Paginate get_cashtoken_addresses, get_unspent, and get_transactions

[yashasvi-ranawat]

Problem

Unbounded search_output queries against the Chaingraph endpoint fail with a database error for addresses or tokens with many UTXOs/transactions. The previous implementation fetched all results in a single request with DEFAULT_TIMEOUT=5s.

Changes

• ChaingraphAPI.py: Added CASHTOKEN_ADDRESSES_PAGE_SIZE, UNSPENT_PAGE_SIZE, and TRANSACTIONS_PAGE_SIZE (all 1000). get_cashtoken_addresses, get_unspent, and get_transactions now use $limit/$offset GraphQL variables and loop with limit/offset pagination until a page returns fewer than PAGE_SIZE rows.
• services.py: Added LONG_DEFAULT_TIMEOUT = 60 and switched all three methods to use it (was DEFAULT_TIMEOUT = 5).

Real-world tests

Moria USD (b38a33f750f84c5c169a6f23cb873e6e79605021585d4f3408789689ed87f366): fetched 4,710 UTXOs → 595 unique addresses in ~57 seconds — 5 pages of 1,000 at 2–19 seconds per page.

bitcoincash:qpnezxnfun4vgwypja3j4f4fzjdyf7ddgvfwvza6f5: fetched 18,823 UTXOs in ~35 seconds via Chaingraph (19 pages) vs 4 seconds via Fulcrum; 41,274 transactions in 3 seconds via Fulcrum.

Note: For plain get_unspent and get_transactions on high-activity addresses, the Fulcrum endpoint is significantly faster — it is purpose-built for address/scripthash lookups. Chaingraph's value is in richer queries such as get_cashtoken_addresses, where token filtering and cross-referencing on-chain data aren't expressible in the Electrum protocol.

[Copilot]

Pull request overview

This PR addresses Chaingraph query failures for high-activity addresses/tokens by adding limit/offset pagination to previously unbounded GraphQL queries, and increasing timeouts to accommodate multi-page fetches.

Changes:

• Added page-size constants and implemented limit/offset pagination loops for Chaingraph get_transactions, get_unspent, and get_cashtoken_addresses.
• Introduced a longer default timeout for potentially large result sets and applied it to the above calls via NetworkAPI.
• Added a docstring note warning that cashtoken holder queries may take over a minute due to pagination.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
bitcash/network/services.py	Adds a long timeout constant and uses it for large-result calls; updates docstring note for cashtoken address lookup.
bitcash/network/APIs/ChaingraphAPI.py	Adds pagination parameters and loops to split large Chaingraph result sets into pages.

Comments suppressed due to low confidence (4)

bitcash/network/APIs/ChaingraphAPI.py:239

• The current de-duplication strategy sorted(set(transactions), key=lambda x: transactions.index(x)) is O(n^2) due to repeated index() calls. With pagination enabling tens of thousands of transactions, this can become a major bottleneck; consider an order-preserving linear-time approach (e.g., tracking a seen set while building the list).

        transactions = [_[0] for _ in transactions][::-1]
        # remove duplicates, when address pays itself, spending tx and locking
        # tx are same transactions
        transactions = sorted(set(transactions), key=lambda x: transactions.index(x))
        return transactions

bitcash/network/APIs/ChaingraphAPI.py:336

• This paginated loop queries block(limit: 1, order_by: {height: desc}, ...) on every page even though the block height is only used as a constant reference. Consider fetching the latest block height once (separate query) and then paginating only the search_output query to reduce per-page load and latency.

query GetUTXO($lb: _text!, $node: String!, $limit: Int!, $offset: Int!) {
  block(
    limit: 1
    order_by: { height: desc }
    where: { accepted_by: { node: { name: { _like: $node } } } }
  ) {

bitcash/network/APIs/ChaingraphAPI.py:360

• Offset/limit pagination is used here, but the search_output query does not specify an order_by. Without deterministic ordering, offset pagination can return overlapping pages or skip rows, causing missing UTXOs. Add a stable order_by (e.g., transaction_hash asc, output_index asc) when paginating.

      ]
    }
    limit: $limit
    offset: $offset
  ) {

bitcash/network/APIs/ChaingraphAPI.py:547

• Offset/limit pagination is used for output holders, but the output query does not specify an order_by. Without deterministic ordering, offset pagination can skip/duplicate rows between pages and produce incomplete address sets. Add a stable order_by (e.g., by transaction_hash/output_index or another unique ordering) when paginating.

      ]
    }
    limit: $limit
    offset: $offset
  ) {

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.