Files
Devour/pkg/tsdocs/parser.go
T
Tomas Dvorak 898a3c303f update
2026-02-24 10:33:59 +01:00

294 lines
7.3 KiB
Go

package tsdocs
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://www.typescriptlang.org",
}
}
func (p *Parser) ParseModulePage(html string, docURL string) (*Module, error) {
doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
if err != nil {
return nil, err
}
module := &Module{
DocURL: docURL,
FetchedAt: time.Now(),
}
module.Name = p.extractModuleName(doc)
module.Doc = p.extractModuleDoc(doc)
module.Interfaces = p.extractInterfaces(doc, module.Name, docURL)
module.Types = p.extractTypeAliases(doc, module.Name, docURL)
module.Functions = p.extractFunctions(doc, module.Name, docURL)
module.Classes = p.extractClasses(doc, module.Name, docURL)
module.Enums = p.extractEnums(doc, module.Name, docURL)
module.Variables = p.extractVariables(doc, module.Name, docURL)
return module, 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, .ais-Hits-item, li.result").Each(func(i int, s *goquery.Selection) {
result := &SearchResult{}
link := s.Find("a").First()
result.Name = strings.TrimSpace(link.Text())
if href, exists := link.Attr("href"); exists {
result.DocURL = resolveURL(p.baseURL, href)
}
result.Doc = strings.TrimSpace(s.Find(".summary, p, .description").First().Text())
if s.HasClass("interface") || strings.Contains(s.Text(), "interface") {
result.Kind = "interface"
} else if s.HasClass("type") || strings.Contains(s.Text(), "type") {
result.Kind = "type"
} else if s.HasClass("function") || strings.Contains(s.Text(), "function") {
result.Kind = "function"
} else if s.HasClass("class") || strings.Contains(s.Text(), "class") {
result.Kind = "class"
} else {
result.Kind = "doc"
}
results = append(results, result)
})
return results, nil
}
func (p *Parser) extractModuleName(doc *goquery.Document) string {
title := doc.Find("h1, .title, .page-title").First().Text()
return strings.TrimSpace(title)
}
func (p *Parser) extractModuleDoc(doc *goquery.Document) string {
docblock := doc.Find(".markdown p:first-of-type, .content p:first-of-type, #main p").First()
return strings.TrimSpace(docblock.Text())
}
func (p *Parser) extractInterfaces(doc *goquery.Document, moduleName string, docURL string) []*Interface {
var interfaces []*Interface
doc.Find("h2, h3, .context-item, .api-item").Each(func(_ int, s *goquery.Selection) {
text := s.Text()
if !strings.Contains(strings.ToLower(text), "interface") {
return
}
iface := &Interface{}
nameEl := s.Find("code, .name, a").First()
if nameEl.Length() == 0 {
nameEl = s
}
iface.Name = strings.TrimSpace(nameEl.Text())
iface.Name = strings.TrimSuffix(iface.Name, "<")
iface.Name = strings.Split(iface.Name, "<")[0]
iface.Name = strings.TrimSpace(iface.Name)
if id, exists := s.Attr("id"); exists {
iface.DocURL = docURL + "#" + id
}
next := s.Next()
for next.Length() > 0 && !next.Is("h2, h3") {
if next.Is("p") && iface.Doc == "" {
iface.Doc = strings.TrimSpace(next.Text())
}
next = next.Next()
}
if iface.Name != "" && iface.Name != "interface" {
interfaces = append(interfaces, iface)
}
})
return interfaces
}
func (p *Parser) extractTypeAliases(doc *goquery.Document, moduleName string, docURL string) []*TypeAlias {
var types []*TypeAlias
doc.Find("h2, h3, .context-item, .api-item").Each(func(_ int, s *goquery.Selection) {
text := s.Text()
if !strings.Contains(strings.ToLower(text), "type") {
return
}
ta := &TypeAlias{}
nameEl := s.Find("code, .name").First()
if nameEl.Length() == 0 {
nameEl = s
}
ta.Name = strings.TrimSpace(nameEl.Text())
ta.Name = strings.TrimSuffix(ta.Name, "<")
ta.Name = strings.Split(ta.Name, "<")[0]
ta.Name = strings.TrimSpace(ta.Name)
if id, exists := s.Attr("id"); exists {
ta.DocURL = docURL + "#" + id
}
if ta.Name != "" && ta.Name != "type" {
types = append(types, ta)
}
})
return types
}
func (p *Parser) extractFunctions(doc *goquery.Document, moduleName string, docURL string) []*Function {
var functions []*Function
doc.Find("h2, h3, .context-item, .api-item, pre code").Each(func(_ int, s *goquery.Selection) {
text := s.Text()
if !strings.Contains(text, "function") && !strings.Contains(text, "(") {
return
}
fn := &Function{}
sigText := text
if idx := strings.Index(sigText, "("); idx > 0 {
prefix := sigText[:idx]
parts := strings.Fields(prefix)
if len(parts) > 0 {
fn.Name = parts[len(parts)-1]
}
fn.Signature = strings.TrimSpace(sigText)
}
if id, exists := s.Attr("id"); exists {
fn.DocURL = docURL + "#" + id
}
next := s.Next()
for next.Length() > 0 && !next.Is("h2, h3, pre") {
if next.Is("p") && fn.Doc == "" {
fn.Doc = strings.TrimSpace(next.Text())
}
next = next.Next()
}
if fn.Name != "" {
functions = append(functions, fn)
}
})
return functions
}
func (p *Parser) extractClasses(doc *goquery.Document, moduleName string, docURL string) []*Class {
var classes []*Class
doc.Find("h2, h3, .context-item, .api-item").Each(func(_ int, s *goquery.Selection) {
text := s.Text()
if !strings.Contains(strings.ToLower(text), "class") {
return
}
class := &Class{}
nameEl := s.Find("code, .name").First()
if nameEl.Length() == 0 {
nameEl = s
}
class.Name = strings.TrimSpace(nameEl.Text())
class.Name = strings.TrimSuffix(class.Name, "<")
class.Name = strings.Split(class.Name, "<")[0]
class.Name = strings.TrimSpace(class.Name)
if id, exists := s.Attr("id"); exists {
class.DocURL = docURL + "#" + id
}
if class.Name != "" && class.Name != "class" {
classes = append(classes, class)
}
})
return classes
}
func (p *Parser) extractEnums(doc *goquery.Document, moduleName string, docURL string) []*Enum {
var enums []*Enum
doc.Find("h2, h3, .context-item").Each(func(_ int, s *goquery.Selection) {
text := s.Text()
if !strings.Contains(strings.ToLower(text), "enum") {
return
}
enum := &Enum{}
nameEl := s.Find("code, .name").First()
if nameEl.Length() == 0 {
nameEl = s
}
enum.Name = strings.TrimSpace(nameEl.Text())
if id, exists := s.Attr("id"); exists {
enum.DocURL = docURL + "#" + id
}
if enum.Name != "" && enum.Name != "enum" {
enums = append(enums, enum)
}
})
return enums
}
func (p *Parser) extractVariables(doc *goquery.Document, moduleName string, docURL string) []*Variable {
var variables []*Variable
doc.Find("pre code").Each(func(_ int, s *goquery.Selection) {
text := s.Text()
if strings.Contains(text, "const ") || strings.Contains(text, "let ") || strings.Contains(text, "var ") {
v := &Variable{}
if idx := strings.Index(text, "="); idx > 0 {
decl := text[:idx]
decl = strings.TrimPrefix(decl, "const")
decl = strings.TrimPrefix(decl, "let")
decl = strings.TrimPrefix(decl, "var")
v.Name = strings.TrimSpace(decl)
}
if v.Name != "" {
variables = append(variables, v)
}
}
})
return variables
}
func resolveURL(base string, href string) string {
return parserutil.ResolveURL(base, href)
}