Preliminary parsing of control structures.

edit/style.go

@@ -41,6 +41,19 @@ var styleForSep = map[string]string{
 	"}":  "1",
 
 	"&": "1",
+
+	"if":    "33",
+	"then":  "33",
+	"elif":  "33",
+	"else":  "33",
+	"fi":    "33",
+	"while": "33",
+	"do":    "33",
+	"done":  "33",
+	"for":   "33",
+	"in":    "33",
+	"begin": "33",
+	"end":   "33",
 }
 
 // Styles for semantic coloring.

eval/compile.go

@@ -194,6 +194,11 @@ func (cp *compiler) form(n *parse.Form) Op {
 		}
 	}
 
+	if n.Control != nil {
+		cp.errorf(n.Control.Begin(), "control structure not yet implemented")
+		return func(ec *EvalCtx) {}
+	}
+
 	headStr, ok := oneString(n.Head)
 	if ok {
 		compileForm, ok := builtinSpecials[headStr]

parse/boilerplate.go

@@ -31,6 +31,11 @@ func (n *Form) addToAssignments(ch *Assignment) {
 	addChild(n, ch)
 }
 
+func (n *Form) setControl(ch *Control) {
+	n.Control = ch
+	addChild(n, ch)
+}
+
 func (n *Form) setHead(ch *Compound) {
 	n.Head = ch
 	addChild(n, ch)
@@ -82,6 +87,49 @@ func parseAssignment(ps *parser) *Assignment {
 	return n
 }
 
+func (n *Control) setCondition(ch *Chunk) {
+	n.Condition = ch
+	addChild(n, ch)
+}
+
+func (n *Control) setIterator(ch *Primary) {
+	n.Iterator = ch
+	addChild(n, ch)
+}
+
+func (n *Control) setArray(ch *Array) {
+	n.Array = ch
+	addChild(n, ch)
+}
+
+func (n *Control) setBody(ch *Chunk) {
+	n.Body = ch
+	addChild(n, ch)
+}
+
+func (n *Control) addToElifConditions(ch *Chunk) {
+	n.ElifConditions = append(n.ElifConditions, ch)
+	addChild(n, ch)
+}
+
+func (n *Control) addToElifBodies(ch *Chunk) {
+	n.ElifBodies = append(n.ElifBodies, ch)
+	addChild(n, ch)
+}
+
+func (n *Control) setElseBody(ch *Chunk) {
+	n.ElseBody = ch
+	addChild(n, ch)
+}
+
+func parseControl(ps *parser, leader string) *Control {
+	n := &Control{node: node{begin: ps.pos}}
+	n.parse(ps, leader)
+	n.end = ps.pos
+	n.sourceText = ps.src[n.begin:n.end]
+	return n
+}
+
 func (n *ExitusRedir) setDest(ch *Compound) {
 	n.Dest = ch
 	addChild(n, ch)

parse/parse.go

@@ -2,17 +2,18 @@
 package parse
 
 //go:generate ./boilerplate.py
-//go:generate stringer -type=PrimaryType,RedirMode -output=string.go
+//go:generate stringer -type=PrimaryType,RedirMode,ControlKinD -output=string.go
 
 import (
 	"bytes"
 	"errors"
+	"fmt"
 	"unicode"
 )
 
 // Parse parses elvish source.
 func Parse(src string) (*Chunk, error) {
-	ps := &parser{src, 0, 0, []map[rune]int{{}}, nil}
+	ps := &parser{src, 0, 0, []map[rune]int{{}}, 0, nil}
 	bn := parseChunk(ps)
 	if ps.pos != len(src) {
 		ps.error(errUnexpectedRune)
@@ -26,19 +27,27 @@ func Parse(src string) (*Chunk, error) {
 
 // Errors.
 var (
-	errUnexpectedRune       = errors.New("unexpected rune")
-	errShouldBeForm         = newError("", "form")
-	errDuplicateExitusRedir = newError("duplicate exitus redir")
-	errBadRedirSign         = newError("bad redir sign", "'<'", "'>'", "'>>'", "'<>'")
-	errShouldBeFD           = newError("", "a composite term representing fd")
-	errShouldBeFilename     = newError("", "a composite term representing filename")
-	errShouldBeArray        = newError("", "spaced")
-	errStringUnterminated   = newError("string not terminated")
-	errInvalidEscape        = newError("invalid escape sequence")
-	errInvalidEscapeOct     = newError("invalid escape sequence", "octal digit")
-	errInvalidEscapeHex     = newError("invalid escape sequence", "hex digit")
-	errInvalidEscapeControl = newError("invalid control sequence", "a rune between @ (0x40) and _(0x5F)")
-	errShouldBePrimary      = newError("",
+	errUnexpectedRune         = errors.New("unexpected rune")
+	errShouldBeForm           = newError("", "form")
+	errDuplicateExitusRedir   = newError("duplicate exitus redir")
+	errShouldBeThen           = newError("", "then")
+	errShouldBeElifOrElseOrFi = newError("", "elif", "else", "fi")
+	errShouldBeFi             = newError("", "fi")
+	errShouldBeDo             = newError("", "do")
+	errShouldBeDone           = newError("", "done")
+	errShouldBeIn             = newError("", "in")
+	errShouldBePipelineSep    = newError("", "';'", "newline")
+	errShouldBeEnd            = newError("", "end")
+	errBadRedirSign           = newError("bad redir sign", "'<'", "'>'", "'>>'", "'<>'")
+	errShouldBeFD             = newError("", "a composite term representing fd")
+	errShouldBeFilename       = newError("", "a composite term representing filename")
+	errShouldBeArray          = newError("", "spaced")
+	errStringUnterminated     = newError("string not terminated")
+	errInvalidEscape          = newError("invalid escape sequence")
+	errInvalidEscapeOct       = newError("invalid escape sequence", "octal digit")
+	errInvalidEscapeHex       = newError("invalid escape sequence", "hex digit")
+	errInvalidEscapeControl   = newError("invalid control sequence", "a rune between @ (0x40) and _(0x5F)")
+	errShouldBePrimary        = newError("",
 		"single-quoted string", "double-quoted string", "bareword")
 	errShouldBeVariableName       = newError("", "variable name")
 	errShouldBeRBracket           = newError("", "']'")
@@ -60,6 +69,18 @@ type Chunk struct {
 func (bn *Chunk) parse(ps *parser) {
 	bn.parseSeps(ps)
 	for startsPipeline(ps.peek()) {
+		leader, starter := findLeader(ps)
+		if leader != "" && !starter && ps.controls > 0 {
+			// We found a non-starting leader and there is a control block that
+			// has not been closed. Stop parsing this chunk. We don't check the
+			// validity of the leader; the checking is done where the control
+			// block is parsed (e.g. (*Form).parseIf).
+			break
+		}
+		// We have more chance to check for validity of the leader, but
+		// eventually it will be checked in (*Form).parse. So we don't check it
+		// here, for more uniform error reporting and recovery.
+
 		bn.addToPipelines(parsePipeline(ps))
 		if bn.parseSeps(ps) == 0 {
 			break
@@ -126,10 +147,27 @@ func startsPipeline(r rune) bool {
 	return startsForm(r)
 }
 
-// Form = { Space } { { Assignment } { Space } } Compound { Space } { ( Compound | MapPair | Redir | ExitusRedir ) { Space } }
+// findLeader look aheads a command leader. It returns the leader and whether
+// it starts a control block.
+func findLeader(ps *parser) (string, bool) {
+	switch leader := ps.findPossibleLeader(); leader {
+	case "if", "while", "for", "do", "begin":
+		// Starting leaders are always legal.
+		return leader, true
+	case "then", "elif", "else", "fi", "done", "end":
+		return leader, false
+	default:
+		// There is no leader.
+		return "", false
+	}
+}
+
+// Form = { Space } { { Assignment } { Space } }
+//        { Compound | Control } { Space } { ( Compound | MapPair | Redir | ExitusRedir ) { Space } }
 type Form struct {
 	node
 	Assignments []*Assignment
+	Control     *Control
 	Head        *Compound
 	Args        []*Compound
 	NamedArgs   []*MapPair
@@ -142,14 +180,24 @@ func (fn *Form) parse(ps *parser) {
 	for fn.tryAssignment(ps) {
 		parseSpaces(fn, ps)
 	}
-	if !startsCompound(ps.peek()) {
+	leader, starter := findLeader(ps)
+	if leader != "" {
+		// Parse Control.
+		if starter {
+			fn.setControl(parseControl(ps, leader))
+		} else {
+			ps.error(fmt.Errorf("bogus command leader %q ignored", leader))
+		}
+	} else if startsCompound(ps.peek()) {
+		// Parse Head.
+		fn.setHead(parseCompound(ps))
+		parseSpaces(fn, ps)
+	} else {
 		if len(fn.Assignments) > 0 {
 			return
 		}
 		ps.error(errShouldBeCompound)
 	}
-	fn.setHead(parseCompound(ps))
-	parseSpaces(fn, ps)
 
 	for {
 		r := ps.peek()
@@ -203,6 +251,9 @@ func (fn *Form) tryAssignment(ps *parser) bool {
 	return true
 }
 
+func (fn *Form) parseIf(ps *parser) {
+}
+
 func startsForm(r rune) bool {
 	return isSpace(r) || startsCompound(r)
 }
@@ -225,6 +276,128 @@ func (an *Assignment) parse(ps *parser) {
 	an.setSrc(parseCompound(ps))
 }
 
+// Control = IfControl | WhileControl | ForControl | BeginControl
+// IfControl = If Chunk Then Chunk { Elif Chunk Then Chunk } [ Else Chunk ] Fi
+// WhileControl = While Chunk Do Chunk [ Else Chunk ] Done
+// ForControl = For Primary In Array PipelineSep Do Chunk [ Else Chunk ] Done
+// BeginControl = Begin Chunk Done
+// If = "if" Space { Space }
+// (Similiar for Then, Elif, Else, Fi, While, Do, Done, For, Begin, End)
+type Control struct {
+	node
+	Kind           ControlKind
+	Condition      *Chunk   // Valid for IfControl and WhileControl.
+	Iterator       *Primary // Valid for ForControl.
+	Array          *Array   // Valid for ForControl.
+	Body           *Chunk   // Valid for all.
+	ElifConditions []*Chunk // Valid for IfControl.
+	ElifBodies     []*Chunk // Valid for IfControl.
+	ElseBody       *Chunk   // Valid for IfControl, WhileControl and ForControl.
+}
+
+// ControlKind identifies which control structure a Control represents.
+type ControlKind int
+
+// Possible values of ControlKind.
+const (
+	BadControl ControlKind = iota
+	IfControl
+	WhileControl
+	ForControl
+	BeginControl
+)
+
+func (ctrl *Control) parse(ps *parser, leader string) {
+	ps.advance(len(leader))
+	addSep(ctrl, ps)
+
+	ps.controls++
+	defer func() { ps.controls-- }()
+
+	consumeLeader := func() string {
+		leader, _ := findLeader(ps)
+		if len(leader) > 0 {
+			ps.advance(len(leader))
+			addSep(ctrl, ps)
+		}
+		return leader
+	}
+
+	doElseDone := func() {
+		if consumeLeader() != "do" {
+			ps.error(errShouldBeDo)
+		}
+		ctrl.setBody(parseChunk(ps))
+		if leader, _ := findLeader(ps); leader == "else" {
+			consumeLeader()
+			ctrl.setElseBody(parseChunk(ps))
+		}
+		if consumeLeader() != "done" {
+			ps.error(errShouldBeDone)
+		}
+	}
+
+	switch leader {
+	case "if":
+		ctrl.Kind = IfControl
+		ctrl.setCondition(parseChunk(ps))
+		if consumeLeader() != "then" {
+			ps.error(errShouldBeThen)
+		}
+		ctrl.setBody(parseChunk(ps))
+	Elifs:
+		for {
+			switch consumeLeader() {
+			case "fi":
+				break Elifs
+			case "elif":
+				ctrl.addToElifConditions(parseChunk(ps))
+				if consumeLeader() != "then" {
+					ps.error(errShouldBeThen)
+				}
+				ctrl.addToElifBodies(parseChunk(ps))
+			case "else":
+				ctrl.setElseBody(parseChunk(ps))
+				if consumeLeader() != "fi" {
+					ps.error(errShouldBeFi)
+				}
+				break Elifs
+			default:
+				ps.error(errShouldBeElifOrElseOrFi)
+				break Elifs
+			}
+		}
+	case "while":
+		ctrl.Kind = WhileControl
+		ctrl.setCondition(parseChunk(ps))
+		doElseDone()
+	case "for":
+		ctrl.Kind = ForControl
+		parseSpaces(ctrl, ps)
+		ctrl.setIterator(parsePrimary(ps))
+		parseSpaces(ctrl, ps)
+		if consumeLeader() != "in" {
+			ps.error(errShouldBeIn)
+		}
+		ctrl.setArray(parseArray(ps))
+		switch ps.peek() {
+		case '\n', ';':
+			ps.next()
+		default:
+			ps.error(errShouldBePipelineSep)
+		}
+		doElseDone()
+	case "begin":
+		ctrl.Kind = BeginControl
+		ctrl.setBody(parseChunk(ps))
+		if consumeLeader() != "end" {
+			ps.error(errShouldBeEnd)
+		}
+	default:
+		ps.error(fmt.Errorf("unknown leader %q; parser error", leader))
+	}
+}
+
 // ExitusRedir = '?' '>' { Space } Compound
 type ExitusRedir struct {
 	node

parse/parse_test.go

@@ -31,6 +31,14 @@ var goodCases = []struct {
 	// Lots of unnecessary whitespaces
 	{"  ;\n\n  ls \t ;\n", ast{"Chunk", fs{"Pipelines": []string{"ls \t "}}}},
 
+	// Control structures.
+	/*
+		{"if true; then echo then; else echo else; fi",
+			ast{"Chunk/Pipeline/Form", fs{
+				"Control": "233",
+			}}},
+	*/
+
 	// Form
 	{"ls x y", ast{"Chunk/Pipeline/Form", fs{
 		"Head": "ls",

parse/parser.go

@@ -13,11 +13,12 @@ import (
 //
 // NOTE: The str member is assumed to be valid UF-8.
 type parser struct {
-	src     string
-	pos     int
-	overEOF int
-	cutsets []map[rune]int
-	errors  *errutil.Errors
+	src      string
+	pos      int
+	overEOF  int
+	cutsets  []map[rune]int
+	controls int
+	errors   *errutil.Errors
 }
 
 const eof rune = -1
@@ -41,6 +42,26 @@ func (ps *parser) hasPrefix(prefix string) bool {
 	return strings.HasPrefix(ps.src[ps.pos:], prefix)
 }
 
+// findWord look aheads a series of runes in [a-z] followed by ' ', '\t' or
+// '\n'. If the lookahead fails, it returns an empty string. It is useful for
+// looking for command leaders.
+func (ps *parser) findPossibleLeader() string {
+	rest := ps.src[ps.pos:]
+	i := strings.IndexFunc(rest, func(r rune) bool {
+		return r < 'a' || r > 'z'
+	})
+	if i == -1 {
+		// The whole rest is just one possible leader.
+		return rest
+	}
+	switch rest[i] {
+	case ' ', '\t', '\n':
+		return rest[:i]
+	default:
+		return ""
+	}
+}
+
 func (ps *parser) next() rune {
 	if ps.pos == len(ps.src) {
 		ps.overEOF++
@@ -63,6 +84,14 @@ func (ps *parser) backup() {
 	ps.pos -= s
 }
 
+func (ps *parser) advance(c int) {
+	ps.pos += c
+	if ps.pos > len(ps.src) {
+		ps.overEOF = ps.pos - len(ps.src)
+		ps.pos = len(ps.src)
+	}
+}
+
 func (ps *parser) error(e error) {
 	if ps.errors == nil {
 		ps.errors = &errutil.Errors{}

parse/string.go

@@ -1,4 +1,4 @@
-// generated by stringer -type=PrimaryType,RedirMode -output=string.go; DO NOT EDIT
+// Code generated by "stringer -type=PrimaryType,RedirMode,ControlKind -output=string.go"; DO NOT EDIT
 
 package parse
 
@@ -9,7 +9,7 @@ const _PrimaryType_name = "BadPrimaryBarewordSingleQuotedDoubleQuotedVariableWil
 var _PrimaryType_index = [...]uint8{0, 10, 18, 30, 42, 50, 58, 63, 75, 88, 92, 98, 101, 107}
 
 func (i PrimaryType) String() string {
-	if i < 0 || i+1 >= PrimaryType(len(_PrimaryType_index)) {
+	if i < 0 || i >= PrimaryType(len(_PrimaryType_index)-1) {
 		return fmt.Sprintf("PrimaryType(%d)", i)
 	}
 	return _PrimaryType_name[_PrimaryType_index[i]:_PrimaryType_index[i+1]]
@@ -20,8 +20,19 @@ const _RedirMode_name = "BadRedirModeReadWriteReadWriteAppend"
 var _RedirMode_index = [...]uint8{0, 12, 16, 21, 30, 36}
 
 func (i RedirMode) String() string {
-	if i < 0 || i+1 >= RedirMode(len(_RedirMode_index)) {
+	if i < 0 || i >= RedirMode(len(_RedirMode_index)-1) {
 		return fmt.Sprintf("RedirMode(%d)", i)
 	}
 	return _RedirMode_name[_RedirMode_index[i]:_RedirMode_index[i+1]]
 }
+
+const _ControlKind_name = "BadControlIfControlWhileControlForControlBeginControl"
+
+var _ControlKind_index = [...]uint8{0, 10, 19, 31, 41, 53}
+
+func (i ControlKind) String() string {
+	if i < 0 || i >= ControlKind(len(_ControlKind_index)-1) {
+		return fmt.Sprintf("ControlKind(%d)", i)
+	}
+	return _ControlKind_name[_ControlKind_index[i]:_ControlKind_index[i+1]]
+}