Experimental io/fs.FS support
There is now experimental support for reading photos and the results of standard places response API calls using the io/fs.FS interfaces.
Individual photos can be read using the Open
method where file names are expected to take the form of.
- A unique numeric identifier for a photo on the Flickr website
- The fully-qualified path (not the whole URL) for an static photo asset hosted by the Flickr webservers.
- A URL-encoded query string followed by the fully-qualified path (not the whole URL) for an static photo asset hosted by the Flickr webservers encoded as a URL fragment.
For example:
fs.Open("6923069836")
Or:
fs.Open("/7129/7070018487_fbd223e965_o.png")
Or:
fs.Open("method=flickr.photosets.getPhotos&photoset_id=72157629455113026&user_id=35034348999%40N01/#/7244/7071114647_b8bcd16b65_o.jpg")
The last form is to account for paths derived from the fs.ReadDir
method. Specifically, the ReadDir
method expects filenames to take the form of URL-escaped URL query parameters to be passed to the Flickr API. For example:
fs.ReadDir("method=flickr.photosets.getPhotos&photoset_id=72157629455113026&user_id=35034348999%40N01")
This method will return zero or more fs.DirEntry
instances whose name (path) will be "#" + the fully qualified URL to the photo matching the query. When combined with the directory "root" which is actually a set of query parameters you end up with things like:
method=flickr.photosets.getPhotos&photoset_id=72157629455113026&user_id=35034348999%40N01/#/7244/7071114647_b8bcd16b65_o.jpg
Which is not ideal but easy enough to account for (which the Open
and ReadFile
methods do automatically.
Tests
All of the tests pass but there may still be "gotchas" or other edge cases. In order to run the tests with calls to the Flickr API you will need to run them with a valid -client-uri
flag. For example:
$> go test -v -run TestFS -client-uri 'oauth1://?consumer_key={CONSUMER_KEY}&consumer_secret={CONSUMER_SECRET}&oauth_token={OAUTH_TOKEN}&oauth_token_secret={OAUTH_SECRET}'
Caching
There is current no caching. Every time you Open
a file it is fetched from the Flickr API and/or photo servers. Some amount of caching would be good.
Example
Error handling has been removed for the sake of brevity.
import (
"context"
"image"
io_fs "io/fs"
"github.com/aaronland/go-flickr-api/client"
"github.com/aaronland/go-flickr-api/fs"
)
func main() {
ctx := context.Background()
// https://github.com/aaronland/go-flickr-api?tab=readme-ov-file#clients
client_uri := "oauth1://..."
cl, _ := client.NewClient(ctx, client_uri)
fs := New(ctx, cl)
r, _ := fs.Open("6923069836")
defer r.Close()
im, _, _ := image.Decode(fl)
// Do something with im here
walk_func := func(path string, d io_fs.DirEntry, err error) error {
r, _ := fs.Open(path)
defer r.Close()
// Do something with r here
return nil
}
u := url.Values{}
u.Set("method", "flickr.photosets.getPhotos")
u.Set("photoset_id", "72157629455113026")
u.Set("user_id", "35034348999@N01")
q := u.Encode()
io_fs.WalkDir(fs, q, walk_func)
)