Python: TwitterSearch

This library allows you easily create a search through the Twitter API without having to know too much about the API details. Based on such a search you can even iterate throughout all tweets reachable via the Twitter Search API. There is an automatic reload of the next pages while using the iteration. TwitterSearch was developed as part of an interdisciplinary project at the Technische Universität München. Reasons to use TwitterSearch

Well, because it can be quite annoying to always parse the search url together and a minor spelling mistake is sometimes hard to find. Not to mention the pain of getting the next page of the results. Why not centralize this process and concentrate on the more important parts of the project?

More than that, TwitterSearch is:

• pretty small (around 500 lines of code currently)
• pretty easy to use, even for beginners
• pretty good at giving you all available information (including meta information)
• pretty iterable without any need to manually reload more results from the API
• pretty wrong values of API arguments are to raise an exception. This is done before the API gets queried and therefore helps to avoid to reach Twitters’ limitations by obviously wrong API calls
• pretty friendly to Python >= 2.7 and Python >= 3.2
• pretty pretty to look at :)

Installation

TwitterSearch is also available on pypi and therefore can be installed via

pip install TwitterSearch

or

easy_install TwitterSearch

If you’d like to work with bleeding edge versions you’re free to clone the devel branch. A manual installation can be done doing by downloading or cloning the repository and running

python setup.py install.

Search Twitter

Everybody knows how much work it is to study at a university. So why not take a small shortcut? So in this example we assume we would like to find out how to copy a doctorate thesis in Germany. Let’s have a look what the Twitter users have to say about Mr Guttenberg.

from TwitterSearch import *

tso = TwitterSearchOrder() # create a TwitterSearchOrder object
tso.set_keywords(['ahok']) # let's define all words we would like to have a look for
tso.set_language('id') # we want to see Indonesian tweets only
tso.set_include_entities(True) # and don't give us all those entity information

# it's about time to create a TwitterSearch object with our secret tokens
ts = TwitterSearch(
    consumer_key = ,
    consumer_secret = ,
    access_token = ,
    access_token_secret = 
    )

# this is where the fun actually starts :)
for tweet in ts.search_tweets_iterable(tso):
    print( '@%s tweeted: %s' % ( tweet['user']['screen_name'], tweet['text'] ) )

Hasilnya kira-kira sebagai berikut,

@Babannic2 tweeted: RT @Beritasatu: Pendeta Stephen Tong Sebut Jokowi dan Ahok Tipe Pemimpin Antikorupsi https://t.co/YJomhnXBVW
@JSJkrysx tweeted: Sekjen PDIP: Ahok-Djarot Konkret, Paslon Lain Retorika https://t.co/Vs6KKYSHiV
@FirmanRenold tweeted: RT @jokoanwar: “Rakyat adalah BOS kami. Kami adalah pelayan rakyat.”

—Ahok https://t.co/cKkR3kOcbK

@Takviri tweeted: RT @AtunSumiatun: Ahok bermain tanpa beban. https://t.co/uyRBQ3j0aS
@NKRIndonesia79 tweeted: Siapapun yang "melawan" Ahok pasti jadi salah? Sepertinya iya... https://t.co/gPyn8HpnkP
@antonkerenz1 tweeted: RT @mantriss: Wah hebat pendukung Pak Ahok bicara kesopanan .... Semoga betul2 sudah berubah.... 😃😃 https://t.co/sh1DDadg5K
@tom_psp tweeted: RT @Beritasatu: #PopulerB1 1: Diajak Sandi Keroyok Ahok, Sylvi Tak Paham https://t.co/0KxkEgr8NJ https://t.co/CGj6qVUHvJ

Access User Timelines

Alternatif lain adalah mengakses timeline seorang user saja, caranya,

from TwitterSearch import *

tuo = TwitterUserOrder('onnowpurbo') # create a TwitterUserOrder

# it's about time to create a TwitterSearch object with our secret tokens
ts = TwitterSearch(
    consumer_key = ,
    consumer_secret = ,
    access_token = ,
    access_token_secret = 
    )

# start asking Twitter about the timeline
for tweet in ts.search_tweets_iterable(tuo):
	print( '@%s tweeted: %s' % ( tweet['user']['screen_name'], tweet['text'] ) )

Hasilnya kira-kira,

@onnowpurbo tweeted: alhamdulillah :) ...  https://t.co/JRQotgThNF
@onnowpurbo tweeted: keren! ... https://t.co/xuCqpVJMAd
@onnowpurbo tweeted: sama2 lihat aktifitas di pulpstone kayanya menarik nih .. https://t.co/PN3nnf0ztW
@onnowpurbo tweeted: openwrt dulunya utk router, cuma dengan kemampuan yang ada bisa utk macem2 :) .. https://t.co/3VZgng1RLd
@onnowpurbo tweeted: bisa diskusi via email onno@indo.net.id, beraba besar komunitas pulpstone? https://t.co/GEksxU49Zy

Detail dari TwitterSearch?

Jika Anda ingin mendapatkan informasi lebih lanjut tentang bagaimana TwitterSearch bekerja secara internal dan bagaimana menggunakannya dengan semua kemungkinan itu kita lihat dokumentasi terbaru. Sebuah changelog juga tersedia dalam repositori https://github.com/ckoepp/TwitterSearch/tree/master/docs Update ke 1.0.0 dan yang lebih baru

If you’re upgrading from a version < 1.0.0 be aware that the API changed! As part of the process to obtain PEP-8 compatibility all methods had to be renamed. The code changes to support the PEP-8 naming scheme are trivial. Just change the old method naming scheme from setKeywords(...) to the new one of set_keywords(...).

Apart from this issue, four other API changes were introduced with version 1.0.0:

• simplified proxy functionality (no usage of dicts but plain strings as only HTTPS proxies can be supported anyway)
• simplified geo-code parameter (TwitterSearchOrder.set_geocode(...,metric=True) renamed to set_geocode(...,imperial_metric=True))
• simplified TwitterSearch.get_statistics() from dict to tuple style ({'queries':<int>, 'tweets':<int>} to (<int>,<int>))
• additional feature: timelines of users can now be accessed using the new class TwitterUserOrder

In total those changes can be done quickly without browsing the documentation.

If you’re unable apply those changes, you might consider using TwitterSearch versions < 1.0.0. Those will stay available through pypi and therefore will be installable in the future using the common installation methods like pip install -I TwitterSearch==0.78.6. Using the release tags is another easy way to navigate through all versions of this library.

Referensi

• https://pypi.python.org/pypi/TwitterSearch/
• https://github.com/ckoepp/TwitterSearch/tree/master/docs