blackfriday/smartypants.go

398 lines
9.1 KiB
Go
Raw Normal View History

2011-05-29 11:17:53 +08:00
//
// Black Friday Markdown Processor
// Originally based on http://github.com/tanoku/upskirt
// by Russ Ross <russ@russross.com>
//
//
//
// SmartyPants rendering
//
//
package blackfriday
import (
"bytes"
)
type smartypantsData struct {
inSingleQuote bool
inDoubleQuote bool
2011-05-29 11:17:53 +08:00
}
func wordBoundary(c byte) bool {
2011-05-29 11:17:53 +08:00
return c == 0 || isspace(c) || ispunct(c)
}
func tolower(c byte) byte {
if c >= 'A' && c <= 'Z' {
return c - 'A' + 'a'
}
return c
}
func isdigit(c byte) bool {
return c >= '0' && c <= '9'
}
func smartQuotesHelper(ob *bytes.Buffer, previousChar byte, nextChar byte, quote byte, isOpen *bool) bool {
2011-05-29 11:17:53 +08:00
// edge of the buffer is likely to be a tag that we don't get to see,
// so we treat it like text sometimes
// enumerate all sixteen possibilities for (previousChar, nextChar)
// each can be one of {0, space, punct, other}
switch {
case previousChar == 0 && nextChar == 0:
// context is not any help here, so toggle
*isOpen = !*isOpen
case isspace(previousChar) && nextChar == 0:
// [ "] might be [ "<code>foo...]
*isOpen = true
case ispunct(previousChar) && nextChar == 0:
// [!"] hmm... could be [Run!"] or [("<code>...]
*isOpen = false
case /* isnormal(previousChar) && */ nextChar == 0:
// [a"] is probably a close
*isOpen = false
case previousChar == 0 && isspace(nextChar):
// [" ] might be [...foo</code>" ]
*isOpen = false
case isspace(previousChar) && isspace(nextChar):
// [ " ] context is not any help here, so toggle
*isOpen = !*isOpen
case ispunct(previousChar) && isspace(nextChar):
// [!" ] is probably a close
*isOpen = false
case /* isnormal(previousChar) && */ isspace(nextChar):
// [a" ] this is one of the easy cases
*isOpen = false
case previousChar == 0 && ispunct(nextChar):
// ["!] hmm... could be ["$1.95] or [</code>"!...]
*isOpen = false
case isspace(previousChar) && ispunct(nextChar):
// [ "!] looks more like [ "$1.95]
*isOpen = true
case ispunct(previousChar) && ispunct(nextChar):
// [!"!] context is not any help here, so toggle
*isOpen = !*isOpen
case /* isnormal(previousChar) && */ ispunct(nextChar):
// [a"!] is probably a close
*isOpen = false
case previousChar == 0 /* && isnormal(nextChar) */ :
// ["a] is probably an open
*isOpen = true
case isspace(previousChar) /* && isnormal(nextChar) */ :
// [ "a] this is one of the easy cases
*isOpen = true
case ispunct(previousChar) /* && isnormal(nextChar) */ :
// [!"a] is probably an open
*isOpen = true
2011-05-29 11:17:53 +08:00
default:
// [a'b] maybe a contraction?
*isOpen = false
2011-05-29 11:17:53 +08:00
}
ob.WriteByte('&')
if *isOpen {
2011-05-29 11:17:53 +08:00
ob.WriteByte('l')
} else {
ob.WriteByte('r')
}
ob.WriteByte(quote)
ob.WriteString("quo;")
return true
}
func smartSquote(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
2011-05-29 11:17:53 +08:00
if len(text) >= 2 {
t1 := tolower(text[1])
if t1 == '\'' {
nextChar := byte(0)
2011-05-29 11:17:53 +08:00
if len(text) >= 3 {
nextChar = text[2]
2011-05-29 11:17:53 +08:00
}
if smartQuotesHelper(ob, previousChar, nextChar, 'd', &smrt.inDoubleQuote) {
2011-05-29 11:17:53 +08:00
return 1
}
}
if (t1 == 's' || t1 == 't' || t1 == 'm' || t1 == 'd') && (len(text) < 3 || wordBoundary(text[2])) {
2011-05-29 11:17:53 +08:00
ob.WriteString("&rsquo;")
return 0
}
if len(text) >= 3 {
t2 := tolower(text[2])
if ((t1 == 'r' && t2 == 'e') || (t1 == 'l' && t2 == 'l') || (t1 == 'v' && t2 == 'e')) && (len(text) < 4 || wordBoundary(text[3])) {
2011-05-29 11:17:53 +08:00
ob.WriteString("&rsquo;")
return 0
}
}
}
nextChar := byte(0)
2011-05-29 11:17:53 +08:00
if len(text) > 1 {
nextChar = text[1]
2011-05-29 11:17:53 +08:00
}
if smartQuotesHelper(ob, previousChar, nextChar, 's', &smrt.inSingleQuote) {
2011-05-29 11:17:53 +08:00
return 0
}
ob.WriteByte(text[0])
return 0
}
func smartParens(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
2011-05-29 11:17:53 +08:00
if len(text) >= 3 {
t1 := tolower(text[1])
t2 := tolower(text[2])
if t1 == 'c' && t2 == ')' {
ob.WriteString("&copy;")
return 2
}
if t1 == 'r' && t2 == ')' {
ob.WriteString("&reg;")
return 2
}
if len(text) >= 4 && t1 == 't' && t2 == 'm' && text[3] == ')' {
ob.WriteString("&trade;")
return 3
}
}
ob.WriteByte(text[0])
return 0
}
func smartDash(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
2011-05-29 11:17:53 +08:00
if len(text) >= 2 {
if text[1] == '-' {
ob.WriteString("&mdash;")
return 1
}
if wordBoundary(previousChar) && wordBoundary(text[1]) {
2011-05-29 11:17:53 +08:00
ob.WriteString("&ndash;")
return 0
}
}
ob.WriteByte(text[0])
return 0
}
func smartDashLatex(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
2011-05-29 11:17:53 +08:00
if len(text) >= 3 && text[1] == '-' && text[2] == '-' {
ob.WriteString("&mdash;")
return 2
}
if len(text) >= 2 && text[1] == '-' {
ob.WriteString("&ndash;")
return 1
}
ob.WriteByte(text[0])
return 0
}
func smartAmp(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
2011-05-29 11:17:53 +08:00
if bytes.HasPrefix(text, []byte("&quot;")) {
nextChar := byte(0)
2011-05-29 11:17:53 +08:00
if len(text) >= 7 {
nextChar = text[6]
2011-05-29 11:17:53 +08:00
}
if smartQuotesHelper(ob, previousChar, nextChar, 'd', &smrt.inDoubleQuote) {
2011-05-29 11:17:53 +08:00
return 5
}
}
if bytes.HasPrefix(text, []byte("&#0;")) {
return 3
}
ob.WriteByte('&')
return 0
}
func smartPeriod(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
2011-05-29 11:17:53 +08:00
if len(text) >= 3 && text[1] == '.' && text[2] == '.' {
ob.WriteString("&hellip;")
return 2
}
if len(text) >= 5 && text[1] == ' ' && text[2] == '.' && text[3] == ' ' && text[4] == '.' {
ob.WriteString("&hellip;")
return 4
}
ob.WriteByte(text[0])
return 0
}
func smartBacktick(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
2011-05-29 11:17:53 +08:00
if len(text) >= 2 && text[1] == '`' {
nextChar := byte(0)
2011-05-29 11:17:53 +08:00
if len(text) >= 3 {
nextChar = text[2]
2011-05-29 11:17:53 +08:00
}
if smartQuotesHelper(ob, previousChar, nextChar, 'd', &smrt.inDoubleQuote) {
2011-05-29 11:17:53 +08:00
return 1
}
}
return 0
}
func smartNumberGeneric(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
if wordBoundary(previousChar) && len(text) >= 3 {
2011-05-29 11:17:53 +08:00
// is it of the form digits/digits(word boundary)?, i.e., \d+/\d+\b
num_end := 0
for len(text) > num_end && isdigit(text[num_end]) {
num_end++
}
if num_end == 0 {
ob.WriteByte(text[0])
2011-05-29 11:17:53 +08:00
return 0
}
if len(text) < num_end+2 || text[num_end] != '/' {
ob.WriteByte(text[0])
2011-05-29 11:17:53 +08:00
return 0
}
den_end := num_end + 1
for len(text) > den_end && isdigit(text[den_end]) {
den_end++
}
if den_end == num_end+1 {
ob.WriteByte(text[0])
2011-05-29 11:17:53 +08:00
return 0
}
if len(text) == den_end || wordBoundary(text[den_end]) {
2011-05-29 11:33:16 +08:00
ob.WriteString("<sup>")
2011-05-29 11:17:53 +08:00
ob.Write(text[:num_end])
2011-05-29 11:33:16 +08:00
ob.WriteString("</sup>&frasl;<sub>")
2011-05-29 11:17:53 +08:00
ob.Write(text[num_end+1 : den_end])
2011-05-29 11:33:16 +08:00
ob.WriteString("</sub>")
2011-05-29 11:17:53 +08:00
return den_end - 1
}
}
ob.WriteByte(text[0])
return 0
}
func smartNumber(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
if wordBoundary(previousChar) && len(text) >= 3 {
2011-05-29 11:17:53 +08:00
if text[0] == '1' && text[1] == '/' && text[2] == '2' {
if len(text) < 4 || wordBoundary(text[3]) {
2011-05-29 11:17:53 +08:00
ob.WriteString("&frac12;")
return 2
}
}
if text[0] == '1' && text[1] == '/' && text[2] == '4' {
if len(text) < 4 || wordBoundary(text[3]) || (len(text) >= 5 && tolower(text[3]) == 't' && tolower(text[4]) == 'h') {
2011-05-29 11:17:53 +08:00
ob.WriteString("&frac14;")
return 2
}
}
if text[0] == '3' && text[1] == '/' && text[2] == '4' {
if len(text) < 4 || wordBoundary(text[3]) || (len(text) >= 6 && tolower(text[3]) == 't' && tolower(text[4]) == 'h' && tolower(text[5]) == 's') {
2011-05-29 11:17:53 +08:00
ob.WriteString("&frac34;")
return 2
}
}
}
ob.WriteByte(text[0])
return 0
}
func smartDquote(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
nextChar := byte(0)
2011-05-29 11:17:53 +08:00
if len(text) > 1 {
nextChar = text[1]
2011-05-29 11:17:53 +08:00
}
if !smartQuotesHelper(ob, previousChar, nextChar, 'd', &smrt.inDoubleQuote) {
2011-05-29 11:17:53 +08:00
ob.WriteString("&quot;")
}
return 0
}
func smartLtag(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int {
2011-05-29 11:17:53 +08:00
i := 0
for i < len(text) && text[i] != '>' {
i++
}
ob.Write(text[:i+1])
return i
}
type smartCallback func(ob *bytes.Buffer, smrt *smartypantsData, previousChar byte, text []byte) int
2011-05-29 11:17:53 +08:00
type SmartypantsRenderer [256]smartCallback
2011-05-29 11:17:53 +08:00
func Smartypants(flags int) *SmartypantsRenderer {
r := new(SmartypantsRenderer)
r['"'] = smartDquote
r['&'] = smartAmp
r['\''] = smartSquote
r['('] = smartParens
2011-05-29 11:17:53 +08:00
if flags&HTML_SMARTYPANTS_LATEX_DASHES == 0 {
r['-'] = smartDash
2011-05-29 11:17:53 +08:00
} else {
r['-'] = smartDashLatex
2011-05-29 11:17:53 +08:00
}
r['.'] = smartPeriod
2011-05-29 11:17:53 +08:00
if flags&HTML_SMARTYPANTS_FRACTIONS == 0 {
r['1'] = smartNumber
r['3'] = smartNumber
2011-05-29 11:17:53 +08:00
} else {
for ch := '1'; ch <= '9'; ch++ {
r[ch] = smartNumberGeneric
2011-05-29 11:17:53 +08:00
}
}
r['<'] = smartLtag
r['`'] = smartBacktick
2011-05-29 11:17:53 +08:00
return r
}
func htmlSmartypants(ob *bytes.Buffer, text []byte, opaque interface{}) {
2011-05-29 11:17:53 +08:00
options := opaque.(*htmlOptions)
smrt := smartypantsData{false, false}
2011-05-29 11:17:53 +08:00
// first do normal entity escaping
var escaped bytes.Buffer
attrEscape(&escaped, text)
text = escaped.Bytes()
2011-05-29 11:17:53 +08:00
mark := 0
for i := 0; i < len(text); i++ {
if action := options.smartypants[text[i]]; action != nil {
if i > mark {
ob.Write(text[mark:i])
2011-05-29 11:17:53 +08:00
}
previousChar := byte(0)
2011-05-29 11:17:53 +08:00
if i > 0 {
previousChar = text[i-1]
2011-05-29 11:17:53 +08:00
}
i += action(ob, &smrt, previousChar, text[i:])
2011-05-29 11:17:53 +08:00
mark = i + 1
}
}
if mark < len(text) {
ob.Write(text[mark:])
2011-05-29 11:17:53 +08:00
}
}