python-entsoe 0.6.0__tar.gz → 0.6.2__tar.gz

python_entsoe-0.6.2/.github/workflows/publish.yml

@@ -0,0 +1,26 @@
+name: Publish to PyPI
+
+on:
+  push:
+    tags: ["v*"]
+
+jobs:
+  publish:
+    runs-on: ubuntu-latest
+    permissions:
+      id-token: write
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install build tools
+        run: pip install build
+
+      - name: Build package
+        run: python -m build
+
+      - name: Publish to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1

python_entsoe-0.6.2/.gitlab-ci.yml

@@ -0,0 +1,22 @@
+stages:
+  - build
+  - publish
+
+default:
+  image: ghcr.io/astral-sh/uv:python3.13-bookworm
+
+build:
+  stage: build
+  script:
+    - uv build --no-sources
+  artifacts:
+    paths:
+      - dist/
+
+publish:
+  stage: publish
+  rules:
+    - if: '$CI_COMMIT_TAG =~ /^v[0-9]+\.[0-9]+\.[0-9]+/'
+  script:
+    - uv build --no-sources
+    - uv publish --publish-url "$CI_API_V4_URL/projects/$CI_PROJECT_ID/packages/pypi" --check-url "$CI_API_V4_URL/projects/$CI_PROJECT_ID/packages/pypi/simple" --username gitlab-ci-token --password "$CI_JOB_TOKEN"

{python_entsoe-0.6.0 → python_entsoe-0.6.2}/CHANGELOG.md

@@ -1,5 +1,22 @@
 # Changelog
 
+## [0.6.2](https://gitlab.com/datons.com/libraries/python-entsoe/-/tags/v0.6.2) (2026-10-15)
+
+
+### Bug Fixes
+
+* **prices:** `day_ahead` mixed the SDAC and EXAA auctions for DE_LU and AT, and the intraday auctions for ES, returning several prices per timestamp. It now returns SDAC; `sequence=2` returns EXAA (CLI: `--sequence 2`).
+* **parser:** points omitted from A03 (variable-sized block) curves are filled with the previous value, so no interval is lost. Affects every endpoint.
+* **parser:** TimeSeries that the API repeats verbatim no longer produce duplicate rows.
+* **cache:** the cache moves to a versioned layout (`~/.cache/entsoe/v2/`). Data cached by earlier versions, which may hold the errors above, is deleted on first use and downloaded again.
+
+## [0.6.1](https://github.com/datons/python-entsoe/compare/python-entsoe-v0.6.0...python-entsoe-v0.6.1) (2026-05-16)
+
+
+### Bug Fixes
+
+* restore catalog imports broken by incomplete refactor in dcc35f7 ([74c0814](https://github.com/datons/python-entsoe/commit/74c0814bbdd3bd8d43981e77a1ab573ddf80a483))
+
 ## [0.6.0](https://github.com/datons/python-entsoe/compare/python-entsoe-v0.5.0...python-entsoe-v0.6.0) (2026-03-04)
 
 

{python_entsoe-0.6.0 → python_entsoe-0.6.2}/CLAUDE.md

@@ -9,19 +9,23 @@ Python client for the ENTSO-E Transparency Platform API.
 - Run scripts: `uv run python <script>`
 - Run tests: `uv run pytest`
 
-## PyPI Publishing
+## Publishing
 
-Publishing is **tag-based only**. Pushing to `main` does NOT trigger a release.
+The repo lives in GitLab (`datons.com/libraries/python-entsoe`, private). Publishing is **tag-based only**: pushing to `main` does NOT release. The CI publishes to the `datons.com/libraries` package registry, which `esios` and the other services consume through the `datons-libraries` index.
 
 ```bash
 # 1. Bump version in pyproject.toml
 # 2. Commit the version bump
 # 3. Tag and push
-git tag v0.3.0
+git tag v0.6.2
 git push origin main --tags
 ```
 
-The GitHub Actions workflow (`.github/workflows/publish.yml`) uses OIDC trusted publishing — no tokens needed.
+Public users install from PyPI, so a fix that matters to them is also released there: push `main` and the same tag to the `github` remote. `.github/workflows/publish.yml` builds and publishes on `v*` tags through PyPI trusted publishing (no token). GitLab stays the source of truth.
+
+```bash
+git push github main --tags
+```
 
 ## Example Notebooks
 

{python_entsoe-0.6.0 → python_entsoe-0.6.2}/PKG-INFO

@@ -1,11 +1,11 @@
-Metadata-Version: 2.4
+Metadata-Version: 2.5
 Name: python-entsoe
-Version: 0.6.0
+Version: 0.6.2
 Summary: Python client for the ENTSO-E Transparency Platform API
-Project-URL: Homepage, https://github.com/datons/python-entsoe
-Project-URL: Repository, https://github.com/datons/python-entsoe
-Project-URL: Issues, https://github.com/datons/python-entsoe/issues
-Project-URL: Changelog, https://github.com/datons/python-entsoe/releases
+Project-URL: Homepage, https://gitlab.com/datons.com/libraries/python-entsoe
+Project-URL: Repository, https://gitlab.com/datons.com/libraries/python-entsoe
+Project-URL: Issues, https://gitlab.com/datons.com/libraries/python-entsoe/-/issues
+Project-URL: Changelog, https://gitlab.com/datons.com/libraries/python-entsoe/-/releases
 Author-email: Jesús López <jesus.lopez@datons.com>
 License-Expression: MIT
 Keywords: api,electricity,energy,entsoe,transparency
@@ -21,7 +21,6 @@ Classifier: Topic :: Scientific/Engineering
 Requires-Python: >=3.10
 Requires-Dist: pandas>=2.0
 Requires-Dist: pyarrow>=23.0.1
-Requires-Dist: pyyaml>=6.0
 Requires-Dist: requests>=2.28
 Requires-Dist: rich>=13.0
 Requires-Dist: typer>=0.9
@@ -94,11 +93,14 @@ df = client.load.forecast(start, end, country="FR")
 
 | Method | Description | Parameters |
 |--------|-------------|------------|
-| `day_ahead(start, end, country)` | Day-ahead market prices (EUR/MWh) | `country`: ISO code |
+| `day_ahead(start, end, country, sequence=1)` | Day-ahead market prices (EUR/MWh) | `country`: ISO code; `sequence`: `1` SDAC, `2` EXAA |
 
 ```python
 df = client.prices.day_ahead(start, end, country="FR")
 # Returns: timestamp, value, currency, price_unit
+
+# DE_LU and AT also publish the separate EXAA auction (10:15 CET)
+exaa = client.prices.day_ahead(start, end, country="DE_LU", sequence=2)
 ```
 
 ### `client.generation`
@@ -252,7 +254,7 @@ See the [`examples/`](examples/) directory for Jupyter notebooks with plotly vis
 ## Development
 
 ```bash
-git clone https://github.com/jsulopzs/python-entsoe.git
+git clone git@gitlab.com:datons.com/libraries/python-entsoe.git
 cd python-entsoe
 uv sync
 

{python_entsoe-0.6.0 → python_entsoe-0.6.2}/README.md

@@ -65,11 +65,14 @@ df = client.load.forecast(start, end, country="FR")
 
 | Method | Description | Parameters |
 |--------|-------------|------------|
-| `day_ahead(start, end, country)` | Day-ahead market prices (EUR/MWh) | `country`: ISO code |
+| `day_ahead(start, end, country, sequence=1)` | Day-ahead market prices (EUR/MWh) | `country`: ISO code; `sequence`: `1` SDAC, `2` EXAA |
 
 ```python
 df = client.prices.day_ahead(start, end, country="FR")
 # Returns: timestamp, value, currency, price_unit
+
+# DE_LU and AT also publish the separate EXAA auction (10:15 CET)
+exaa = client.prices.day_ahead(start, end, country="DE_LU", sequence=2)
 ```
 
 ### `client.generation`
@@ -223,7 +226,7 @@ See the [`examples/`](examples/) directory for Jupyter notebooks with plotly vis
 ## Development
 
 ```bash
-git clone https://github.com/jsulopzs/python-entsoe.git
+git clone git@gitlab.com:datons.com/libraries/python-entsoe.git
 cd python-entsoe
 uv sync
 

{python_entsoe-0.6.0 → python_entsoe-0.6.2}/docs/data-availability.md

@@ -92,6 +92,28 @@ environment variable.
 export $(cat .env | xargs) && uv run jupyter nbconvert --execute ...
 ```
 
+## 7. Day-ahead documents (A44) mix several auctions
+
+**Affected endpoints:** `prices.day_ahead`
+
+**Problem:** One A44 response can hold several price series for the same hours:
+
+- DE_LU and AT publish the SDAC auction (gate closure 12:00 CET) as `classificationSequence_AttributeInstanceComponent.position` 1 and the EXAA auction (10:15 CET) as position 2.
+- ES also returns the intraday auctions (IDA1–3) under `contract_MarketAgreement.type` A07, numbered by the same sequence element.
+- Some zones (FR, AT) repeat an identical TimeSeries.
+
+Parsing every series gives two or more prices per timestamp.
+
+**Solution:** Keep only daily contracts (`A01`, or no contract type) whose sequence matches the requested one, treating a missing sequence as 1. `day_ahead(..., sequence=2)` returns EXAA. Identical rows are dropped.
+
+## 8. Curve type A03 omits repeated points
+
+**Affected endpoints:** all (prices, load, generation, transmission)
+
+**Problem:** With `<curveType>A03</curveType>` (variable-sized blocks), a Point applies until the next Point, and positions whose value equals the previous one are left out. Reading only the Points present loses those intervals: a day of quarter-hourly prices can come back with 95 rows, and a flat generation series with a single row.
+
+**Solution:** Fill every position up to the Period's `timeInterval/end` with the last value seen. The Period end tracks published data, so recent partial days are not extended.
+
 ## Summary table
 
 | Issue | Endpoint | Symptom | Fix |
@@ -102,3 +124,5 @@ export $(cat .env | xargs) && uv run jupyter nbconvert --execute ...
 | Route availability | net_transfer_capacity | `NoDataError` for FR→DE_LU | Use FR→ES, FR→GB, FR→IT_NORTH |
 | Value tag name | imbalance_prices | All values are `None` | Parse `imbalance_Price.amount` tag |
 | Env vars in notebooks | all (via nbconvert) | `ValueError: API key required` | Export env vars before execution |
+| Mixed auctions | prices.day_ahead | Duplicate timestamps (DE_LU, AT, ES) | Filter by contract type and sequence |
+| A03 omitted points | all | Missing intervals | Forward-fill to Period end |

{python_entsoe-0.6.0 → python_entsoe-0.6.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "python-entsoe"
-version = "0.6.0"
+version = "0.6.2"
 description = "Python client for the ENTSO-E Transparency Platform API"
 readme = "README.md"
 license = "MIT"
@@ -22,7 +22,6 @@ classifiers = [
 ]
 dependencies = [
     "pandas>=2.0",
-    "pyyaml>=6.0",
     "requests>=2.28",
     "typer>=0.9",
     "rich>=13.0",
@@ -33,10 +32,10 @@ dependencies = [
 entsoe = "entsoe.cli:app"
 
 [project.urls]
-Homepage = "https://github.com/datons/python-entsoe"
-Repository = "https://github.com/datons/python-entsoe"
-Issues = "https://github.com/datons/python-entsoe/issues"
-Changelog = "https://github.com/datons/python-entsoe/releases"
+Homepage = "https://gitlab.com/datons.com/libraries/python-entsoe"
+Repository = "https://gitlab.com/datons.com/libraries/python-entsoe"
+Issues = "https://gitlab.com/datons.com/libraries/python-entsoe/-/issues"
+Changelog = "https://gitlab.com/datons.com/libraries/python-entsoe/-/releases"
 
 [build-system]
 requires = ["hatchling"]
@@ -47,9 +46,6 @@ packages = ["src/entsoe"]
 
 [tool.pytest.ini_options]
 testpaths = ["tests"]
-markers = [
-    "integration: tests that hit real APIs (deselect with -m 'not integration')",
-]
 
 [dependency-groups]
 dev = [

{python_entsoe-0.6.0 → python_entsoe-0.6.2}/src/entsoe/_mappings.py

@@ -399,7 +399,6 @@ NEIGHBOURS: dict[str, set[str]] = {
     "BG": {"RO", "RS", "MK", "GR", "TR"},
     "MK": {"RS", "BG", "GR", "AL"},
     "GR": {"BG", "MK", "AL", "IT_SUD", "TR"},
-    "IT_SUD": {"GR"},
     "AL": {"RS", "MK", "GR", "ME"},
     "ME": {"RS", "BA", "AL"},
     "TR": {"BG", "GR"},

{python_entsoe-0.6.0 → python_entsoe-0.6.2}/src/entsoe/_xml.py

@@ -15,6 +15,7 @@ from __future__ import annotations
 import re
 import xml.etree.ElementTree as ET
 from datetime import timedelta
+from typing import Callable
 
 import pandas as pd
 
@@ -79,12 +80,42 @@ def _extract_point_value(point: ET.Element) -> float | None:
     return None
 
 
-def parse_timeseries(xml_text: str) -> pd.DataFrame:
+#: Element holding the sequence of a TimeSeries within a document, e.g. the
+#: SDAC (1) and EXAA (2) day-ahead auctions both published for DE_LU and AT.
+SEQUENCE_TAG = "classificationSequence_AttributeInstanceComponent.position"
+
+
+def timeseries_sequence(ts: ET.Element) -> int | None:
+    """Return the classification sequence of a TimeSeries, or None if absent."""
+    for el in ts.iter():
+        if _strip_ns(el.tag) == SEQUENCE_TAG and el.text:
+            return int(el.text)
+    return None
+
+
+def timeseries_contract_type(ts: ET.Element) -> str | None:
+    """Return the contract_MarketAgreement.type of a TimeSeries (A01 daily, A07 intraday)."""
+    return _find_text(ts, "contract_MarketAgreement.type")
+
+
+def parse_timeseries(
+    xml_text: str,
+    ts_filter: Callable[[ET.Element], bool] | None = None,
+) -> pd.DataFrame:
     """Parse an ENTSO-E XML response into a DataFrame.
 
     Handles multiple TimeSeries (e.g., generation per type) and multiple
     Periods within each TimeSeries (e.g., multi-day responses).
 
+    Points omitted from a curveType A03 (variable-sized block) Period carry
+    the previous point's value, so they are filled in. TimeSeries the API
+    repeats verbatim produce duplicate rows, which are dropped.
+
+    Args:
+        xml_text: Raw XML response.
+        ts_filter: Optional predicate on each TimeSeries element; series for
+            which it returns False are skipped.
+
     Returns a DataFrame with columns:
     - timestamp: tz-aware UTC datetime
     - value: numeric value (MW or EUR/MWh depending on query)
@@ -107,6 +138,9 @@ def parse_timeseries(xml_text: str) -> pd.DataFrame:
     rows: list[dict] = []
 
     for ts in timeseries_list:
+        if ts_filter is not None and not ts_filter(ts):
+            continue
+
         # Extract TimeSeries metadata
         ts_meta: dict = {}
 
@@ -163,23 +197,38 @@ def parse_timeseries(xml_text: str) -> pd.DataFrame:
             period_start = pd.Timestamp(start_text)
             resolution = _parse_resolution(resolution_text)
 
+            values: dict[int, float | None] = {}
             for point in _findall(period, "Point"):
                 position_text = _find_text(point, "position")
                 if position_text is None:
                     continue
+                values[int(position_text)] = _extract_point_value(point)
 
-                position = int(position_text)
-                value = _extract_point_value(point)
+            if not values:
+                continue
 
+            positions = sorted(values)
+            if _find_text(ts, "curveType") == "A03":
+                end_text = _find_text(time_interval, "end")
+                if end_text is not None:
+                    n_slots = int((pd.Timestamp(end_text) - period_start) / resolution)
+                else:
+                    n_slots = positions[-1]
+                filled: dict[int, float | None] = {}
+                last = None
+                for pos in range(positions[0], max(n_slots, positions[-1]) + 1):
+                    last = values.get(pos, last)
+                    filled[pos] = last
+                values, positions = filled, list(filled)
+
+            for position in positions:
                 timestamp = period_start + resolution * (position - 1)
-
-                row = {"timestamp": timestamp, "value": value, **ts_meta}
-                rows.append(row)
+                rows.append({"timestamp": timestamp, "value": values[position], **ts_meta})
 
     if not rows:
         raise NoDataError()
 
-    df = pd.DataFrame(rows)
+    df = pd.DataFrame(rows).drop_duplicates()
     df["timestamp"] = pd.to_datetime(df["timestamp"], utc=True)
     df = df.sort_values("timestamp").reset_index(drop=True)
     return df

{python_entsoe-0.6.0 → python_entsoe-0.6.2}/src/entsoe/cache.py

@@ -6,7 +6,7 @@ published (~48h), so caching is safe and enabled by default.
 
 Storage layout::
 
-    {cache_dir}/
+    {cache_dir}/v{CACHE_LAYOUT_VERSION}/
     ├── prices/day_ahead/{ES,FR,...}/data.parquet
     ├── load/{actual,forecast}/{ES,FR,...}/data.parquet
     ├── generation/{actual,forecast,capacity}/{ES,...}/data.parquet
@@ -18,6 +18,9 @@ Each partition stores a single wide-format parquet file: the DatetimeIndex
 holds timestamps, and columns are values (e.g. "value", or per-type
 columns like "solar", "wind_onshore").
 
+The version directory changes whenever cached values from earlier releases
+are known to be wrong; data from older layouts is removed on first use.
+
 Gap detection works per-column: requesting specific columns checks whether
 those columns have data for the requested range, not whether any column does.
 """
@@ -42,6 +45,13 @@ _DEFAULT_CACHE_DIR = Path(
     os.environ.get("XDG_CACHE_HOME", Path.home() / ".cache")
 ) / "entsoe"
 
+# Bumped when previously cached values are known to be wrong.
+# v2 (0.6.2): A03 curves lost their omitted points; day-ahead mixed auctions.
+CACHE_LAYOUT_VERSION = 2
+
+# Domain directories written by the unversioned (<= 0.6.1) layout.
+_LEGACY_DOMAINS = ("prices", "load", "generation", "transmission", "balancing")
+
 # Data older than this (hours) is considered final and won't be re-fetched
 _DEFAULT_RECENT_TTL_HOURS = 48
 
@@ -71,16 +81,37 @@ class CacheStore:
 
     def __init__(self, config: CacheConfig):
         self.config = config
+        self._purge_stale_layouts()
+
+    @property
+    def root(self) -> Path:
+        """Directory holding the current cache layout."""
+        return self.config.cache_dir / f"v{CACHE_LAYOUT_VERSION}"
+
+    def _purge_stale_layouts(self) -> None:
+        """Delete data cached by older layouts, which may hold wrong values."""
+        base = self.config.cache_dir
+        if not base.is_dir():
+            return
+        stale = [base / d for d in _LEGACY_DOMAINS]
+        stale += [
+            p for p in base.glob("v*")
+            if p.name[1:].isdigit() and int(p.name[1:]) < CACHE_LAYOUT_VERSION
+        ]
+        for path in stale:
+            if path.is_dir():
+                logger.info("Removing outdated ENTSO-E cache: %s", path)
+                shutil.rmtree(path, ignore_errors=True)
 
     # -- Path resolution -------------------------------------------------------
 
     def _parquet_path(self, domain: str, topic: str, partition: str) -> Path:
-        """Data file: {cache_dir}/{domain}/{topic}/{partition}/data.parquet"""
-        return self.config.cache_dir / domain / topic / partition / "data.parquet"
+        """Data file: {root}/{domain}/{topic}/{partition}/data.parquet"""
+        return self.root / domain / topic / partition / "data.parquet"
 
     def _meta_path(self, domain: str, topic: str, partition: str) -> Path:
-        """Metadata file: {cache_dir}/{domain}/{topic}/{partition}/meta.json"""
-        return self.config.cache_dir / domain / topic / partition / "meta.json"
+        """Metadata file: {root}/{domain}/{topic}/{partition}/meta.json"""
+        return self.root / domain / topic / partition / "meta.json"
 
     # -- Data Read / Write -----------------------------------------------------
 
@@ -291,13 +322,13 @@ class CacheStore:
         count = 0
 
         if domain and topic and partition:
-            target = self.config.cache_dir / domain / topic / partition
+            target = self.root / domain / topic / partition
         elif domain and topic:
-            target = self.config.cache_dir / domain / topic
+            target = self.root / domain / topic
         elif domain:
-            target = self.config.cache_dir / domain
+            target = self.root / domain
         else:
-            target = self.config.cache_dir
+            target = self.root
 
         if target.exists():
             count = sum(1 for f in target.rglob("*") if f.is_file())
@@ -307,7 +338,7 @@ class CacheStore:
 
     def status(self) -> dict:
         """Return cache statistics."""
-        cache_dir = self.config.cache_dir
+        cache_dir = self.root
         if not cache_dir.exists():
             return {"path": str(cache_dir), "files": 0, "size_mb": 0.0, "domains": {}}