Use symmetric Python linkage with one-to-one assignment

.github/workflows/test.yml

@@ -53,12 +53,6 @@ jobs:
       with:
         python-version: ${{ matrix.python-version }}
 
-    - name: Set up Java
-      uses: actions/setup-java@v6.0.1
-      with:
-        distribution: 'temurin'
-        java-version: 22
-
     - name: Download package
       uses: actions/download-artifact@v8
       with:
@@ -71,5 +65,7 @@ jobs:
         uv pip install --system "$(ls dist/*.whl)[dev]"
 
     - name: Test with pytest
+      env:
+        MPLBACKEND: Agg
       run: |
         pytest

MANIFEST.in

@@ -3,6 +3,4 @@
 # SPDX-License-Identifier: MIT
 
 include powerplantmatching/package_data/*
-include powerplantmatching/package_data/duke_binaries/*
 include README.md LICENSE
-include requirements.yaml

README.md

@@ -105,7 +105,6 @@ and/or the current release stored on Zenodo with a release-specific DOI:
 
 `powerplantmatching` is released as free software under the [MIT](LICENSES/MIT.txt) license.
 The default output data [powerplants.csv](powerplants.csv) generated by the package is released under [CC BY 4.0](LICENSES/CC-BY-4.0.txt).
-Parts of the repository may be licensed under different licenses, especially dependent package binaries for `duke` being licensed under [Apache 2.0 license](https://github.com/PyPSA/powerplantmatching/tree/master/LICENSES/Apache-2.0.txt).
 
 This repository uses the [REUSE](https://reuse.software/) conventions to indicate the licenses that apply to individual files and parts of the repository.
 For details on the licenses that apply, see the the header information of the respective files and [REUSE.toml](REUSE.toml) for details.

REUSE.toml

@@ -32,11 +32,3 @@ path = [
 ]
 SPDX-FileCopyrightText = "Contributors to powerplantmatching <https://github.com/pypsa/powerplantmatching>"
 SPDX-License-Identifier = "CC0-1.0"
-
-
-[[annotations]]
-path = [
-    "powerplantmatching/package_data/duke_binaries/*.jar",
-]
-SPDX-FileCopyrightText = "Lars Marius Garshol <larsga@garshol.priv.no>"
-SPDX-License-Identifier = "Apache-2.0"

docs/api-core.md

@@ -8,6 +8,6 @@ SPDX-License-Identifier: MIT
 
 ::: powerplantmatching.core
 
-::: powerplantmatching.duke
+::: powerplantmatching.linkage
 
 ::: powerplantmatching.accessor

docs/license.md

@@ -8,7 +8,6 @@ SPDX-License-Identifier: MIT
 
 `powerplantmatching` is released as free software under the [MIT](LICENSES/MIT.txt) license.
 The default output data [powerplants.csv](powerplants.csv) generated by the package is released under [CC BY 4.0](LICENSES/CC-BY-4.0.txt).
-Parts of the repository may be licensed under different licenses, especially dependent package binaries for `duke` being licensed under [Apache 2.0 license](https://github.com/PyPSA/powerplantmatching/tree/master/LICENSES/Apache-2.0.txt).
 
 This repository uses the [REUSE](https://reuse.software/) conventions to indicate the licenses that apply to individual files and parts of the repository.
 For details on the licenses that apply, see the the header information of the respective files and [REUSE.toml](REUSE.toml) for details.

docs/release-notes.md

@@ -8,7 +8,10 @@ SPDX-License-Identifier: MIT
 
 ## Upcoming Version
 
-*
+* Replace Java DUKE with a blocked Python matcher using rapidfuzz. Java, the bundled jars, XML matching files and `powerplantmatching.duke` are removed. Matching results change.
+* Score plant names symmetrically and select fuzzy links with sparse maximum-score one-to-one assignment. Exact EIC matching still runs first and accepts only isolated one-to-one pairs.
+* Preserve EICs as sorted unique lists throughout unit aggregation, reduction and cached reload. Missing EICs are `[]`.
+* Rename `parallel_duke_processes` to `parallel_processes`. Update custom configuration keys; no compatibility alias is provided.
 
 ## [v0.9.0](https://github.com/PyPSA/powerplantmatching/releases/tag/v0.9.0) (3rd October 2026)
 

powerplantmatching/accessor.py

@@ -88,12 +88,12 @@ def set_name(self, name):
     def get_name(self):
         return self._obj.columns.name
 
-    def match_with(self, df, labels=None, config=None, reduced=True, **dukeargs):
+    def match_with(self, df, labels=None, config=None, reduced=True, **matchargs):
         from .matching import combine_multiple_datasets, reduce_matched_dataframe
         from .utils import to_list_if_other
 
         dfs = [self._obj] + to_list_if_other(df)
-        res = combine_multiple_datasets(dfs, labels, config=config, **dukeargs)
+        res = combine_multiple_datasets(dfs, labels, config=config, **matchargs)
         if reduced:
             return res.pipe(reduce_matched_dataframe, config=config)
         return res

powerplantmatching/cleaning.py

@@ -16,8 +16,8 @@
 from deprecation import deprecated
 
 from .core import PANDAS_V3, get_config, get_obj_if_Acc
-from .duke import duke
-from .utils import get_name, set_column_name
+from .linkage import match
+from .utils import collect_eic_codes, get_name, set_column_name
 
 logger = logging.getLogger(__name__)
 
@@ -44,7 +44,7 @@ def mode(x):
     "DateOut": "max",
     "File": mode,
     "projectID": set,
-    "EIC": set,
+    "EIC": collect_eic_codes,
     "Duration": "sum",  # note this is weighted sum
     "Volume_Mm3": "sum",
     "DamHeight_m": "sum",
@@ -505,11 +505,11 @@ def aggregate_units(
         country_query = "Country == @c"
         query = " and ".join(filter(None, [agg_query, block_query, country_query]))
         duplicates = pd.concat(
-            [duke(df.query(query), threads=threads) for c in countries]
+            [match(df.query(query), threads=threads) for c in countries]
         )
     else:
         query = " and ".join(filter(None, [agg_query, block_query]))
-        duplicates = duke(df.query(query) if query else df, threads=threads)
+        duplicates = match(df.query(query) if query else df, threads=threads)
 
     df = cliques(df, duplicates)
     df = df.groupby("grouped").agg(props_for_groups)

powerplantmatching/collection.py

@@ -31,7 +31,7 @@ def collect(
     update=False,
     reduced=True,
     config=None,
-    **dukeargs,
+    **matchargs,
 ):
     """
     Return the collection for a given list of datasets in matched or
@@ -47,7 +47,7 @@ def collect(
         Switch as to return the reduced (True) or matched (False) dataset.
     config : dict
         Configuration file of powerplantmatching
-    **dukeargs : keyword-args for duke
+    **matchargs : keyword arguments for matching
     """
 
     from . import data
@@ -88,7 +88,7 @@ def df_by_name(name):
 
     if update:
         dfs = parmap(df_by_name, datasets)
-        matched = combine_multiple_datasets(dfs, datasets, config=config, **dukeargs)
+        matched = combine_multiple_datasets(dfs, datasets, config=config, **matchargs)
         (
             matched.assign(projectID=lambda df: df.projectID.astype(str)).to_csv(
                 outfn_matched, index_label="id"

powerplantmatching/core.py

@@ -110,6 +110,11 @@ def get_config(filename=None, **overrides):
             config.update(yaml.load(f, Loader=yaml.FullLoader))
     config.update(overrides)
 
+    if "parallel_duke_processes" in config:
+        raise ValueError(
+            "Rename parallel_duke_processes to parallel_processes in your configuration"
+        )
+
     sha1digest = sha1(cPickle.dumps(overrides)).digest()
     if len(dict(**overrides)) == 0:
         config["hash"] = "default"