abstract_solana

Python-first Solana SDK — schema-driven RPC bodies, DB-backed response caching, explicit rate-limiting with fallback URLs, and PumpFun bonding-curve math. No magic defaults. No hidden globals.

---

What It Is

abstract_solana is a structured Python wrapper around the Solana JSON-RPC API. It treats every RPC call as a typed schema, every response as a cacheable record, and every endpoint as an explicitly wired URL — not something discovered at runtime.

The design priorities, in order:

1. Schemas over ad-hoc dicts — each RPC method maps to a dedicated body-builder function with typed parameters and validated conversions.
2. Registries over globals — RPC endpoints, DB table configs, and rate-limit state are held in explicit registries, not module-level singletons.
3. DB caching over redundant calls — responses are persisted to PostgreSQL (JSONB) keyed by the natural lookup column (pubkey, slot, commitment, etc.). Repeat calls hit the DB first.
4. Explicit rate limiting with URL fallback — a RateLimiter tracks per-method call counts and automatically falls back to a secondary RPC URL on 429s.
5. PumpFun as a first-class citizen — bonding curve derivation and associated token account resolution are built in, not bolted on.

---

Package Structure

abstract_solana/
├── __init__.py                  # re-exports all submodules
├── abstract_rpcs/
│   ├── get_body.py              # per-method RPC body builders
│   ├── db_templates.py          # DB table configs (CREATE, INSERT, SELECT per method)
│   ├── solana_rpc_client.py     # get_rpc_dict, make_call, abstract_solana_rate_limited_call
│   └── rate_limiter.py          # RateLimiter — tracks usage, owns URL selection
├── pumpFun/
│   ├── bondingCurves.py         # derive_bonding_curve, derive_bonding_curve_accounts
│   ├── pubKeyUtils.py           # get_pubkey, pubkey_find_program_address
│   └── genesis_functions.py
├── rpc_utils/                   # lower-level HTTP/RPC helpers
└── abstract_solana_utils/       # shared utility functions

---

Installation

pip install abstract_solana

Dependencies: solana, solders, abstract_solcatcher, abstract_utilities

---

Core Modules

abstract_rpcs — The RPC Engine

This is the heart of the package. It composes three responsibilities cleanly:

Body construction (get_body.py)

Each Solana RPC method (getBalance, getAccountInfo, getBlock, etc.) has a dedicated Python function that accepts typed arguments, validates them, converts pubkeys/signatures to their proper types, fills in defaults, and returns a well-formed JSON-RPC body dict.

from abstract_solana.abstract_rpcs import get_rpc_dict

body = get_rpc_dict("getBalance", pubkey="YourWalletAddressHere")
# {'jsonrpc': '2.0', 'method': 'getBalance', 'params': [...], 'id': 1}

get_rpc_dict uses inspect.signature to discover each body-builder's parameters, then runs get_conversions to normalize and type-coerce every argument. Pubkeys and signatures are converted via get_pubkey / get_sigkey — no raw strings passed through.

DB caching (db_templates.py)

get_insert_list() returns a registry of table configs — one per RPC method — each specifying:

Key	Purpose
tableName	PostgreSQL table name
columnSearch	The lookup key (e.g. pubkey, slot, commitment)
insertName	Column name for the response payload
type	Solana type name for the response
searchQuery	Parameterized SELECT
insertQuery	Parameterized INSERT
table	CREATE TABLE IF NOT EXISTS DDL

Covered methods include: isconnected, getbalance, getaccountinfo, getaccountinfojsonparsed, getblockcommitment, getblocktime, getclusternodes, getblock, getrecentperformancesamples, getblockheight, getblocks, and many more.

Rate-limited execution (solana_rpc_client.py + rate_limiter.py)

from abstract_solana.abstract_rpcs import abstract_solana_rate_limited_call

response = abstract_solana_rate_limited_call("getBalance", pubkey="YourAddressHere")

Under the hood:

1. Builds the request body via get_rpc_dict.
2. Asks rate_limiter.get_url(method) which URL to use — primary or fallback.
3. Makes the POST via make_call (which injects headers and handles retries).
4. Logs the response and retry_after header back to the rate limiter.
5. If the primary URL returns 429, automatically retries against the secondary URL.

The RateLimiter owns all URL state. Nothing is a module-level global.

---

pumpFun — Bonding Curve Math

PumpFun's bonding curves are Program Derived Addresses (PDAs). This module wraps the derivation cleanly.

from abstract_solana.pumpFun import (
    derive_bonding_curve_accounts,
    get_bonding_curve,
    get_associated_bonding_curve,
    isOnCurve,
)

accounts = derive_bonding_curve_accounts("TokenMintAddressHere")
# {
#   'bonding_curve': <Pubkey>,
#   'associated_bonding_curve': <Pubkey>
# }

# Check validity before any downstream call
if isOnCurve("TokenMintAddressHere"):
    curve = get_bonding_curve("TokenMintAddressHere")

derive_bonding_curve_accounts guards against invalid mints: if mint.is_on_curve() returns False, it returns an empty dict rather than propagating a bad PDA derivation.

derive_bonding_curve and derive_associated_bonding_curve both accept an optional programId — defaulting to PUMP_FUN_PROGRAM — so the same logic works against custom forks of the program without monkey-patching.

---

Type Conversion Pipeline

A consistent type-coercion layer runs across all RPC calls:

# pubkeys and signatures are detected by key name, not by caller convention
pubkeys    = ['address', 'account', 'pubkeys', 'pubkey', 'mint', 'owner', 'delegate']
signatures = ['until', 'before', 'tx_sig', 'signature', 'signatures']

If a kwarg key is in pubkeys, the value is passed through get_pubkey. If it's in signatures, through get_sigkey. Lists are handled element-wise. This means callers can pass raw strings; the SDK handles wrapping into solders/solana-py types.

---

Supported RPC Methods (DB-Cached)

The DB template registry covers the following Solana RPC methods out of the box:

isConnected · getBalance · getAccountInfo · getAccountInfoJsonParsed · getBlockCommitment · getBlockTime · getClusterNodes · getBlock · getRecentPerformanceSamples · getBlockHeight · getBlocks · and more defined in get_insert_list()

Each maps to its own PostgreSQL table with an auto-generated schema. Tables are created idempotently on first use.

---

Design Notes

Why PostgreSQL for caching? Solana RPC data is structured and queryable. JSONB lets you store the full response without schema churn while still indexing on lookup keys. A simple WHERE pubkey = %s is cheaper than a repeat RPC call on a rate-limited endpoint.

Why explicit fresh_call flags? Each table config carries fresh_call: True, making cache-bypass an explicit, per-method decision rather than a hidden TTL policy. You can override this per call — it's a field, not a behavior.

Why inspect.signature for param discovery? Body builders are plain functions. Using inspect.signature to drive get_conversions means adding a new RPC method is just adding a function — no registration boilerplate, no decorator magic.

---

Author & License

Author: putkoff — partners@abstractendeavors.com
License: MIT
Version: 0.0.2.141 (Alpha)