Files
arsam f65143a3fd Import source: transport
Import the existing local project snapshot. This commit organizes existing files; it does not represent newly developed functionality.
2026-09-18 23:19:02 +03:30

69 lines
1.8 KiB
Go

package reactor_crw
import (
"fmt"
"io"
"net/http"
)
// Headers defines a simple wrapper for client headers.
type Headers map[string]string
// HttpTransport allows making a network request over HTTP protocol. It is a
// wrapper over HTTP.Client with a set of default params required by the crawler.
type HttpTransport struct {
headers Headers
client *http.Client
}
// NewHttpTransport creates a new *HttpTransport with provided client and custom
// headers. Custom headers can override the default ones.
func NewHttpTransport(c *http.Client, h Headers) *HttpTransport {
var defaultHeaders = Headers{
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:92.0) Gecko/20100101 Firefox/92.0",
"Referer": "http://joyreactor.cc/",
"DNT": "1",
}
t := &HttpTransport{
headers: defaultHeaders,
client: c,
}
for k, v := range h {
t.headers[k] = v
}
return t
}
// FetchData makes an HTTP request using provided URL and returns the response
// body as io.ReadCloser interface. Each request will be prepared with provided
// headers list. The end client is responsible for closing the response body.
func (t *HttpTransport) FetchData(url string) (io.ReadCloser, error) {
req, err := t.prepareRequest(http.MethodGet, url)
if err != nil {
return nil, err
}
res, err := t.client.Do(req)
if err != nil {
return nil, fmt.Errorf("cannot make request to %s: %w", url, err)
}
return res.Body, nil
}
func (t *HttpTransport) prepareRequest(method, url string) (*http.Request, error) {
req, err := http.NewRequest(method, url, nil)
if err != nil {
return nil, fmt.Errorf("cannot prepare request to %s: %w", url, err)
}
for k, v := range t.headers {
req.Header.Add(k, v)
}
return req, nil
}