Calculating statistics for the number of steemit users using python

pinkwonder(61)

Published in

#steemit

Words

186

Reading

1 min

Listen

Play

I want to experiment with Steemit API, and I decide to create a small script for getting information about users from blockchain.

To collect statistics, I used the Python language, Jupyter Notebook, Steem-Python library to access the Steemit blockchain API and Matplotlib for plotting.

The installation of libraries is trivial with standard Python pip utility (an example can be found here), so I will immediately give examples of code to collect data.

For the beginning, we need to import modules and create an instance of Steem class.

from steem import *
from steem.instance import set_shared_steemd_instance
nodes = [
'https://steemd.steemitstage.com/', # working nodes can be changed in future
]
mySteem = Steem(nodes=nodes)
set_shared_steemd_instance(mySteem.steemd) # need to set instance for future operations with API.

Next, we download all steemit accounts' usernames.

def downloadAllUsernames():
    batch = mySteem.lookup_accounts(-1, 1000) #1000 is maximum names for one batch request
    i = 0
    while len(batch) != 1:
        batch = [x+"\n" for x in batch]
        with open("data/usernames%d.txt"%i, "wt") as f:
            f.writelines(batch)
        i += 1
        batch = mySteem.lookup_accounts(batch[-1], 1000)
downloadAllUsernames() # actually call the function

We save usernames to text files for getting intermediate results, so we can pause script and continue running it later.

After this step finished, we can collect extra information for all users, for example, we can retrieve creation date and posts count for every user.

import time
def applyToAllAccounts(func, lastUser = -1):
    batch = mySteem.lookup_accounts(lastUser, 1000)
    i = 0
    while len(batch) != 1:
        func(batch)
        time.sleep(0.5) #pause for waiting and not get heavy requests to node, otherwise node return fake accounts instead real
        i += 1
        batch = mySteem.lookup_accounts(batch[-1], 1000)

Collecting all data

import datetime, dateutil.parser
def calcPerMonth(accList):
    def convertDataToYearMonthString(date):
        d = dateutil.parser.parse(date)
        return d.strftime('%Y/%m')
    accs = mySteem.get_accounts(accList)
    #debug log
    print(accs[0]["name"])
    for acc in accs:
        created = acc["created"]
        isMiner = acc["mined"]
        isActive = acc["post_count"] > 0
        createdMonth = convertDataToYearMonthString(created)
        if createdMonth not in accsPerMonthCount:
            accsPerMonthCount[createdMonth] = 0
            accsPerMonthActive[createdMonth] = 0
            accsPerMonthMiners[createdMonth] = 0
        accsPerMonthCount[createdMonth] += 1
        if isMiner:
            accsPerMonthMiners[createdMonth] += 1
        if isActive:
            accsPerMonthActive[createdMonth] += 1
accsPerMonthCount  = {} #total users
accsPerMonthActive = {} #users with active posts
accsPerMonthMiners = {} #users with "mined" flag
applyToAllAccounts(calcPerMonth)

This step can take 30-60 minutes.

And, finally, we can display the diagrams with the number of accounts by month of creation.

def drawNewUsersPlot(dictToPlot, title):
    #del dictToPlot["1970/01"]
    listItems = list(dictToPlot.items())
    listItems.sort()
    dates, users = tuple(zip(*listItems))
    totalUsers = [sum(users[:i]) for i in range(1, len(users)+1)]
    plt.close('all')
    fig, ax = plt.subplots(1, figsize=(20, 10))
    plt.bar(range(len(totalUsers)), totalUsers, align='center', width = 0.9)
    plt.xticks(range(len(dates)), dates)
    fig.autofmt_xdate()
    plt.title(title)
    plt.show()
drawNewUsersPlot(accsPerMonthCount, "Steemit total accounts count per month")
drawNewUsersPlot(accsPerMonthActive, "Steemit active accounts count per month")