This commit is contained in:
Tomas Dvorak
2026-02-22 15:41:27 +01:00
parent 0b88627e54
commit 409acd2e08
84 changed files with 65382 additions and 27475 deletions
+254
View File
@@ -0,0 +1,254 @@
package scraper
import (
"context"
"crypto/sha256"
"encoding/hex"
"fmt"
"io"
"net/http"
"strings"
"time"
"github.com/yourorg/devour/pkg/javadocs"
)
type JavaDocsScraper struct {
config *Config
parser *javadocs.Parser
client *http.Client
}
func NewJavaDocsScraper(config *Config) *JavaDocsScraper {
return &JavaDocsScraper{
config: config,
parser: javadocs.NewParser(),
client: &http.Client{
Timeout: config.Timeout,
},
}
}
func (s *JavaDocsScraper) Scrape(ctx context.Context, source *Source) ([]*Document, error) {
var documents []*Document
if source.URL == "" {
return nil, fmt.Errorf("URL is required for Java docs scraper")
}
html, err := s.fetchPage(ctx, source.URL)
if err != nil {
return nil, fmt.Errorf("failed to fetch page: %w", err)
}
pkg, err := s.parser.ParsePackagePage(html, source.URL)
if err != nil {
return nil, fmt.Errorf("failed to parse package: %w", err)
}
mainDoc := s.packageToDocument(pkg, source.Name)
documents = append(documents, mainDoc)
for _, class := range pkg.Classes {
doc := s.classToDocument(class, pkg, source.Name)
documents = append(documents, doc)
}
for _, iface := range pkg.Interfaces {
doc := s.interfaceToDocument(iface, pkg, source.Name)
documents = append(documents, doc)
}
for _, enum := range pkg.Enums {
doc := s.enumToDocument(enum, pkg, source.Name)
documents = append(documents, doc)
}
for _, exc := range pkg.Exceptions {
doc := s.exceptionToDocument(exc, pkg, source.Name)
documents = append(documents, doc)
}
return documents, nil
}
func (s *JavaDocsScraper) DetectChanges(ctx context.Context, source *Source, lastHash string) (bool, string, error) {
html, err := s.fetchPage(ctx, source.URL)
if err != nil {
return false, "", err
}
hash := s.generateHash(html)
changed := hash != lastHash
return changed, hash, nil
}
func (s *JavaDocsScraper) fetchPage(ctx context.Context, url string) (string, error) {
req, err := http.NewRequestWithContext(ctx, "GET", url, nil)
if err != nil {
return "", err
}
req.Header.Set("User-Agent", s.config.UserAgent)
resp, err := s.client.Do(req)
if err != nil {
return "", err
}
defer resp.Body.Close()
if resp.StatusCode != http.StatusOK {
return "", fmt.Errorf("HTTP %d", resp.StatusCode)
}
body, err := io.ReadAll(resp.Body)
if err != nil {
return "", err
}
return string(body), nil
}
func (s *JavaDocsScraper) generateHash(content string) string {
hash := sha256.Sum256([]byte(content))
return hex.EncodeToString(hash[:])
}
func (s *JavaDocsScraper) packageToDocument(pkg *javadocs.Package, sourceName string) *Document {
var content strings.Builder
fmt.Fprintf(&content, "# Package %s\n\n", pkg.Name)
fmt.Fprintf(&content, "%s\n", pkg.Doc)
metadata := map[string]interface{}{
"package": pkg.Name,
"doc_url": pkg.DocURL,
"doc_type": "java-package",
}
return &Document{
ID: generateDocID(pkg.DocURL),
Source: sourceName,
Type: "java-package",
Title: pkg.Name,
Content: content.String(),
URL: pkg.DocURL,
Metadata: metadata,
Hash: s.generateHash(content.String()),
Timestamp: time.Now(),
}
}
func (s *JavaDocsScraper) classToDocument(class *javadocs.Class, pkg *javadocs.Package, sourceName string) *Document {
var content strings.Builder
fmt.Fprintf(&content, "# %s\n\n", class.QualifiedName)
fmt.Fprintf(&content, "Kind: %s\n\n", class.Kind)
fmt.Fprintf(&content, "%s\n", class.Doc)
if len(class.Methods) > 0 {
fmt.Fprintf(&content, "\n## Methods\n")
for _, m := range class.Methods {
fmt.Fprintf(&content, "- `%s`\n", m.Signature)
}
}
metadata := map[string]interface{}{
"package": pkg.Name,
"qualified_name": class.QualifiedName,
"kind": string(class.Kind),
"doc_url": class.DocURL,
}
return &Document{
ID: generateDocID(class.DocURL),
Source: sourceName,
Type: "java-class",
Title: class.QualifiedName,
Content: content.String(),
URL: class.DocURL,
Metadata: metadata,
Hash: s.generateHash(content.String()),
Timestamp: time.Now(),
}
}
func (s *JavaDocsScraper) interfaceToDocument(iface *javadocs.Class, pkg *javadocs.Package, sourceName string) *Document {
var content strings.Builder
fmt.Fprintf(&content, "# %s (interface)\n\n", iface.QualifiedName)
fmt.Fprintf(&content, "%s\n", iface.Doc)
metadata := map[string]interface{}{
"package": pkg.Name,
"qualified_name": iface.QualifiedName,
"kind": "interface",
"doc_url": iface.DocURL,
}
return &Document{
ID: generateDocID(iface.DocURL),
Source: sourceName,
Type: "java-interface",
Title: iface.QualifiedName,
Content: content.String(),
URL: iface.DocURL,
Metadata: metadata,
Hash: s.generateHash(content.String()),
Timestamp: time.Now(),
}
}
func (s *JavaDocsScraper) enumToDocument(enum *javadocs.Enum, pkg *javadocs.Package, sourceName string) *Document {
var content strings.Builder
fmt.Fprintf(&content, "# %s (enum)\n\n", enum.QualifiedName)
fmt.Fprintf(&content, "%s\n", enum.Doc)
if len(enum.Constants) > 0 {
fmt.Fprintf(&content, "\n## Constants\n")
for _, c := range enum.Constants {
fmt.Fprintf(&content, "- `%s`\n", c.Name)
}
}
metadata := map[string]interface{}{
"package": pkg.Name,
"qualified_name": enum.QualifiedName,
"kind": "enum",
"doc_url": enum.DocURL,
}
return &Document{
ID: generateDocID(enum.DocURL),
Source: sourceName,
Type: "java-enum",
Title: enum.QualifiedName,
Content: content.String(),
URL: enum.DocURL,
Metadata: metadata,
Hash: s.generateHash(content.String()),
Timestamp: time.Now(),
}
}
func (s *JavaDocsScraper) exceptionToDocument(exc *javadocs.Class, pkg *javadocs.Package, sourceName string) *Document {
var content strings.Builder
fmt.Fprintf(&content, "# %s (exception)\n\n", exc.QualifiedName)
fmt.Fprintf(&content, "%s\n", exc.Doc)
metadata := map[string]interface{}{
"package": pkg.Name,
"qualified_name": exc.QualifiedName,
"kind": "exception",
"doc_url": exc.DocURL,
}
return &Document{
ID: generateDocID(exc.DocURL),
Source: sourceName,
Type: "java-exception",
Title: exc.QualifiedName,
Content: content.String(),
URL: exc.DocURL,
Metadata: metadata,
Hash: s.generateHash(content.String()),
Timestamp: time.Now(),
}
}