Move json log file reader to ButlerLogRecords

This makes it available to test code and to other users without
having to limit reading to butler formatter.

python/lsst/daf/butler/core/logging.py

@@ -28,7 +28,7 @@
 import datetime
 import traceback
 from collections import defaultdict
-from typing import List, Union, Optional, ClassVar, Iterable, Iterator, Dict
+from typing import List, Union, Optional, ClassVar, Iterable, Iterator, Dict, IO
 
 from logging import LogRecord, StreamHandler, Formatter
 from pydantic import BaseModel, ValidationError
@@ -230,6 +230,56 @@ def from_records(cls, records: Iterable[ButlerLogRecord]) -> ButlerLogRecords:
         """
         return cls(__root__=list(records))
 
+    @classmethod
+    def from_file(cls, filename: str) -> ButlerLogRecords:
+        """Read records from file.
+
+        Parameters
+        ----------
+        filename : `str`
+            Name of file containing the JSON records.
+
+        Notes
+        -----
+        Works with one-record-per-line format JSON files and a direct
+        serialization of the Pydantic model.
+        """
+        with open(filename, "r") as fd:
+            return cls.from_stream(fd)
+
+    @classmethod
+    def from_stream(cls, stream: IO) -> ButlerLogRecords:
+        """Read records from I/O stream.
+
+        Parameters
+        ----------
+        stream : `typing.IO`
+            Stream from which to read JSON records.
+
+        Notes
+        -----
+        Works with one-record-per-line format JSON files and a direct
+        serialization of the Pydantic model.
+        """
+        first_line = stream.readline()
+
+        # Allow byte or str streams since pydantic supports either.
+        first_char = str(first_line[0])
+        if first_char == "[":
+            # This is a ButlerLogRecords model serialization.
+            all = first_line + stream.read()
+            return cls.parse_raw(all)
+
+        # A stream of records with one record per line.
+        if first_char != "{":
+            raise RuntimeError(f"Unrecognized JSON log format. First lines is '{first_line}'")
+        records = [ButlerLogRecord.parse_raw(first_line)]
+        for line in stream:
+            if line:  # Filter out blank lines.
+                records.append(ButlerLogRecord.parse_raw(line))
+
+        return cls.from_records(records)
+
     @property
     def log_format(self) -> str:
         if self._log_format is None:

python/lsst/daf/butler/formatters/logs.py

@@ -66,22 +66,7 @@ def _readFile(self, path: str, pytype: Optional[Type[Any]] = None) -> Any:
         elif not issubclass(pytype, ButlerLogRecords):
             raise RuntimeError(f"Python type {pytype} does not seem to be a ButlerLogRecords type")
 
-        with open(path, "r") as fd:
-            first = fd.readline()
-            if first.startswith("["):
-                # This is a ButlerLogRecords serialization.
-                all = first + fd.read()
-                return pytype.parse_raw(all)
-
-            # A stream of records with one record per line.
-            if not first.startswith("{"):
-                raise RuntimeError(f"Unrecognized JSON log format. First lines is '{first}'")
-            records = [ButlerLogRecord.parse_raw(first)]
-            for line in fd:
-                if line:  # Filter out blank lines.
-                    records.append(ButlerLogRecord.parse_raw(line))
-
-            return pytype.from_records(records)
+        return pytype.from_file(path)
 
     def _fromBytes(self, serializedDataset: bytes, pytype: Optional[Type[Any]] = None) -> Any:
         """Read the bytes object as a python object.
@@ -99,7 +84,7 @@ def _fromBytes(self, serializedDataset: bytes, pytype: Optional[Type[Any]] = Non
             The requested data as a Python object or None if the string could
             not be read.
         """
-        # Duplicates some of the logic from readFile above.
+        # Duplicates some of the logic from ButlerLogRecords.from_file
         if pytype is None:
             pytype = ButlerLogRecords
         elif not issubclass(pytype, ButlerLogRecords):

tests/test_logging.py

@@ -109,7 +109,8 @@ def testJsonLogStream(self):
 
         # Rewind the stream and pull messages out of it.
         stream.seek(0)
-        records = [ButlerLogRecord.parse_raw(line) for line in stream]
+        records = ButlerLogRecords.from_stream(stream)
+        self.assertIsInstance(records[0], ButlerLogRecord)
         self.assertEqual(records[0].message, "A message")
         self.assertEqual(records[1].levelname, "WARNING")