wikitext-parser/wikitext.go

package wikitext_parser

import (
	"strings"
)

func NormalizeWikiTitle(title string) string {
	return strings.Replace(title, " ", "_", -1)
}

type NewLineToken struct {
}

type GetInterfaceSliceStringValueOptions 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
}

func (o *GetInterfaceSliceStringValueOptions) Default() {
	o.Trim = true
	o.StringHandler = func(value string, opt *GetInterfaceSliceStringValueOptions) []string {
		return []string{value}
	}
	o.HTMLHandler = func(value *HTML, opt *GetInterfaceSliceStringValueOptions) []string {
		return []string{value.Tag.String()}
	}
	o.NewLineHandler = func(opt *GetInterfaceSliceStringValueOptions) []string {
		return []string{"\n"}
	}
	o.LinkHandler = func(value *Link, opt *GetInterfaceSliceStringValueOptions) (result []string) {
		if len(value.Name) > 0 {
			result = append(result, GetWikiStringValue(value.Name, opt)...)
		} else {
			result = append(result, value.URL)
		}
		result = append(result, GetWikiStringValue(value.Name, opt)...)
		return
	}
	o.TemplateLinkHandler = func(value *Template, opt *GetInterfaceSliceStringValueOptions) (result []string) {
		output := 0
		for _, vv := range value.Parameters {
			for _, vvv := range GetWikiStringValue(vv, opt) {
				vvv = strings.TrimSpace(vvv)
				if len(vvv) > 0 {
					output++
					result = append(result, vvv)
				}
			}
		}
		if output == 0 {
			result = append(result, value.Name)
		}

		return
	}
	o.TemplateHandler = func(value *Template, opt *GetInterfaceSliceStringValueOptions) (result []string) {
		switch strings.ToUpper(value.Name) {
		case "PAGENAME", "SUBPAGENAME":
			result = append(result, opt.PageName)
		default:
			result = append(result, value.Name)
		}

		return
	}
	o.UnorderedListHandler = func(value *UnorderedList, opt *GetInterfaceSliceStringValueOptions) []string {
		return GetWikiStringValue(value.Entries, opt)
	}
	o.DescriptionListHandler = func(value *DescriptionList, opt *GetInterfaceSliceStringValueOptions) []string {
		return []string{strings.Join(GetWikiStringValue(value.Name, opt), ", ") + ": " + strings.Join(GetWikiStringValue(value.Entries, opt), ", ")}
	}
}

func GetWikiStringValue(v []interface{}, opts *GetInterfaceSliceStringValueOptions) (r []string) {
	var result []string
	for _, value := range v {

		if text, ok := value.(string); ok {
			result = append(result, opts.StringHandler(text, opts)...)
		} else if template, ok := value.(*Template); ok {
			if template.IsLink {
				result = append(result, opts.TemplateLinkHandler(template, opts)...)
			} else {
				result = append(result, opts.TemplateHandler(template, opts)...)
			}
		} else if html, ok := value.(*HTML); ok && html.Tag != nil {
			result = append(result, opts.HTMLHandler(html, opts)...)
		} else if _, ok := value.(NewLineToken); ok {
			result = append(result, opts.NewLineHandler(opts)...)
		} else if link, ok := value.(*Link); ok {
			result = append(result, opts.LinkHandler(link, opts)...)
		} else if unorderedList, ok := value.(*UnorderedList); ok {
			result = append(result, opts.UnorderedListHandler(unorderedList, opts)...)
		} else if descriptionList, ok := value.(*DescriptionList); ok {
			result = append(result, opts.DescriptionListHandler(descriptionList, opts)...)
		}
	}

	r = make([]string, 0, len(result))

	for _, e := range result {
		if opts.Trim {
			e = strings.TrimSpace(e)
		}
		if len(e) > 0 {
			r = append(r, e)
		}
	}
	return
}

//ParseWikiText small WikiText parser that extracts text, Templates, and its arguments/parameters
func ParseWikiText(text string) (result []interface{}) {
	index := 0

	for index < len(text) {
		templateIndex := strings.Index(text[index:], "{{")
		linkIndex := strings.Index(text[index:], "[[")
		if templateIndex == -1 && linkIndex == -1 {
			t := strings.TrimSpace(text[index:])
			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
				}
			}

			t := strings.TrimSpace(text[index : index+bestIndex])
			if len(t) > 0 {
				result = append(result, text[index:index+bestIndex])
			}
			var tpl *Template
			index, tpl = ParseTemplate(text, index+bestIndex+2, 0, text[index+bestIndex])
			if tpl != nil {
				result = append(result, tpl)
			}
		}
	}

	return
}