tweetf0rm_1_0
branch
The old version is in the - The old version hasn't been updated for several reasons. Primarily because (1) it's too tedious to setup
redis
for this; and (2) using proxies don't work well unless you have massive private premium proxy servers. - If you want to see the old version, you can go old.
Note
- These are based on my use cases, which primarily for batch processing (e.g., continuously monitoring a set of public users and fetch their timelines).
- If there are missing functions, you are welcome to contribute and make pull requests.
- I do have a huge collection of tweets, see below Datasets section, but Twitter license (or at least the company's position on this) does not allow me redistribute the crawled data (e.g., someone asked the question a while back: https://dev.twitter.com/discussions/8232). If you want to get a hand on this dataset (e.g., through collaboration), contact me at [email protected].
- If you need
geocode
Twitter users (e.g., figure out where the user is from based on thelocation
string in their profile), you can take a look at this TwitterUserGeocoder - Post collect process tweeta is a set of convenience functions that might help you parse raw json tweets (and give you a
TweetaTweet
object so that you can access the tweet through functions (e.g.,tweet.tweet_id()
andtweet.created_at()
)). - Please cite any of these:
- Bian J, Zhao Y, Salloum RG, Guo Y, Wang M, Prosperi M, Zhang H, Du X, Ramirez-Diaz LJ, He Z, Sun Y. Using Social Media Data to Understand the Impact of Promotional Information on Laypeople’s Discussions: A Case Study of Lynch Syndrome. J Med Internet Res 2017;19(12):e414. DOI: 10.2196/jmir.9266. PMID: 29237586
- Bian J, Yoshigoe K, Hicks A, Yuan J, He Z, Xie M, Guo Y, Prosperi M, Salluom R, Modave F. Mining Twitter to assess the public perception of the "Internet of things". PLoS One. 2016 Jul 8;11(7):e0158450. doi: 10.1371/journal.pone.0158450. eCollection 2016. PMID: 27391760
- Hicks A, Hogan WR, Rutherford M, Malin B, Xie M, Fellbaum C, Yin Z, Fabbri D, Hanna J, Bian J. Mining Twitter as a First Step toward Assessing the Adequacy of Gender Identification Terms on Intake Forms. AMIA Annu Symp Proc. 2015;2015:611-620. PMID: 26958196*
Installation
None... just clone this and start using it.
git clone git://github.com/bianjiang/tweetf0rm.git
Dependencies
Just do:
pip install -r requirements.txt
Usage
- First, you'll want to login the twitter dev site and create an applciation at https://dev.twitter.com/apps to have access to the Twitter API!
- After you register, create an access token and grab your applications
Consumer Key
,Consumer Secret
,Access token
andAccess token secret
from the OAuth tool tab. Put these information into aconfig.json
underapikeys
(see an example below).
{
"apikeys": {
"i0mf0rmer01": {
"app_key": "APP_KEY",
"app_secret": "APP_SECRET",
"oauth_token": "OAUTH_TOKEN",
"oauth_token_secret": "OAUTH_TOKEN_SECRET"
}
}
}
Command line options
In general,
-c
: the config file for Twitter API keys-o
: the output folder (where you want to hold your data)-cmd
: the command you want to run-cc
: the config file for the command (each command often needs different config files, see examples below)-wait
: waitx
secs between calls (only in REST API access)
Streaming API access
Public sample
- statuses/sample
-cmd
:sample
(this is default)
# Get public tweets using streaming API
python twitter_streamer.py -c ../twittertracker-config/config_i0mf0rmer01.json -o /mnt/data2/twitter/sample/ -cmd sample
geo
Filter by - statuses/filter
-cmd
:locations
-cc
: e.g.,test_data/geo/US_BY_STATE_1.json
# Streaming API: get tweets within geo boundries defined in -cc test_data/geo/US_BY_STATE_1.json
python twitter_streamer.py -c ../twittertracker-config/config_i0mf0rmer02.json -o /mnt/data2/twitter/US_BY_STATE -cmd locations -cc test_data/geo/US_BY_STATE_1.json
REST APIs
Search and monitor a list of keywords
- search/tweets
-cmd
:search
-cc
:test_data/search.json
{
"keyword_list_0":{
"geocode":null,
"terms":[
"\"cervarix\"",
"\"cervical cancer\"",
"\"cervical #cancer\"",
"\"#cervical cancer\"",
"\"cervicalcancer\"",
"\"#cervicalcancer\""
],
"since_id":1,
"querystring":"(\"cervarix\") OR (\"cervical cancer\") OR (\"cervical #cancer\") OR (\"#cervical cancer\") OR (\"cervicalcancer\") OR (\"#cervicalcancer\")"
},
"keyword_list_1": {
"geocode": [
"dona_ana_nm",
"32.41906196127472,-106.82334114385034,51.93959956432837mi"
],
"querystring": "(\"cancer #cervical\") OR (\"cancercervical\") OR (\"#cancercervical\")",
"since_id": 0,
"terms": [
"cancer #cervical",
"cancercervical",
"#cancercervical"
]
}
}
# Search using a search config file
python twitter_tracker.py -c ../twittertracker-config/config_i0mf0rmer08.json -cmd search -o data/search_query -cc test_data/search.json -wait 5
- It will output the file into a folder with the current timestamp ('YYYYYMMDD') with a filename derived from md5(querystring).
- This one has no end, it will continuously query Twitter for any new tweets matching the query.
- The reason that I'm using
search/tweets
rather than the streaming APIstatuses/filters
(with thetrack
option) is that often time I want to get old tweets as well (even through it's just a few days old. Twitter only provide roughly a week old tweets when you do your search; whilestatues/filters
does not provide anyold
tweets at all). - The other caveat is that you can only track a limited number of keywords with
statues/filter
. So, if you have a lot to track, you will need to have a lot of separate instances, each tracking different part of the keywords. - With
search/tweets
, you can just search a portion of the keyword list at a time (when this happens take a look at thetest_scripts/generate_search_json.py
, which break a long list of keywords down into small portions, and generate the necessary config file for this). - Note that you can also set the
geocode
field to constrain the search within that areas.
Monitor and fetch users' timelines
- statuses/user_timelines
-cmd
:user_timelines
-cc
:test_data/user_timelines.json
{
"2969995619":{
"remove":false,
"user_id":2969995619,
"since_id":1
}
}
remove
is used to track users whose timelines cannot be pulled (e.g., private, etc.), and it will not crawlremoved
user ids.
# Monitor and fetch users' timelines
python twitter_tracker.py -c ../twittertracker-config/config_i0mf0rmer08.json -cmd user_timelines -o data/timelines -cc test_data/user_timelines.json -wait 5
Get tweets by a list of ids
- statues/lookup
-cmd
:tweets_by_ids
-cc
: see below
{"current_ix": 0, "tweet_ids": ["911333326765441025", "890608763698200577"]}
- It grabs upto 100 (per Twitter API limit) number of tweets from the
tweet_ids
list. - It assumes the
tweet_ids
is unique, and if it stops (e.g., 'CTRL+C', it will remember itcurrent_ix
, when you restart, it starts from there)
python twitter_tracker.py -c ../twittertracker-config/config_i0mf0rmer08.json -o data/tweets_by_ids -cmd tweets_by_ids -cc test_data/tweet_ids.json
Get tweets by an id range
- statues/lookup
-cmd
:tweets_by_id_range
-cc
: see below
{"end_id": 299, "current_id": 0}
- We can use this to fetch historical data, e.g,.
search_history.json
as shown above, which starts at tweet_id = 0, and fetch 100 tweets in each iteratation and move the current_id to += 100, until it reachesend_id
. Note, it does NOT fetchtweet_id == end_id
(up to end_id - 1)
python twitter_tracker.py -c ../twittertracker-config/config_i0mf0rmer08.json -o data/tweets_id_range -cmd tweets_by_id_range -cc test_data/tweets_id_range.json
Get user objects by user ids
- users/lookup
-cmd
:users_by_ids
-cc
: see blow
{"current_ix": 0, "users": ["2969995619"]}
python twitter_tracker.py -c ../twittertracker-config/config_i0mf0rmer08.json -o data/users_by_ids -cmd users_by_ids -cc test_data/user_ids.json
Get user objects by screen names
- users/lookup
-cmd
:users_by_screen_names
-cc
: see blow
{"current_ix": 0, "users": ["meetpacific"]}
python twitter_tracker.py -c ../twittertracker-config/config_i0mf0rmer08.json -o data/users_by_screen_names -cmd users_by_screen_names -cc test_data/user_screen_names.json
Collect ['/friends/ids', '/friends/list', '/followers/ids', '/followers/list']
{"current_ix": 0, "users": ["2969995619"]}
# /friends/ids
python twitter_tracker.py -c ../twittertracker-config/config_i0mf0rmer08.json -o data/friends_ids -cmd '/friends/ids' -cc test_data/user_ids.json
# /followers/ids
python twitter_tracker.py -c ../twittertracker-config/config_i0mf0rmer08.json -o data/followers_ids -cmd '/followers/ids' -cc test_data/user_ids.json
*/ids
only fetchesids
(5,000 at a time), while*/list
fetches the actualuser
objects (100 at a time).- Note the differences between this and fetching user timelines.
- In
timeline
, we want to keep tracking the users and get their latest tweets; while in this case, we only care about a snapshot of their relations. So, this will stop when it has looped through the entire list.
- In
place
by either a query
or ip
Search for - geo/search
-cmd
:collect_places_by_[query/ip]
-cc
: see blow
{"current_ix": 0, "places": ["Gainesville, FL", "Shanghai, China"]}
{"current_ix": 0, "places": ["74.125.19.104"]}
# Search for `place` objects based on place names (query)
python twitter_tracker.py -c ../twittertracker-config/config_i0mf0rmer08.json -o data/places_by_queries -cmd places_by_queries -cc test_data/place_names.json
# Search for `place` based on `ip` addresses
python twitter_tracker.py -c ../twittertracker-config/config_i0mf0rmer08.json -o data/places_by_ips -cmd places_by_ips -cc test_data/ips.json
- Note, you will need a user context for these API endpoints (which means you have to provide
OAUTH_TOKEN
andOAUTH_TOKEN_SECRET
in your config).
Datasets
Twitter license (or at least the company's position on this) does not allow me redistribute the crawled data (e.g., someone asked the question a while back: https://dev.twitter.com/discussions/8232). If you want to get a hand on this dataset (e.g., through collaboration), contact me at [email protected]. But, here is what I have:
- Random sample since 2014 (update: since 2009): I have been crawling tweets using GET statuses/sample since 2014, nonstop... except a few days the server went down...
- Tweets within US by states: Using POST statuses/filter with a
locations
filter by US states, since 10/16/2016. - Tweets related to HPV: HPV related tweets using keywords such as “Human Papillomavirus”, “HPV”, “Gardasil” and “Cervarix” with the Twitter Search API, since 02/2016 (as of today, 2/18/2017, it is still running). I do have a similar dataset from 11/2/2015 till 02/2016, but that's from a friend.
- Tweets related to transgender: Tweets collected using keywords related to transgender (e.g., trans*, transmale, etc.) between 01/17/2015 and 05/12/2015; and then user timelines of whom are self-identified as trans. This is published here, "Hicks A, Hogan WR, Rutherford M, Malin B, Xie M, Fellbaum C, Yin Z, Fabbri D, Hanna J, Bian J. Mining Twitter as a First Step toward Assessing the Adequacy of Gender Identification Terms on Intake Forms. AMIA Annu Symp Proc. 2015;2015:611-620. PMID: 26958196."
- Tweets related to lynch syndrome: Bian J, Zhao Y, Salloum RG, Guo Y, Wang M, Prosperi M, Zhang H, Du X, Ramirez-Diaz LJ, He Z, Sun Y. Using Social Media Data to Understand the Impact of Promotional Information on Laypeople’s Discussions: A Case Study of Lynch Syndrome. J Med Internet Res 2017;19(12):e414"
- A few other MISC data sets
License
The MIT License (MIT) Copyright (c) 2018 Jiang Bian ([email protected])
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.