wp-go/plugin/digest/digest.go

207 lines
4.3 KiB
Go
Raw Permalink Normal View History

2023-01-12 12:42:16 +00:00
package digest
2022-09-26 08:35:38 +00:00
import (
2023-01-21 11:31:23 +00:00
"github.com/fthvgb1/wp-go/helper/html"
"github.com/fthvgb1/wp-go/helper/slice"
str "github.com/fthvgb1/wp-go/helper/strings"
2022-09-26 08:35:38 +00:00
"regexp"
"strings"
"unicode/utf8"
)
2024-04-09 15:51:45 +00:00
var quto = regexp.MustCompile(`"*|&*|<*|>*| *|[*|]*| *`)
func SetQutos(reg string) {
quto = regexp.MustCompile(reg)
}
type SpecialSolveConf struct {
Num int
ChuckOvered bool
EscapeCharacter map[rune]SpecialSolve
Tags map[string]SpecialSolve
}
type SpecialSolve struct {
Num int
ChuckOvered bool
}
2023-04-18 13:05:37 +00:00
func StripTags(content, allowTag string) string {
2022-10-08 08:37:27 +00:00
content = strings.Trim(content, " \t\n\r\000\x0B")
content = strings.Replace(content, "]]>", "]]>", -1)
2023-04-18 13:05:37 +00:00
content = html.StripTagsX(content, allowTag)
return content
2022-10-08 08:37:27 +00:00
}
2024-04-09 15:51:45 +00:00
func Html(content string, limit int) (string, string) {
2023-04-18 13:05:37 +00:00
closeTag := ""
2022-09-26 08:35:38 +00:00
length := utf8.RuneCountInString(content) + 1
2023-04-18 13:05:37 +00:00
if length <= limit {
return content, ""
}
index := quto.FindAllStringIndex(content, -1)
2024-04-07 12:02:39 +00:00
var runeIndex [][]int
if len(index) > 0 {
runeIndex = slice.Map(index, func(t []int) []int {
return slice.Map(t, func(i int) int {
return utf8.RuneCountInString(content[:i])
})
})
}
2023-04-18 13:05:37 +00:00
end := 0
ru := []rune(content)
tagIn := false
total := len(ru)
l, r := '<', '>'
i := -1
for {
i++
2024-04-09 15:51:45 +00:00
if end >= limit || i >= total {
break
}
for len(runeIndex) > 0 && i == runeIndex[0][0] {
i = runeIndex[0][1]
runeIndex = runeIndex[1:]
end++
continue
2022-09-26 08:35:38 +00:00
}
2023-04-18 13:05:37 +00:00
if end >= limit || i >= total {
break
}
if ru[i] == l {
tagIn = true
continue
} else if ru[i] == r {
tagIn = false
continue
}
2024-04-10 14:31:08 +00:00
if tagIn == false {
2023-04-18 13:05:37 +00:00
end++
2022-09-26 08:35:38 +00:00
}
}
2023-04-18 13:05:37 +00:00
if i > total {
i = total
}
content = string(ru[:i])
closeTag = html.CloseTag(content)
return content, closeTag
2022-09-26 08:35:38 +00:00
}
2024-04-09 15:51:45 +00:00
func CustomizeHtml(content string, limit int, m map[string]SpecialSolveConf) (string, string) {
closeTag := ""
length := utf8.RuneCountInString(content) + 1
if length <= limit {
return content, ""
}
index := quto.FindAllStringIndex(content, -1)
var runeIndex [][]int
if len(index) > 0 {
runeIndex = slice.Map(index, func(t []int) []int {
return slice.Map(t, func(i int) int {
return utf8.RuneCountInString(content[:i])
})
})
}
count := 0
runeContent := []rune(content)
tagIn := false
runeTotal := len(runeContent)
l, r := '<', '>'
i := -1
2024-04-10 14:31:08 +00:00
selfCloseTags := html.GetSelfCloseTags()
2024-04-09 15:51:45 +00:00
var currentTag, parentTag string
var allTags = []string{"<top>"}
var tag []rune
var tagLocal = 0
for {
i++
if count >= limit || i >= runeTotal {
break
}
for len(runeIndex) > 0 && i == runeIndex[0][0] {
i = runeIndex[0][1]
runeIndex = runeIndex[1:]
count++
continue
}
if count >= limit || i >= runeTotal {
break
}
if runeContent[i] == l {
tagLocal = i
tagIn = true
continue
}
if tagIn && runeContent[i] == r {
tagIn = false
tags := str.Join("<", string(tag), ">")
if strings.Contains(tags, " ") {
tags = str.Join("<", strings.Split(string(tag), " ")[0], ">")
}
currentTag = tags
rawTag := strings.ReplaceAll(strings.Trim(tags, "<>"), "/", "")
_, ok := selfCloseTags[rawTag]
if !ok {
if '/' == tags[1] {
parentTag = allTags[len(allTags)-2]
allTags = allTags[:len(allTags)-1]
} else {
parentTag = allTags[len(allTags)-1]
allTags = append(allTags, currentTag)
}
} else {
parentTag = allTags[len(allTags)-1]
}
tag = tag[:0]
if len(m) > 0 {
nn, ok := m[parentTag]
if ok {
if n, ok := nn.Tags[tags]; ok {
if (count+n.Num) > limit && n.ChuckOvered {
i = tagLocal
break
}
count += n.Num
continue
}
}
if n, ok := m[tags]; ok {
if (count+n.Num) > limit && n.ChuckOvered {
i = tagLocal
break
}
count += n.Num
}
}
continue
}
if tagIn {
tag = append(tag, runeContent[i])
continue
}
currentTags := allTags[len(allTags)-1]
mm, ok := m[currentTags]
2024-04-10 15:48:23 +00:00
if !ok || len(mm.EscapeCharacter) < 1 {
2024-04-09 15:51:45 +00:00
count++
2024-04-10 15:48:23 +00:00
continue
}
if n, ok := mm.EscapeCharacter[runeContent[i]]; ok {
if (count+n.Num) > limit && n.ChuckOvered {
break
2024-04-09 15:51:45 +00:00
}
2024-04-10 15:48:23 +00:00
count += n.Num
} else {
count++
2024-04-09 15:51:45 +00:00
}
}
if i > runeTotal {
i = runeTotal
}
content = string(runeContent[:i])
closeTag = html.CloseTag(content)
return content, closeTag
}