Ton Analytics code

Various code for TON blockchain analysis.

ETL

Initial blockchain data extraction is performed with ton-indexer. PostgreSQL is not good enough for huge analytical workload so data loaded into S3-based datalake. Airflow is used for incremental ETL process:

• E: extract data from psql sliced by date intervals
• T: convert results to parquet file format
• L: load it into S3 object storage

Datalake tables:

Table	Prefix
accounts	dwh/staging/accounts/date=YYYYMM/
transactions	dwh/staging/transactions/date=YYYYMM/
messages	dwh/staging/messages/date=YYYYMM/

After each incremental upload log entry created on psql table ``increment_state```

Parsers

Smart-contract data/messages parser for extracting meaningful data, for example, NFT owners. Example:

POSTGRES_URI=postgres://postgres:pass@localhost:5432/ton_index \
  python3 contracts_parser.py -c jetton -d cache/ -e http://localhost:9090/execute \
  jetton_contract_data.csv jetton_parsed.csv

Supported contract types:

• DNS contract (TEP-81)
• Jetton master contract (TEP-74) with TEP-64 support

Utils

• contracts_executor - wrapper around ton-contract-executor for executing arbitrary get methods on contracts providing code and data cells. Used in parser to extract information.

Analysis

Some analysis scripts:

• TON_events_plot.ipynb - plot gantt chart visualising TON scam (for this research)