Google Reader API: item ids base10 vs base16

[lwindolf]

Maybe I'm overlooking something, but I believe the Google Reader API behaviour to be incorrect.

When I fetch item ids like this

  curl -H "Authorization: GoogleLogin auth=test/XXXXXXXXXXXXXXXXX" "http://admin@localhost:8080/reader/api/0/stream/items/ids?s=feed/2&output=json&n=10"

I get a JSON with base10 item ids. For example:

  {"itemRefs":[{"id":"2924"},{"id":"2931"},{"id":"2921"},{"id":"2932"},{"id":"2923"},{"id":"2928"},{"id":"2922"},{"id":"2927"},{"id":"2929"},{"id":"2926"}]}

When I use one of the ids to fetch the item

 curl -d "source=client&T=test/XXXXXXXXXXXXXXXXXX&output=json&i=2923" http://admin@localhost:8080/reader/api/0/stream/items/contents

then when fetching the most recent id I get

     {"error_message":"no items returned from the database"}

My understanding is the item ids are to be passed to /stream/items/contents as they are given by /stream/items/ids. But current the code does not do this. In googlereader/handler.go:348 base16 is applied.

func getItemIDs(r *http.Request) ([]int64, error) {
    [...]

    for i, item := range items {
	    var itemID int64
	    _, err := fmt.Sscanf(item, EntryIDLong, &itemID)
	    if err != nil {
		    itemID, err = strconv.ParseInt(item, 16, 64)       <<<<------- base16 

This is also visible in the log. The requested decimal item ids are treated as hex values causing much larger internal values which then do not exist. For comparison I also checked the FeedHQ Google Reader API implementation which does not indicate base16 and also original documentation mirros like https://raw.githubusercontent.com/mihaip/google-reader-api/master/wiki/ApiStreamItemsContents.wiki which also indicate decimal input values.

Please advise if my debugging is correct then I provide a PR.