// Copyright 2011 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

package template

import (
	
	
)

// transitionFunc is the array of context transition functions for text nodes.
// A transition function takes a context and template text input, and returns
// the updated context and the number of bytes consumed from the front of the
// input.
var transitionFunc = [...]func(context, []byte) (context, int){
	stateText:           tText,
	stateTag:            tTag,
	stateAttrName:       tAttrName,
	stateAfterName:      tAfterName,
	stateBeforeValue:    tBeforeValue,
	stateHTMLCmt:        tHTMLCmt,
	stateRCDATA:         tSpecialTagEnd,
	stateAttr:           tAttr,
	stateURL:            tURL,
	stateSrcset:         tURL,
	stateJS:             tJS,
	stateJSDqStr:        tJSDelimited,
	stateJSSqStr:        tJSDelimited,
	stateJSRegexp:       tJSDelimited,
	stateJSTmplLit:      tJSTmpl,
	stateJSBlockCmt:     tBlockCmt,
	stateJSLineCmt:      tLineCmt,
	stateJSHTMLOpenCmt:  tLineCmt,
	stateJSHTMLCloseCmt: tLineCmt,
	stateCSS:            tCSS,
	stateCSSDqStr:       tCSSStr,
	stateCSSSqStr:       tCSSStr,
	stateCSSDqURL:       tCSSStr,
	stateCSSSqURL:       tCSSStr,
	stateCSSURL:         tCSSStr,
	stateCSSBlockCmt:    tBlockCmt,
	stateCSSLineCmt:     tLineCmt,
	stateError:          tError,
}

var commentStart = []byte("<!--")
var commentEnd = []byte("-->")

// tText is the context transition function for the text state.
func tText( context,  []byte) (context, int) {
	 := 0
	for {
		 :=  + bytes.IndexByte([:], '<')
		if  <  || +1 == len() {
			return , len()
		} else if +4 <= len() && bytes.Equal(commentStart, [:+4]) {
			return context{state: stateHTMLCmt},  + 4
		}
		++
		 := false
		if [] == '/' {
			if +1 == len() {
				return , len()
			}
			,  = true, +1
		}
		,  := eatTagName(, )
		if  !=  {
			if  {
				 = elementNone
			}
			// We've found an HTML tag.
			return context{state: stateTag, element: }, 
		}
		 = 
	}
}

var elementContentType = [...]state{
	elementNone:     stateText,
	elementScript:   stateJS,
	elementStyle:    stateCSS,
	elementTextarea: stateRCDATA,
	elementTitle:    stateRCDATA,
}

// tTag is the context transition function for the tag state.
func tTag( context,  []byte) (context, int) {
	// Find the attribute name.
	 := eatWhiteSpace(, 0)
	if  == len() {
		return , len()
	}
	if [] == '>' {
		return context{
			state:   elementContentType[.element],
			element: .element,
		},  + 1
	}
	,  := eatAttrName(, )
	if  != nil {
		return context{state: stateError, err: }, len()
	}
	,  := stateTag, attrNone
	if  ==  {
		return context{
			state: stateError,
			err:   errorf(ErrBadHTML, nil, 0, "expected space, attr name, or end of tag, but got %q", [:]),
		}, len()
	}

	 := strings.ToLower(string([:]))
	if .element == elementScript &&  == "type" {
		 = attrScriptType
	} else {
		switch attrType() {
		case contentTypeURL:
			 = attrURL
		case contentTypeCSS:
			 = attrStyle
		case contentTypeJS:
			 = attrScript
		case contentTypeSrcset:
			 = attrSrcset
		}
	}

	if  == len() {
		 = stateAttrName
	} else {
		 = stateAfterName
	}
	return context{state: , element: .element, attr: }, 
}

// tAttrName is the context transition function for stateAttrName.
func tAttrName( context,  []byte) (context, int) {
	,  := eatAttrName(, 0)
	if  != nil {
		return context{state: stateError, err: }, len()
	} else if  != len() {
		.state = stateAfterName
	}
	return , 
}

// tAfterName is the context transition function for stateAfterName.
func tAfterName( context,  []byte) (context, int) {
	// Look for the start of the value.
	 := eatWhiteSpace(, 0)
	if  == len() {
		return , len()
	} else if [] != '=' {
		// Occurs due to tag ending '>', and valueless attribute.
		.state = stateTag
		return , 
	}
	.state = stateBeforeValue
	// Consume the "=".
	return ,  + 1
}

var attrStartStates = [...]state{
	attrNone:       stateAttr,
	attrScript:     stateJS,
	attrScriptType: stateAttr,
	attrStyle:      stateCSS,
	attrURL:        stateURL,
	attrSrcset:     stateSrcset,
}

// tBeforeValue is the context transition function for stateBeforeValue.
func tBeforeValue( context,  []byte) (context, int) {
	 := eatWhiteSpace(, 0)
	if  == len() {
		return , len()
	}
	// Find the attribute delimiter.
	 := delimSpaceOrTagEnd
	switch [] {
	case '\'':
		,  = delimSingleQuote, +1
	case '"':
		,  = delimDoubleQuote, +1
	}
	.state, .delim = attrStartStates[.attr], 
	return , 
}

// tHTMLCmt is the context transition function for stateHTMLCmt.
func tHTMLCmt( context,  []byte) (context, int) {
	if  := bytes.Index(, commentEnd);  != -1 {
		return context{},  + 3
	}
	return , len()
}

// specialTagEndMarkers maps element types to the character sequence that
// case-insensitively signals the end of the special tag body.
var specialTagEndMarkers = [...][]byte{
	elementScript:   []byte("script"),
	elementStyle:    []byte("style"),
	elementTextarea: []byte("textarea"),
	elementTitle:    []byte("title"),
}

var (
	specialTagEndPrefix = []byte("</")
	tagEndSeparators    = []byte("> \t\n\f/")
)

// tSpecialTagEnd is the context transition function for raw text and RCDATA
// element states.
func tSpecialTagEnd( context,  []byte) (context, int) {
	if .element != elementNone {
		// script end tags ("</script") within script literals are ignored, so that
		// we can properly escape them.
		if .element == elementScript && (isInScriptLiteral(.state) || isComment(.state)) {
			return , len()
		}
		if  := indexTagEnd(, specialTagEndMarkers[.element]);  != -1 {
			return context{}, 
		}
	}
	return , len()
}

// indexTagEnd finds the index of a special tag end in a case insensitive way, or returns -1
func indexTagEnd( []byte,  []byte) int {
	 := 0
	 := len(specialTagEndPrefix)
	for len() > 0 {
		// Try to find the tag end prefix first
		 := bytes.Index(, specialTagEndPrefix)
		if  == -1 {
			return 
		}
		 = [+:]
		// Try to match the actual tag if there is still space for it
		if len() <= len() && bytes.EqualFold(, [:len()]) {
			 = [len():]
			// Check the tag is followed by a proper separator
			if len() > 0 && bytes.IndexByte(tagEndSeparators, [0]) != -1 {
				return  + 
			}
			 += len()
		}
		 +=  + 
	}
	return -1
}

// tAttr is the context transition function for the attribute state.
func tAttr( context,  []byte) (context, int) {
	return , len()
}

// tURL is the context transition function for the URL state.
func tURL( context,  []byte) (context, int) {
	if bytes.ContainsAny(, "#?") {
		.urlPart = urlPartQueryOrFrag
	} else if len() != eatWhiteSpace(, 0) && .urlPart == urlPartNone {
		// HTML5 uses "Valid URL potentially surrounded by spaces" for
		// attrs: https://www.w3.org/TR/html5/index.html#attributes-1
		.urlPart = urlPartPreQuery
	}
	return , len()
}

// tJS is the context transition function for the JS state.
func tJS( context,  []byte) (context, int) {
	 := bytes.IndexAny(, "\"`'/{}<-#")
	if  == -1 {
		// Entire input is non string, comment, regexp tokens.
		.jsCtx = nextJSCtx(, .jsCtx)
		return , len()
	}
	.jsCtx = nextJSCtx([:], .jsCtx)
	switch [] {
	case '"':
		.state, .jsCtx = stateJSDqStr, jsCtxRegexp
	case '\'':
		.state, .jsCtx = stateJSSqStr, jsCtxRegexp
	case '`':
		.state, .jsCtx = stateJSTmplLit, jsCtxRegexp
	case '/':
		switch {
		case +1 < len() && [+1] == '/':
			.state,  = stateJSLineCmt, +1
		case +1 < len() && [+1] == '*':
			.state,  = stateJSBlockCmt, +1
		case .jsCtx == jsCtxRegexp:
			.state = stateJSRegexp
		case .jsCtx == jsCtxDivOp:
			.jsCtx = jsCtxRegexp
		default:
			return context{
				state: stateError,
				err:   errorf(ErrSlashAmbig, nil, 0, "'/' could start a division or regexp: %.32q", [:]),
			}, len()
		}
	// ECMAScript supports HTML style comments for legacy reasons, see Appendix
	// B.1.1 "HTML-like Comments". The handling of these comments is somewhat
	// confusing. Multi-line comments are not supported, i.e. anything on lines
	// between the opening and closing tokens is not considered a comment, but
	// anything following the opening or closing token, on the same line, is
	// ignored. As such we simply treat any line prefixed with "<!--" or "-->"
	// as if it were actually prefixed with "//" and move on.
	case '<':
		if +3 < len() && bytes.Equal(commentStart, [:+4]) {
			.state,  = stateJSHTMLOpenCmt, +3
		}
	case '-':
		if +2 < len() && bytes.Equal(commentEnd, [:+3]) {
			.state,  = stateJSHTMLCloseCmt, +2
		}
	// ECMAScript also supports "hashbang" comment lines, see Section 12.5.
	case '#':
		if +1 < len() && [+1] == '!' {
			.state,  = stateJSLineCmt, +1
		}
	case '{':
		// We only care about tracking brace depth if we are inside of a
		// template literal.
		if len(.jsBraceDepth) == 0 {
			return ,  + 1
		}
		.jsBraceDepth[len(.jsBraceDepth)-1]++
	case '}':
		if len(.jsBraceDepth) == 0 {
			return ,  + 1
		}
		// There are no cases where a brace can be escaped in the JS context
		// that are not syntax errors, it seems. Because of this we can just
		// count "\}" as "}" and move on, the script is already broken as
		// fully fledged parsers will just fail anyway.
		.jsBraceDepth[len(.jsBraceDepth)-1]--
		if .jsBraceDepth[len(.jsBraceDepth)-1] >= 0 {
			return ,  + 1
		}
		.jsBraceDepth = .jsBraceDepth[:len(.jsBraceDepth)-1]
		.state = stateJSTmplLit
	default:
		panic("unreachable")
	}
	return ,  + 1
}

func tJSTmpl( context,  []byte) (context, int) {
	var  int
	for {
		 :=  + bytes.IndexAny([:], "`\\$")
		if  <  {
			break
		}
		switch [] {
		case '\\':
			++
			if  == len() {
				return context{
					state: stateError,
					err:   errorf(ErrPartialEscape, nil, 0, "unfinished escape sequence in JS string: %q", ),
				}, len()
			}
		case '$':
			if len() >= +2 && [+1] == '{' {
				.jsBraceDepth = append(.jsBraceDepth, 0)
				.state = stateJS
				return ,  + 2
			}
		case '`':
			// end
			.state = stateJS
			return ,  + 1
		}
		 =  + 1
	}

	return , len()
}

// tJSDelimited is the context transition function for the JS string and regexp
// states.
func tJSDelimited( context,  []byte) (context, int) {
	 := `\"`
	switch .state {
	case stateJSSqStr:
		 = `\'`
	case stateJSRegexp:
		 = `\/[]`
	}

	,  := 0, false
	for {
		 :=  + bytes.IndexAny([:], )
		if  <  {
			break
		}
		switch [] {
		case '\\':
			++
			if  == len() {
				return context{
					state: stateError,
					err:   errorf(ErrPartialEscape, nil, 0, "unfinished escape sequence in JS string: %q", ),
				}, len()
			}
		case '[':
			 = true
		case ']':
			 = false
		case '/':
			// If "</script" appears in a regex literal, the '/' should not
			// close the regex literal, and it will later be escaped to
			// "\x3C/script" in escapeText.
			if  > 0 && +7 <= len() && bytes.Equal(bytes.ToLower([-1:+7]), []byte("</script")) {
				++
			} else if ! {
				.state, .jsCtx = stateJS, jsCtxDivOp
				return ,  + 1
			}
		default:
			// end delimiter
			if ! {
				.state, .jsCtx = stateJS, jsCtxDivOp
				return ,  + 1
			}
		}
		 =  + 1
	}

	if  {
		// This can be fixed by making context richer if interpolation
		// into charsets is desired.
		return context{
			state: stateError,
			err:   errorf(ErrPartialCharset, nil, 0, "unfinished JS regexp charset: %q", ),
		}, len()
	}

	return , len()
}

var blockCommentEnd = []byte("*/")

// tBlockCmt is the context transition function for /*comment*/ states.
func tBlockCmt( context,  []byte) (context, int) {
	 := bytes.Index(, blockCommentEnd)
	if  == -1 {
		return , len()
	}
	switch .state {
	case stateJSBlockCmt:
		.state = stateJS
	case stateCSSBlockCmt:
		.state = stateCSS
	default:
		panic(.state.String())
	}
	return ,  + 2
}

// tLineCmt is the context transition function for //comment states, and the JS HTML-like comment state.
func tLineCmt( context,  []byte) (context, int) {
	var  string
	var  state
	switch .state {
	case stateJSLineCmt, stateJSHTMLOpenCmt, stateJSHTMLCloseCmt:
		,  = "\n\r\u2028\u2029", stateJS
	case stateCSSLineCmt:
		,  = "\n\f\r", stateCSS
		// Line comments are not part of any published CSS standard but
		// are supported by the 4 major browsers.
		// This defines line comments as
		//     LINECOMMENT ::= "//" [^\n\f\d]*
		// since https://www.w3.org/TR/css3-syntax/#SUBTOK-nl defines
		// newlines:
		//     nl ::= #xA | #xD #xA | #xD | #xC
	default:
		panic(.state.String())
	}

	 := bytes.IndexAny(, )
	if  == -1 {
		return , len()
	}
	.state = 
	// Per section 7.4 of EcmaScript 5 : https://es5.github.io/#x7.4
	// "However, the LineTerminator at the end of the line is not
	// considered to be part of the single-line comment; it is
	// recognized separately by the lexical grammar and becomes part
	// of the stream of input elements for the syntactic grammar."
	return , 
}

// tCSS is the context transition function for the CSS state.
func tCSS( context,  []byte) (context, int) {
	// CSS quoted strings are almost never used except for:
	// (1) URLs as in background: "/foo.png"
	// (2) Multiword font-names as in font-family: "Times New Roman"
	// (3) List separators in content values as in inline-lists:
	//    <style>
	//    ul.inlineList { list-style: none; padding:0 }
	//    ul.inlineList > li { display: inline }
	//    ul.inlineList > li:before { content: ", " }
	//    ul.inlineList > li:first-child:before { content: "" }
	//    </style>
	//    <ul class=inlineList><li>One<li>Two<li>Three</ul>
	// (4) Attribute value selectors as in a[href="http://example.com/"]
	//
	// We conservatively treat all strings as URLs, but make some
	// allowances to avoid confusion.
	//
	// In (1), our conservative assumption is justified.
	// In (2), valid font names do not contain ':', '?', or '#', so our
	// conservative assumption is fine since we will never transition past
	// urlPartPreQuery.
	// In (3), our protocol heuristic should not be tripped, and there
	// should not be non-space content after a '?' or '#', so as long as
	// we only %-encode RFC 3986 reserved characters we are ok.
	// In (4), we should URL escape for URL attributes, and for others we
	// have the attribute name available if our conservative assumption
	// proves problematic for real code.

	 := 0
	for {
		 :=  + bytes.IndexAny([:], `("'/`)
		if  <  {
			return , len()
		}
		switch [] {
		case '(':
			// Look for url to the left.
			 := bytes.TrimRight([:], "\t\n\f\r ")
			if endsWithCSSKeyword(, "url") {
				 := len() - len(bytes.TrimLeft([+1:], "\t\n\f\r "))
				switch {
				case  != len() && [] == '"':
					.state,  = stateCSSDqURL, +1
				case  != len() && [] == '\'':
					.state,  = stateCSSSqURL, +1
				default:
					.state = stateCSSURL
				}
				return , 
			}
		case '/':
			if +1 < len() {
				switch [+1] {
				case '/':
					.state = stateCSSLineCmt
					return ,  + 2
				case '*':
					.state = stateCSSBlockCmt
					return ,  + 2
				}
			}
		case '"':
			.state = stateCSSDqStr
			return ,  + 1
		case '\'':
			.state = stateCSSSqStr
			return ,  + 1
		}
		 =  + 1
	}
}

// tCSSStr is the context transition function for the CSS string and URL states.
func tCSSStr( context,  []byte) (context, int) {
	var  string
	switch .state {
	case stateCSSDqStr, stateCSSDqURL:
		 = `\"`
	case stateCSSSqStr, stateCSSSqURL:
		 = `\'`
	case stateCSSURL:
		// Unquoted URLs end with a newline or close parenthesis.
		// The below includes the wc (whitespace character) and nl.
		 = "\\\t\n\f\r )"
	default:
		panic(.state.String())
	}

	 := 0
	for {
		 :=  + bytes.IndexAny([:], )
		if  <  {
			,  := tURL(, decodeCSS([:]))
			return ,  + 
		}
		if [] == '\\' {
			++
			if  == len() {
				return context{
					state: stateError,
					err:   errorf(ErrPartialEscape, nil, 0, "unfinished escape sequence in CSS string: %q", ),
				}, len()
			}
		} else {
			.state = stateCSS
			return ,  + 1
		}
		, _ = tURL(, decodeCSS([:+1]))
		 =  + 1
	}
}

// tError is the context transition function for the error state.
func tError( context,  []byte) (context, int) {
	return , len()
}

// eatAttrName returns the largest j such that s[i:j] is an attribute name.
// It returns an error if s[i:] does not look like it begins with an
// attribute name, such as encountering a quote mark without a preceding
// equals sign.
func eatAttrName( []byte,  int) (int, *Error) {
	for  := ;  < len(); ++ {
		switch [] {
		case ' ', '\t', '\n', '\f', '\r', '=', '>':
			return , nil
		case '\'', '"', '<':
			// These result in a parse warning in HTML5 and are
			// indicative of serious problems if seen in an attr
			// name in a template.
			return -1, errorf(ErrBadHTML, nil, 0, "%q in attribute name: %.32q", [:+1], )
		default:
			// No-op.
		}
	}
	return len(), nil
}

var elementNameMap = map[string]element{
	"script":   elementScript,
	"style":    elementStyle,
	"textarea": elementTextarea,
	"title":    elementTitle,
}

// asciiAlpha reports whether c is an ASCII letter.
func asciiAlpha( byte) bool {
	return 'A' <=  &&  <= 'Z' || 'a' <=  &&  <= 'z'
}

// asciiAlphaNum reports whether c is an ASCII letter or digit.
func asciiAlphaNum( byte) bool {
	return asciiAlpha() || '0' <=  &&  <= '9'
}

// eatTagName returns the largest j such that s[i:j] is a tag name and the tag type.
func eatTagName( []byte,  int) (int, element) {
	if  == len() || !asciiAlpha([]) {
		return , elementNone
	}
	 :=  + 1
	for  < len() {
		 := []
		if asciiAlphaNum() {
			++
			continue
		}
		// Allow "x-y" or "x:y" but not "x-", "-y", or "x--y".
		if ( == ':' ||  == '-') && +1 < len() && asciiAlphaNum([+1]) {
			 += 2
			continue
		}
		break
	}
	return , elementNameMap[strings.ToLower(string([:]))]
}

// eatWhiteSpace returns the largest j such that s[i:j] is white space.
func eatWhiteSpace( []byte,  int) int {
	for  := ;  < len(); ++ {
		switch [] {
		case ' ', '\t', '\n', '\f', '\r':
			// No-op.
		default:
			return 
		}
	}
	return len()
}