Support doc link for the sklearn module. (#10287)

python-package/xgboost/core.py

@@ -190,6 +190,27 @@ def _register_log_callback(lib: ctypes.CDLL) -> None:
         raise XGBoostError(lib.XGBGetLastError())
 
 
+def _parse_version(ver: str) -> Tuple[Tuple[int, int, int], str]:
+    """Avoid dependency on packaging (PEP 440)."""
+    # 2.0.0-dev, 2.0.0, 2.0.0.post1, or 2.0.0rc1
+    if ver.find("post") != -1:
+        major, minor, patch = ver.split(".")[:-1]
+        postfix = ver.split(".")[-1]
+    elif "-dev" in ver:
+        major, minor, patch = ver.split("-")[0].split(".")
+        postfix = "dev"
+    else:
+        major, minor, patch = ver.split(".")
+        rc = patch.find("rc")
+        if rc != -1:
+            postfix = patch[rc:]
+            patch = patch[:rc]
+        else:
+            postfix = ""
+
+    return (int(major), int(minor), int(patch)), postfix
+
+
 def _load_lib() -> ctypes.CDLL:
     """Load xgboost Library."""
     lib_paths = find_lib_path()
@@ -237,17 +258,8 @@ Error message(s): {os_error_list}
         )
     _register_log_callback(lib)
 
-    def parse(ver: str) -> Tuple[int, int, int]:
-        """Avoid dependency on packaging (PEP 440)."""
-        # 2.0.0-dev, 2.0.0, or 2.0.0rc1
-        major, minor, patch = ver.split("-")[0].split(".")
-        rc = patch.find("rc")
-        if rc != -1:
-            patch = patch[:rc]
-        return int(major), int(minor), int(patch)
-
     libver = _lib_version(lib)
-    pyver = parse(_py_version())
+    pyver, _ = _parse_version(_py_version())
 
     # verify that we are loading the correct binary.
     if pyver != libver:

python-package/xgboost/sklearn.py

@@ -40,6 +40,8 @@ from .core import (
     XGBoostError,
     _deprecate_positional_args,
     _parse_eval_str,
+    _parse_version,
+    _py_version,
 )
 from .data import _is_cudf_df, _is_cudf_ser, _is_cupy_alike, _is_pandas_df
 from .training import train
@@ -795,6 +797,32 @@ class XGBModel(XGBModelBase):
     def __sklearn_is_fitted__(self) -> bool:
         return hasattr(self, "_Booster")
 
+    @property
+    def _doc_link_module(self) -> str:
+        return "xgboost"
+
+    @property
+    def _doc_link_template(self) -> str:
+        ver = _py_version()
+        (major, minor, _), post = _parse_version(ver)
+
+        if post == "dev":
+            rel = "latest"
+        else:
+            # RTD tracks the release branch. We don't have independent branches for
+            # patch releases.
+            rel = f"release_{major}.{minor}.0"
+
+        module = self.__class__.__module__
+        # All sklearn estimators are forwarded to the top level module in both source
+        # code and sphinx api doc.
+        if module == "xgboost.sklearn":
+            module = module.split(".")[0]
+        name = self.__class__.__name__
+
+        base = "https://xgboost.readthedocs.io/en"
+        return f"{base}/{rel}/python/python_api.html#{module}.{name}"
+
     def get_booster(self) -> Booster:
         """Get the underlying xgboost Booster of this model.
 

tests/ci_build/conda_env/linux_cpu_test.yml

@@ -14,7 +14,7 @@ dependencies:
 - pylint
 - numpy
 - scipy
-- scikit-learn
+- scikit-learn>=1.4.1
 - pandas
 - matplotlib
 - dask

tests/ci_build/conda_env/macos_cpu_test.yml

@@ -11,7 +11,7 @@ dependencies:
 - numpy
 - scipy
 - llvm-openmp
-- scikit-learn
+- scikit-learn>=1.4.1
 - pandas
 - matplotlib
 - dask

tests/ci_build/conda_env/win64_cpu_test.yml

@@ -6,7 +6,7 @@ dependencies:
 - wheel
 - numpy
 - scipy
-- scikit-learn
+- scikit-learn>=1.4.1
 - pandas
 - matplotlib
 - dask

tests/python/test_basic.py

@@ -9,6 +9,7 @@ import pytest
 
 import xgboost as xgb
 from xgboost import testing as tm
+from xgboost.core import _parse_version
 
 dpath = "demo/data/"
 rng = np.random.RandomState(1994)
@@ -315,3 +316,14 @@ class TestBasicPathLike:
         """An invalid model_file path should raise XGBoostError."""
         with pytest.raises(xgb.core.XGBoostError):
             xgb.Booster(model_file=Path("invalidpath"))
+
+
+def test_parse_ver() -> None:
+    (major, minor, patch), post = _parse_version("2.1.0")
+    assert post == ""
+    (major, minor, patch), post = _parse_version("2.1.0-dev")
+    assert post == "dev"
+    (major, minor, patch), post = _parse_version("2.1.0rc1")
+    assert post == "rc1"
+    (major, minor, patch), post = _parse_version("2.1.0.post1")
+    assert post == "post1"

tests/python/test_with_sklearn.py

@@ -1484,3 +1484,16 @@ def test_tags() -> None:
 
     tags = xgb.XGBRanker()._more_tags()
     assert "multioutput" not in tags
+
+
+def test_doc_link() -> None:
+    for est in [
+        xgb.XGBRegressor(),
+        xgb.XGBClassifier(),
+        xgb.XGBRanker(),
+        xgb.XGBRFRegressor(),
+        xgb.XGBRFClassifier(),
+    ]:
+        name = est.__class__.__name__
+        link = est._get_doc_link()
+        assert f"xgboost.{name}" in link

tests/test_distributed/test_with_dask/test_with_dask.py

@@ -35,6 +35,7 @@ from hypothesis import HealthCheck, assume, given, note, settings
 from sklearn.datasets import make_classification, make_regression
 
 import xgboost as xgb
+from xgboost import dask as dxgb
 from xgboost import testing as tm
 from xgboost.data import _is_cudf_df
 from xgboost.testing.params import hist_cache_strategy, hist_parameter_strategy
@@ -2324,3 +2325,16 @@ async def test_worker_restarted(c, s, a, b):
             d_train,
             evals=[(d_train, "train")],
         )
+
+
+def test_doc_link() -> None:
+    for est in [
+        dxgb.DaskXGBRegressor(),
+        dxgb.DaskXGBClassifier(),
+        dxgb.DaskXGBRanker(),
+        dxgb.DaskXGBRFRegressor(),
+        dxgb.DaskXGBRFClassifier(),
+    ]:
+        name = est.__class__.__name__
+        link = est._get_doc_link()
+        assert f"xgboost.dask.{name}" in link