S.O.N.G/wikitext-parser
S.O.N.G/wikitext-parser
3
0
Fork 0
Code Issues Pull requests Releases Activity

Full parsing of templates and lists on main body

Browse source
This commit is contained in:
DataHoarder 2022-02-20 22:10:24 +01:00
parent a0cd37f626
commit 5c99a0ad20
2 changed files with 100 additions and 42 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
template.go
View file

@ -157,7 +157,7 @@ func ParseTemplate(text string, index int, depth int, startCharacter byte) (i in
}
lastToken = scanIndex
i = scanIndex - 1
} else if c == ';' {
} else if afterNewLine && c == ';' {
addValue()
var list *DescriptionList
var scanIndex int

140
wikitext.go
View file

@ -5,37 +5,37 @@ import (
)
func NormalizeWikiTitle(title string) string {
return strings.Replace(title, " ", "_", -1)
return strings.ReplaceAll(title, " ", "_")
}
type NewLineToken struct {
}
type GetInterfaceSliceStringValueOptions struct {
type WikiStringValueOptions struct {
PageName string
Trim bool
StringHandler func(value string, opt *GetInterfaceSliceStringValueOptions) []string
HTMLHandler func(value *HTML, opt *GetInterfaceSliceStringValueOptions) []string
LinkHandler func(value *Link, opt *GetInterfaceSliceStringValueOptions) []string
TemplateLinkHandler func(value *Template, opt *GetInterfaceSliceStringValueOptions) []string
TemplateHandler func(value *Template, opt *GetInterfaceSliceStringValueOptions) []string
UnorderedListHandler func(value *UnorderedList, opt *GetInterfaceSliceStringValueOptions) []string
DescriptionListHandler func(value *DescriptionList, opt *GetInterfaceSliceStringValueOptions) []string
NewLineHandler func(opt *GetInterfaceSliceStringValueOptions) []string
StringHandler func(value string, opt *WikiStringValueOptions) []string
HTMLHandler func(value *HTML, opt *WikiStringValueOptions) []string
LinkHandler func(value *Link, opt *WikiStringValueOptions) []string
TemplateLinkHandler func(value *Template, opt *WikiStringValueOptions) []string
TemplateHandler func(value *Template, opt *WikiStringValueOptions) []string
UnorderedListHandler func(value *UnorderedList, opt *WikiStringValueOptions) []string
DescriptionListHandler func(value *DescriptionList, opt *WikiStringValueOptions) []string
NewLineHandler func(opt *WikiStringValueOptions) []string
}
func (o *GetInterfaceSliceStringValueOptions) Default() {
func (o *WikiStringValueOptions) Default() {
o.Trim = true
o.StringHandler = func(value string, opt *GetInterfaceSliceStringValueOptions) []string {
o.StringHandler = func(value string, opt *WikiStringValueOptions) []string {
return []string{value}
}
o.HTMLHandler = func(value *HTML, opt *GetInterfaceSliceStringValueOptions) []string {
o.HTMLHandler = func(value *HTML, opt *WikiStringValueOptions) []string {
return []string{value.Tag.String()}
}
o.NewLineHandler = func(opt *GetInterfaceSliceStringValueOptions) []string {
o.NewLineHandler = func(opt *WikiStringValueOptions) []string {
return []string{"\n"}
}
o.LinkHandler = func(value *Link, opt *GetInterfaceSliceStringValueOptions) (result []string) {
o.LinkHandler = func(value *Link, opt *WikiStringValueOptions) (result []string) {
if len(value.Name) > 0 {
result = append(result, GetWikiStringValue(value.Name, opt)...)
} else {
@ -44,7 +44,7 @@ func (o *GetInterfaceSliceStringValueOptions) Default() {
result = append(result, GetWikiStringValue(value.Name, opt)...)
return
}
o.TemplateLinkHandler = func(value *Template, opt *GetInterfaceSliceStringValueOptions) (result []string) {
o.TemplateLinkHandler = func(value *Template, opt *WikiStringValueOptions) (result []string) {
output := 0
for _, vv := range value.Parameters {
for _, vvv := range GetWikiStringValue(vv, opt) {
@ -61,7 +61,7 @@ func (o *GetInterfaceSliceStringValueOptions) Default() {
return
}
o.TemplateHandler = func(value *Template, opt *GetInterfaceSliceStringValueOptions) (result []string) {
o.TemplateHandler = func(value *Template, opt *WikiStringValueOptions) (result []string) {
switch strings.ToUpper(value.Name) {
case "PAGENAME", "SUBPAGENAME":
result = append(result, opt.PageName)
@ -71,15 +71,15 @@ func (o *GetInterfaceSliceStringValueOptions) Default() {
return
}
o.UnorderedListHandler = func(value *UnorderedList, opt *GetInterfaceSliceStringValueOptions) []string {
o.UnorderedListHandler = func(value *UnorderedList, opt *WikiStringValueOptions) []string {
return GetWikiStringValue(value.Entries, opt)
}
o.DescriptionListHandler = func(value *DescriptionList, opt *GetInterfaceSliceStringValueOptions) []string {
o.DescriptionListHandler = func(value *DescriptionList, opt *WikiStringValueOptions) []string {
return []string{strings.Join(GetWikiStringValue(value.Name, opt), ", ") + ": " + strings.Join(GetWikiStringValue(value.Entries, opt), ", ")}
}
}
func GetWikiStringValue(v []interface{}, opts *GetInterfaceSliceStringValueOptions) (r []string) {
func GetWikiStringValue(v []interface{}, opts *WikiStringValueOptions) (r []string) {
var result []string
for _, value := range v {
@ -120,35 +120,93 @@ func GetWikiStringValue(v []interface{}, opts *GetInterfaceSliceStringValueOptio
//ParseWikiText small WikiText parser that extracts text, Templates, and its arguments/parameters
func ParseWikiText(text string) (result []interface{}) {
index := 0
var i int
for index < len(text) {
templateIndex := strings.Index(text[index:], "{{")
linkIndex := strings.Index(text[index:], "[[")
if templateIndex == -1 && linkIndex == -1 {
t := strings.TrimSpace(text[index:])
var c byte
lastToken := index
addValue := func() int {
if lastToken < len(text) && i-lastToken > 0 {
t := strings.TrimSpace(text[lastToken:i])
if len(t) > 0 {
result = append(result, text[index:])
}
break
} else {
bestIndex := templateIndex
if templateIndex == -1 {
bestIndex = linkIndex
} else {
if linkIndex != -1 && linkIndex < bestIndex {
bestIndex = linkIndex
}
result = append(result, t)
}
t := strings.TrimSpace(text[index : index+bestIndex])
if len(t) > 0 {
result = append(result, text[index:index+bestIndex])
}
return len(t)
}
return 0
}
afterNewLine := false
for i = index; i < len(text); i++ {
c = text[i]
if (c == '{' && i < len(text)-1 && text[i+1] == '{') || (c == '[' && i < len(text)-1 && text[i+1] == '[') {
addValue()
var tpl *Template
index, tpl = ParseTemplate(text, index+bestIndex+2, 0, text[index+bestIndex])
var scanIndex int
scanIndex, tpl = ParseTemplate(text, i+2, 1, c)
if tpl != nil {
result = append(result, tpl)
}
lastToken = scanIndex
i = scanIndex - 1
} else if (c == '{' && i < len(text)-1 && text[i+1] != '{' && text[i+1] != '[') || (c == '[' && i < len(text)-1 && text[i+1] != '[' && text[i+1] != '{') {
addValue()
var link *Link
var scanIndex int
scanIndex, link = ParseLink(text, i+1, 1, c)
if link != nil {
result = append(result, link)
}
lastToken = scanIndex
i = scanIndex - 1
} else if c == '<' { //html trigger
addValue()
var html *HTML
var scanIndex int
scanIndex, html = ParseHTML(text, i, 1)
if html != nil {
result = append(result, html)
}
lastToken = scanIndex
i = scanIndex - 1
} else if c == '\n' {
addValue()
lastToken = i + 1
afterNewLine = true
result = append(result, NewLineToken{})
} else if afterNewLine && (c == '*' || c == '#') {
addValue()
var list *UnorderedList
var scanIndex int
scanIndex, list = ParseUnorderedList(text, i, 1, 1, c)
if list != nil {
result = append(result, list)
}
lastToken = scanIndex
i = scanIndex - 1
} else if afterNewLine && c == ';' {
addValue()
var list *DescriptionList
var scanIndex int
scanIndex, list = ParseDescriptionList(text, i+1, 1)
if list != nil {
result = append(result, list)
}
lastToken = scanIndex
i = scanIndex - 1
} else if afterNewLine && c == ':' {
addValue()
lastToken = i + 1
}
if afterNewLine && c != '\n' && c != ' ' && c != '\t' {
afterNewLine = false
}
}