Devour/pkg/vuedocs/parser.go

package vuedocs

import (
	"strings"
	"time"

	"github.com/PuerkitoBio/goquery"
	"github.com/yourorg/devour/pkg/parserutil"
)

type Parser struct {
	baseURL string
}

func NewParser() *Parser {
	return &Parser{
		baseURL: "https://vuejs.org",
	}
}

func (p *Parser) ParseReferencePage(html string, docURL string) (*Reference, error) {
	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
	if err != nil {
		return nil, err
	}

	ref := &Reference{
		DocURL:    docURL,
		FetchedAt: time.Now(),
	}

	ref.GlobalAPI = p.extractGlobalAPI(doc, docURL)
	ref.Composition = p.extractCompositionAPI(doc, docURL)
	ref.Options = p.extractOptionsAPI(doc, docURL)
	ref.Directives = p.extractDirectives(doc, docURL)
	ref.Components = p.extractComponents(doc, docURL)
	ref.SpecialAttrs = p.extractSpecialAttrs(doc, docURL)

	return ref, nil
}

func (p *Parser) ParseSearchResults(html string) ([]*SearchResult, error) {
	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
	if err != nil {
		return nil, err
	}

	var results []*SearchResult

	doc.Find(".search-result, a[href*='/api/'], .nav-link, .api-link").Each(func(i int, s *goquery.Selection) {
		result := &SearchResult{}

		result.Name = strings.TrimSpace(s.Text())

		if href, exists := s.Attr("href"); exists {
			result.DocURL = resolveURL(p.baseURL, href)

			if strings.Contains(href, "/composition-") {
				result.Kind = "composition"
			} else if strings.Contains(href, "/options-") {
				result.Kind = "options"
			} else if strings.Contains(href, "/directive") {
				result.Kind = "directive"
			} else if strings.Contains(href, "/component") {
				result.Kind = "component"
			} else {
				result.Kind = "api"
			}
		}

		if result.Name != "" {
			results = append(results, result)
		}
	})

	return results, nil
}

func (p *Parser) extractGlobalAPI(doc *goquery.Document, docURL string) []*API {
	var apis []*API

	doc.Find("h2, h3, .api-item, [id]").Each(func(_ int, s *goquery.Selection) {
		id, _ := s.Attr("id")
		text := s.Text()

		globalAPIs := []string{"createApp", "createSSRApp", "defineComponent", "defineAsyncComponent", "resolveComponent", "resolveDirective", "withDirectives", "withKeys", "withModifiers"}
		isGlobal := false
		for _, name := range globalAPIs {
			if id == name || strings.Contains(text, name) {
				isGlobal = true
				break
			}
		}

		if !isGlobal {
			return
		}

		api := &API{}

		nameEl := s.Find("code, .name").First()
		if nameEl.Length() == 0 {
			nameEl = s
		}
		api.Name = strings.TrimSpace(nameEl.Text())
		api.Name = strings.TrimSuffix(api.Name, "(")

		api.DocURL = docURL + "#" + api.Name

		next := s.Next()
		for next.Length() > 0 && !next.Is("h2, h3, h4") {
			if next.Is("p") && api.Doc == "" {
				api.Doc = strings.TrimSpace(next.Text())
			}
			next = next.Next()
		}

		if api.Name != "" {
			apis = append(apis, api)
		}
	})

	return apis
}

func (p *Parser) extractCompositionAPI(doc *goquery.Document, docURL string) []*Composition {
	var compos []*Composition

	doc.Find("h2, h3, h4, .api-item, [id^='ref'], [id^='reactive'], [id^='computed'], [id^='watch'], [id^='on']").Each(func(_ int, s *goquery.Selection) {
		id, _ := s.Attr("id")
		text := s.Text()

		if id == "" && !strings.Contains(text, "(") {
			return
		}

		compNames := []string{"ref", "reactive", "computed", "watch", "watchEffect", "watchPostEffect", "watchSyncEffect", "onMounted", "onUpdated", "onUnmounted", "onBeforeMount", "onBeforeUpdate", "onBeforeUnmount", "onErrorCaptured", "onRenderTracked", "onRenderTriggered", "provide", "inject", "toRef", "toRefs", "toValue", "unref", "isRef", "shallowRef", "triggerRef", "customRef", "shallowReactive", "readonly", "shallowReadonly", "markRaw", "toRaw"}
		isComp := false
		for _, name := range compNames {
			if id == name || strings.HasPrefix(id, name+"-") || strings.Contains(text, name+"(") {
				isComp = true
				break
			}
		}

		if !isComp {
			return
		}

		comp := &Composition{}

		nameEl := s.Find("code, .name").First()
		if nameEl.Length() == 0 {
			nameEl = s
		}
		comp.Name = strings.TrimSpace(nameEl.Text())
		comp.Name = strings.TrimSuffix(comp.Name, "(")

		if comp.Name == "" && id != "" {
			comp.Name = id
		}

		comp.DocURL = docURL + "#" + comp.Name

		next := s.Next()
		for next.Length() > 0 && !next.Is("h2, h3, h4") {
			if next.Is("p") && comp.Doc == "" {
				comp.Doc = strings.TrimSpace(next.Text())
			}
			next = next.Next()
		}

		if comp.Name != "" {
			compos = append(compos, comp)
		}
	})

	return compos
}

func (p *Parser) extractOptionsAPI(doc *goquery.Document, docURL string) []*Options {
	var opts []*Options

	optionNames := []string{"data", "props", "computed", "methods", "watch", "emits", "expose", "setup", "name", "components", "directives", "inheritAttrs", "extends", "mixins", "provide", "inject", "template", "render", "renderTracked", "renderTriggered", "errorCaptured", "beforeCreate", "created", "beforeMount", "mounted", "beforeUpdate", "updated", "beforeUnmount", "unmounted", "activated", "deactivated", "serverPrefetch"}

	doc.Find("h2, h3, h4, .api-item").Each(func(_ int, s *goquery.Selection) {
		id, _ := s.Attr("id")
		text := s.Text()

		for _, name := range optionNames {
			if id == name || strings.Contains(strings.ToLower(text), name) {
				opt := &Options{
					Name: name,
				}

				opt.DocURL = docURL + "#" + name

				next := s.Next()
				for next.Length() > 0 && !next.Is("h2, h3, h4") {
					if next.Is("p") && opt.Doc == "" {
						opt.Doc = strings.TrimSpace(next.Text())
					}
					next = next.Next()
				}

				opts = append(opts, opt)
				break
			}
		}
	})

	return opts
}

func (p *Parser) extractDirectives(doc *goquery.Document, docURL string) []*Directive {
	var directives []*Directive

	directiveNames := []string{"v-bind", "v-on", "v-model", "v-if", "v-else", "v-else-if", "v-show", "v-for", "v-slot", "v-text", "v-html", "v-cloak", "v-once", "v-pre", "v-memo"}

	doc.Find("h2, h3, h4, .api-item").Each(func(_ int, s *goquery.Selection) {
		id, _ := s.Attr("id")
		text := s.Text()

		for _, name := range directiveNames {
			if id == name || strings.Contains(text, name) {
				dir := &Directive{
					Name: name,
				}

				dir.DocURL = docURL + "#" + name

				next := s.Next()
				for next.Length() > 0 && !next.Is("h2, h3, h4") {
					if next.Is("p") && dir.Doc == "" {
						dir.Doc = strings.TrimSpace(next.Text())
					}
					next = next.Next()
				}

				directives = append(directives, dir)
				break
			}
		}
	})

	return directives
}

func (p *Parser) extractComponents(doc *goquery.Document, docURL string) []*Component {
	var components []*Component

	componentNames := []string{"Transition", "TransitionGroup", "KeepAlive", "Teleport", "Suspense"}

	doc.Find("h2, h3, h4, .api-item").Each(func(_ int, s *goquery.Selection) {
		id, _ := s.Attr("id")
		text := s.Text()

		for _, name := range componentNames {
			if id == name || strings.Contains(text, name) {
				comp := &Component{
					Name: name,
				}

				comp.DocURL = docURL + "#" + name

				next := s.Next()
				for next.Length() > 0 && !next.Is("h2, h3, h4") {
					if next.Is("p") && comp.Doc == "" {
						comp.Doc = strings.TrimSpace(next.Text())
					}
					next = next.Next()
				}

				components = append(components, comp)
				break
			}
		}
	})

	return components
}

func (p *Parser) extractSpecialAttrs(doc *goquery.Document, docURL string) []*SpecialAttr {
	var attrs []*SpecialAttr

	attrNames := []string{"key", "ref", "is"}

	doc.Find("h2, h3, h4, .api-item").Each(func(_ int, s *goquery.Selection) {
		id, _ := s.Attr("id")
		text := s.Text()

		for _, name := range attrNames {
			if id == name+"-attribute" || id == name || strings.Contains(text, name+" attribute") {
				attr := &SpecialAttr{
					Name: name,
				}

				attr.DocURL = docURL + "#" + name

				next := s.Next()
				for next.Length() > 0 && !next.Is("h2, h3, h4") {
					if next.Is("p") && attr.Doc == "" {
						attr.Doc = strings.TrimSpace(next.Text())
					}
					next = next.Next()
				}

				attrs = append(attrs, attr)
				break
			}
		}
	})

	return attrs
}

func resolveURL(base string, href string) string {
	return parserutil.ResolveURL(base, href)
}