hakurei/internal/pkg/net.go

package pkg

import (
	"crypto/sha512"
	"errors"
	"io"
	"net/http"
	"os"
	"sync"

	"hakurei.app/container/check"
)

// An httpArtifact is an [Artifact] backed by an [http] request.
type httpArtifact struct {
	// Caller-supplied request.
	req *http.Request

	// Caller-supplied checksum of the response body, also used as the
	// identifier. This is validated during curing.
	id ID

	// doFunc is the Do method of [http.Client] supplied by the caller.
	doFunc func(req *http.Request) (*http.Response, error)

	// Instance of [Cache] to submit the cured artifact to.
	c *Cache
	// Response body read to EOF.
	data []byte
	// Populated when submitting to or loading from [Cache].
	pathname *check.Absolute

	// Synchronises access to pathname and data.
	mu sync.Mutex
}

// NewHTTP returns a new [File] backed by the supplied client and request. If
// c is nil, [http.DefaultClient] is used instead.
func (c *Cache) NewHTTP(hc *http.Client, req *http.Request, checksum Checksum) File {
	if hc == nil {
		hc = http.DefaultClient
	}
	return &httpArtifact{req: req, id: checksum, doFunc: hc.Do, c: c}
}

// NewHTTPGet returns a new [File] backed by the supplied client. A GET request
// is set up for url. If c is nil, [http.DefaultClient] is used instead.
func (c *Cache) NewHTTPGet(hc *http.Client, url string, checksum Checksum) (File, error) {
	req, err := http.NewRequest(http.MethodGet, url, nil)
	if err != nil {
		return nil, err
	}
	return c.NewHTTP(hc, req, checksum), nil
}

// ID returns the caller-supplied hash of the response body.
func (a *httpArtifact) ID() ID { return a.id }

// ResponseStatusError is returned for a response returned by an [http.Client]
// with a status code other than [http.StatusOK].
type ResponseStatusError int

func (e ResponseStatusError) Error() string {
	return "the requested URL returned non-OK status: " + http.StatusText(int(e))
}

// do sends the caller-supplied request on the caller-supplied [http.Client]
// and reads its response body to EOF and returns the resulting bytes.
func (a *httpArtifact) do() (data []byte, err error) {
	var resp *http.Response
	if resp, err = a.doFunc(a.req); err != nil {
		return
	}

	if resp.StatusCode != http.StatusOK {
		_ = resp.Body.Close()
		return nil, ResponseStatusError(resp.StatusCode)
	}

	if data, err = io.ReadAll(resp.Body); err != nil {
		_ = resp.Body.Close()
		return
	}

	err = resp.Body.Close()
	return
}

// Hash cures the [Artifact] and returns its hash. The return value is always
// identical to that of the ID method.
func (a *httpArtifact) Hash() (Checksum, error) { _, err := a.Pathname(); return a.id, err }

// Pathname cures the [Artifact] and returns its pathname in the [Cache].
func (a *httpArtifact) Pathname() (pathname *check.Absolute, err error) {
	a.mu.Lock()
	defer a.mu.Unlock()

	if a.pathname != nil {
		return a.pathname, nil
	}

	if a.data != nil {
		pathname, err = a.c.StoreFile(
			a.id, a.data,
			(*Checksum)(&a.id),
			true,
		)
		if err == nil {
			a.pathname = pathname
		}
		return
	} else {
		a.pathname, a.data, _, err = a.c.LoadOrStoreFile(
			a.id, a.do,
			(*Checksum)(&a.id),
			true,
		)
		if err != nil {
			a.pathname, a.data = nil, nil
		}
		return a.pathname, err
	}
}

// Data completes the http request and returns the resulting response body read
// to EOF. Data does not write to the underlying [Cache].
func (a *httpArtifact) Data() (data []byte, err error) {
	a.mu.Lock()
	defer a.mu.Unlock()

	if a.data != nil {
		// validated by cache or a previous call to Data
		return a.data, nil
	}

	if a.pathname, a.data, err = a.c.LoadFile(a.id); err == nil {
		return a.data, nil
	} else {
		a.pathname, a.data = nil, nil
		if !errors.Is(err, os.ErrNotExist) {
			return
		}
	}

	if data, err = a.do(); err != nil {
		return
	}

	h := sha512.New384()
	h.Write(data)
	if got := (Checksum)(h.Sum(nil)); got != a.id {
		return nil, &ChecksumMismatchError{got, a.id}
	}
	a.data = data
	return
}