fix: address bugs in gen-build-spec

Signed-off-by: Abhinav Pradeep <[email protected]>

Author: AbhinavPradeep

src/macaron/build_spec_generator/build_spec_generator.py

@@ -97,8 +97,8 @@ def gen_build_spec_for_purl(
             case BuildSpecFormat.DOCKERFILE:
                 try:
                     build_spec_content = gen_dockerfile(build_spec)
-                except ValueError as error:
-                    logger.error("Error while serializing the build spec: %s.", error)
+                except GenerateBuildSpecError as error:
+                    logger.error("Error while generating the build spec: %s.", error)
                     return os.EX_DATAERR
                 build_spec_file_path = os.path.join(build_spec_dir_path, "dockerfile.buildspec")
 

src/macaron/build_spec_generator/common_spec/core.py

@@ -378,7 +378,7 @@ def gen_generic_build_spec(
             "purl": str(purl),
             "language": target_language,
             "build_tools": build_tool_names,
-            "build_commands": [selected_build_command],
+            "build_commands": [selected_build_command] if selected_build_command else [],
         }
     )
     ECOSYSTEMS[purl.type.upper()].value(base_build_spec_dict).resolve_fields(purl)

src/macaron/build_spec_generator/common_spec/pypi_spec.py

@@ -155,6 +155,16 @@ def resolve_fields(self, purl: PackageURL) -> None:
                         chronologically_likeliest_version = (
                             pypi_package_json.get_chronologically_suitable_setuptools_version()
                         )
+                        try:
+                            # Get information from the wheel file name.
+                            logger.debug(pypi_package_json.wheel_filename)
+                            _, _, _, tags = parse_wheel_filename(pypi_package_json.wheel_filename)
+                            for tag in tags:
+                                wheel_name_python_version_list.append(tag.interpreter)
+                                wheel_name_platforms.add(tag.platform)
+                            logger.debug(python_version_set)
+                        except InvalidWheelFilename:
+                            logger.debug("Could not parse wheel file name to extract version")
                 except SourceCodeError:
                     logger.debug("Could not find pure wheel matching this PURL")
 
@@ -214,17 +224,6 @@ def resolve_fields(self, purl: PackageURL) -> None:
                     except (InvalidRequirement, InvalidSpecifier) as error:
                         logger.debug("Malformed requirement encountered %s : %s", requirement, error)
 
-                try:
-                    # Get information from the wheel file name.
-                    logger.debug(pypi_package_json.wheel_filename)
-                    _, _, _, tags = parse_wheel_filename(pypi_package_json.wheel_filename)
-                    for tag in tags:
-                        wheel_name_python_version_list.append(tag.interpreter)
-                        wheel_name_platforms.add(tag.platform)
-                    logger.debug(python_version_set)
-                except InvalidWheelFilename:
-                    logger.debug("Could not parse wheel file name to extract version")
-
                 self.data["language_version"] = list(python_version_set) or wheel_name_python_version_list
 
                 # Use the default build command for pure Python packages.
@@ -243,9 +242,18 @@ def resolve_fields(self, purl: PackageURL) -> None:
 
         if not patched_build_commands:
             # Resolve and patch build commands.
-            selected_build_commands = self.data["build_commands"] or self.get_default_build_commands(
-                self.data["build_tools"]
-            )
+
+            # To ensure that selected_build_commands is never empty, we seed with the fallback
+            # command of python -m build --wheel -n
+            if self.data["build_commands"]:
+                selected_build_commands = self.data["build_commands"]
+            else:
+                self.data["build_commands"] = ["python -m build --wheel -n".split()]
+                selected_build_commands = (
+                    self.get_default_build_commands(self.data["build_tools"]) or self.data["build_commands"]
+                )
+
+            logger.debug(selected_build_commands)
 
             patched_build_commands = (
                 patch_commands(

src/macaron/build_spec_generator/dockerfile/pypi_dockerfile_output.py

@@ -4,6 +4,7 @@
 """This module implements the logic to generate a dockerfile from a Python buildspec."""
 
 import logging
+import re
 from textwrap import dedent
 
 from packaging.specifiers import InvalidSpecifier, SpecifierSet
@@ -35,8 +36,7 @@ def gen_dockerfile(buildspec: BaseBuildSpecDict) -> str:
     """
     language_version: str | None = pick_specific_version(buildspec)
     if language_version is None:
-        logger.debug("Could not derive a specific interpreter version.")
-        raise GenerateBuildSpecError("Could not derive specific interpreter version.")
+        raise GenerateBuildSpecError("Could not derive specific interpreter version")
     backend_install_commands: str = " && ".join(build_backend_commands(buildspec))
     build_tool_install: str = ""
     if (
@@ -124,8 +124,18 @@ def pick_specific_version(buildspec: BaseBuildSpecDict) -> str | None:
         try:
             version_set &= SpecifierSet(version)
         except InvalidSpecifier as error:
-            logger.debug("Malformed interpreter version encountered: %s (%s)", version, error)
-            return None
+            logger.debug("Non-standard interpreter version encountered: %s (%s)", version, error)
+            # Whilst the Python tags specify interpreter implementation
+            # as well as version, with no standard way to parse out the
+            # implementation, we can attempt to heuristically:
+            try_parse_version = infer_interpreter_version(version)
+            if try_parse_version:
+                try:
+                    version_set &= SpecifierSet(f">={try_parse_version}")
+                except InvalidSpecifier as error_for_retry:
+                    logger.debug("Could not parse interpreter version from: %s (%s)", version, error_for_retry)
+
+    logger.debug(version_set)
 
     # Now to get the latest acceptable one, we can step through all interpreter
     # versions. For the most accurate result, we can query python.org for a
@@ -141,6 +151,31 @@ def pick_specific_version(buildspec: BaseBuildSpecDict) -> str | None:
     return None
 
 
+def infer_interpreter_version(tag: str) -> str | None:
+    """Infer interpreter version from Python-tag.
+
+    Parameters
+    ----------
+    tag: Python-tag, likely inferred from wheel name.
+
+
+    Returns
+    -------
+    str: interpreter version inferred from Python-tag
+    """
+    # We will parse the interpreter version of CPython or just
+    # whatever generic Python version is specified.
+    pattern = re.compile(r"^(py|cp)(\d{1,3})$")
+    parsed_tag = pattern.match(tag)
+    if parsed_tag:
+        digits = parsed_tag.group(2)
+        # As match succeeded len(digits) \in {1,2,3}
+        if len(digits) == 1:
+            return parsed_tag.group(2)
+        return f"{digits[0]}.{digits[1:]}"
+    return None
+
+
 def build_backend_commands(buildspec: BaseBuildSpecDict) -> list[str]:
     """Generate the installation commands for each inferred build backend.