dccd 3.6.0__tar.gz → 3.6.2__tar.gz

{dccd-3.6.0 → dccd-3.6.2}/CHANGELOG.md

@@ -16,6 +16,27 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Removed
 
+## [3.6.2] - 2026-06-20
+
+### Fixed
+
+- Kraken adapter maps pairs by **altname** (`{base}{quote}`, with `BTC→XBT` and
+  `DOGE→XDG`) instead of legacy X/Z-prefixed codes, so OHLC and trades for modern
+  Kraken assets (TRX, DOT, BNB, …) and Dogecoin no longer fail with `Unknown asset
+  pair`. Legacy pairs are unaffected — Kraken accepts altnames universally and the
+  response is parsed by its code-key fallback. (#169)
+
+## [3.6.1] - 2026-06-19
+
+### Fixed
+
+- `ParquetStore.save()` now rejects rows whose timestamp is null or `<= 0` at the
+  storage write boundary (one central guard for every adapter and data type), so a
+  lost/epoch-0 timestamp can no longer be persisted or poison gap detection — a
+  single `TS=0` bar otherwise dragged `inventory()` `min_ts` to `1970` and inflated
+  `expected_rows`/`missing_rows` to a bogus ~89 %. Dropped rows are logged, not
+  raised, so one bad bar can't abort a good page. (#165)
+
 ## [3.6.0] - 2026-06-13
 
 ### Added

{dccd-3.6.0 → dccd-3.6.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dccd
-Version: 3.6.0
+Version: 3.6.2
 Summary: Download Crypto Currency Data — hexagonal architecture, async-first.
 Author-email: Arthur Bernard <arthur.bernard.92@gmail.com>
 License: MIT

{dccd-3.6.0 → dccd-3.6.2}/dccd/sources/kraken.py

@@ -37,28 +37,31 @@ logger = logging.getLogger(__name__)
 
 _BASE = "https://api.kraken.com/0/public"
 
+# Canonical symbol name → Kraken altname for assets whose ticker differs.
+_KRAKEN_ALIASES: dict[str, str] = {
+    "BTC": "XBT",
+    "DOGE": "XDG",
+}
+
 
 def _kraken_pair(symbol: Symbol) -> str:
-    """Convert a canonical Symbol to a Kraken REST pair string.
+    """Convert a canonical Symbol to a Kraken REST altname pair string.
+
+    Kraken accepts altnames (e.g. ``XBTUSD``, ``TRXUSD``) for all assets,
+    including modern ones that lack legacy X/Z-prefixed codes.  ``BTC`` is
+    aliased to ``XBT`` and ``DOGE`` to ``XDG`` on both base and quote.
 
     Examples
     --------
     >>> from dccd.domain.symbol import Symbol
     >>> _kraken_pair(Symbol(base='BTC', quote='USD'))
-    'XXBTZUSD'
+    'XBTUSD'
     >>> _kraken_pair(Symbol(base='ETH', quote='BTC'))
-    'XETHXXBT'
+    'ETHXBT'
     """
-    # Kraken names BTC as XBT — for the quote too, not just the base, so a
-    # crypto/crypto pair like ETH/BTC must become XETHXXBT (not XETHXBTC,
-    # which Kraken rejects with "Unknown asset pair").
-    base = "XBT" if symbol.base == "BTC" else symbol.base
-    quote = "XBT" if symbol.quote == "BTC" else symbol.quote
-    if base in ("BCH", "DASH"):
-        return f"{base}{quote}"
-    if quote in ("EUR", "USD", "CAD", "JPY", "GBP"):
-        return f"X{base}Z{quote}"
-    return f"X{base}X{quote}"
+    base = _KRAKEN_ALIASES.get(symbol.base, symbol.base)
+    quote = _KRAKEN_ALIASES.get(symbol.quote, symbol.quote)
+    return f"{base}{quote}"
 
 
 def _ws_pair(symbol: Symbol) -> str:

{dccd-3.6.0 → dccd-3.6.2}/dccd/storage/parquet.py

@@ -288,6 +288,20 @@ class ParquetStore:
         if len(df) == 0:
             return 0
 
+        # Reject bars with an invalid timestamp (null or <= 0). TS is ns UTC
+        # int64; 0 is the Unix epoch (1970) — always corrupt for crypto market
+        # data (real history starts ~2009). One such row poisons gap detection
+        # (inventory min_ts → 1970, expected_rows balloons). Drop, don't raise,
+        # so one bad bar can't abort a good page. Seen in prod: a Kraken OHLC bar
+        # with a null time parsed to 0 (audit 2026-06-19).
+        n_before = len(df)
+        df = df.filter(pl.col("TS").is_not_null() & (pl.col("TS") > 0))
+        dropped = n_before - len(df)
+        if dropped:
+            logger.warning("save(%s): dropped %d row(s) with invalid TS<=0", ds, dropped)
+        if len(df) == 0:
+            return 0
+
         fmt = self._period_fmt(ds)
         df_with_period = df.with_columns(
             pl.from_epoch("TS", time_unit="ns").dt.strftime(fmt).alias("_period")

{dccd-3.6.0 → dccd-3.6.2}/dccd/tests/v3/test_sources.py

@@ -114,11 +114,11 @@ class TestKrakenCapabilities:
         self.src = KrakenSource()
 
     def test_render_symbol(self):
-        assert self.src.render_symbol(BTC_USD) == "XXBTZUSD"
+        assert self.src.render_symbol(BTC_USD) == "XBTUSD"
 
     def test_render_symbol_crypto_quote_btc(self):
-        # ETH/BTC: BTC→XBT for the *quote* too, else Kraken rejects "XETHXBTC".
-        assert self.src.render_symbol(Symbol(base="ETH", quote="BTC")) == "XETHXXBT"
+        # ETH/BTC: BTC→XBT via _KRAKEN_ALIASES on the quote side → ETHXBT.
+        assert self.src.render_symbol(Symbol(base="ETH", quote="BTC")) == "ETHXBT"
 
     def test_ohlc_cap_history_recent(self):
         cap = self.src.capability_for(DataType.OHLC, "rest", "historical")
@@ -132,6 +132,46 @@ class TestKrakenCapabilities:
         assert cap.history == "full"
 
 
+class TestKrakenPairMapping:
+    """Unit tests for _kraken_pair altname construction."""
+
+    def test_legacy_btc_usd(self):
+        from dccd.sources.kraken import _kraken_pair
+        assert _kraken_pair(Symbol(base="BTC", quote="USD")) == "XBTUSD"
+
+    def test_legacy_eth_btc(self):
+        from dccd.sources.kraken import _kraken_pair
+        assert _kraken_pair(Symbol(base="ETH", quote="BTC")) == "ETHXBT"
+
+    def test_legacy_xrp_usd(self):
+        from dccd.sources.kraken import _kraken_pair
+        assert _kraken_pair(Symbol(base="XRP", quote="USD")) == "XRPUSD"
+
+    def test_legacy_xrp_btc(self):
+        from dccd.sources.kraken import _kraken_pair
+        assert _kraken_pair(Symbol(base="XRP", quote="BTC")) == "XRPXBT"
+
+    def test_modern_trx_usd(self):
+        from dccd.sources.kraken import _kraken_pair
+        assert _kraken_pair(Symbol(base="TRX", quote="USD")) == "TRXUSD"
+
+    def test_modern_dot_btc(self):
+        from dccd.sources.kraken import _kraken_pair
+        assert _kraken_pair(Symbol(base="DOT", quote="BTC")) == "DOTXBT"
+
+    def test_modern_bnb_usd(self):
+        from dccd.sources.kraken import _kraken_pair
+        assert _kraken_pair(Symbol(base="BNB", quote="USD")) == "BNBUSD"
+
+    def test_doge_usd(self):
+        from dccd.sources.kraken import _kraken_pair
+        assert _kraken_pair(Symbol(base="DOGE", quote="USD")) == "XDGUSD"
+
+    def test_doge_btc(self):
+        from dccd.sources.kraken import _kraken_pair
+        assert _kraken_pair(Symbol(base="DOGE", quote="BTC")) == "XDGXBT"
+
+
 class TestBybitCapabilities:
     def setup_method(self):
         self.src = BybitSource()

{dccd-3.6.0 → dccd-3.6.2}/dccd/tests/v3/test_storage.py

@@ -41,7 +41,7 @@ class TestParquetStore:
     def test_save_and_load_ohlc(self, tmp_store, ohlc_ds):
         bars = [
             OHLCBar(ts=i * 3600 * NS, open=1.0, high=2.0, low=0.5, close=1.5, volume=10.0)
-            for i in range(10)
+            for i in range(1, 11)  # start at 1 — TS=0 is the Unix epoch and is rejected
         ]
         n = tmp_store.save(ohlc_ds, bars)
         assert n > 0
@@ -53,7 +53,7 @@ class TestParquetStore:
     def test_save_and_load_trades(self, tmp_store, trades_ds):
         trades = [
             Trade(ts=i * 1000 * NS, price=50000.0 + i, amount=0.1, side="buy")
-            for i in range(5)
+            for i in range(1, 6)  # start at 1 — TS=0 is the Unix epoch and is rejected
         ]
         n = tmp_store.save(trades_ds, trades)
         assert n > 0
@@ -92,10 +92,11 @@ class TestParquetStore:
         assert any(d["exchange"] == "binance" for d in inv)
 
     def test_inventory_bytes_and_gaps(self, tmp_store, ohlc_ds):
-        # 10 hourly bars but with one hole (skip index 5) → 9 stored, 10 expected.
+        # 10 hourly bars but with one hole (skip index 6) → 9 stored, 10 expected.
+        # Start at 1 — TS=0 is the Unix epoch and is rejected by the store guard.
         bars = [
             OHLCBar(ts=i * 3600 * NS, open=1.0, high=2.0, low=0.5, close=1.5, volume=10.0)
-            for i in range(10) if i != 5
+            for i in range(1, 11) if i != 6
         ]
         tmp_store.save(ohlc_ds, bars)
         entry = next(d for d in tmp_store.inventory() if d["data_type"] == "ohlc")
@@ -114,7 +115,7 @@ class TestParquetStore:
     def test_load_with_range(self, tmp_store, ohlc_ds):
         bars = [
             OHLCBar(ts=i * 3600 * NS, open=1.0, high=2.0, low=0.5, close=1.5, volume=10.0)
-            for i in range(10)
+            for i in range(1, 11)  # start at 1 — TS=0 is the Unix epoch and is rejected
         ]
         tmp_store.save(ohlc_ds, bars)
         df = tmp_store.load(ohlc_ds, start_ns=3 * 3600 * NS, end_ns=6 * 3600 * NS)
@@ -132,11 +133,54 @@ class TestParquetStore:
                 bids=[OrderBookLevel(price=100.0, amount=1.0)],
                 asks=[OrderBookLevel(price=100.1, amount=0.5)],
             )
-            for i in range(3)
+            for i in range(1, 4)  # start at 1 — TS=0 is the Unix epoch and is rejected
         ]
         n = tmp_store.save(book_ds, snaps)
         assert n > 0
 
+    def test_save_rejects_nonpositive_ts(self, tmp_store, ohlc_ds, trades_ds):
+        """Bars with TS=0 or TS<0 are silently dropped; valid bars land on disk.
+
+        Regression guard for the 2026-06-19 prod incident: a Kraken OHLC bar
+        with a null-parsed timestamp (TS=0) inflated inventory expected_rows
+        to ~89% missing by dragging min_ts to 1970-01-01.
+        """
+        # Use timestamps well into BTC history (2017-ish, in ns).
+        base_ts = 1_500_000_000 * NS  # 2017-07-14 in ns
+
+        # OHLC: mix 3 valid bars with one TS=0 and one TS=-1.
+        bars = [
+            OHLCBar(ts=base_ts + i * 3600 * NS, open=1.0, high=2.0, low=0.5, close=1.5, volume=10.0)
+            for i in range(3)
+        ] + [
+            OHLCBar(ts=0, open=60882.4, high=60900.0, low=60800.0, close=60867.9, volume=5.0),
+            OHLCBar(ts=-1, open=1.0, high=2.0, low=0.5, close=1.5, volume=1.0),
+        ]
+        n = tmp_store.save(ohlc_ds, bars)
+
+        # Only the 3 valid rows should be counted.
+        assert n == 3
+
+        df = tmp_store.load(ohlc_ds)
+        assert len(df) == 3
+        assert df["TS"].min() > 0, "no TS<=0 row must reach disk"
+
+        # No 1970.parquet partition should exist.
+        ohlc_dir = tmp_store.directory(ohlc_ds)
+        assert not (ohlc_dir / "1970.parquet").exists()
+
+        # Trades: same guard applies — one valid trade + one with TS=0.
+        trades = [
+            Trade(ts=base_ts, price=50000.0, amount=0.1, side="buy"),
+            Trade(ts=0, price=60000.0, amount=0.2, side="sell"),
+        ]
+        n_trades = tmp_store.save(trades_ds, trades)
+        assert n_trades == 1
+
+        df_trades = tmp_store.load(trades_ds)
+        assert len(df_trades) == 1
+        assert df_trades["TS"].min() > 0
+
 
 class TestRunsStore:
     @pytest.fixture

{dccd-3.6.0 → dccd-3.6.2}/dccd/tests/v3/test_storage_extended.py

@@ -50,10 +50,11 @@ class TestMissingIntervalsPartialYear:
         assert leading, f"Expected leading gap starting at 0, got: {gaps}"
 
     def test_trailing_gap_detected(self, store: ParquetStore, ohlc_ds: DatasetId) -> None:
-        """Save bars [0, 4]; request [0, 10] → trailing gap after last bar."""
+        """Save bars [1, 5]; request [0, 10] → trailing gap after last bar."""
+        # Start at 1 — TS=0 is the Unix epoch and is rejected by the store guard.
         bars = [
             OHLCBar(ts=i * 3600 * NS, open=1.0, high=1.0, low=1.0, close=1.0, volume=1.0)
-            for i in range(5)
+            for i in range(1, 6)
         ]
         store.save(ohlc_ds, bars)
 
@@ -64,17 +65,18 @@ class TestMissingIntervalsPartialYear:
         assert trailing, f"Expected trailing gap ending at {end_ns}, got: {gaps}"
 
     def test_no_gap_when_complete(self, store: ParquetStore, ohlc_ds: DatasetId) -> None:
-        """Exact coverage → no gaps."""
+        """Exact coverage → no gaps within stored range."""
+        # Start at 1 — TS=0 is the Unix epoch and is rejected by the store guard.
         bars = [
             OHLCBar(ts=i * 3600 * NS, open=1.0, high=1.0, low=1.0, close=1.0, volume=1.0)
-            for i in range(5)
+            for i in range(1, 6)
         ]
         store.save(ohlc_ds, bars)
-        end_ns = 4 * 3600 * NS
-        gaps = store.missing_intervals(ohlc_ds, 0, end_ns)
+        end_ns = 5 * 3600 * NS
+        gaps = store.missing_intervals(ohlc_ds, 1 * 3600 * NS, end_ns)
         # May still have a trailing gap if end_ns > last bar
-        # No gap should start after the last bar within the stored range
-        assert all(g[0] >= 4 * 3600 * NS for g in gaps), f"Unexpected gap: {gaps}"
+        # No gap should start before the last stored bar
+        assert all(g[0] >= 5 * 3600 * NS for g in gaps), f"Unexpected gap: {gaps}"
 
 
 # ---------------------------------------------------------------------------
@@ -206,9 +208,10 @@ class TestFileStatsCache:
         self, store: ParquetStore, ohlc_ds: DatasetId, monkeypatch: pytest.MonkeyPatch
     ) -> None:
         """After save() (mtime changes) the cache is invalidated and new rows appear."""
+        # Start at 1 — TS=0 is the Unix epoch and is rejected by the store guard.
         bars_first = [
             OHLCBar(ts=i * 3600 * NS, open=1.0, high=1.0, low=1.0, close=1.0, volume=1.0)
-            for i in range(5)
+            for i in range(1, 6)
         ]
         store.save(ohlc_ds, bars_first)
 
@@ -228,7 +231,7 @@ class TestFileStatsCache:
         monkeypatch.undo()
         bars_second = [
             OHLCBar(ts=i * 3600 * NS, open=1.0, high=1.0, low=1.0, close=1.0, volume=1.0)
-            for i in range(5, 10)
+            for i in range(6, 11)  # non-overlapping with first batch (1..5)
         ]
         store.save(ohlc_ds, bars_second)
 
@@ -343,17 +346,18 @@ class TestFileStatsValueIdentity:
         self, store: ParquetStore, ohlc_ds: DatasetId
     ) -> None:
         """inventory() rows/min_ts/max_ts/expected_rows/missing_rows match stored data."""
-        # 10 hourly bars with index 5 missing → 9 rows, expected=10, missing=1
+        # 10 hourly bars (indices 1..10) with index 6 missing → 9 rows, expected=10, missing=1.
+        # Start at 1 — TS=0 is the Unix epoch and is rejected by the store guard.
         bars = [
             OHLCBar(ts=i * 3600 * NS, open=1.0, high=2.0, low=0.5, close=1.5, volume=10.0)
-            for i in range(10) if i != 5
+            for i in range(1, 11) if i != 6
         ]
         store.save(ohlc_ds, bars)
         entry = next(d for d in store.inventory() if d["data_type"] == "ohlc")
 
         assert entry["rows"] == 9
-        assert entry["min_ts"] == 0
-        assert entry["max_ts"] == 9 * 3600 * NS
+        assert entry["min_ts"] == 1 * 3600 * NS
+        assert entry["max_ts"] == 10 * 3600 * NS
         assert entry["expected_rows"] == 10
         assert entry["missing_rows"] == 1
         assert entry["bytes"] > 0
@@ -362,16 +366,17 @@ class TestFileStatsValueIdentity:
         self, store: ParquetStore, trades_ds: DatasetId
     ) -> None:
         """Trades inventory has no gap fields, correct rows/min_ts/max_ts."""
+        # Start at 1 — TS=0 is the Unix epoch and is rejected by the store guard.
         trades = [
             Trade(ts=i * 1000 * NS, price=50000.0, amount=0.1, side="buy")
-            for i in range(5)
+            for i in range(1, 6)
         ]
         store.save(trades_ds, trades)
         entry = next(d for d in store.inventory() if d["data_type"] == "trades")
 
         assert entry["rows"] == 5
-        assert entry["min_ts"] == 0
-        assert entry["max_ts"] == 4 * 1000 * NS
+        assert entry["min_ts"] == 1 * 1000 * NS
+        assert entry["max_ts"] == 5 * 1000 * NS
         assert entry["missing_rows"] is None
         assert entry["expected_rows"] is None
         assert entry["bytes"] > 0

{dccd-3.6.0 → dccd-3.6.2}/dccd.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dccd
-Version: 3.6.0
+Version: 3.6.2
 Summary: Download Crypto Currency Data — hexagonal architecture, async-first.
 Author-email: Arthur Bernard <arthur.bernard.92@gmail.com>
 License: MIT

{dccd-3.6.0 → dccd-3.6.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "dccd"
-version = "3.6.0"
+version = "3.6.2"
 description = "Download Crypto Currency Data — hexagonal architecture, async-first."
 readme = "README.md"
 license = { text = "MIT" }