fix: route reference URLs through source_kind, not just github.com

src/repo2rlenv/pipelines/code_instruct.py

@@ -72,6 +72,7 @@
     task_fingerprints,
 )
 from repo2rlenv.pipelines.base import PipelineResult
+from repo2rlenv.sources import blob_reference_url
 from repo2rlenv.spec.input import GenerationInput, PipelineName
 from repo2rlenv.spec.options import CodeInstructOptions
 
@@ -492,15 +493,24 @@ def _build_task(self, seed: Seed, parsed: ParsedTask, *, test_filename: str) ->
         )
         dockerfile = build_code_instruct_dockerfile(image_ref)
 
+        reference = blob_reference_url(
+            self.input.repo.source_kind,
+            owner,
+            name,
+            self.input.repo.ref,
+            seed.relative_path,
+            seed.start_line,
+            seed.end_line,
+        )
+
         repo2env = {
             "pipeline": "code_instruct",
             "pipeline_version": "0.6.2",
             "repo": f"{owner}/{name}",
             "ref": self.input.repo.ref,
-            "reference": (
-                f"https://github.com/{owner}/{name}/blob/{self.input.repo.ref}/"
-                f"{seed.relative_path}#L{seed.start_line}-L{seed.end_line}"
-            ),
+            # Omitted, not None, for a local checkout: TOML has no null and
+            # tomli_w rejects it outright.
+            **({"reference": reference} if reference is not None else {}),
             "source_access": self.input.repo.access,
             "built_at": datetime.now(UTC).isoformat(),
             "synthesis_llm": self.input.llm.qualified_name,

src/repo2rlenv/pipelines/equivalence_tests.py

@@ -68,6 +68,7 @@
 from repo2rlenv.pipelines._function_extractor import FunctionCandidate, walk_repo
 from repo2rlenv.pipelines._oss_instruct import check_equivalence_test_strength
 from repo2rlenv.pipelines.base import PipelineResult
+from repo2rlenv.sources import blob_reference_url
 from repo2rlenv.spec.input import GenerationInput, PipelineName
 from repo2rlenv.spec.options import EquivalenceTestsOptions
 
@@ -657,15 +658,24 @@ def _build_task(
 
         instruction = _build_instruction(cand)
 
+        reference = blob_reference_url(
+            self.input.repo.source_kind,
+            owner,
+            name,
+            self.input.repo.ref,
+            cand.relative_path,
+            cand.lineno,
+            cand.end_lineno,
+        )
+
         repo2env = {
             "pipeline": "equivalence_tests",
             "pipeline_version": "0.7.1",
             "repo": f"{owner}/{name}",
             "ref": self.input.repo.ref,
-            "reference": (
-                f"https://github.com/{owner}/{name}/blob/{self.input.repo.ref}/"
-                f"{cand.relative_path}#L{cand.lineno}-L{cand.end_lineno}"
-            ),
+            # Omitted, not None, for a local checkout: TOML has no null and
+            # tomli_w rejects it outright.
+            **({"reference": reference} if reference is not None else {}),
             "source_access": self.input.repo.access,
             "built_at": datetime.now(UTC).isoformat(),
             "synthesis_llm": self.input.llm.qualified_name,

src/repo2rlenv/sources.py

@@ -67,3 +67,25 @@ def detect_source_kind(url: str) -> SourceKind:
 
 def capabilities_for(kind: SourceKind) -> frozenset[Capability]:
     return _CAPABILITIES[kind]
+
+
+def blob_reference_url(
+    kind: SourceKind, owner: str, name: str, ref: str, path: str, start_line: int, end_line: int
+) -> str | None:
+    """Browsable link to a line range in ``path`` at ``ref``, or ``None`` for
+    a local checkout (no host to link to, so callers must omit rather than
+    null the field, since TOML has no null).
+
+    GitHub and GitLab differ in both the path segment and the line-range
+    anchor: GitHub is ``/blob/`` with ``#L{start}-L{end}``; GitLab's
+    canonical form is ``/-/blob/`` with ``#L{start}-{end}`` (no second
+    ``L``).
+    """
+    if kind is SourceKind.LOCAL:
+        return None
+    host = "gitlab.com" if kind is SourceKind.GITLAB else "github.com"
+    blob_path = "-/blob" if kind is SourceKind.GITLAB else "blob"
+    anchor = (
+        f"L{start_line}-{end_line}" if kind is SourceKind.GITLAB else f"L{start_line}-L{end_line}"
+    )
+    return f"https://{host}/{owner}/{name}/{blob_path}/{ref}/{path}#{anchor}"

tests/test_pipeline_code_instruct.py

@@ -9,11 +9,13 @@
 import pytest
 
 from repo2rlenv.pipelines._eval_script import all_tests_passed as _all_tests_passed
+from repo2rlenv.pipelines._oss_instruct import ParsedTask, Seed
 from repo2rlenv.pipelines.code_instruct import (
     CodeInstructPipeline,
     build_code_instruct_dockerfile,
     make_solution_diff,
 )
+from repo2rlenv.sources import SourceKind
 from repo2rlenv.spec.options import CodeInstructOptions
 
 # ---------------------------------------------------------------------------
@@ -123,3 +125,86 @@ def test_code_instruct_options_defaults():
     assert opts.seed_max_loc == 200
     assert opts.require_test_fails_without_oracle is True
     assert opts.require_test_passes_with_oracle is True
+
+
+# ---------------------------------------------------------------------------
+# _build_task: reference URL must follow source_kind, not always github.com
+# ---------------------------------------------------------------------------
+
+
+def _stub_pipeline_for_build_task(source_kind=SourceKind.GITHUB):
+    """A pipeline instance with just enough scaffolding to call `_build_task`."""
+    from types import SimpleNamespace
+    from unittest.mock import MagicMock
+
+    pipe = CodeInstructPipeline.__new__(CodeInstructPipeline)
+    pipe._llm_cost_usd = 0.0
+    pipe.bootstrap = SimpleNamespace(
+        image_tag="local/r2e-bootstrap/o__r:abc",
+        image_digest="local/r2e-bootstrap/o__r:abc",
+        pushed_to_registry=False,
+        language=SimpleNamespace(value="python"),
+    )
+    pipe.input = MagicMock()
+    pipe.input.repo.owner_name = ("o", "r")
+    pipe.input.repo.ref = "main"
+    pipe.input.repo.access = "auto"
+    pipe.input.repo.source_kind = source_kind
+    pipe.input.output.org = "default"
+    pipe.input.llm.qualified_name = "test-provider/test-model"
+    pipe._progress_cb = None
+    return pipe
+
+
+def _seed_and_parsed():
+    seed = Seed(
+        relative_path="src/calc.py",
+        start_line=10,
+        end_line=20,
+        text="def add(x, y):\n    return x + y\n",
+    )
+    parsed = ParsedTask(
+        problem="Implement add(x, y).",
+        test_code="def test_add():\n    assert add(1, 2) == 3\n",
+        solution_code="def add(x, y):\n    return x + y\n",
+    )
+    return seed, parsed
+
+
+@pytest.mark.parametrize(
+    ("source_kind", "expected_reference"),
+    [
+        (SourceKind.GITHUB, "https://github.com/o/r/blob/main/src/calc.py#L10-L20"),
+        (SourceKind.GITLAB, "https://gitlab.com/o/r/-/blob/main/src/calc.py#L10-20"),
+        (SourceKind.LOCAL, None),
+    ],
+)
+def test_build_task_reference_host_matches_source(source_kind, expected_reference):
+    """The 'reference' provenance URL must point at the seed's actual host,
+    not always github.com: code_instruct sets no required_capabilities and
+    runs on any source (sources.py), and a GitLab- or local-sourced task
+    previously got a github.com link that does not resolve."""
+    pipe = _stub_pipeline_for_build_task(source_kind=source_kind)
+    seed, parsed = _seed_and_parsed()
+    task = pipe._build_task(seed, parsed, test_filename="test_r2e_deadbeef.py")
+    if expected_reference is None:
+        # TOML has no null, so a local checkout must drop the key, not null it.
+        assert "reference" not in task.repo2env
+    else:
+        assert task.repo2env["reference"] == expected_reference
+
+
+def test_build_task_local_source_writes_a_loadable_task_toml(tmp_path):
+    """A None reference would crash tomli_w.dumps (TOML has no null) inside
+    write_harbor_task. Drive the real emitter, not just the dict, so a
+    regression here fails loudly instead of only at push time."""
+    import tomllib
+
+    from repo2rlenv.emitter.harbor import write_harbor_task
+
+    pipe = _stub_pipeline_for_build_task(source_kind=SourceKind.LOCAL)
+    seed, parsed = _seed_and_parsed()
+    task = pipe._build_task(seed, parsed, test_filename="test_r2e_deadbeef.py")
+    task_path = write_harbor_task(task, tmp_path)
+    loaded = tomllib.loads((task_path / "task.toml").read_text())
+    assert "reference" not in loaded["metadata"]["repo2env"]

tests/test_pipeline_equivalence_tests.py

@@ -30,6 +30,7 @@
     build_equivalence_dockerfile,
     uses_both_names,
 )
+from repo2rlenv.sources import SourceKind
 from repo2rlenv.spec.options import EquivalenceTestsOptions
 
 
@@ -285,3 +286,76 @@ def test_equivalence_tests_options_defaults():
     assert opts.max_loc == 60
     assert opts.require_test_fails_with_stub is True
     assert opts.require_test_passes_with_oracle is True
+
+
+# ---------------------------------------------------------------------------
+# _build_task: reference URL must follow source_kind, not always github.com
+# ---------------------------------------------------------------------------
+
+
+def _stub_pipeline_for_build_task(source_kind=SourceKind.GITHUB):
+    """A pipeline instance with just enough scaffolding to call `_build_task`."""
+    from types import SimpleNamespace
+    from unittest.mock import MagicMock
+
+    pipe = EquivalenceTestsPipeline.__new__(EquivalenceTestsPipeline)
+    pipe._llm_cost_usd = 0.0
+    pipe.bootstrap = SimpleNamespace(
+        image_tag="local/r2e-bootstrap/o__r:abc",
+        image_digest="local/r2e-bootstrap/o__r:abc",
+        pushed_to_registry=False,
+        language=SimpleNamespace(value="python"),
+    )
+    pipe.input = MagicMock()
+    pipe.input.repo.owner_name = ("o", "r")
+    pipe.input.repo.ref = "main"
+    pipe.input.repo.access = "auto"
+    pipe.input.repo.source_kind = source_kind
+    pipe.input.output.org = "default"
+    pipe.input.llm.qualified_name = "test-provider/test-model"
+    pipe._progress_cb = None
+    return pipe
+
+
+@pytest.mark.parametrize(
+    ("source_kind", "expected_reference"),
+    [
+        (SourceKind.GITHUB, "https://github.com/o/r/blob/main/src/calc.py#L1-L2"),
+        (SourceKind.GITLAB, "https://gitlab.com/o/r/-/blob/main/src/calc.py#L1-2"),
+        (SourceKind.LOCAL, None),
+    ],
+)
+def test_build_task_reference_host_matches_source(source_kind, expected_reference):
+    """The 'reference' provenance URL must point at the candidate's actual
+    host, not always github.com: equivalence_tests sets no
+    required_capabilities and runs on any source (sources.py), and a
+    GitLab- or local-sourced task previously got a github.com link that
+    does not resolve."""
+    pipe = _stub_pipeline_for_build_task(source_kind=source_kind)
+    cand = _candidate()
+    task = pipe._build_task(
+        cand, "def test_add():\n    assert add(1, 2) == 3\n", test_filename="test_r2e_deadbeef.py"
+    )
+    if expected_reference is None:
+        # TOML has no null, so a local checkout must drop the key, not null it.
+        assert "reference" not in task.repo2env
+    else:
+        assert task.repo2env["reference"] == expected_reference
+
+
+def test_build_task_local_source_writes_a_loadable_task_toml(tmp_path):
+    """A None reference would crash tomli_w.dumps (TOML has no null) inside
+    write_harbor_task. Drive the real emitter, not just the dict, so a
+    regression here fails loudly instead of only at push time."""
+    import tomllib
+
+    from repo2rlenv.emitter.harbor import write_harbor_task
+
+    pipe = _stub_pipeline_for_build_task(source_kind=SourceKind.LOCAL)
+    cand = _candidate()
+    task = pipe._build_task(
+        cand, "def test_add():\n    assert add(1, 2) == 3\n", test_filename="test_r2e_deadbeef.py"
+    )
+    task_path = write_harbor_task(task, tmp_path)
+    loaded = tomllib.loads((task_path / "task.toml").read_text())
+    assert "reference" not in loaded["metadata"]["repo2env"]