Fix a bug in backtrace parsing; add extractor for python-level backtraces

2 files changed, 85 insertions, 17 deletions

diff --git a/backtrace.py b/backtrace.py
index 57e907e..cebba2f 100644
--- a/backtrace.py
+++ b/backtrace.py
@@ -13,6 +13,26 @@ class Frame(object):
         self.function = function
         self.info = info
 
+    def as_python_frame(self):
+        '''For relevant frames, return (file, line, funcname, args) tuple; otherwise None'''
+        if self.function != 'PyEval_EvalFrameEx':
+            return None
+        return PythonFrame(self)
+    #result += '#%i %s\n' % (frame.index, frame.info)
+
+class PythonFrame(object):
+    '''
+    Class representating an interpretation of a C stack frame as a Python stack frame
+    '''
+    def __init__(self, c_frame):
+        self.c_frame = c_frame
+        m = re.match('\(f=Frame 0x[0-9a-f]+, for file (\S+), line ([0-9]+), in (\S+) \((.*), throwflag=0\) at (\S+)',
+                     c_frame.info)
+        self.file = m.group(1)
+        self.linenum = m.group(2)
+        self.funcname = m.group(3)
+        self.vars = m.group(4)
+
 class Thread(object):
     '''
     Class representing a thread within a backtrace
@@ -23,10 +43,30 @@ class Thread(object):
         self.frames = {}
         self.framelist = []
 
+    def as_python_backtrace(self, with_vars):
+        '''Generate a textual guess at a python-level backtrace'''
+        result = ''
+        for frame in self.framelist:
+            pyframe = frame.as_python_frame()
+            if pyframe:
+                result += '#%i %s:%s %s\n' % (frame.index,
+                                              pyframe.file,
+                                              pyframe.linenum,
+                                              pyframe.funcname)
+                if with_vars:
+                    result += '    %s\n' % (pyframe.vars, )
+        return result
+
 class Backtrace(object):
     '''
     Class representing a parsed backtrace
     '''
+    @classmethod
+    def from_text_file(cls, filename, debug=False):
+        with open(filename, 'r') as f:
+            text = f.read()
+        return Backtrace(text, debug)
+
     def __init__(self, string, debug=False):
         '''
         Parse the given string (from gdb)
@@ -51,7 +91,7 @@ class Backtrace(object):
                 self._frame = None
                 continue
 
-            m = re.match('^#([0-9]+)\s+(?:0x([0-9a-f]+) in)? (\S+) (.*)$', line)
+            m = re.match('^#([0-9]+)\s+(?:0x([0-9a-f]+) in )?(\S+) (.*)$', line)
             if m:
                 if debug:
                     print 'STACK FRAME:', m.groups()
@@ -64,18 +104,8 @@ class Backtrace(object):
                           address,
                           m.group(3),
                           m.group(4))
-
-                if self._thread is None:
-                    if debug:
-                        print 'GOT CRASH SITE'
-                    self._crash_site = f
-                    self._frame = None
-                    continue
-                else:
-                    self._thread.frames[f.index] = f
-                    self._thread.framelist.append(f)
-                    self._frame = f
-                    continue
+                self.__on_new_frame(f, debug)
+                continue
 
             if line.startswith(' '):
                 if self._frame:
@@ -83,6 +113,17 @@ class Backtrace(object):
 
         #pprint(self._threads[5].frames[6].__dict__)
 
+    def __on_new_frame(self, f, debug):
+        if self._thread is None:
+            if debug:
+                print 'GOT CRASH SITE'
+            self._crash_site = f
+            self._frame = None
+        else:
+            self._thread.frames[f.index] = f
+            self._thread.framelist.append(f)
+            self._frame = f
+
     def get_crash_site(self):
         '''Get a (Thread, Frame) pair for where the crash happened (or None)'''
         debug = False
@@ -96,3 +137,8 @@ class Backtrace(object):
                     print t.frames[0].__dict__
                 if t.frames[0].address == self._crash_site.address:
                     return (t, t.frames[0])
+
+    def as_python_backtrace(self, with_vars):
+        '''Generate a textual guess at a python-level backtrace'''
+        thread, frame = self.get_crash_site()
+        return thread.as_python_backtrace(with_vars)
diff --git a/test_backtrace_parser.py b/test_backtrace_parser.py
index e2d831f..713eadb 100644
--- a/test_backtrace_parser.py
+++ b/test_backtrace_parser.py
@@ -3,14 +3,16 @@ from backtrace import Backtrace
 
 class TestParser(unittest.TestCase):
     def test_bz677051(self):
-        text = open('data/rhbz-677051-attachment-478410.txt', 'r').read()
-        bt = Backtrace(text)
+        bt = Backtrace.from_text_file('data/rhbz-677051-attachment-478410.txt')
         
         thread, frame = bt.get_crash_site()
         self.assertEquals(thread.index, 1)
         self.assertEquals(thread.extra, '0xb78516c0 (LWP 17355)')
-        #self.assertEquals(len(thread.frames), 59)
-        #self.assertEquals(len(thread.framelist), 59)
+        self.assertEquals(len(thread.frames), 60)
+        self.assertEquals(len(thread.framelist), 60)
+
+        # Frame #13 wasn't handled by an earlier version of the parser:
+        self.assert_(13 in thread.frames)
 
         self.assertEquals(frame.index, 0)
         self.assertEquals(frame.address, 0x4d2ade0a)
@@ -24,6 +26,26 @@ class TestParser(unittest.TestCase):
                           "        base = <optimized out>\n"
                           "        dict = <optimized out>\n"
                           "        h = <optimized out>")
+
+
+
+    def test_python_backtrace(self):
+        bt = Backtrace.from_text_file('data/rhbz-677051-attachment-478410.txt')
+
+        pybt = bt.as_python_backtrace(False)
+
+        self.assertEquals(pybt,
+                          '#4 /usr/lib/python2.7/site-packages/yum/sqlitesack.py:1362 searchPrco\n'
+                          '#9 /usr/lib/python2.7/site-packages/yum/sqlitesack.py:46 newFunc\n'
+                          '#13 /usr/lib/python2.7/site-packages/yum/sqlitesack.py:1400 searchProvides\n'
+                          '#23 /usr/lib/python2.7/site-packages/yum/packageSack.py:598 _computeAggregateListResult\n'
+                          '#27 /usr/lib/python2.7/site-packages/yum/packageSack.py:426 searchProvides\n'
+                          '#30 /usr/lib/python2.7/site-packages/yum/__init__.py:2990 returnPackagesByDep\n'
+                          '#33 /usr/lib/python2.7/site-packages/yum/__init__.py:3313 install\n'
+                          '#37 /usr/sbin/yum-complete-transaction:199 main\n'
+                          '#40 /usr/sbin/yum-complete-transaction:118 __init__\n'
+                          '#51 /usr/sbin/yum-complete-transaction:256 <module>\n')
+
         
 if __name__ == '__main__':
     unittest.main()