Fast transactions paging using block_time

endpoints/get_address_transactions.py

@@ -6,6 +6,7 @@
 from pydantic import BaseModel
 from sqlalchemy import text, func
 from sqlalchemy.future import select
+from starlette.responses import Response
 
 from dbsession import async_session
 from endpoints import sql_db_only
@@ -129,6 +130,70 @@ async def get_full_transactions_for_address(
                                          resolve_previous_outpoints)
 
 
+@app.get("/addresses/{kaspaAddress}/full-transactions-page",
+         response_model=List[TxModel],
+         response_model_exclude_unset=True,
+         tags=["Kaspa addresses"])
+@sql_db_only
+async def get_full_transactions_for_address_page(
+        response: Response,
+        kaspaAddress: str = Path(
+            description="Kaspa address as string e.g. "
+                        "kaspa:pzhh76qc82wzduvsrd9xh4zde9qhp0xc8rl7qu2mvl2e42uvdqt75zrcgpm00",
+            regex="^kaspa\:[a-z0-9]{61,63}$"),
+        limit: int = Query(
+            description="The max number of records to get. "
+                        "For paging combine with using 'before' from oldest previous result, "
+                        "repeat until an **empty** resultset is returned."
+                        "The actual number of transactions returned can be higher if there are transactions with the same block time at the limit.",
+            ge=1,
+            le=500,
+            default=50),
+        before: int = Query(
+            description="Only include transactions with block time before this (epoch-millis)",
+            ge=0,
+            default=0),
+        fields: str = "",
+        resolve_previous_outpoints: PreviousOutpointLookupMode =
+        Query(default="no",
+              description=DESC_RESOLVE_PARAM)):
+    """
+    Get all transactions for a given address from database.
+    And then get their related full transaction data
+    """
+
+    async with async_session() as s:
+        # Doing it this way as opposed to adding it directly in the IN clause
+        # so I can re-use the same result in tx_list, TxInput and TxOutput
+        before = int(time.time() * 1000) if before == 0 else before
+        tx_within_limit_before = await s.execute(select(TxAddrMapping.transaction_id,
+                                                        TxAddrMapping.block_time)
+                                                 .filter(TxAddrMapping.address == kaspaAddress)
+                                                 .filter(TxAddrMapping.block_time < before)
+                                                 .limit(limit)
+                                                 .order_by(TxAddrMapping.block_time.desc())
+                                                 )
+
+        tx_ids_and_block_times = [(x.transaction_id, x.block_time) for x in tx_within_limit_before.all()]
+        tx_ids = {tx_id for tx_id, block_time in tx_ids_and_block_times}
+        oldest_block_time = tx_ids_and_block_times[-1][1]
+
+        if len(tx_ids_and_block_times) == limit:
+            # To avoid gaps when transactions with the same block_time are at the boundry between pages.
+            # Get the time of the last transaction and fetch additional transactions for the same address and timestamp
+            tx_with_same_block_time = await s.execute(select(TxAddrMapping.transaction_id)
+                                                      .filter(TxAddrMapping.address == kaspaAddress)
+                                                      .filter(TxAddrMapping.block_time == oldest_block_time))
+            tx_ids.update([x for x in tx_with_same_block_time.scalars().all()])
+
+    response.headers["X-Current-Page"] = str(len(tx_ids))
+    response.headers["X-Oldest-Epoch-Millis"] = str(oldest_block_time)
+
+    return await search_for_transactions(TxSearch(transactionIds=list(tx_ids)),
+                                         fields,
+                                         resolve_previous_outpoints)
+
+
 @app.get("/addresses/{kaspaAddress}/transactions-count",
          response_model=TransactionCount,
          tags=["Kaspa addresses"])