working on integrating position. identify cmd assignment words. make group words consistent. always copy Raw runes. fixup command whitespace

pkg/shparse/extend.go

@@ -64,30 +64,26 @@ func (w *WordType) writeRune(ch rune) {
 	return
 }
 
-func (w *WordType) cloneRaw() {
-	if len(w.Raw) == 0 {
-		return
-	}
-	buf := make([]rune, 0, len(w.Raw))
-	w.Raw = append(buf, w.Raw...)
-}
-
 type extendContext struct {
+	Input     []*WordType
+	InputPos  int
 	QC        QuoteContext
 	Rtn       []*WordType
 	CurWord   *WordType
 	Intention string
 }
 
-func makeExtendContext(qc QuoteContext, w *WordType) *extendContext {
-	rtn := &extendContext{QC: qc, Intention: WordTypeLit}
-	if w != nil {
-		w.cloneRaw()
-		rtn.Rtn = []*WordType{w}
-		rtn.CurWord = w
-		rtn.Intention = w.Type
+func makeExtendContext(qc QuoteContext, word *WordType) *extendContext {
+	rtn := &extendContext{QC: qc}
+	if word == nil {
+		rtn.Intention = WordTypeLit
+		return rtn
+	} else {
+		rtn.Intention = word.Type
+		rtn.Rtn = []*WordType{word}
+		rtn.CurWord = word
+		return rtn
 	}
-	return rtn
 }
 
 func (ec *extendContext) appendWord(w *WordType) {

pkg/shparse/shparse.go

@@ -3,6 +3,8 @@ package shparse
 import (
 	"bytes"
 	"fmt"
+
+	"github.com/scripthaus-dev/sh2-server/pkg/utilfn"
 )
 
 //
@@ -103,33 +105,37 @@ var bashNoneRW = []string{
 type wordMeta struct {
 	Type         string
 	EmptyWord    []rune
+	PrefixLen    int
 	SuffixLen    int
 	CanExtend    bool
 	QuoteContext bool
 }
 
-func makeWordMeta(wtype string, emptyWord string, suffixLen int, canExtend bool, quoteContext bool) {
-	wordMetaMap[wtype] = wordMeta{wtype, []rune(emptyWord), suffixLen, canExtend, quoteContext}
+func makeWordMeta(wtype string, emptyWord string, prefixLen int, suffixLen int, canExtend bool, quoteContext bool) {
+	if len(emptyWord) != prefixLen+suffixLen {
+		panic(fmt.Sprintf("invalid empty word %s %d %d", emptyWord, prefixLen, suffixLen))
+	}
+	wordMetaMap[wtype] = wordMeta{wtype, []rune(emptyWord), prefixLen, suffixLen, canExtend, quoteContext}
 }
 
 func init() {
 	wordMetaMap = make(map[string]wordMeta)
-	makeWordMeta(WordTypeRaw, "", 0, false, false)
-	makeWordMeta(WordTypeLit, "", 0, true, false)
-	makeWordMeta(WordTypeOp, "", 0, false, false)
-	makeWordMeta(WordTypeKey, "", 0, false, false)
-	makeWordMeta(WordTypeGroup, "", 0, false, false)
-	makeWordMeta(WordTypeSimpleVar, "$", 0, true, false)
-	makeWordMeta(WordTypeVarBrace, "${}", 1, true, true)
-	makeWordMeta(WordTypeDQ, `""`, 1, true, true)
-	makeWordMeta(WordTypeDDQ, `$""`, 1, true, true)
-	makeWordMeta(WordTypeDP, "$()", 1, false, false)
-	makeWordMeta(WordTypeBQ, "``", 1, false, false)
-	makeWordMeta(WordTypeSQ, "''", 1, true, false)
-	makeWordMeta(WordTypeDSQ, "$''", 1, true, false)
-	makeWordMeta(WordTypeDPP, "$(())", 2, false, false)
-	makeWordMeta(WordTypePP, "(())", 2, false, false)
-	makeWordMeta(WordTypeDB, "$[]", 1, false, false)
+	makeWordMeta(WordTypeRaw, "", 0, 0, false, false)
+	makeWordMeta(WordTypeLit, "", 0, 0, true, false)
+	makeWordMeta(WordTypeOp, "", 0, 0, false, false)
+	makeWordMeta(WordTypeKey, "", 0, 0, false, false)
+	makeWordMeta(WordTypeGroup, "", 0, 0, false, false)
+	makeWordMeta(WordTypeSimpleVar, "$", 1, 0, true, false)
+	makeWordMeta(WordTypeVarBrace, "${}", 2, 1, true, true)
+	makeWordMeta(WordTypeDQ, `""`, 1, 1, true, true)
+	makeWordMeta(WordTypeDDQ, `$""`, 2, 1, true, true)
+	makeWordMeta(WordTypeDP, "$()", 2, 1, false, false)
+	makeWordMeta(WordTypeBQ, "``", 1, 1, false, false)
+	makeWordMeta(WordTypeSQ, "''", 1, 1, true, false)
+	makeWordMeta(WordTypeDSQ, "$''", 2, 1, true, false)
+	makeWordMeta(WordTypeDPP, "$(())", 3, 2, false, false)
+	makeWordMeta(WordTypePP, "(())", 2, 2, false, false)
+	makeWordMeta(WordTypeDB, "$[]", 2, 1, false, false)
 }
 
 func MakeEmptyWord(wtype string, qc QuoteContext, offset int) *WordType {
@@ -169,48 +175,72 @@ func makeRepeatStr(ch byte, slen int) string {
 	return string(rtn)
 }
 
+func (w *WordType) isBlank() bool {
+	return w.Type == WordTypeLit && len(w.Raw) == 0
+}
+
+func (w *WordType) stringWithPos(pos int) string {
+	notCompleteFlag := " "
+	if !w.Complete {
+		notCompleteFlag = "*"
+	}
+	str := string(w.Raw)
+	if pos != -1 {
+		str = utilfn.StrWithPos{Str: str, Pos: pos}.String()
+	}
+	return fmt.Sprintf("%-4s[%3d]%s %s%q", w.Type, w.Offset, notCompleteFlag, makeRepeatStr('_', len(w.Prefix)), str)
+}
+
 func (w *WordType) String() string {
 	notCompleteFlag := " "
 	if !w.Complete {
 		notCompleteFlag = "*"
 	}
-	return fmt.Sprintf("%4s[%3d]%s %s%q", w.Type, w.Offset, notCompleteFlag, makeRepeatStr('_', len(w.Prefix)), string(w.FullRawString()))
+	return fmt.Sprintf("%-4s[%3d]%s %s%q", w.Type, w.Offset, notCompleteFlag, makeRepeatStr('_', len(w.Prefix)), string(w.Raw))
 }
 
-func dumpWords(words []*WordType, indentStr string) {
+// offset = -1 for don't show
+func dumpWords(words []*WordType, indentStr string, offset int) {
+	wrotePos := false
 	for _, word := range words {
-		fmt.Printf("%s%s\n", indentStr, word.String())
+		posInWord := false
+		if !wrotePos && offset != -1 && offset <= word.Offset {
+			fmt.Printf("%s*   [%3d] [*]\n", indentStr, offset)
+			wrotePos = true
+		}
+		if !wrotePos && offset != -1 && offset < word.Offset+len(word.Raw) {
+			fmt.Printf("%s%s\n", indentStr, word.stringWithPos(offset-word.Offset))
+			wrotePos = true
+			posInWord = true
+		} else {
+			fmt.Printf("%s%s\n", indentStr, word.String())
+		}
 		if len(word.Subs) > 0 {
-			dumpWords(word.Subs, indentStr+"  ")
+			if posInWord {
+				wmeta := wordMetaMap[word.Type]
+				dumpWords(word.Subs, indentStr+"  ", offset-word.Offset-wmeta.PrefixLen)
+			} else {
+				dumpWords(word.Subs, indentStr+"  ", -1)
+			}
 		}
 	}
 }
 
-func dumpCommands(cmds []*CmdType, indentStr string) {
+func dumpCommands(cmds []*CmdType, indentStr string, pos *CmdPos) {
 	for _, cmd := range cmds {
 		fmt.Printf("%sCMD: %s [%d]\n", indentStr, cmd.Type, len(cmd.Words))
-		dumpWords(cmd.Words, indentStr+"  ")
+		dumpWords(cmd.AssignmentWords, indentStr+" *", -1)
+		dumpWords(cmd.Words, indentStr+"  ", -1)
 	}
 }
 
-func (w *WordType) FullRawString() []rune {
-	if w.Type == WordTypeGroup {
-		var rtn []rune
-		for _, sw := range w.Subs {
-			rtn = append(rtn, sw.FullRawString()...)
-		}
-		return rtn
-	}
-	return w.Raw
-}
-
 func wordsToStr(words []*WordType) string {
 	var buf bytes.Buffer
 	for _, word := range words {
 		if len(word.Prefix) > 0 {
 			buf.WriteString(string(word.Prefix))
 		}
-		buf.WriteString(string(word.FullRawString()))
+		buf.WriteString(string(word.Raw))
 	}
 	return buf.String()
 }
@@ -442,6 +472,132 @@ func identifyReservedWords(words []*WordType) {
 	}
 }
 
+type CmdPos struct {
+	CmdPos    int
+	CmdOffset int
+
+	CurWord       *WordType // nil if between words
+	CurWordOffset int
+
+	CmdWordPos   int
+	OffsetInWord int // if BetweenWords is set, this offset can be negative (position is inside of prefix)
+	BetweenWords bool
+}
+
+// func FindCmdPos(cmds []*CmdType, offset int) CmdPos {
+// 	if len(words) == 0 {
+// 		return WordsPos{[]int{0}, 0, true}
+// 	}
+// 	pos := 0
+// 	for idx, word := range words {
+// 		if offset <= word.Offset+len(word.Raw) {
+// 			if offset <= word.Offset {
+// 				// in the prefix, so we are between-words with a possibly negative offset
+// 				return WordPos{WordPos: idx, OffsetInWord: offset - word.Offset, BetweenWords: true}
+// 			}
+// 			if offset == pos+fullWordLen {
+// 				return WordPos{WordPos: idx + 1, OffsetInWord: 0, BetweenWords: true}
+// 			}
+// 			return WordPos{WordPos: idx, OffsetInWord: offset - word.Offset, BetweenWords: false}
+// 		}
+// 		pos += fullWordLen
+// 	}
+// 	return WordPos{WordPos: []int{len(words)}, OffsetInWord: 0, BetweenWords: true}
+// }
+
+func ResetWordOffsets(words []*WordType) {
+	pos := 0
+	for _, word := range words {
+		pos += len(word.Prefix)
+		word.Offset = pos
+		if len(word.Subs) > 0 {
+			ResetWordOffsets(word.Subs)
+		}
+		pos += len(word.Raw)
+	}
+}
+
+func CommandsToWords(cmds []*CmdType) []*WordType {
+	var rtn []*WordType
+	for _, cmd := range cmds {
+		rtn = append(rtn, cmd.Words...)
+	}
+	return rtn
+}
+
+func (c *CmdType) stripPrefix() []rune {
+	if len(c.AssignmentWords) > 0 {
+		w := c.AssignmentWords[0]
+		prefix := w.Prefix
+		w.Prefix = nil
+		return prefix
+	}
+	if len(c.Words) > 0 {
+		w := c.Words[0]
+		prefix := w.Prefix
+		w.Prefix = nil
+		return prefix
+	}
+	return nil
+}
+
+func (c *CmdType) isEmpty() bool {
+	return len(c.AssignmentWords) == 0 && len(c.Words) == 0
+}
+
+func (c *CmdType) lastWord() *WordType {
+	if len(c.Words) > 0 {
+		return c.Words[len(c.Words)-1]
+	}
+	if len(c.AssignmentWords) > 0 {
+		return c.AssignmentWords[len(c.AssignmentWords)-1]
+	}
+	return nil
+}
+
+func (c *CmdType) endOffset() int {
+	lastWord := c.lastWord()
+	if lastWord == nil {
+		return 0
+	}
+	return lastWord.Offset + len(lastWord.Raw)
+}
+
+func indexInRunes(arr []rune, ch rune) int {
+	for idx, r := range arr {
+		if r == ch {
+			return idx
+		}
+	}
+	return -1
+}
+
+func isAssignmentWord(w *WordType) bool {
+	if w.Type == WordTypeLit || w.Type == WordTypeGroup {
+		eqIdx := indexInRunes(w.Raw, '=')
+		if eqIdx == -1 {
+			return false
+		}
+		prefix := w.Raw[0:eqIdx]
+		return isSimpleVarName(prefix)
+	}
+	return false
+}
+
+// simple commands steal whitespace from subsequent commands
+func cmdWhitespaceFixup(cmds []*CmdType) {
+	for idx := 0; idx < len(cmds)-1; idx++ {
+		cmd := cmds[idx]
+		if cmd.Type != CmdTypeSimple || cmd.isEmpty() {
+			continue
+		}
+		nextCmd := cmds[idx+1]
+		nextPrefix := nextCmd.stripPrefix()
+		blankWord := &WordType{Type: WordTypeLit, QC: cmd.lastWord().QC, Offset: cmd.endOffset(), Prefix: nextPrefix, Complete: true}
+		cmd.Words = append(cmd.Words, blankWord)
+	}
+}
+
 func ParseCommands(words []*WordType) []*CmdType {
 	identifyReservedWords(words)
 	state := parseCmdState{Input: words}
@@ -466,8 +622,13 @@ func ParseCommands(words []*WordType) []*CmdType {
 			state.Cur = &CmdType{Type: CmdTypeSimple}
 			state.Rtn = append(state.Rtn, state.Cur)
 		}
-		state.Cur.Words = append(state.Cur.Words, word)
+		if len(state.Cur.Words) == 0 && isAssignmentWord(word) {
+			state.Cur.AssignmentWords = append(state.Cur.AssignmentWords, word)
+		} else {
+			state.Cur.Words = append(state.Cur.Words, word)
+		}
 		state.InputPos++
 	}
+	cmdWhitespaceFixup(state.Rtn)
 	return state.Rtn
 }

pkg/shparse/shparse_test.go

@@ -15,14 +15,13 @@ import (
 func testParse(t *testing.T, s string) {
 	words := Tokenize(s)
 
-	fmt.Printf("%s\n", s)
-	dumpWords(words, "  ")
-	fmt.Printf("\n")
-
+	fmt.Printf("parse <<\n%s\n>>\n", s)
+	dumpWords(words, "  ", 8)
 	outStr := wordsToStr(words)
 	if outStr != s {
 		t.Errorf("tokenization output does not match input: %q => %q", s, outStr)
 	}
+	fmt.Printf("------\n\n")
 }
 
 func Test1(t *testing.T) {
@@ -46,6 +45,7 @@ func Test1(t *testing.T) {
 	testParse(t, "echo `ls $x \"hello $x\" \\`ls\\`; ./foo`")
 	testParse(t, `echo $"hello $x $(ls)"`)
 	testParse(t, "echo 'hello'\nls\n")
+	testParse(t, "echo 'hello'abc$'\a'")
 }
 
 func lastWord(words []*WordType) *WordType {
@@ -88,15 +88,17 @@ func testParseCommands(t *testing.T, str string) {
 	fmt.Printf("parse: %q\n", str)
 	words := Tokenize(str)
 	cmds := ParseCommands(words)
-	dumpCommands(cmds, "  ")
+	dumpCommands(cmds, "  ", nil)
 	fmt.Printf("\n")
 }
 
 func TestCmd(t *testing.T) {
 	testParseCommands(t, "ls foo")
+	testParseCommands(t, "function foo () { echo hello; }")
 	testParseCommands(t, "ls foo && ls bar; ./run $x hello | xargs foo; ")
 	testParseCommands(t, "if [[ 2 > 1 ]]; then echo hello\nelse echo world; echo next; done")
 	testParseCommands(t, "case lots of stuff; i don\\'t know how to parse; esac; ls foo")
-	testParseCommands(t, "(ls & ./x); for x in $vars 3; do { echo $x; ls foo; } done")
-	testParseCommands(t, "function foo () { echo hello; }")
+	testParseCommands(t, "(ls & ./x \n\n); for x in $vars 3; do { echo $x; ls foo; } done")
+	testParseCommands(t, `ls f"oo" "${x:"hello$y"}"`)
+	testParseCommands(t, `x="foo $y" z=10 ls`)
 }

pkg/shparse/tokenize.go

@@ -16,6 +16,13 @@ type tokenizeOutputState struct {
 	SavedPrefix []rune
 }
 
+func copyRunes(rarr []rune) []rune {
+	if len(rarr) == 0 {
+		return nil
+	}
+	return append([]rune(nil), rarr...)
+}
+
 // does not set CurWord
 func (state *tokenizeOutputState) appendStandaloneWord(word *WordType) {
 	state.delimitCurWord()
@@ -36,7 +43,9 @@ func (state *tokenizeOutputState) appendWord(word *WordType) {
 		return
 	}
 	state.ensureGroupWord()
+	word.Offset = word.Offset - state.CurWord.Offset
 	state.CurWord.Subs = append(state.CurWord.Subs, word)
+	state.CurWord.Raw = append(state.CurWord.Raw, word.Raw...)
 }
 
 func (state *tokenizeOutputState) ensureGroupWord() {
@@ -46,29 +55,34 @@ func (state *tokenizeOutputState) ensureGroupWord() {
 	if state.CurWord.Type == WordTypeGroup {
 		return
 	}
-	// moves the prefix from CurWord to the new group word
+	// moves the prefix from CurWord to the new group word, resets offsets
 	groupWord := &WordType{
 		Type:     WordTypeGroup,
 		Offset:   state.CurWord.Offset,
 		QC:       state.CurWord.QC,
+		Raw:      copyRunes(state.CurWord.Raw),
 		Complete: true,
 		Prefix:   state.CurWord.Prefix,
 	}
 	state.CurWord.Prefix = nil
+	state.CurWord.Offset = 0
 	groupWord.Subs = []*WordType{state.CurWord}
 	state.CurWord = groupWord
 }
 
-func ungroupWord(w *WordType) []*WordType {
-	if w.Type != WordTypeGroup {
-		return []*WordType{w}
+func ungroupWord(groupWord *WordType) []*WordType {
+	if groupWord.Type != WordTypeGroup {
+		return []*WordType{groupWord}
 	}
-	rtn := w.Subs
-	if len(w.Prefix) > 0 && len(rtn) > 0 {
-		newPrefix := append([]rune{}, w.Prefix...)
+	rtn := groupWord.Subs
+	if len(groupWord.Prefix) > 0 && len(rtn) > 0 {
+		newPrefix := append([]rune{}, groupWord.Prefix...)
 		newPrefix = append(newPrefix, rtn[0].Prefix...)
 		rtn[0].Prefix = newPrefix
 	}
+	for _, word := range rtn {
+		word.Offset = word.Offset + groupWord.Offset
+	}
 	return rtn
 }
 
@@ -89,6 +103,7 @@ func (state *tokenizeOutputState) ensureLitCurWord(pc *parseContext) {
 			panic("invalid state, there can be no saved prefix")
 		}
 		litWord := pc.makeWord(WordTypeLit, 0, true)
+		litWord.Offset = litWord.Offset - state.CurWord.Offset
 		state.CurWord.Subs = append(state.CurWord.Subs, litWord)
 	}
 }
@@ -115,6 +130,7 @@ func (state *tokenizeOutputState) appendLiteral(pc *parseContext, ch rune) {
 			panic(fmt.Sprintf("invalid curword type (group) %q", state.CurWord.Type))
 		}
 		lastWord.Raw = append(lastWord.Raw, ch)
+		state.CurWord.Raw = append(state.CurWord.Raw, ch)
 	} else {
 		panic(fmt.Sprintf("invalid curword type %q", state.CurWord.Type))
 	}
@@ -355,7 +371,7 @@ func (c *parseContext) makeWord(t string, length int, complete bool) *WordType {
 	rtn := &WordType{Type: t}
 	rtn.Offset = c.Pos
 	rtn.QC = c.QC
-	rtn.Raw = c.Input[c.Pos : c.Pos+length]
+	rtn.Raw = copyRunes(c.Input[c.Pos : c.Pos+length])
 	rtn.Complete = complete
 	c.Pos += length
 	return rtn
@@ -565,6 +581,19 @@ func (c *parseContext) parseSimpleVarName(offset int) int {
 	}
 }
 
+func isSimpleVarName(rstr []rune) bool {
+	if len(rstr) == 0 {
+		return false
+	}
+	for idx, ch := range rstr {
+		if (ch == '_' || (ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z')) || ((idx != 0) && ch >= '0' && ch <= '9') {
+			continue
+		}
+		return false
+	}
+	return true
+}
+
 func Tokenize(cmd string) []*WordType {
 	c := &parseContext{Input: []rune(cmd)}
 	rtn, _ := c.tokenizeRaw()

pkg/utilfn/utilfn.go

@@ -118,3 +118,26 @@ func ContainsStr(strs []string, test string) bool {
 	}
 	return false
 }
+
+type StrWithPos struct {
+	Str string
+	Pos int
+}
+
+func (sp StrWithPos) String() string {
+	return strWithCursor(sp.Str, sp.Pos)
+}
+
+func strWithCursor(str string, pos int) string {
+	if pos < 0 {
+		return "[*]_" + str
+	}
+	if pos >= len(str) {
+		if pos > len(str) {
+			return str + "_[*]"
+		}
+		return str + "[*]"
+	} else {
+		return str[:pos] + "[*]" + str[pos:]
+	}
+}