Merge pull request #40 from uclahs-cds/nwiltsie-mask-versions

Add ability to mask absolute version numbers in Nextflow tests

run-nextflow-tests/README.md

@@ -16,7 +16,8 @@ Configuration tests are self-contained JSON files named `configtest*.json` with
 | nf_params | A map of command-line parameters to pass to Nextflow (`nextflow --<key>=<value>`) |
 | envvars | A map of environment variables to set (`KEY=VALUE nextflow ...`) |
 | mocks | Method names to be mocked, mapped to the objects they should return |
-| dated_field | A list of JSONPath-like keys indicating values in the rendered configuration that contain datestamps |
+| dated_fields | A list of JSONPath-like keys indicating values in the rendered configuration that contain datestamps |
+| version_fields | A list of JSONPath-like keys indicating values in the rendered configuration that contain the pipeline version number |
 | expected_results | The expected output of the test |
 
 For each test, this Action parses the configuration and runs a modified version of [`nextflow config`](https://www.nextflow.io/docs/latest/cli.html#config), comparing the results against `expected_results` and warning about any differences.
@@ -153,6 +154,14 @@ jobs:
     "trace.file",
     "params.date"
   ],
+  "version_fields": [
+    "manifest.version",
+    "params.log_output_dir",
+    "params.output_dir_base",
+    "report.file",
+    "trace.file",
+    "timeline.file"
+  ],
   "expected_result": {}
 }
 ```
@@ -162,6 +171,7 @@ jobs:
 The true Nextflow configuration output is slightly modified for usability:
 
 * Every field listed in `dated_fields` has timestamps matching the format `YYYYMMDDTHHMMSSZ` replaced with the static value `19970704T165655Z` ([Pathfinder's landing](https://science.nasa.gov/mission/mars-pathfinder/)).
+* Every field listed in `version_fields` has sub-strings matching the `manifest.version` value replaced with the static value `VER.SI.ON`.
 * Every value that looks like a Java object (e.g. `[Ljava.lang.String;@49c7b90e`) has the hash code replaced with the static value `dec0ded`.
     * These should not appear in test files. When they do, it is a sign that the corresponding variable is missing a `def` in the configuration file.
 * Closures are expressed as the first valid item in this list:

run-nextflow-tests/configtest.py

@@ -27,6 +27,20 @@ class NextflowConfigTest:
     # pylint: disable=too-many-instance-attributes
     SENTINEL: ClassVar = "=========SENTINEL_OUTPUT=========="
 
+    # python3.7 doesn't support `kw_only` and other useful dataclass features.
+    # These two fields are workarounds for that.
+    OPTIONAL_DICTS: ClassVar = {
+        "nf_params",
+        "envvars",
+        "mocks",
+        "dated_fields",
+        "version_fields",
+    }
+    OPTIONAL_LISTS: ClassVar = {
+        "dated_fields",
+        "version_fields",
+    }
+
     pipeline: Path = dataclasses.field(init=False, compare=False)
     filepath: Path = dataclasses.field(init=False, compare=False)
 
@@ -42,6 +56,7 @@ class NextflowConfigTest:
     mocks: Dict
 
     dated_fields: List[str]
+    version_fields: List[str]
 
     expected_result: Dict
 
@@ -55,6 +70,12 @@ def from_file(cls: Type[T], pipeline: Path, filepath: Path) -> T:
         data.pop("empty_files", None)
         data.pop("mapped_files", None)
 
+        for fieldname in cls.OPTIONAL_DICTS:
+            data.setdefault(fieldname, {})
+
+        for fieldname in cls.OPTIONAL_LISTS:
+            data.setdefault(fieldname, [])
+
         result = cls(**data)
         result.pipeline = pipeline
         result.filepath = filepath.resolve()
@@ -82,6 +103,11 @@ def to_file(self):
         data = dataclasses.asdict(self)
         data.pop("pipeline")
 
+        # Strip any empty optional fields from the output
+        for field in self.OPTIONAL_DICTS | self.OPTIONAL_LISTS:
+            if not data[field]:
+                data.pop(field)
+
         with data.pop("filepath").open(mode="w") as outfile:
             json.dump(
                 data,
@@ -181,12 +207,14 @@ def _run_test(self):
         config_text = config_output.rsplit(self.SENTINEL, maxsplit=1)[-1]
 
         try:
-            return parse_config(config_text, self.dated_fields)
+            return parse_config(
+                config_text, self.dated_fields, self.version_fields
+            )
         except Exception:
             print(config_output)
             raise
 
-    def print_diffs(self, other: T):
+    def print_diffs(self, other: "NextflowConfigTest"):
         "Print the diff results to the console."
         diff_process = subprocess.run(
             ["diff", self.filepath, other.filepath],

run-nextflow-tests/utils.py

@@ -153,9 +153,12 @@ def parse_value(value_str: str) -> Any:
     return value
 
 
-def parse_config(config_str: str, dated_fields: List[str]) -> dict:
+def parse_config(config_str: str,
+                 dated_fields: List[str],
+                 version_fields: List[str]) -> dict:
     "Parse a string of Java properties."
     param_re = re.compile(r"^(?P<key>\S+?[^\\])=(?P<value>.*)$")
+    version_fields = list(version_fields)
 
     def assign_value(closure, key, value):
         if "." not in key:
@@ -170,6 +173,17 @@ def assign_value(closure, key, value):
 
             assign_value(closure[local_key], remainder, value)
 
+    # Parse out the current manifest version
+    try:
+        version_str = re.search(
+            r"^manifest.version=(.*)$",
+            config_str,
+            re.MULTILINE
+        ).group(1)
+
+    except AttributeError:
+        version_str = None
+
     config: dict[str, Any] = {}
 
     for line in config_str.splitlines():
@@ -188,6 +202,10 @@ def assign_value(closure, key, value):
             # Replace the date with Pathfinder's landing
             value = DATE_RE.sub("19970704T165655Z", value)
 
+        if escaped_key in version_fields and version_str:
+            # Replace the version with an obvious weird value
+            value = value.replace(version_str, "VER.SI.ON")
+
         assign_value(config, escaped_key, value)
 
     # Specifically sort the config