Allow raw string literals in shortcode params

This commit is contained in:
Joshua Wong 2020-01-14 13:36:33 +08:00 committed by Bjørn Erik Pedersen
parent 0c0bb37285
commit da81455656
2 changed files with 62 additions and 7 deletions

View file

@ -85,6 +85,8 @@ func lexShortcodeRightDelim(l *pageLexer) stateFunc {
// 2. "param" or "param\" // 2. "param" or "param\"
// 3. param="123" or param="123\" // 3. param="123" or param="123\"
// 4. param="Some \"escaped\" text" // 4. param="Some \"escaped\" text"
// 5. `param`
// 6. param=`123`
func lexShortcodeParam(l *pageLexer, escapedQuoteStart bool) stateFunc { func lexShortcodeParam(l *pageLexer, escapedQuoteStart bool) stateFunc {
first := true first := true
@ -95,15 +97,21 @@ func lexShortcodeParam(l *pageLexer, escapedQuoteStart bool) stateFunc {
for { for {
r = l.next() r = l.next()
if first { if first {
if r == '"' { if r == '"' || (r == '`' && !escapedQuoteStart) {
// a positional param with quotes // a positional param with quotes
if l.paramElements == 2 { if l.paramElements == 2 {
return l.errorf("got quoted positional parameter. Cannot mix named and positional parameters") return l.errorf("got quoted positional parameter. Cannot mix named and positional parameters")
} }
l.paramElements = 1 l.paramElements = 1
l.backup() l.backup()
if r == '"' {
return lexShortcodeQuotedParamVal(l, !escapedQuoteStart, tScParam) return lexShortcodeQuotedParamVal(l, !escapedQuoteStart, tScParam)
} }
return lexShortCodeParamRawStringVal(l, tScParam)
} else if r == '`' && escapedQuoteStart {
return l.errorf("unrecognized escape character")
}
first = false first = false
} else if r == '=' { } else if r == '=' {
// a named param // a named param
@ -143,6 +151,32 @@ func lexShortcodeParamVal(l *pageLexer) stateFunc {
return lexInsideShortcode return lexInsideShortcode
} }
func lexShortCodeParamRawStringVal(l *pageLexer, typ ItemType) stateFunc {
openBacktickFound := false
Loop:
for {
switch r := l.next(); {
case r == '`':
if openBacktickFound {
l.backup()
break Loop
} else {
openBacktickFound = true
l.ignore()
}
case r == eof, r == '\n':
return l.errorf("unterminated raw string in shortcode parameter-argument: '%s'", l.current())
}
}
l.emitString(typ)
l.next()
l.ignore()
return lexInsideShortcode
}
func lexShortcodeQuotedParamVal(l *pageLexer, escapedQuotedValuesAllowed bool, typ ItemType) stateFunc { func lexShortcodeQuotedParamVal(l *pageLexer, escapedQuotedValuesAllowed bool, typ ItemType) stateFunc {
openQuoteFound := false openQuoteFound := false
escapedInnerQuoteFound := false escapedInnerQuoteFound := false
@ -161,6 +195,8 @@ Loop:
escapedInnerQuoteFound = true escapedInnerQuoteFound = true
escapedQuoteState = 1 escapedQuoteState = 1
} }
} else if l.peek() == '`' {
return l.errorf("unrecognized escape character")
} }
case r == eof, r == '\n': case r == eof, r == '\n':
return l.errorf("unterminated quoted string in shortcode parameter-argument: '%s'", l.current()) return l.errorf("unterminated quoted string in shortcode parameter-argument: '%s'", l.current())
@ -177,7 +213,6 @@ Loop:
} else { } else {
escapedQuoteState = 0 escapedQuoteState = 0
} }
} }
} }
@ -284,6 +319,8 @@ func lexInsideShortcode(l *pageLexer) stateFunc {
peek := l.peek() peek := l.peek()
if peek == '"' || peek == '\\' { if peek == '"' || peek == '\\' {
return lexShortcodeQuotedParamVal(l, peek != '\\', tScParamVal) return lexShortcodeQuotedParamVal(l, peek != '\\', tScParamVal)
} else if peek == '`' {
return lexShortCodeParamRawStringVal(l, tScParamVal)
} }
return lexShortcodeParamVal return lexShortcodeParamVal
case r == '/': case r == '/':
@ -295,10 +332,10 @@ func lexInsideShortcode(l *pageLexer) stateFunc {
l.emit(tScClose) l.emit(tScClose)
case r == '\\': case r == '\\':
l.ignore() l.ignore()
if l.peek() == '"' { if l.peek() == '"' || l.peek() == '`' {
return lexShortcodeParam(l, true) return lexShortcodeParam(l, true)
} }
case l.elementStepNum > 0 && (isAlphaNumericOrHyphen(r) || r == '"'): // positional params can have quotes case l.elementStepNum > 0 && (isAlphaNumericOrHyphen(r) || r == '"' || r == '`'): // positional params can have quotes
l.backup() l.backup()
return lexShortcodeParam(l, false) return lexShortcodeParam(l, false)
case isAlphaNumeric(r): case isAlphaNumeric(r):

View file

@ -13,7 +13,9 @@
package pageparser package pageparser
import "testing" import (
"testing"
)
var ( var (
tstEOF = nti(tEOF, "") tstEOF = nti(tEOF, "")
@ -77,13 +79,16 @@ var shortCodeLexerTests = []lexerTest{
tstLeftNoMD, tstSC1, nti(tScParam, "3.14"), tstRightNoMD, tstEOF}}, tstLeftNoMD, tstSC1, nti(tScParam, "3.14"), tstRightNoMD, tstEOF}},
{"float param, named", `{{< sc1 param1=3.14 >}}`, []Item{ {"float param, named", `{{< sc1 param1=3.14 >}}`, []Item{
tstLeftNoMD, tstSC1, tstParam1, nti(tScParamVal, "3.14"), tstRightNoMD, tstEOF}}, tstLeftNoMD, tstSC1, tstParam1, nti(tScParamVal, "3.14"), tstRightNoMD, tstEOF}},
{"named param, raw string", `{{< sc1 param1=` + "`" + "Hello World" + "`" + " >}}", []Item{
tstLeftNoMD, tstSC1, tstParam1, nti(tScParamVal, "Hello World"), tstRightNoMD, tstEOF}},
{"float param, named, space before", `{{< sc1 param1= 3.14 >}}`, []Item{ {"float param, named, space before", `{{< sc1 param1= 3.14 >}}`, []Item{
tstLeftNoMD, tstSC1, tstParam1, nti(tScParamVal, "3.14"), tstRightNoMD, tstEOF}}, tstLeftNoMD, tstSC1, tstParam1, nti(tScParamVal, "3.14"), tstRightNoMD, tstEOF}},
{"Youtube id", `{{< sc1 -ziL-Q_456igdO-4 >}}`, []Item{ {"Youtube id", `{{< sc1 -ziL-Q_456igdO-4 >}}`, []Item{
tstLeftNoMD, tstSC1, nti(tScParam, "-ziL-Q_456igdO-4"), tstRightNoMD, tstEOF}}, tstLeftNoMD, tstSC1, nti(tScParam, "-ziL-Q_456igdO-4"), tstRightNoMD, tstEOF}},
{"non-alphanumerics param quoted", `{{< sc1 "-ziL-.%QigdO-4" >}}`, []Item{ {"non-alphanumerics param quoted", `{{< sc1 "-ziL-.%QigdO-4" >}}`, []Item{
tstLeftNoMD, tstSC1, nti(tScParam, "-ziL-.%QigdO-4"), tstRightNoMD, tstEOF}}, tstLeftNoMD, tstSC1, nti(tScParam, "-ziL-.%QigdO-4"), tstRightNoMD, tstEOF}},
{"raw string", `{{< sc1` + "`" + "Hello World" + "`" + ` >}}`, []Item{
tstLeftNoMD, tstSC1, nti(tScParam, "Hello World"), tstRightNoMD, tstEOF}},
{"two params", `{{< sc1 param1 param2 >}}`, []Item{ {"two params", `{{< sc1 param1 param2 >}}`, []Item{
tstLeftNoMD, tstSC1, tstParam1, tstParam2, tstRightNoMD, tstEOF}}, tstLeftNoMD, tstSC1, tstParam1, tstParam2, tstRightNoMD, tstEOF}},
// issue #934 // issue #934
@ -137,11 +142,24 @@ var shortCodeLexerTests = []lexerTest{
{"escaped quotes inside nonescaped quotes in positional param", {"escaped quotes inside nonescaped quotes in positional param",
`{{< sc1 "Hello \"escaped\" World" >}}`, []Item{ `{{< sc1 "Hello \"escaped\" World" >}}`, []Item{
tstLeftNoMD, tstSC1, nti(tScParam, `Hello "escaped" World`), tstRightNoMD, tstEOF}}, tstLeftNoMD, tstSC1, nti(tScParam, `Hello "escaped" World`), tstRightNoMD, tstEOF}},
{"escaped raw string, named param", `{{< sc1 param1=` + `\` + "`" + "Hello World" + `\` + "`" + ` >}}`, []Item{
tstLeftNoMD, tstSC1, tstParam1, nti(tError, "unrecognized escape character")}},
{"escaped raw string, positional param", `{{< sc1 param1 ` + `\` + "`" + "Hello World" + `\` + "`" + ` >}}`, []Item{
tstLeftNoMD, tstSC1, tstParam1, nti(tError, "unrecognized escape character")}},
{"two raw string params", `{{< sc1` + "`" + "Hello World" + "`" + "`" + "Second Param" + "`" + ` >}}`, []Item{
tstLeftNoMD, tstSC1, nti(tScParam, "Hello World"), nti(tScParam, "Second Param"), tstRightNoMD, tstEOF}},
{"unterminated quote", `{{< sc1 param2="Hello World>}}`, []Item{ {"unterminated quote", `{{< sc1 param2="Hello World>}}`, []Item{
tstLeftNoMD, tstSC1, tstParam2, nti(tError, "unterminated quoted string in shortcode parameter-argument: 'Hello World>}}'")}}, tstLeftNoMD, tstSC1, tstParam2, nti(tError, "unterminated quoted string in shortcode parameter-argument: 'Hello World>}}'")}},
{"unterminated raw string", `{{< sc1` + "`" + "Hello World" + ` >}}`, []Item{
tstLeftNoMD, tstSC1, nti(tError, "unterminated raw string in shortcode parameter-argument: 'Hello World >}}'")}},
{"unterminated raw string in second argument", `{{< sc1` + "`" + "Hello World" + "`" + "`" + "Second Param" + ` >}}`, []Item{
tstLeftNoMD, tstSC1, nti(tScParam, "Hello World"), nti(tError, "unterminated raw string in shortcode parameter-argument: 'Second Param >}}'")}},
{"one named param, one not", `{{< sc1 param1="Hello World" p2 >}}`, []Item{ {"one named param, one not", `{{< sc1 param1="Hello World" p2 >}}`, []Item{
tstLeftNoMD, tstSC1, tstParam1, tstVal, tstLeftNoMD, tstSC1, tstParam1, tstVal,
nti(tError, "got positional parameter 'p2'. Cannot mix named and positional parameters")}}, nti(tError, "got positional parameter 'p2'. Cannot mix named and positional parameters")}},
{"one named param, one quoted positional param, both raw strings", `{{< sc1 param1=` + "`" + "Hello World" + "`" + "`" + "Second Param" + "`" + ` >}}`, []Item{
tstLeftNoMD, tstSC1, tstParam1, tstVal,
nti(tError, "got quoted positional parameter. Cannot mix named and positional parameters")}},
{"one named param, one quoted positional param", `{{< sc1 param1="Hello World" "And Universe" >}}`, []Item{ {"one named param, one quoted positional param", `{{< sc1 param1="Hello World" "And Universe" >}}`, []Item{
tstLeftNoMD, tstSC1, tstParam1, tstVal, tstLeftNoMD, tstSC1, tstParam1, tstVal,
nti(tError, "got quoted positional parameter. Cannot mix named and positional parameters")}}, nti(tError, "got quoted positional parameter. Cannot mix named and positional parameters")}},