Indexer For Blockchain Analysis

pip install pandas
pip install dateutil
pip install py-algorand-sdk

import os
import json
from time import sleep
from datetime import timezone

from dateutil.rrule import HOURLY, rrule
from dateutil.parser import parse
import pandas as pd

from algosdk.v2client import indexer

pd.set_option('display.max_columns', None)
pd.set_option('display.max_rows', None)

indexer_token = os.environ.get('TOKEN')
myindexer = indexer.IndexerClient(indexer_token=indexer_token, indexer_address='https://mainnet-algorand.api.purestake.io/idx2')

def get_txn_response(start_time, end_time):
    """
    Returns all transactions added to the blockchain between 'start_time' and 'end_time'

    """
    # The indexer expacting time inputs to be in RFC 3339 format
    start_time = start_time.astimezone(timezone.utc).isoformat('T')
    end_time = end_time.astimezone(timezone.utc).isoformat('T')

    nexttoken = ""
    numtx = 1

    responses = []

    # Retrieve up-to 1000 transactions at each request.
    while numtx > 0:
        response = myindexer.search_transactions(start_time=start_time, end_time=end_time,
                                                 next_page=nexttoken, limit=1000)
        transactions = response['transactions']
        responses += transactions
        numtx = len(transactions)
        if numtx > 0:
            # pointer to the next chunk of requests
            nexttoken = response['next-token']
    return responses

start = "20200801T000000"
end = "20200808T000000"

# get a list of dates between start date and end date with 6 hours intervals
dates = list(rrule(HOURLY,
                   dtstart=parse(start),
                   until=parse(end),
                   interval=6))

responses = []
for start_time, end_time in zip(dates[:-1], dates[1:]):
    responses += get_txn_response(start_time, end_time)
df = pd.DataFrame(responses)
df.head()

df = pd.concat([df.drop(['asset-transfer-transaction'], axis=1),
                df['asset-transfer-transaction'].apply(pd.Series).add_suffix('-asset-txn').drop(['0-asset-txn'], axis=1, errors='ignore')], axis=1)

df = pd.concat([df.drop(['payment-transaction'], axis=1), 
                df['payment-transaction'].apply(pd.Series).add_suffix('-pay-txn').drop(['0-pay-txn'], axis=1, errors='ignore')], axis=1)

df = pd.concat([df.drop(['asset-config-transaction'], axis=1),
                df['asset-config-transaction'].apply(pd.Series).add_suffix('-asst-cnfg-txn').drop(['0-asst-cnfg-txn'], axis=1, errors='ignore')], axis=1)

df = pd.concat([df.drop(['signature'], axis=1), 
                df['signature'].apply(pd.Series).add_suffix('-sig').drop(['0-sig'], axis=1, errors='ignore')], axis=1)

# format the unix seconds to human readble date format
df['date'] = pd.to_datetime(df['round-time'], unit='s').dt.date

df.head()

len(df)

df['tx-type'].value_counts()

df['amount-pay-txn'].sum()

df['day'] = pd.to_datetime(df['date']).dt.day_name()
df['day'].value_counts().plot()

df['asset-id-asset-txn'].nunique()

df['asset-id-asset-txn'].value_counts()

def get_txn_response():
    nexttoken = ""
    numtx = 1

    responses = []

    # Retrieve up-to 1000 assets at each request.
    while numtx > 0:
        response = myindexer.search_assets(next_page=nexttoken)
        assets = response['assets']
        responses += assets
        numtx = len(assets)
        if numtx > 0:
            # pointer to the next chunk of requests
            nexttoken = response['next-token']
    return responses

assets = pd.DataFrame(get_txn_response())
assets.head()

assets = pd.concat([assets.drop(['params'], axis=1),
                    assets['params'].apply(pd.Series)], axis=1)
assets.head()

assets_counts = pd.DataFrame(df['asset-id-asset-txn'].value_counts()).join(assets[['index', 'name', 'unit-name']].set_index('index'))

assets_counts.head()

pd.DataFrame(assets_counts.groupby('unit-name')['asset-id-asset-txn'].sum()).nlargest(10, 'asset-id-asset-txn').plot()