updated header parsing to load a string just after is_32bit byte, adapted tests to check correct behaviour

Author: planrich

jitlog/objects.py

@@ -345,9 +345,10 @@ def read_python_source(file):
         return data
 
 class TraceForest(object):
-    def __init__(self, version, is_32bit=False):
+    def __init__(self, version, is_32bit=False, machine=None):
         self.word_size = 4 if is_32bit else 8
         self.version = version
+        self.machine = machine
         self.roots = []
         self.traces = {}
         self.addrs = {}

jitlog/parser.py

@@ -1,6 +1,7 @@
 import sys
 from jitlog import constants as const, marks
 from jitlog.objects import TraceForest
+from vmprof.binary import read_string
 if sys.version_info[0] >= 3:
     from io import StringIO
 else:
@@ -27,7 +28,8 @@ def _parse_jitlog(fileobj):
     is_jit_log = fileobj.read(1) == const.MARK_JITLOG_HEADER
     version = ord(fileobj.read(1)) | (ord(fileobj.read(1)) << 8)
     is_32bit = ord(fileobj.read(1))
-    forest = TraceForest(version, is_32bit)
+    machine = read_string(fileobj, True)
+    forest = TraceForest(version, is_32bit, machine)
     assert is_jit_log, "Missing header. Data might not be a jitlog!"
     assert version >= JITLOG_MIN_VERSION, \
             "Version does not match. Log is version %d%d is not satisfied" % \

jitlog/test/test_jitlog.py

@@ -240,17 +240,20 @@ def test_counter_points():
     assert len(d) == 2
 
 def test_32bit_log_header():
-    fobj = FileObj([const.MARK_JITLOG_HEADER+ b"\x01\x00\x01"])
+    fobj = FileObj([const.MARK_JITLOG_HEADER+ b"\x01\x00\x01"+\
+                    encode_str('ppc64le')])
     forest = _parse_jitlog(fobj)
     assert forest.version == 1
     assert forest.word_size == 4
+    assert forest.machine == 'ppc64le'
 
 def test_32bit_read_trace():
-    fobj = FileObj([const.MARK_JITLOG_HEADER+ b"\x01\x00\x01",
+    fobj = FileObj([const.MARK_JITLOG_HEADER+ b"\x01\x00\x01"+encode_str('s390x'),
                     const.MARK_START_TRACE, encode_le_u32(0x15), encode_str('loop'), encode_le_u32(0),
                    ])
     forest = _parse_jitlog(fobj)
     assert forest.version == 1
     assert forest.word_size == 4
     assert len(forest.traces) == 1
+    assert forest.machine == 's390x'