.errorf("bad character %#U", r)
}
l.emit(typ)
return lexInsideAction
return l.emit(typ)
}
// atTerminator reports whether the input is at valid termination character to
@@ -541,13 +556,7 @@ func (l *lexer) atTerminator() bool {
case eof, '.', ',', '|', ':', ')', '(':
return true
}
// Does r start the delimiter? This can be ambiguous (with delim=="//", $x/2 will
// succeed but should fail) but only in extremely rare cases caused by willfully
// bad choice of delimiter.
if rd, _ := utf8.DecodeRuneInString(l.rightDelim); rd == r {
return true
}
return false
return strings.HasPrefix(l.input[l.pos:], l.rightDelim)
}
// lexChar scans a character constant. The initial quote is already
@@ -567,8 +576,7 @@ Loop:
break Loop
}
}
l.emit(itemCharConstant)
return lexInsideAction
return l.emit(itemCharConstant)
}
// lexNumber scans a number: decimal, octal, hex, float, or imaginary. This
@@ -584,11 +592,9 @@ func lexNumber(l *lexer) stateFn {
if !l.scanNumber() || l.input[l.pos-1] != 'i' {
return l.errorf("bad number syntax: %q", l.input[l.start:l.pos])
}
l.emit(itemComplex)
} else {
l.emit(itemNumber)
return l.emit(itemComplex)
}
return lexInsideAction
return l.emit(itemNumber)
}
func (l *lexer) scanNumber() bool {
@@ -644,8 +650,7 @@ Loop:
break Loop
}
}
l.emit(itemString)
return lexInsideAction
return l.emit(itemString)
}
// lexRawQuote scans a raw quoted string.
@@ -659,8 +664,7 @@ Loop:
break Loop
}
}
l.emit(itemRawString)
return lexInsideAction
return l.emit(itemRawString)
}
// isSpace reports whether r is a space character.

View File

@@ -362,8 +362,7 @@ var lexTests = []lexTest{
{"extra right paren", "{{3)}}", []item{
tLeft,
mkItem(itemNumber, "3"),
tRpar,
mkItem(itemError, `unexpected right paren U+0029 ')'`),
mkItem(itemError, "unexpected right paren"),
}},
// Fixed bugs
@@ -397,7 +396,12 @@ var lexTests = []lexTest{
// collect gathers the emitted items into a slice.
func collect(t *lexTest, left, right string) (items []item) {
l := lex(t.name, t.input, left, right, true)
l := lex(t.name, t.input, left, right)
l.options = lexOptions{
emitComment: true,
breakOK: true,
continueOK: true,
}
for {
item := l.nextItem()
items = append(items, item)
@@ -434,7 +438,9 @@ func TestLex(t *testing.T) {
items := collect(&test, "", "")
if !equal(items, test.items, false) {
t.Errorf("%s: got\n\t%+v\nexpected\n\t%v", test.name, items, test.items)
return // TODO
}
t.Log(test.name, "OK")
}
}
@@ -472,6 +478,39 @@ func TestDelims(t *testing.T) {
}
}
func TestDelimsAlphaNumeric(t *testing.T) {
test := lexTest{"right delimiter with alphanumeric start", "{{hub .host hub}}", []item{
mkItem(itemLeftDelim, "{{hub"),
mkItem(itemSpace, " "),
mkItem(itemField, ".host"),
mkItem(itemSpace, " "),
mkItem(itemRightDelim, "hub}}"),
tEOF,
}}
items := collect(&test, "{{hub", "hub}}")
if !equal(items, test.items, false) {
t.Errorf("%s: got\n\t%v\nexpected\n\t%v", test.name, items, test.items)
}
}
func TestDelimsAndMarkers(t *testing.T) {
test := lexTest{"delims that look like markers", "{{- .x -}} {{- - .x - -}}", []item{
mkItem(itemLeftDelim, "{{- "),
mkItem(itemField, ".x"),
mkItem(itemRightDelim, " -}}"),
mkItem(itemLeftDelim, "{{- "),
mkItem(itemField, ".x"),
mkItem(itemRightDelim, " -}}"),
tEOF,
}}
items := collect(&test, "{{- ", " -}}")
if !equal(items, test.items, false) {
t.Errorf("%s: got\n\t%v\nexpected\n\t%v", test.name, items, test.items)
}
}
var lexPosTests = []lexTest{
{"empty", "", []item{{itemEOF, 0, "", 1}}},
{"punctuation", "{{,@%#}}", []item{
@@ -533,22 +572,6 @@ func TestPos(t *testing.T) {
}
}
// Test that an error shuts down the lexing goroutine.
func TestShutdown(t *testing.T) {
// We need to duplicate template.Parse here to hold on to the lexer.
const text = "erroneous{{define}}{{else}}1234"
lexer := lex("foo", text, "{{", "}}", false)
_, err := New("root").parseLexer(lexer)
if err == nil {
t.Fatalf("expected error")
}
// The error should have drained the input. Therefore, the lexer should be shut down.
token, ok := <-lexer.items
if ok {
t.Fatalf("input was not drained; got %v", token)
}
}
// parseLexer is a local version of parse that lets us pass in the lexer instead of building it.
// We expect an error, so the tree set and funcs list are explicitly nil.
func (t *Tree) parseLexer(lex *lexer) (tree *Tree, err error) {

View File

@@ -210,7 +210,6 @@ func (t *Tree) recover(errp *error) {
panic(e)
}
if t != nil {
t.lex.drain()
t.stopParse()
}
*errp = e.(error)
@@ -224,8 +223,6 @@ func (t *Tree) startParse(funcs []map[string]any, lex *lexer, treeSet map[string
t.vars = []string{"$"}
t.funcs = funcs
t.treeSet = treeSet
lex.breakOK = !t.hasFunction("break")
lex.continueOK = !t.hasFunction("continue")
}
// stopParse terminates parsing.
@@ -243,8 +240,13 @@ func (t *Tree) stopParse() {
func (t *Tree) Parse(text, leftDelim, rightDelim string, treeSet map[string]*Tree, funcs ...map[string]any) (tree *Tree, err error) {
defer t.recover(&err)
t.ParseName = t.Name
emitComment := t.Mode&ParseComments != 0
t.startParse(funcs, lex(t.Name, text, leftDelim, rightDelim, emitComment), treeSet)
lexer := lex(t.Name, text, leftDelim, rightDelim)
lexer.options = lexOptions{
emitComment: t.Mode&ParseComments != 0,
breakOK: !t.hasFunction("break"),
continueOK: !t.hasFunction("continue"),
}
t.startParse(funcs, lexer, treeSet)
t.text = text
t.parse()
t.add()
@@ -341,7 +343,9 @@ func (t *Tree) parseDefinition() {
}
// itemList:
//
// textOrAction*
//
// Terminates at {{end}} or {{else}}, returned separately.
func (t *Tree) itemList() (list *ListNode, next Node) {
list = t.newList(t.peekNonSpace().pos)
@@ -358,6 +362,7 @@ func (t *Tree) itemList() (list *ListNode, next Node) {
}
// textOrAction:
//
// text | comment | action
func (t *Tree) textOrAction() Node {
switch token := t.nextNonSpace(); token.typ {
@@ -380,8 +385,10 @@ func (t *Tree) clearActionLine() {
}
// Action:
//
// control
// command ("|" command)*
//
// Left delim is past. Now get actions.
// First word could be a keyword such as range.
func (t *Tree) action() (n Node) {
@@ -412,7 +419,9 @@ func (t *Tree) action() (n Node) {
}
// Break:
//
// {{break}}
//
// Break keyword is past.
func (t *Tree) breakControl(pos Pos, line int) Node {
if token := t.nextNonSpace(); token.typ != itemRightDelim {
@@ -425,7 +434,9 @@ func (t *Tree) breakControl(pos Pos, line int) Node {
}
// Continue:
//
// {{continue}}
//
// Continue keyword is past.
func (t *Tree) continueControl(pos Pos, line int) Node {
if token := t.nextNonSpace(); token.typ != itemRightDelim {
@@ -438,6 +449,7 @@ func (t *Tree) continueControl(pos Pos, line int) Node {
}
// Pipeline:
//
// declarations? command ('|' command)*
func (t *Tree) pipeline(context string, end itemType) (pipe *PipeNode) {
token := t.peekNonSpace()
@@ -549,16 +561,20 @@ func (t *Tree) parseControl(allowElseIf bool, context string) (pos Pos, line int
}
// If:
//
// {{if pipeline}} itemList {{end}}
// {{if pipeline}} itemList {{else}} itemList {{end}}
//
// If keyword is past.
func (t *Tree) ifControl() Node {
return t.newIf(t.parseControl(true, "if"))
}
// Range:
//
// {{range pipeline}} itemList {{end}}
// {{range pipeline}} itemList {{else}} itemList {{end}}
//
// Range keyword is past.
func (t *Tree) rangeControl() Node {
r := t.newRange(t.parseControl(false, "range"))
@@ -566,22 +582,28 @@ func (t *Tree) rangeControl() Node {
}
// With:
//
// {{with pipeline}} itemList {{end}}
// {{with pipeline}} itemList {{else}} itemList {{end}}
//
// If keyword is past.
func (t *Tree) withControl() Node {
return t.newWith(t.parseControl(false, "with"))
}
// End:
//
// {{end}}
//
// End keyword is past.
func (t *Tree) endControl() Node {
return t.newEnd(t.expect(itemRightDelim, "end").pos)
}
// Else:
//
// {{else}}
//
// Else keyword is past.
func (t *Tree) elseControl() Node {
// Special case for "else if".
@@ -595,7 +617,9 @@ func (t *Tree) elseControl() Node {
}
// Block:
//
// {{block stringValue pipeline}}
//
// Block keyword is past.
// The name must be something that can evaluate to a string.
// The pipeline is mandatory.
@@ -623,7 +647,9 @@ func (t *Tree) blockControl() Node {
}
// Template:
//
// {{template stringValue pipeline}}
//
// Template keyword is past. The name must be something that can evaluate
// to a string.
func (t *Tree) templateControl() Node {
@@ -654,7 +680,9 @@ func (t *Tree) parseTemplateName(token item, context string) (name string) {
}
// command:
//
// operand (space operand)*
//
// space-separated arguments up to a pipeline character or right delimiter.
// we consume the pipe character but leave the right delim to terminate the action.
func (t *Tree) command() *CommandNode {
@@ -684,7 +712,9 @@ func (t *Tree) command() *CommandNode {
}
// operand:
//
// term .Field*
//
// An operand is a space-separated component of a command,
// a term possibly followed by field accesses.
// A nil return means the next item is not an operand.
@@ -718,12 +748,14 @@ func (t *Tree) operand() Node {
}
// term:
//
// literal (number, string, nil, boolean)
// function (identifier)
// .
// .Field
// $
// '(' pipeline ')'
//
// A term is a simple "expression".
// A nil return means the next item is not a term.
func (t *Tree) term() Node {

View File

@@ -492,7 +492,7 @@ var errorTests = []parseTest{
hasError, `unclosed left paren`},
{"rparen",
"{{.X 1 2 3 ) }}",
hasError, `unexpected ")" in command`},
hasError, "unexpected right paren"},
{"rparen2",
"{{(.X 1 2 3",
hasError, `unclosed action`},
@@ -600,7 +600,8 @@ func TestBlock(t *testing.T) {
}
func TestLineNum(t *testing.T) {
const count = 100
// const count = 100
const count = 3
text := strings.Repeat("{{printf 1234}}\n", count)
tree, err := New("bench").Parse(text, "", "", make(map[string]*Tree), builtins)
if err != nil {
@@ -614,11 +615,11 @@ func TestLineNum(t *testing.T) {
// Action first.
action := nodes[i].(*ActionNode)
if action.Line != line {
t.Fatalf("line %d: action is line %d", line, action.Line)
t.Errorf("line %d: action is line %d", line, action.Line)
}
pipe := action.Pipe
if pipe.Line != line {
t.Fatalf("line %d: pipe is line %d", line, pipe.Line)
t.Errorf("line %d: pipe is line %d", line, pipe.Line)
}
}
}