WIPpython

lisa/trace.py

@@ -427,6 +427,17 @@ def get_metadata(self, key):
             super().get_metadata(key=key)
 
 
+class TraceDumpError(Exception):
+    """
+    Exception containing errors forwarded from the trace-dump parser
+    """
+    def __init__(self, errors):
+        self.errors = sorted(set(errors))
+
+    def __str__(self):
+        return '\n'.join(self.errors)
+
+
 class TraceDumpTraceParser(TraceParserBase):
     """
     trace.dat parser shipped by LISA
@@ -436,7 +447,7 @@ def __init__(self, path, events, needed_metadata=None, **kwargs):
         super().__init__(events=events, needed_metadata=needed_metadata, **kwargs)
         self._trace_path = str(Path(path).resolve())
         self._metadata = {}
-        self._parquet_dir = None
+        self._temp_dir = None
 
     def __enter__(self):
         events = self._requested_events
@@ -447,38 +458,43 @@ def __enter__(self):
         )
         meta = {}
 
+        temp_dir = tempfile.TemporaryDirectory()
+        self._temp_dir = temp_dir
+
         # time-range will not be available in the basic metadata, this requires
         # a full parse
         if events or ('time-range' in needed_metadata):
-            parquet_dir = tempfile.TemporaryDirectory()
-            self._parquet_dir = parquet_dir
-
             meta = self._make_parquets(
                 path=path,
                 events=events,
-                parquet_dir=parquet_dir.name
+                temp_dir=temp_dir.name
             )
         elif needed_metadata:
-            meta = self._make_metadata(path=path)
+            meta = self._make_metadata(
+                path=path,
+                temp_dir=temp_dir.name
+            )
 
         self._metadata.update(meta)
         return self
 
     def __exit__(self, *args, **kwargs):
-        if (parquet_dir := self._parquet_dir):
-            parquet_dir.__exit__(*args, **kwargs)
+        if (temp_dir := self._temp_dir):
+            temp_dir.__exit__(*args, **kwargs)
 
     @classmethod
     def from_dat(cls, path, events, **kwargs):
         return cls(path=path, events=events, **kwargs)
 
     @classmethod
-    def _run(cls, cli_args, run_kwargs=None):
+    def _run(cls, cli_args, temp_dir, run_kwargs=None):
         logger = cls.get_logger()
 
         trace_dump = get_bin('trace-dump')
+        errors_path = Path(temp_dir) / 'errors.json'
         cmd = (
             trace_dump,
+            '--errors-json', errors_path,
             *cli_args,
         )
         pretty_cmd = ' '.join(map(shlex.quote, map(str, cmd)))
@@ -501,6 +517,15 @@ def log(stdout, stderr):
             stdout = completed.stdout
             log(stdout, completed.stderr)
             return stdout
+        finally:
+            try:
+                with open(errors_path) as f:
+                    errors = json.load(f)
+            except FileNotFoundError:
+                pass
+            else:
+                if errors := errors['errors']:
+                    raise TraceDumpError(errors)
 
     @classmethod
     def _process_metadata(cls, meta):
@@ -519,20 +544,21 @@ def _process_metadata(cls, meta):
         return meta
 
     @classmethod
-    def _make_metadata(cls, path):
+    def _make_metadata(cls, path, temp_dir):
         stdout = cls._run(
-            (
+            cli_args=(
                 '--trace', path,
                 'metadata',
             ),
+            temp_dir=temp_dir,
         )
         return cls._process_metadata(json.loads(stdout))
 
     @classmethod
-    def _make_parquets(cls, events, path, parquet_dir):
-        parquet_dir = Path(parquet_dir)
+    def _make_parquets(cls, events, path, temp_dir):
+        temp_dir = Path(temp_dir)
         cls._run(
-            (
+            cli_args=(
                 '--trace', path,
                 'parquet',
                 '--unique-timestamps',
@@ -546,11 +572,12 @@ def _make_parquets(cls, events, path, parquet_dir):
                     for arg in ('--events', _event)
                 )
             ),
-            dict(cwd=parquet_dir),
+            run_kwargs=dict(cwd=temp_dir),
+            temp_dir=temp_dir,
         )
 
-        with open(parquet_dir / 'meta.json') as metaf:
-            meta = json.load(metaf)
+        with open(temp_dir / 'meta.json') as f:
+            meta = json.load(f)
 
         meta = cls._process_metadata(meta)
         return meta
@@ -580,9 +607,9 @@ def __missing__(self, _):
     def _parse_event(self, event):
         desc = self._event_descs[event]
         pid_comms = self._pid_comms
-        parquet_dir = Path(self._parquet_dir.name)
+        temp_dir = Path(self._temp_dir.name)
 
-        df = pd.read_parquet(parquet_dir / desc['path'])
+        df = pd.read_parquet(temp_dir / desc['path'])
         df = self._fixup_df(
             df=df,
             pid_comms=pid_comms,