support colon paths

Author: Dingway98

hyperbrowser/client/managers/sandboxes/dockerfile_analysis.py

@@ -11,7 +11,6 @@
 import re
 import shlex
 from typing import Dict, List, Optional, Tuple
-from urllib.parse import urlsplit
 
 
 _KNOWN_INSTRUCTIONS = frozenset(
@@ -291,8 +290,4 @@ def _is_official_dockerfile_frontend(reference: str) -> bool:
 def _is_remote_add_source(source: str) -> bool:
     if _SCP_GIT_SOURCE_PATTERN.match(source):
         return True
-    try:
-        parsed = urlsplit(source)
-    except ValueError:
-        return False
-    return bool(parsed.scheme and parsed.scheme.lower() != "file")
+    return source.startswith(("http://", "https://", "git://", "ssh://"))

tests/fixtures/docker_context_parity.json

@@ -85,7 +85,14 @@
     {
       "name": "add-arbitrary-uri-scheme",
       "dockerfile": "FROM scratch\nADD oci-layout://example/image /image/\n",
-      "goSourceGroups": [],
+      "goSourceGroups": [["oci-layout://example/image"]],
+      "goFallback": "",
+      "pythonExpectation": "exact"
+    },
+    {
+      "name": "add-colon-paths-are-local",
+      "dockerfile": "FROM scratch\nADD assets:latest /asset/\nADD http:archive /archive/\n",
+      "goSourceGroups": [["assets:latest"], ["http:archive"]],
       "goFallback": "",
       "pythonExpectation": "exact"
     },

tests/test_dockerfile_analysis.py

@@ -64,6 +64,21 @@
         ),
         pytest.param(
             """\
+FROM scratch
+ADD assets:latest /colon/
+ADD http:archive.tar /http-colon/
+ADD oci-layout://example/image /unsupported-scheme/
+""",
+            [
+                ["assets:latest"],
+                ["http:archive.tar"],
+                ["oci-layout://example/image"],
+            ],
+            "",
+            id="add-colon-paths-and-unsupported-schemes-are-local",
+        ),
+        pytest.param(
+            """\
 FROM scratch AS generated
 RUN --mount=type=bind,source=src,target=/src true
 RUN --mount=source=vendor,target=/vendor true

tests/test_sandbox_image_build_helpers.py

@@ -282,6 +282,25 @@ def test_remote_dockerfile_context_is_deterministic_and_sparse(tmp_path):
         second.cleanup()
 
 
+def test_remote_dockerfile_context_includes_colon_named_add_sources(tmp_path):
+    (tmp_path / "Dockerfile").write_text(
+        "FROM scratch\nADD assets:latest /assets/\nADD http:archive /archive/\n"
+    )
+    (tmp_path / "assets:latest").write_text("asset\n")
+    (tmp_path / "http:archive").write_text("archive\n")
+
+    packaged = image_build.package_docker_build_context_manifest(tmp_path)
+    try:
+        assert packaged.manifest.context_mode == "sparse"
+        archived_names = set()
+        for artifact in packaged.bundles.values():
+            with tarfile.open(artifact.path, "r:gz") as archive:
+                archived_names.update(name.rstrip("/") for name in archive.getnames())
+        assert {"assets:latest", "http:archive"} <= archived_names
+    finally:
+        packaged.cleanup()
+
+
 def test_remote_dockerfile_context_falls_back_for_variable_source(tmp_path):
     (tmp_path / "Dockerfile").write_text("FROM scratch\nCOPY $SOURCE /app/\n")
     (tmp_path / "payload.txt").write_text("payload\n")