Beispiel #1
0
func (im *imp) ServeCallback(w http.ResponseWriter, r *http.Request, ctx *importer.SetupContext) {
	tempToken := ctx.AccountNode.Attr(importer.AcctAttrTempToken)
	tempSecret := ctx.AccountNode.Attr(importer.AcctAttrTempSecret)
	if tempToken == "" || tempSecret == "" {
		log.Printf("twitter: no temp creds in callback")
		httputil.BadRequestError(w, "no temp creds in callback")
		return
	}
	if tempToken != r.FormValue("oauth_token") {
		log.Printf("unexpected oauth_token: got %v, want %v", r.FormValue("oauth_token"), tempToken)
		httputil.BadRequestError(w, "unexpected oauth_token")
		return
	}
	oauthClient, err := ctx.NewOAuthClient(oAuthURIs)
	if err != nil {
		err = fmt.Errorf("error getting OAuth client: %v", err)
		httputil.ServeError(w, r, err)
		return
	}
	tokenCred, vals, err := oauthClient.RequestToken(
		ctxutil.Client(ctx),
		&oauth.Credentials{
			Token:  tempToken,
			Secret: tempSecret,
		},
		r.FormValue("oauth_verifier"),
	)
	if err != nil {
		httputil.ServeError(w, r, fmt.Errorf("Error getting request token: %v ", err))
		return
	}
	userid := vals.Get("user_id")
	if userid == "" {
		httputil.ServeError(w, r, fmt.Errorf("Couldn't get user id: %v", err))
		return
	}
	if err := ctx.AccountNode.SetAttrs(
		importer.AcctAttrAccessToken, tokenCred.Token,
		importer.AcctAttrAccessTokenSecret, tokenCred.Secret,
	); err != nil {
		httputil.ServeError(w, r, fmt.Errorf("Error setting token attributes: %v", err))
		return
	}

	u, err := getUserInfo(importer.OAuthContext{ctx.Context, oauthClient, tokenCred})
	if err != nil {
		httputil.ServeError(w, r, fmt.Errorf("Couldn't get user info: %v", err))
		return
	}
	if err := ctx.AccountNode.SetAttrs(
		importer.AcctAttrUserID, u.ID,
		importer.AcctAttrName, u.Name,
		importer.AcctAttrUserName, u.ScreenName,
		nodeattr.Title, fmt.Sprintf("%s's Twitter Account", u.ScreenName),
	); err != nil {
		httputil.ServeError(w, r, fmt.Errorf("Error setting attribute: %v", err))
		return
	}
	http.Redirect(w, r, ctx.AccountURL(), http.StatusFound)
}
Beispiel #2
0
// Lookup returns rectangles for the given address. Currently the only
// implementation is the Google geocoding service.
func Lookup(ctx context.Context, address string) ([]Rect, error) {
	mu.RLock()
	rects, ok := cache[address]
	mu.RUnlock()
	if ok {
		return rects, nil
	}

	rectsi, err := sf.Do(address, func() (interface{}, error) {
		// TODO: static data files from OpenStreetMap, Wikipedia, etc?
		urlStr := "https://maps.googleapis.com/maps/api/geocode/json?address=" + url.QueryEscape(address) + "&sensor=false"
		res, err := ctxhttp.Get(ctx, ctxutil.Client(ctx), urlStr)
		if err != nil {
			return nil, err
		}
		defer res.Body.Close()
		rects, err := decodeGoogleResponse(res.Body)
		log.Printf("Google geocode lookup (%q) = %#v, %v", address, rects, err)
		if err == nil {
			mu.Lock()
			cache[address] = rects
			mu.Unlock()
		}
		return rects, err
	})
	if err != nil {
		return nil, err
	}
	return rectsi.([]Rect), nil
}
Beispiel #3
0
func (imp) getUserInfo(ctx context.Context) (*userInfo, error) {
	u, err := picago.GetUser(ctxutil.Client(ctx), "default")
	if err != nil {
		return nil, err
	}
	return &userInfo{ID: u.ID, Name: u.Name}, nil
}
Beispiel #4
0
func (im *imp) ServeSetup(w http.ResponseWriter, r *http.Request, ctx *importer.SetupContext) error {
	oauthClient, err := ctx.NewOAuthClient(oAuthURIs)
	if err != nil {
		err = fmt.Errorf("error getting OAuth client: %v", err)
		httputil.ServeError(w, r, err)
		return err
	}
	tempCred, err := oauthClient.RequestTemporaryCredentials(ctxutil.Client(ctx), ctx.CallbackURL(), nil)
	if err != nil {
		err = fmt.Errorf("Error getting temp cred: %v", err)
		httputil.ServeError(w, r, err)
		return err
	}
	if err := ctx.AccountNode.SetAttrs(
		importer.AcctAttrTempToken, tempCred.Token,
		importer.AcctAttrTempSecret, tempCred.Secret,
	); err != nil {
		err = fmt.Errorf("Error saving temp creds: %v", err)
		httputil.ServeError(w, r, err)
		return err
	}

	authURL := oauthClient.AuthorizationURL(tempCred, nil)
	http.Redirect(w, r, authURL, 302)
	return nil
}
Beispiel #5
0
// urlFileRef slurps urlstr from the net, writes to a file and returns its
// fileref or "" on error
func (r *run) urlFileRef(urlstr, filename string) string {
	im := r.im
	im.mu.Lock()
	if br, ok := im.imageFileRef[urlstr]; ok {
		im.mu.Unlock()
		return br.String()
	}
	im.mu.Unlock()

	res, err := ctxutil.Client(r).Get(urlstr)
	if err != nil {
		log.Printf("couldn't get image: %v", err)
		return ""
	}
	defer res.Body.Close()

	fileRef, err := schema.WriteFileFromReader(r.Host.Target(), filename, res.Body)
	if err != nil {
		r.errorf("couldn't write file: %v", err)
		return ""
	}

	im.mu.Lock()
	defer im.mu.Unlock()
	im.imageFileRef[urlstr] = fileRef
	return fileRef.String()
}
Beispiel #6
0
func (im extendedOAuth2) ServeCallback(w http.ResponseWriter, r *http.Request, ctx *importer.SetupContext) {
	if im.getUserInfo == nil {
		panic("No getUserInfo is provided, don't use the default ServeCallback!")
	}

	oauthConfig, err := im.auth(ctx)
	if err != nil {
		httputil.ServeError(w, r, fmt.Errorf("Error getting oauth config: %v", err))
		return
	}

	if r.Method != "GET" {
		http.Error(w, "Expected a GET", 400)
		return
	}
	code := r.FormValue("code")
	if code == "" {
		http.Error(w, "Expected a code", 400)
		return
	}

	// picago calls take an *http.Client, so we need to provide one which already
	// has a transport set up correctly wrt to authentication. In particular, it
	// needs to have the access token that is obtained during Exchange.
	transport := &oauth.Transport{
		Config:    oauthConfig,
		Transport: notOAuthTransport(ctxutil.Client(ctx)),
	}
	token, err := transport.Exchange(code)
	log.Printf("Token = %#v, error %v", token, err)
	if err != nil {
		log.Printf("Token Exchange error: %v", err)
		httputil.ServeError(w, r, fmt.Errorf("token exchange error: %v", err))
		return
	}

	picagoCtx, cancel := context.WithCancel(context.WithValue(ctx, ctxutil.HTTPClient, transport.Client()))
	defer cancel()

	userInfo, err := im.getUserInfo(picagoCtx)
	if err != nil {
		log.Printf("Couldn't get username: %v", err)
		httputil.ServeError(w, r, fmt.Errorf("can't get username: %v", err))
		return
	}

	if err := ctx.AccountNode.SetAttrs(
		importer.AcctAttrUserID, userInfo.ID,
		importer.AcctAttrGivenName, userInfo.FirstName,
		importer.AcctAttrFamilyName, userInfo.LastName,
		acctAttrOAuthToken, encodeToken(token),
	); err != nil {
		httputil.ServeError(w, r, fmt.Errorf("Error setting attribute: %v", err))
		return
	}
	http.Redirect(w, r, ctx.AccountURL(), http.StatusFound)
}
Beispiel #7
0
func (imp) ServeCallback(w http.ResponseWriter, r *http.Request, ctx *importer.SetupContext) {
	tempToken := ctx.AccountNode.Attr(importer.AcctAttrTempToken)
	tempSecret := ctx.AccountNode.Attr(importer.AcctAttrTempSecret)
	if tempToken == "" || tempSecret == "" {
		log.Printf("flicker: no temp creds in callback")
		httputil.BadRequestError(w, "no temp creds in callback")
		return
	}
	if tempToken != r.FormValue("oauth_token") {
		log.Printf("unexpected oauth_token: got %v, want %v", r.FormValue("oauth_token"), tempToken)
		httputil.BadRequestError(w, "unexpected oauth_token")
		return
	}
	oauthClient, err := ctx.NewOAuthClient(oAuthURIs)
	if err != nil {
		err = fmt.Errorf("error getting OAuth client: %v", err)
		httputil.ServeError(w, r, err)
		return
	}
	tokenCred, vals, err := oauthClient.RequestToken(
		ctxutil.Client(ctx),
		&oauth.Credentials{
			Token:  tempToken,
			Secret: tempSecret,
		},
		r.FormValue("oauth_verifier"),
	)
	if err != nil {
		httputil.ServeError(w, r, fmt.Errorf("Error getting request token: %v ", err))
		return
	}
	userID := vals.Get("user_nsid")
	if userID == "" {
		httputil.ServeError(w, r, fmt.Errorf("Couldn't get user id: %v", err))
		return
	}
	username := vals.Get("username")
	if username == "" {
		httputil.ServeError(w, r, fmt.Errorf("Couldn't get user name: %v", err))
		return
	}

	// TODO(mpl): get a few more bits of info (first name, last name etc) like I did for twitter, if possible.
	if err := ctx.AccountNode.SetAttrs(
		importer.AcctAttrAccessToken, tokenCred.Token,
		importer.AcctAttrAccessTokenSecret, tokenCred.Secret,
		importer.AcctAttrUserID, userID,
		importer.AcctAttrUserName, username,
	); err != nil {
		httputil.ServeError(w, r, fmt.Errorf("Error setting basic account attributes: %v", err))
		return
	}
	http.Redirect(w, r, ctx.AccountURL(), http.StatusFound)
}
Beispiel #8
0
// Get fetches through octx the resource defined by url and the values in form.
func (octx OAuthContext) Get(url string, form url.Values) (*http.Response, error) {
	if octx.Creds == nil {
		return nil, errors.New("No OAuth credentials. Not logged in?")
	}
	if octx.Client == nil {
		return nil, errors.New("No OAuth client.")
	}
	res, err := octx.Client.Get(ctxutil.Client(octx.Ctx), octx.Creds, url, form)
	if err != nil {
		return nil, fmt.Errorf("Error fetching %s: %v", url, err)
	}
	if res.StatusCode != http.StatusOK {
		return nil, fmt.Errorf("Get request on %s failed with: %s", url, res.Status)
	}
	return res, nil
}
Beispiel #9
0
func doGet(ctx context.Context, url string, form url.Values) (*http.Response, error) {
	requestURL := url + "?" + form.Encode()
	req, err := http.NewRequest("GET", requestURL, nil)
	if err != nil {
		return nil, err
	}
	res, err := ctxutil.Client(ctx).Do(req)
	if err != nil {
		log.Printf("Error fetching %s: %v", url, err)
		return nil, err
	}
	if res.StatusCode != http.StatusOK {
		return nil, fmt.Errorf("Get request on %s failed with: %s", requestURL, res.Status)
	}
	return res, nil
}
Beispiel #10
0
func doGet(ctx context.Context, url string) ([]byte, error) {
	req, err := http.NewRequest("GET", url, nil)
	if err != nil {
		return nil, err
	}
	res, err := ctxutil.Client(ctx).Do(req)
	if err != nil {
		log.Printf("Error fetching %s: %v", url, err)
		return nil, err
	}
	defer res.Body.Close()
	if res.StatusCode != http.StatusOK {
		return nil, fmt.Errorf("Get request on %s failed with: %s", url, res.Status)
	}
	return ioutil.ReadAll(io.LimitReader(res.Body, 8<<20))
}
Beispiel #11
0
func (imp) Run(ctx *importer.RunContext) error {
	clientId, secret, err := ctx.Credentials()
	if err != nil {
		return err
	}
	acctNode := ctx.AccountNode()
	ocfg := baseOAuthConfig
	ocfg.ClientId, ocfg.ClientSecret = clientId, secret
	token := decodeToken(acctNode.Attr(acctAttrOAuthToken))
	transport := &oauth.Transport{
		Config:    &ocfg,
		Token:     &token,
		Transport: notOAuthTransport(ctxutil.Client(ctx)),
	}
	ctx.Context = context.WithValue(ctx.Context, ctxutil.HTTPClient, transport.Client())

	root := ctx.RootNode()
	if root.Attr(nodeattr.Title) == "" {
		if err := root.SetAttr(nodeattr.Title,
			fmt.Sprintf("%s %s - Google/Picasa Photos",
				acctNode.Attr(importer.AcctAttrGivenName),
				acctNode.Attr(importer.AcctAttrFamilyName))); err != nil {
			return err
		}
	}

	r := &run{
		RunContext:  ctx,
		incremental: !forceFullImport && acctNode.Attr(importer.AcctAttrCompletedVersion) == runCompleteVersion,
		photoGate:   syncutil.NewGate(3),
	}
	if err := r.importAlbums(); err != nil {
		return err
	}

	r.mu.Lock()
	anyErr := r.anyErr
	r.mu.Unlock()
	if !anyErr {
		if err := acctNode.SetAttrs(importer.AcctAttrCompletedVersion, runCompleteVersion); err != nil {
			return err
		}
	}

	return nil
}
Beispiel #12
0
// EnumerateObjects lists the objects in a bucket.
// This function relies on the ctx oauth2.HTTPClient value being set to an OAuth2
// authorized and authenticated HTTP client.
// If after is non-empty, listing will begin with lexically greater object names.
// If limit is non-zero, the length of the list will be limited to that number.
func EnumerateObjects(ctx context.Context, bucket, after string, limit int) ([]*storage.ObjectAttrs, error) {
	// Build url, with query params
	var params []string
	if after != "" {
		params = append(params, "marker="+url.QueryEscape(after))
	}
	if limit > 0 {
		params = append(params, fmt.Sprintf("max-keys=%v", limit))
	}
	query := ""
	if len(params) > 0 {
		query = "?" + strings.Join(params, "&")
	}

	req, err := simpleRequest("GET", gsAccessURL+"/"+bucket+"/"+query)
	if err != nil {
		return nil, err
	}
	req.Cancel = ctx.Done()
	res, err := ctxutil.Client(ctx).Do(req)
	if err != nil {
		return nil, err
	}
	defer res.Body.Close()
	if res.StatusCode != http.StatusOK {
		return nil, fmt.Errorf("gcsutil: bad enumerate response code: %v", res.Status)
	}

	var xres struct {
		Contents []SizedObject
	}
	if err = xml.NewDecoder(res.Body).Decode(&xres); err != nil {
		return nil, err
	}

	objAttrs := make([]*storage.ObjectAttrs, len(xres.Contents))
	for k, o := range xres.Contents {
		objAttrs[k] = &storage.ObjectAttrs{
			Name: o.Key,
			Size: o.Size,
		}
	}

	return objAttrs, nil
}
Beispiel #13
0
func (r *run) importAlbums(ctx context.Context) error {
	albums, err := picago.GetAlbums(ctxutil.Client(ctx), "default")
	if err != nil {
		return fmt.Errorf("importAlbums: error listing albums: %v", err)
	}
	albumsNode, err := r.getTopLevelNode("albums", "Albums")
	for _, album := range albums {
		select {
		case <-ctx.Done():
			return ctx.Err()
		default:
		}
		if err := r.importAlbum(ctx, albumsNode, album); err != nil {
			return fmt.Errorf("picasa importer: error importing album %s: %v", album, err)
		}
	}
	return nil
}
Beispiel #14
0
func init() {
	importer.Register("picasa", imp{
		newExtendedOAuth2(
			baseOAuthConfig,
			func(ctx context.Context) (*userInfo, error) {

				u, err := picago.GetUser(ctxutil.Client(ctx), "default")
				if err != nil {
					return nil, err
				}
				firstName, lastName := u.Name, ""
				i := strings.LastIndex(u.Name, " ")
				if i >= 0 {
					firstName, lastName = u.Name[:i], u.Name[i+1:]
				}
				return &userInfo{
					ID:        u.ID,
					FirstName: firstName,
					LastName:  lastName,
				}, nil
			}),
	})
}
Beispiel #15
0
// GetPartialObject fetches part of a Google Cloud Storage object.
// This function relies on the ctx ctxutil.HTTPClient value being set to an OAuth2
// authorized and authenticated HTTP client.
// If length is negative, the rest of the object is returned.
// It returns ErrInvalidRange if the server replies with http.StatusRequestedRangeNotSatisfiable.
// The caller must call Close on the returned value.
func GetPartialObject(ctx context.Context, obj Object, offset, length int64) (io.ReadCloser, error) {
	if offset < 0 {
		return nil, errors.New("invalid negative offset")
	}
	if err := obj.valid(); err != nil {
		return nil, err
	}

	req, err := simpleRequest("GET", gsAccessURL+"/"+obj.Bucket+"/"+obj.Key)
	if err != nil {
		return nil, err
	}
	if length >= 0 {
		req.Header.Set("Range", fmt.Sprintf("bytes=%d-%d", offset, offset+length-1))
	} else {
		req.Header.Set("Range", fmt.Sprintf("bytes=%d-", offset))
	}
	req.Cancel = ctx.Done()
	res, err := ctxutil.Client(ctx).Do(req)
	if err != nil {
		return nil, fmt.Errorf("GET (offset=%d, length=%d) failed: %v\n", offset, length, err)
	}
	if res.StatusCode == http.StatusNotFound {
		res.Body.Close()
		return nil, os.ErrNotExist
	}
	if !(res.StatusCode == http.StatusPartialContent || (offset == 0 && res.StatusCode == http.StatusOK)) {
		res.Body.Close()
		if res.StatusCode == http.StatusRequestedRangeNotSatisfiable {
			return nil, ErrInvalidRange
		}
		return nil, fmt.Errorf("GET (offset=%d, length=%d) got failed status: %v\n", offset, length, res.Status)
	}

	return res.Body, nil
}
Beispiel #16
0
func (r *run) importBatch(authToken string, parent *importer.Object) (keepTrying bool, err error) {
	sleepDuration := r.nextAfter.Sub(time.Now())
	// block until we either get canceled or until it is time to run
	select {
	case <-r.Context().Done():
		log.Printf("pinboard: Importer interrupted.")
		return false, r.Context().Err()
	case <-time.After(sleepDuration):
		// just proceed
	}
	start := time.Now()

	u := fmt.Sprintf(fetchUrl, authToken, batchLimit, r.nextCursor)
	resp, err := ctxutil.Client(r.Context()).Get(u)
	if err != nil {
		return false, err
	}
	defer resp.Body.Close()
	switch {
	case resp.StatusCode == StatusTooManyRequests:
		r.lastPause = r.lastPause * 2
		r.nextAfter = time.Now().Add(r.lastPause)
		return true, nil
	case resp.StatusCode != http.StatusOK:
		return false, fmt.Errorf("Unexpected status code %v fetching %v", resp.StatusCode, u)
	}

	body, err := ioutil.ReadAll(resp.Body)
	if err != nil {
		return false, err
	}

	var postBatch []apiPost
	if err = json.Unmarshal(body, &postBatch); err != nil {
		return false, err
	}

	if err != nil {
		return false, err
	}

	postCount := len(postBatch)
	if postCount == 0 {
		// we are done!
		return false, nil
	}

	log.Printf("pinboard: Importing %d posts...", postCount)
	var grp syncutil.Group
	for _, post := range postBatch {
		select {
		case <-r.Context().Done():
			log.Printf("pinboard: Importer interrupted")
			return false, r.Context().Err()
		default:
		}

		post := post
		r.postGate.Start()
		grp.Go(func() error {
			defer r.postGate.Done()
			return r.importPost(&post, parent)
		})
	}

	log.Printf("pinboard: Imported batch of %d posts in %s.", postCount, time.Now().Sub(start))

	r.nextCursor = postBatch[postCount-1].Time
	r.lastPause = pauseInterval
	r.nextAfter = time.Now().Add(pauseInterval)
	tryAgain := postCount == batchLimit
	return tryAgain, grp.Err()
}
Beispiel #17
0
// viaAPI is true if it came via the REST API, or false if it came via a zip file.
func (r *run) importTweet(parent *importer.Object, tweet tweetItem, viaAPI bool) (dup bool, err error) {
	select {
	case <-r.Context().Done():
		r.errorf("Twitter importer: interrupted")
		return false, r.Context().Err()
	default:
	}
	id := tweet.ID()
	tweetNode, err := parent.ChildPathObject(id)
	if err != nil {
		return false, err
	}

	// Because the zip format and the API format differ a bit, and
	// might diverge more in the future, never use the zip content
	// to overwrite data fetched via the API. If we add new
	// support for different fields in the future, we might want
	// to revisit this decision.  Be wary of flip/flopping data if
	// modifying this, though.
	if tweetNode.Attr(attrImportMethod) == "api" && !viaAPI {
		return true, nil
	}

	// e.g. "2014-06-12 19:11:51 +0000"
	createdTime, err := timeParseFirstFormat(tweet.CreatedAt(), time.RubyDate, "2006-01-02 15:04:05 -0700")
	if err != nil {
		return false, fmt.Errorf("could not parse time %q: %v", tweet.CreatedAt(), err)
	}

	url := fmt.Sprintf("https://twitter.com/%s/status/%v",
		r.AccountNode().Attr(importer.AcctAttrUserName),
		id)

	attrs := []string{
		"twitterId", id,
		nodeattr.Type, "twitter.com:tweet",
		nodeattr.StartDate, schema.RFC3339FromTime(createdTime),
		nodeattr.Content, tweet.Text(),
		nodeattr.URL, url,
	}
	if lat, long, ok := tweet.LatLong(); ok {
		attrs = append(attrs,
			nodeattr.Latitude, fmt.Sprint(lat),
			nodeattr.Longitude, fmt.Sprint(long),
		)
	}
	if viaAPI {
		attrs = append(attrs, attrImportMethod, "api")
	} else {
		attrs = append(attrs, attrImportMethod, "zip")
	}

	for i, m := range tweet.Media() {
		filename := m.BaseFilename()
		if tweetNode.Attr("camliPath:"+filename) != "" && (i > 0 || tweetNode.Attr("camliContentImage") != "") {
			// Don't re-import media we've already fetched.
			continue
		}
		tried, gotMedia := 0, false
		for _, mediaURL := range m.URLs() {
			tried++
			res, err := ctxutil.Client(r.Context()).Get(mediaURL)
			if err != nil {
				return false, fmt.Errorf("Error fetching %s for tweet %s : %v", mediaURL, url, err)
			}
			if res.StatusCode == http.StatusNotFound {
				continue
			}
			if res.StatusCode != 200 {
				return false, fmt.Errorf("HTTP status %d fetching %s for tweet %s", res.StatusCode, mediaURL, url)
			}
			if !viaAPI {
				log.Printf("For zip tweet %s, reading %v", url, mediaURL)
			}
			fileRef, err := schema.WriteFileFromReader(r.Host.Target(), filename, res.Body)
			res.Body.Close()
			if err != nil {
				return false, fmt.Errorf("Error fetching media %s for tweet %s: %v", mediaURL, url, err)
			}
			attrs = append(attrs, "camliPath:"+filename, fileRef.String())
			if i == 0 {
				attrs = append(attrs, "camliContentImage", fileRef.String())
			}
			log.Printf("Slurped %s as %s for tweet %s (%v)", mediaURL, fileRef.String(), url, tweetNode.PermanodeRef())
			gotMedia = true
			break
		}
		if !gotMedia && tried > 0 {
			return false, fmt.Errorf("All media URLs 404s for tweet %s", url)
		}
	}

	changes, err := tweetNode.SetAttrs2(attrs...)
	if err == nil && changes {
		log.Printf("Imported tweet %s", url)
	}
	return !changes, err
}
Beispiel #18
0
func (r *run) updatePhotoInAlbum(ctx context.Context, albumNode *importer.Object, photo picago.Photo) (ret error) {
	if photo.ID == "" {
		return errors.New("photo has no ID")
	}

	getMediaBytes := func() (io.ReadCloser, error) {
		log.Printf("Importing media from %v", photo.URL)
		resp, err := ctxutil.Client(ctx).Get(photo.URL)
		if err != nil {
			return nil, fmt.Errorf("importing photo %s: %v", photo.ID, err)
		}
		if resp.StatusCode != http.StatusOK {
			resp.Body.Close()
			return nil, fmt.Errorf("importing photo %s: status code = %d", photo.ID, resp.StatusCode)
		}
		return resp.Body, nil
	}

	var fileRefStr string
	idFilename := photo.ID + "-" + photo.Filename
	photoNode, err := albumNode.ChildPathObjectOrFunc(idFilename, func() (*importer.Object, error) {
		h := blob.NewHash()
		rc, err := getMediaBytes()
		if err != nil {
			return nil, err
		}
		fileRef, err := schema.WriteFileFromReader(r.Host.Target(), photo.Filename, io.TeeReader(rc, h))
		if err != nil {
			return nil, err
		}
		fileRefStr = fileRef.String()
		wholeRef := blob.RefFromHash(h)
		if pn, err := findExistingPermanode(r.Host.Searcher(), wholeRef); err == nil {
			return r.Host.ObjectFromRef(pn)
		}
		return r.Host.NewObject()
	})
	if err != nil {
		return err
	}

	const attrMediaURL = "picasaMediaURL"
	if fileRefStr == "" {
		fileRefStr = photoNode.Attr(nodeattr.CamliContent)
		// Only re-download the source photo if its URL has changed.
		// Empirically this seems to work: cropping a photo in the
		// photos.google.com UI causes its URL to change. And it makes
		// sense, looking at the ugliness of the URLs with all their
		// encoded/signed state.
		if !mediaURLsEqual(photoNode.Attr(attrMediaURL), photo.URL) {
			rc, err := getMediaBytes()
			if err != nil {
				return err
			}
			fileRef, err := schema.WriteFileFromReader(r.Host.Target(), photo.Filename, rc)
			rc.Close()
			if err != nil {
				return err
			}
			fileRefStr = fileRef.String()
		}
	}

	title := strings.TrimSpace(photo.Description)
	if strings.Contains(title, "\n") {
		title = title[:strings.Index(title, "\n")]
	}
	if title == "" && schema.IsInterestingTitle(photo.Filename) {
		title = photo.Filename
	}

	// TODO(tgulacsi): add more attrs (comments ?)
	// for names, see http://schema.org/ImageObject and http://schema.org/CreativeWork
	attrs := []string{
		nodeattr.CamliContent, fileRefStr,
		attrPicasaId, photo.ID,
		nodeattr.Title, title,
		nodeattr.Description, photo.Description,
		nodeattr.LocationText, photo.Location,
		nodeattr.DateModified, schema.RFC3339FromTime(photo.Updated),
		nodeattr.DatePublished, schema.RFC3339FromTime(photo.Published),
		nodeattr.URL, photo.PageURL,
	}
	if photo.Latitude != 0 || photo.Longitude != 0 {
		attrs = append(attrs,
			nodeattr.Latitude, fmt.Sprintf("%f", photo.Latitude),
			nodeattr.Longitude, fmt.Sprintf("%f", photo.Longitude),
		)
	}
	if err := photoNode.SetAttrs(attrs...); err != nil {
		return err
	}
	if err := photoNode.SetAttrValues("tag", photo.Keywords); err != nil {
		return err
	}
	if photo.Position > 0 {
		if err := albumNode.SetAttr(
			nodeattr.CamliPathOrderColon+strconv.Itoa(photo.Position-1),
			photoNode.PermanodeRef().String()); err != nil {
			return err
		}
	}

	// Do this last, after we're sure the "camliContent" attribute
	// has been saved successfully, because this is the one that
	// causes us to do it again in the future or not.
	if err := photoNode.SetAttrs(attrMediaURL, photo.URL); err != nil {
		return err
	}
	return nil
}
Beispiel #19
0
func (r *run) importAlbum(ctx context.Context, albumsNode *importer.Object, album picago.Album) (ret error) {
	if album.ID == "" {
		return errors.New("album has no ID")
	}
	albumNode, err := albumsNode.ChildPathObject(album.ID)
	if err != nil {
		return fmt.Errorf("importAlbum: error listing album: %v", err)
	}

	dateMod := schema.RFC3339FromTime(album.Updated)

	// Data reference: https://developers.google.com/picasa-web/docs/2.0/reference
	// TODO(tgulacsi): add more album info
	changes, err := albumNode.SetAttrs2(
		attrPicasaId, album.ID,
		nodeattr.Type, "picasaweb.google.com:album",
		nodeattr.Title, album.Title,
		nodeattr.DatePublished, schema.RFC3339FromTime(album.Published),
		nodeattr.LocationText, album.Location,
		nodeattr.Description, album.Description,
		nodeattr.URL, album.URL,
	)
	if err != nil {
		return fmt.Errorf("error setting album attributes: %v", err)
	}
	if !changes && r.incremental && albumNode.Attr(nodeattr.DateModified) == dateMod {
		return nil
	}
	defer func() {
		// Don't update DateModified on the album node until
		// we've successfully imported all the photos.
		if ret == nil {
			ret = albumNode.SetAttr(nodeattr.DateModified, dateMod)
		}
	}()

	log.Printf("Importing album %v: %v/%v (published %v, updated %v)", album.ID, album.Name, album.Title, album.Published, album.Updated)

	// TODO(bradfitz): GetPhotos does multiple HTTP requests to
	// return a slice of all photos. My "InstantUpload/Auto
	// Backup" album has 6678 photos (and growing) and this
	// currently takes like 40 seconds. Fix.
	photos, err := picago.GetPhotos(ctxutil.Client(ctx), "default", album.ID)
	if err != nil {
		return err
	}

	log.Printf("Importing %d photos from album %q (%s)", len(photos), albumNode.Attr(nodeattr.Title),
		albumNode.PermanodeRef())

	var grp syncutil.Group
	for i := range photos {
		select {
		case <-ctx.Done():
			return ctx.Err()
		default:
		}
		photo := photos[i]
		r.photoGate.Start()
		grp.Go(func() error {
			defer r.photoGate.Done()
			return r.updatePhotoInAlbum(ctx, albumNode, photo)
		})
	}
	return grp.Err()
}