We try to use Server-Side API to push custom events to our pixel in order to create event-based Custom Audiences in Facebook Ads (https://developers.facebook.com/docs/marketing-api/facebook-pixel/server-side-api/)
We use _fbp cookie value to match users (it's a first party cookie created on our website by FB pixel).
For example (Python):
from facebook_business.api import FacebookAdsApi
from facebook_business.adobjects.adspixel import AdsPixel
my_app_id = 'X'
my_app_secret = 'X'
my_access_token = 'X'
my_pixel_id = 'X'
FacebookAdsApi.init(access_token=my_access_token, app_id=my_app_id, app_secret=my_app_secret)
fields = []
params = {
'data': [{'event_name': 'icrm_test_20191113_fbp_1m', 'event_time': 1573230217, 'user_data':{'fbp': 'fb.1.1558571054389.1098115397'}}]
}
print(AdsPixel(my_pixel_id).create_event(fields=fields, params=params))
The problem is, when we create a Custom Audience in Facebook Ads, the size of the list is always < 1000, even if we push hundreds of thousands of cookie IDs, which means Facebook matched a very low % of cookies, which were sent.
Custom Audience definition based on server-side event:
List size is always <1000, no matter how many fbp cookies are sent:
It seems like there is some kind of an issue matching _fbp cookies to Facebook user profiles. Is there any known way of improving/fixing matching results? We can't use hashes of sensitive data.
External_id matching (https://developers.facebook.com/docs/marketing-api/facebook-pixel/server-side-api/parameters) also gave us similar results.
The event time you are using there translates to 11/08/2019 # 4:23pm (UTC).
So they're not gonna get included in your 30 day window.
Try importing time
import time
Then set
'event_time': int(time.time())
Related
I am using the Twitter API from Matlab, specifically by means of the twitter class from the Datafeed Toolbox.
I have essentially followed the example code from the official documentation. I created a Twitter app in my Twitter developer page, and obtained its API keys and access tokens. With those I can use the Twitter Standard search API from Matlab:
c = twitter(consumerkey,consumersecret,accesstoken,accesstokensecret);
% The variables 'consumerkey' etc are defined as character vectors
s = search(c,tweetquery,'count',100); % this works
Now I want to use the Premium search API. This has two endpoints for accessing Tweets:
30-day endpoint: provides Tweets from the previous 30 days.
Full-archive endpoint: provides complete and instant access to Tweets dating all the way back to the first Tweet in March 2006.
In addition, the Premium API has two tiers of access:
Free Sandbox access that enables initial testing and development.
Paid Premium access that provides increased access.
The link above specifies the restrictions associated to sandbox as compared with paid access.
I am trying to use the full-archive endpoint with sandbox access. For that I had to create a developer environment on Twitter, which I named dev.
The search method in Matlab's twitter class (which worked for the Standard access, as described above) doesn't seem to work with the Premium access. But I noticed that search actually calls getdata, and the latter does work for Premium access as follows. First, the Premium access URL needs to be defined:
c.URL = 'https://api.twitter.com/1.1/tweets/search/fullarchive/dev.json';
and then the following syntax works:
s = getdata(c,c.URL,'query','Jimi Hendrix'); % this works
I have also been able to add operators within the query string, for example to specify a range of geographical positions or to restrict the search to tweets that contain images:
s = getdata(c,c.URL,'query','place:"Palo Alto"'); % this works
s = getdata(c,c.URL,'query','Robert Smith bounding_box:[-0.2 51.4 0.1 51.6]') % this works
However—and this is my question—, I haven't been able to use additional request parameters defined in the Twitter API to refine the search, such as fromDate, toDate or maxResults:
s = getdata(c,c.URL,'query','John Frusciante', 'fromDate', '201708130000') % doesn't work
s = getdata(c,c.URL,'query','Rob Scallon', ...
'fromDate', '201708130000', 'toDate', '201708150000') % doesn't work
s = getdata(c,c.URL,'query','Michael Lemmo', 'maxResults', '20') % doesn't work
All of the above return an HTTP/1.1 422 Unprocessable Entity error.
Is my syntax not correct? Maybe the fromDate etc parameters have to be part of the query string? Or maybe the sandbox tier of the Premium search doesn't support those parameters?
For context, I don't really know what all those terms like endpoint, tier, developer environment and token mean, but still I'd like to make this work.
Going by the description at https://developer.twitter.com/en/docs/tweets/search/api-reference/premium-search#DataParameters, what you call 'addition request parameters' are defined for requests of type POST /search/:product. These are HTTP POST requests, can you try using postdata (https://in.mathworks.com/help/datafeed/twitter.postdata.html) instead of getdata. Their usage is almost identical.
I am looking to find the follower count of a Scratch user using the Scratch API. I already know how to get their message count, with https://api.scratch.mit.edu/users/[USER]/messages/count/.
This answer targets the Scratch REST API, documented here.
You get the user's followers by requesting them: https://api.scratch.mit.edu/users/some_username/following where some_username is to be replaced by the actual username.
This will return 0 to 20 results (20 is the default limit of objects returned by the REST API). If there's less than 20 results, then you're done. The amount of followers is simply the count of the objects returned.
If there's 20 objects returned, we can't be certain we've requested all the user's friends as there might be more to come. Therefore, we skip the first 20 followers of that user by supplying the ?offset= parameter: https://api.scratch.mit.edu/users/some_username/following?offset=20
This retrieves the second 'page' of friends. Now we simply loop through the procedure described above, incrementing offset by 20 each time until either less than 20 results are returned or no results are returned. The amount of friends of that user is the cumulative count of the objects returned.
As mentioned by _nix on this forum thread, there is currently no API to achieve this. However, he/she rightly points out that the number can be obtained from a user's profile page.
You may write a script (in JavaScript, for example) to parse the HTML and get the follower count in the brackets at the top of the page.
Hope this helps!
There is a solution in Python:
import requests
import re
def followers(self,user):
followers = int(re.search(r'Followers \(([0-9]+)\)', requests.get(f'https://scratch.mit.edu/users/{user}/followers').text, re.I).group(1))
return f'{followers} on [scratch](https://scratch.mit.edu/users/{user}/followers)'
Credit goes to 12944qwerty, in his code (adapted to remove some implementation specific stuff).
use ScratchDB
var user = "username here";
fetch(`https://scratchdb.lefty.one/v3/user/info/${user}`).then(res => res.json()).then(data => {
console.log(`${user} has ` + data["followers"].toString() + " followers");
}
(Edit: this is javascript btw, I prefer Python but Python doesn't have a cloud.set function and this is how I did it)
Use ScratchDB (I used httpx, but you can GET with anything):
import httpx
import json
user = "griffpatch"
response = httpx.get(f"https://scratchdb.lefty.one/v3/user/info/{ user }")
userData = json.loads(response.text)
followers = userData["statistics"]["followers"]
https://api.scratch.mit.edu/users/griffpatch/followers
this gives the follower names, scratch staus(scratch team or not), pfp, everything in their profile
I am trying to bring comments made on a particular event by targeting this URL: https://graph.facebook.com/1466384840257158/comments
I am passing the user_access_token
I have two comments at present on this event made on the same
day(2014-03-29)
Now I try to pass a date which should bring an empty data result/object
like this: https://graph.facebook.com/1466384840257158/comments?since=2011-01-01&until=2014-01-10
This request has no effect, it still shows me the two comment made
on the 29th
I have tried the same kind of date range on my user-id/feed and it
gave me an empty data object.
Finally i tried event-id/feed (before trying date filter) and it
gave me the following error
.
{
"error": {
"message": "An unexpected error has occurred. Please retry your request later.",
"type": "OAuthException",
"code": 2
}
}
Could you please guide me about date filter on that particular query (point4) or if you have any other idea to use date filter on comments made for an event.
Comments use Cursor-based Pagination, so you cannot use since or until on the comments endpoint (these parameters would work f.ex. for the feed endpoint).
To get the comments in a time range you have to fetch all comments from NOW to the start of the time range, f.ex. with https://graph.facebook.com/1466384840257158/comments?filter=stream&limit=1000+paging (the filter=stream will order the result with the timestamp).
USING SINCE UNTIL FOR COMMENTS on GROUP
If you want to use since and until for comments, it is not possible directly for a group. So, First you can apply it for status(feed) and then get the comments for that feed.
This works for me:
{group_id}/?fields=feed.since(08/25/2016).until(08/31/2016){from,comments{from,message}}
Why don't you try first to filter by notifications?... notifications allows you to add parameters like since. For example (using Facebook pages):
https://graph.facebook.com/PAGEID?fields=notifications.since(2015-3-31 00:00:00).limit(250).include_read(true)&{id,created_time,updated_time,unread,object,link}&access_token=ACCESSTOKEN
Once you got the json data, loop through data, get the ID and send a second request but this time using the PAGEID_POSTID edge. Something like this:
https://graph.facebook.com/PAGEID_POSTID/comments?fields=id,from{name,id},message,can_remove,created_time&limit=1000
Voahla!... there's no need to read every comment!...
Note 1: A Page access token is required, along with the manage_pages permission
Note 2: Use the parameter/field include_read to get all the notifications, even the already readed
Note 3: In the second request, use the parameter/field "filter=stream" to order the posts and get the comments made in the name of your page
Note 4: Don't forget to control the asynchronicity once you loop!
Note 5: Notifications duplicate posts, use an array to avoid to read more than one time the postUse the parameter/field include_read to get all the notifications, even the already readed
I do not know if it's too late. But, Yeah it works in the graph api version 3.3.
for example: if you wanna get comments on a post of a Facebook page you can do it like this:
You have to use page Access-token
The get Graph Request : post_id/comments?since=some_date
In the context of a given Facebook app, suppose User A invited user B to start using it. Once User B accepts to use the app, is there any way to retrieve the ID of User A programmatically (via either PHP/JS SDK) ? This doesn't seem quite documented.
For what it's worth, A/B users are friends, if it's any use.
when user comes following the app request, you can get request id's using
$_GET['request_ids']
then retrieve all the request ids with which you can call graph api to get the corresponding request details like below:
if(isset($_GET['request_ids']))
{
$request_ids = $_GET['request_ids'];
}
$request_ids = explode(",", $request_ids);
foreach($request_ids as $request_id)
{
$request_object = $facebook->api($request_id);
//this $request_object have sender facebook id in the field uid_from
}
If you look here:
http://developers.facebook.com/docs/reference/dialogs/requests/
You can see the object layout. Of note is the data property:
Optional, additional data you may pass for tracking. This will be
stored as part of the request objects created. The maximum length is
255 characters.
In this object you can add your referring UserId and then when the request is claimed, you can then process it on your end.
Hope this helps.
I have the following working perfectly for calls to Facebook pages with simple URLS like: www.facebook.com/TurbonegroHQ
def get_facebook
#artist = Artist.accessible_by(current_ability).find(params[:id])
if #artist.facebook_url.present?
require 'open-uri'
require 'json'
result = JSON.parse(open("https://graph.facebook.com/"<<#artist.facebook_url).read)
#hometown = result["hometown"]
#band_members = result["band_members"]
#likes = result["likes"]
end
end
However, when a Facebook page URL isn't in this format, but something like https://www.facebook.com/pages/Clutch-The-Bakerton-Group-Weathermaker-Music/16637738637 it fails. Any ideas how I can get around this?
Thanks in advance?
TurbonegroHQ is the username of the page - this is interchangeable with the object ID in the Graph API.
If the page doesn't have a username you need to access it via the ID instead. In that case https://graph.facebook.com/16637738637 is the URL to access
Usually you'll have IDs rather than the URLs as the ID is returned in the API on all the endpoints I can think of