Fix empty item handling in Split function, rewrite Words function
Test: treehugger Bug: 172923994 Change-Id: Ic7ee7b1af6e1438df5cf06754b9bec7038b624f2
This commit is contained in:
Sasha Smundak
parent
c08f602f31
commit
cbc17ee62d
2 changed files with 209 additions and 107 deletions
|
|
@ -15,8 +15,10 @@
|
|||
package parser
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
"strings"
|
||||
"unicode"
|
||||
"unicode/utf8"
|
||||
)
|
||||
|
||||
// A MakeString is a string that may contain variable substitutions in it.
|
||||
|
|
@ -130,8 +132,85 @@ func (ms *MakeString) SplitN(sep string, n int) []*MakeString {
|
|||
})
|
||||
}
|
||||
|
||||
// Words splits MakeString into multiple makeStrings separated by whitespace.
|
||||
// Thus, " a $(X)b c " will be split into ["a", "$(X)b", "c"].
|
||||
// Splitting a MakeString consisting solely of whitespace yields empty array.
|
||||
func (ms *MakeString) Words() []*MakeString {
|
||||
return ms.splitNFunc(-1, splitWords)
|
||||
var ch rune // current character
|
||||
const EOF = -1 // no more characters
|
||||
const EOS = -2 // at the end of a string chunk
|
||||
|
||||
// Next character's chunk and position
|
||||
iString := 0
|
||||
iChar := 0
|
||||
|
||||
var words []*MakeString
|
||||
word := SimpleMakeString("", ms.Pos())
|
||||
|
||||
nextChar := func() {
|
||||
if iString >= len(ms.Strings) {
|
||||
ch = EOF
|
||||
} else if iChar >= len(ms.Strings[iString]) {
|
||||
iString++
|
||||
iChar = 0
|
||||
ch = EOS
|
||||
} else {
|
||||
var w int
|
||||
ch, w = utf8.DecodeRuneInString(ms.Strings[iString][iChar:])
|
||||
iChar += w
|
||||
}
|
||||
}
|
||||
|
||||
appendVariableAndAdvance := func() {
|
||||
if iString-1 < len(ms.Variables) {
|
||||
word.appendVariable(ms.Variables[iString-1])
|
||||
}
|
||||
nextChar()
|
||||
}
|
||||
|
||||
appendCharAndAdvance := func(c rune) {
|
||||
if c != EOF {
|
||||
word.appendString(string(c))
|
||||
}
|
||||
nextChar()
|
||||
}
|
||||
|
||||
nextChar()
|
||||
for ch != EOF {
|
||||
// Skip whitespace
|
||||
for ch == ' ' || ch == '\t' {
|
||||
nextChar()
|
||||
}
|
||||
if ch == EOS {
|
||||
// "... $(X)... " case. The current word should be empty.
|
||||
if !word.Empty() {
|
||||
panic(fmt.Errorf("%q: EOS while current word %q is not empty, iString=%d",
|
||||
ms.Dump(), word.Dump(), iString))
|
||||
}
|
||||
appendVariableAndAdvance()
|
||||
}
|
||||
// Copy word
|
||||
for ch != EOF {
|
||||
if ch == ' ' || ch == '\t' {
|
||||
words = append(words, word)
|
||||
word = SimpleMakeString("", ms.Pos())
|
||||
break
|
||||
}
|
||||
if ch == EOS {
|
||||
// "...a$(X)..." case. Append variable to the current word
|
||||
appendVariableAndAdvance()
|
||||
} else {
|
||||
if ch == '\\' {
|
||||
appendCharAndAdvance('\\')
|
||||
}
|
||||
appendCharAndAdvance(ch)
|
||||
}
|
||||
}
|
||||
}
|
||||
if !word.Empty() {
|
||||
words = append(words, word)
|
||||
}
|
||||
return words
|
||||
}
|
||||
|
||||
func (ms *MakeString) splitNFunc(n int, splitFunc func(s string, n int) []string) []*MakeString {
|
||||
|
|
@ -166,9 +245,7 @@ func (ms *MakeString) splitNFunc(n int, splitFunc func(s string, n int) []string
|
|||
}
|
||||
}
|
||||
|
||||
if !curMs.Empty() {
|
||||
ret = append(ret, curMs)
|
||||
}
|
||||
ret = append(ret, curMs)
|
||||
return ret
|
||||
}
|
||||
|
||||
|
|
@ -219,44 +296,6 @@ func splitAnyN(s, sep string, n int) []string {
|
|||
return ret
|
||||
}
|
||||
|
||||
func splitWords(s string, n int) []string {
|
||||
ret := []string{}
|
||||
preserve := ""
|
||||
for n == -1 || n > 1 {
|
||||
index := strings.IndexAny(s, " \t")
|
||||
if index == 0 && len(preserve) == 0 {
|
||||
s = s[1:]
|
||||
} else if index >= 0 {
|
||||
escapeCount := 0
|
||||
for i := index - 1; i >= 0; i-- {
|
||||
if s[i] != '\\' {
|
||||
break
|
||||
}
|
||||
escapeCount += 1
|
||||
}
|
||||
|
||||
if escapeCount%2 == 1 {
|
||||
preserve += s[0 : index+1]
|
||||
s = s[index+1:]
|
||||
continue
|
||||
}
|
||||
|
||||
ret = append(ret, preserve+s[0:index])
|
||||
s = s[index+1:]
|
||||
preserve = ""
|
||||
if n > 0 {
|
||||
n--
|
||||
}
|
||||
} else {
|
||||
break
|
||||
}
|
||||
}
|
||||
if preserve != "" || s != "" || len(ret) == 0 {
|
||||
ret = append(ret, preserve+s)
|
||||
}
|
||||
return ret
|
||||
}
|
||||
|
||||
func unescape(s string) string {
|
||||
ret := ""
|
||||
for {
|
||||
|
|
|
|||
|
|
@ -26,64 +26,53 @@ var splitNTestCases = []struct {
|
|||
n int
|
||||
}{
|
||||
{
|
||||
in: &MakeString{
|
||||
Strings: []string{
|
||||
"a b c",
|
||||
"d e f",
|
||||
" h i j",
|
||||
},
|
||||
Variables: []Variable{
|
||||
Variable{Name: SimpleMakeString("var1", NoPos)},
|
||||
Variable{Name: SimpleMakeString("var2", NoPos)},
|
||||
},
|
||||
},
|
||||
// "a b c$(var1)d e f$(var2) h i j"
|
||||
in: genMakeString("a b c", "var1", "d e f", "var2", " h i j"),
|
||||
sep: " ",
|
||||
n: -1,
|
||||
expected: []*MakeString{
|
||||
SimpleMakeString("a", NoPos),
|
||||
SimpleMakeString("b", NoPos),
|
||||
&MakeString{
|
||||
Strings: []string{"c", "d"},
|
||||
Variables: []Variable{
|
||||
Variable{Name: SimpleMakeString("var1", NoPos)},
|
||||
},
|
||||
},
|
||||
SimpleMakeString("e", NoPos),
|
||||
&MakeString{
|
||||
Strings: []string{"f", ""},
|
||||
Variables: []Variable{
|
||||
Variable{Name: SimpleMakeString("var2", NoPos)},
|
||||
},
|
||||
},
|
||||
SimpleMakeString("h", NoPos),
|
||||
SimpleMakeString("i", NoPos),
|
||||
SimpleMakeString("j", NoPos),
|
||||
genMakeString("a"),
|
||||
genMakeString("b"),
|
||||
genMakeString("c", "var1", "d"),
|
||||
genMakeString("e"),
|
||||
genMakeString("f", "var2", ""),
|
||||
genMakeString("h"),
|
||||
genMakeString("i"),
|
||||
genMakeString("j"),
|
||||
},
|
||||
},
|
||||
{
|
||||
in: &MakeString{
|
||||
Strings: []string{
|
||||
"a b c",
|
||||
"d e f",
|
||||
" h i j",
|
||||
},
|
||||
Variables: []Variable{
|
||||
Variable{Name: SimpleMakeString("var1", NoPos)},
|
||||
Variable{Name: SimpleMakeString("var2", NoPos)},
|
||||
},
|
||||
},
|
||||
// "a b c$(var1)d e f$(var2) h i j"
|
||||
in: genMakeString("a b c", "var1", "d e f", "var2", " h i j"),
|
||||
sep: " ",
|
||||
n: 3,
|
||||
expected: []*MakeString{
|
||||
SimpleMakeString("a", NoPos),
|
||||
SimpleMakeString("b", NoPos),
|
||||
&MakeString{
|
||||
Strings: []string{"c", "d e f", " h i j"},
|
||||
Variables: []Variable{
|
||||
Variable{Name: SimpleMakeString("var1", NoPos)},
|
||||
Variable{Name: SimpleMakeString("var2", NoPos)},
|
||||
},
|
||||
},
|
||||
genMakeString("a"),
|
||||
genMakeString("b"),
|
||||
genMakeString("c", "var1", "d e f", "var2", " h i j"),
|
||||
},
|
||||
},
|
||||
{
|
||||
// "$(var1) $(var2)"
|
||||
in: genMakeString("", "var1", " ", "var2", ""),
|
||||
sep: " ",
|
||||
n: -1,
|
||||
expected: []*MakeString{
|
||||
genMakeString("", "var1", ""),
|
||||
genMakeString("", "var2", ""),
|
||||
},
|
||||
},
|
||||
{
|
||||
// "a,,b,c,"
|
||||
in: genMakeString("a,,b,c,"),
|
||||
sep: ",",
|
||||
n: -1,
|
||||
expected: []*MakeString{
|
||||
genMakeString("a"),
|
||||
genMakeString(""),
|
||||
genMakeString("b"),
|
||||
genMakeString("c"),
|
||||
genMakeString(""),
|
||||
},
|
||||
},
|
||||
}
|
||||
|
|
@ -104,15 +93,15 @@ var valueTestCases = []struct {
|
|||
expected string
|
||||
}{
|
||||
{
|
||||
in: SimpleMakeString("a b", NoPos),
|
||||
in: genMakeString("a b"),
|
||||
expected: "a b",
|
||||
},
|
||||
{
|
||||
in: SimpleMakeString("a\\ \\\tb\\\\", NoPos),
|
||||
in: genMakeString("a\\ \\\tb\\\\"),
|
||||
expected: "a \tb\\",
|
||||
},
|
||||
{
|
||||
in: SimpleMakeString("a\\b\\", NoPos),
|
||||
in: genMakeString("a\\b\\"),
|
||||
expected: "a\\b\\",
|
||||
},
|
||||
}
|
||||
|
|
@ -131,31 +120,88 @@ var splitWordsTestCases = []struct {
|
|||
expected []*MakeString
|
||||
}{
|
||||
{
|
||||
in: SimpleMakeString("", NoPos),
|
||||
in: genMakeString(""),
|
||||
expected: []*MakeString{},
|
||||
},
|
||||
{
|
||||
in: SimpleMakeString(" a b\\ c d", NoPos),
|
||||
in: genMakeString(` a b\ c d`),
|
||||
expected: []*MakeString{
|
||||
SimpleMakeString("a", NoPos),
|
||||
SimpleMakeString("b\\ c", NoPos),
|
||||
SimpleMakeString("d", NoPos),
|
||||
genMakeString("a"),
|
||||
genMakeString(`b\ c`),
|
||||
genMakeString("d"),
|
||||
},
|
||||
},
|
||||
{
|
||||
in: SimpleMakeString(" a\tb\\\t\\ c d ", NoPos),
|
||||
in: SimpleMakeString(" a\tb"+`\`+"\t"+`\ c d `, NoPos),
|
||||
expected: []*MakeString{
|
||||
SimpleMakeString("a", NoPos),
|
||||
SimpleMakeString("b\\\t\\ c", NoPos),
|
||||
SimpleMakeString("d", NoPos),
|
||||
genMakeString("a"),
|
||||
genMakeString("b" + `\` + "\t" + `\ c`),
|
||||
genMakeString("d"),
|
||||
},
|
||||
},
|
||||
{
|
||||
in: SimpleMakeString(`a\\ b\\\ c d`, NoPos),
|
||||
in: genMakeString(`a\\ b\\\ c d`),
|
||||
expected: []*MakeString{
|
||||
SimpleMakeString(`a\\`, NoPos),
|
||||
SimpleMakeString(`b\\\ c`, NoPos),
|
||||
SimpleMakeString("d", NoPos),
|
||||
genMakeString(`a\\`),
|
||||
genMakeString(`b\\\ c`),
|
||||
genMakeString("d"),
|
||||
},
|
||||
},
|
||||
{
|
||||
in: genMakeString(`\\ a`),
|
||||
expected: []*MakeString{
|
||||
genMakeString(`\\`),
|
||||
genMakeString("a"),
|
||||
},
|
||||
},
|
||||
{
|
||||
// " "
|
||||
in: &MakeString{
|
||||
Strings: []string{" \t \t"},
|
||||
Variables: nil,
|
||||
},
|
||||
expected: []*MakeString{},
|
||||
},
|
||||
{
|
||||
// " a $(X)b c "
|
||||
in: genMakeString(" a ", "X", "b c "),
|
||||
expected: []*MakeString{
|
||||
genMakeString("a"),
|
||||
genMakeString("", "X", "b"),
|
||||
genMakeString("c"),
|
||||
},
|
||||
},
|
||||
{
|
||||
// " a b$(X)c d"
|
||||
in: genMakeString(" a b", "X", "c d"),
|
||||
expected: []*MakeString{
|
||||
genMakeString("a"),
|
||||
genMakeString("b", "X", "c"),
|
||||
genMakeString("d"),
|
||||
},
|
||||
},
|
||||
{
|
||||
// "$(X) $(Y)"
|
||||
in: genMakeString("", "X", " ", "Y", ""),
|
||||
expected: []*MakeString{
|
||||
genMakeString("", "X", ""),
|
||||
genMakeString("", "Y", ""),
|
||||
},
|
||||
},
|
||||
{
|
||||
// " a$(X) b"
|
||||
in: genMakeString(" a", "X", " b"),
|
||||
expected: []*MakeString{
|
||||
genMakeString("a", "X", ""),
|
||||
genMakeString("b"),
|
||||
},
|
||||
},
|
||||
{
|
||||
// "a$(X) b$(Y) "
|
||||
in: genMakeString("a", "X", " b", "Y", " "),
|
||||
expected: []*MakeString{
|
||||
genMakeString("a", "X", ""),
|
||||
genMakeString("b", "Y", ""),
|
||||
},
|
||||
},
|
||||
}
|
||||
|
|
@ -180,3 +226,20 @@ func dumpArray(a []*MakeString) string {
|
|||
|
||||
return strings.Join(ret, "|||")
|
||||
}
|
||||
|
||||
// generates MakeString from alternating string chunks and variable names,
|
||||
// e.g., genMakeString("a", "X", "b") returns MakeString for "a$(X)b"
|
||||
func genMakeString(items ...string) *MakeString {
|
||||
n := len(items) / 2
|
||||
if len(items) != (2*n + 1) {
|
||||
panic("genMakeString expects odd number of arguments")
|
||||
}
|
||||
|
||||
ms := &MakeString{Strings: make([]string, n+1), Variables: make([]Variable, n)}
|
||||
ms.Strings[0] = items[0]
|
||||
for i := 1; i <= n; i++ {
|
||||
ms.Variables[i-1] = Variable{Name: SimpleMakeString(items[2*i-1], NoPos)}
|
||||
ms.Strings[i] = items[2*i]
|
||||
}
|
||||
return ms
|
||||
}
|
||||
|
|
|
|||
Loading…
Reference in a new issue