Merge most recent version from my master branch

.gitignore

@@ -0,0 +1,2 @@
+.idea/
+*.pyc

README.rst

@@ -20,3 +20,12 @@ Usage
     raven-harakiri <uwsgi_log_string_with_harakiri> --dsn=<your_sentry_dsn>
 
 First argument (or stdin from other process, if piped) should be text with python tracebacker log.
+
+Here's an example of a typical usage of --watch for use under
+[supervisord](http://supervisord.org/).
+
+    [program:raven-harakiri]
+    user=user
+    command=/path/to/a/virtual/env/bin/raven-harakiri --watch /tmp/uwsgi.log --thread-regex 'uWSGIWorker.*'
+    redirect_stderr=true
+    autorestart=true

raven_harakiri.py

@@ -8,14 +8,15 @@
     :license: BSD, see LICENSE for more details.
 """
 from __future__ import print_function
+import subprocess
 import fileinput
 import logging
 import re
 import sys
 from optparse import OptionParser
 from raven import get_version, os, Client
 from raven.utils import json
-from raven.utils.six import StringIO
+from six import StringIO
 from raven.utils.stacks import get_lines_from_file
 
 
@@ -28,7 +29,7 @@ def store_json(option, opt_str, value, parser):
     setattr(parser.values, option.dest, value)
 
 
-def convert_traceback(uwsgi_traceback):
+def convert_traceback(uwsgi_traceback, thread_regex=None):
     """ Convert uwsgi traceback string with following pattern to raven protocol traceback
             thread_id = %s
             filename = %s
@@ -39,7 +40,11 @@ def convert_traceback(uwsgi_traceback):
         More info: http://sentry.readthedocs.org/en/latest/developer/client/#building-the-json-packet
                    http://uwsgi-docs.readthedocs.org/en/latest/Tracebacker.html
 
-        thread_id = MainThread filename = /home/belonika/.pythonz/pythons/CPython-2.6.8/lib/python2.6/socket.py lineno = 554 function = create_connection line = sock.connect(sa)
+        thread_id = MainThread
+        filename = /home/belonika/.pythonz/pythons/CPython-2.6.8/lib/python2.6/socket.py
+        lineno = 554
+        function = create_connection
+        line = sock.connect(sa)
     """
     variables = ('thread_id', 'filename', 'lineno', 'function', 'line')
     regexp = r' '.join(('%s = (?P<%s>.+?)' % (var, var) for var in variables))
@@ -48,6 +53,9 @@ def convert_traceback(uwsgi_traceback):
         match = re.match(r'^%s$' % regexp, line)
         values = match.groupdict() if match else None
         if values:
+            if thread_regex and\
+                    not re.match(thread_regex, values['thread_id']):
+                continue
             frame_result = {
                 'abs_path': values['filename'],
                 'context_line': values['line'],
@@ -60,10 +68,12 @@ def convert_traceback(uwsgi_traceback):
                 'pre_context': [],
                 'vars': {}
             }
-            pre_context, context_line, post_context = get_lines_from_file(frame_result['abs_path'], frame_result['lineno'], 5)
+            pre_context, context_line, post_context = get_lines_from_file(frame_result['abs_path'],
+                                                                          frame_result['lineno'] - 1, 5)
             if context_line is not None:
                 frame_result.update({
                     'pre_context': pre_context,
+                    'context_line': context_line,
                     'post_context': post_context,
                 })
 
@@ -72,32 +82,44 @@ def convert_traceback(uwsgi_traceback):
     return traceback
 
 
-def send_message(client, options, traceback):
-    print("Client configuration:")
-    for k in ('servers', 'project', 'public_key', 'secret_key'):
-        print('  %-15s: %s' % (k, getattr(client, k)))
-    print()
+def extract_http(log):
+    for line in log.split('\n'):
+        match = re.match(
+            r'^[^-]+- HARAKIRI \[core (?P<core>.+)\] (?P<remote_addr>.+) - '
+            r'(?P<method>.+) (?P<url>.+) since (?P<begin_time>.+)$', line)
+        if match:
+            break
+    if match:
+        values = match.groupdict()
+        return values['method'], values['url'], values['remote_addr']
+    else:
+        return None
 
-    if not all([client.servers, client.project, client.public_key, client.secret_key]):
-        print("Error: All values must be set!")
-        sys.exit(1)
 
+def send_message(client, options, log):
     if not client.is_enabled():
         print('Error: Client reports as being disabled!')
         sys.exit(1)
 
+    http_info = extract_http(log)
+
     data = {
         'logger': 'uwsgi.harakiri',
-        'sentry.interfaces.Stacktrace': {
-            'frames': convert_traceback(traceback)
+        'stacktrace': {
+            'frames': convert_traceback(log, options.get('thread_regex'))
         },
-        'sentry.interfaces.Http': {
-            'method': 'GET',
-            'url': 'http://example.com',
-        }
     }
 
-    print('Sending a message...',)
+    if http_info:
+        method, url, remote_addr = http_info
+        data['request'] = {
+            'method': method,
+            'url': url,
+            'env': {
+                'REMOTE_ADDR': remote_addr
+            }
+        }
+
     ident = client.get_ident(client.captureMessage(
         message='uWSGI harakiri',
         data=data,
@@ -107,11 +129,32 @@ def send_message(client, options, traceback):
     ))
 
     if client.state.did_fail():
-        print('error!')
         return False
 
-    print('success!')
-    print('Event ID was %r' % (ident,))
+    return ident
+
+
+def group_log(client, options, to_follow, proc=None):
+    in_harakiri = False
+    harakiri_lines = []
+    harakiris_seen = 0
+    while 1:
+        line = to_follow.readline()
+        if not line:
+            if proc and proc.returncode:
+                print("Error: tail exited uncleanly -- you must use GNU tail")
+                os.exit(1)
+            break
+        if 'HARAKIRI ON WORKER' in line:
+            in_harakiri = True
+        if in_harakiri:
+            harakiri_lines.append(line)
+            harakiris_seen += 'HARAKIRI' in line
+            if harakiris_seen == 4:
+                send_message(client, options, log="".join(harakiri_lines))
+                in_harakiri = False
+                harakiri_lines = []
+                harakiris_seen = 0
 
 
 def main():
@@ -123,9 +166,21 @@ def main():
     parser.add_option("--tags", action="callback", callback=store_json, type="string", nargs=1, dest="tags")
     parser.add_option("--verbose", action="store_true")
     parser.add_option("--dsn")
+    parser.add_option("--tail", action="store_true")
+    parser.add_option("--watch", action="store_true")
+    parser.add_option("--thread-regex")
     opts, args = parser.parse_args()
 
     dsn = opts.dsn or os.environ.get('SENTRY_DSN')
+
+    if not dsn:
+        try:
+            from django.conf import settings
+        except ImportError:
+            pass
+        else:
+            dsn = settings.RAVEN_CONFIG['dsn']
+
     if not dsn:
         print("Error: No configuration detected!")
         print("You must either pass a DSN to the command, or set the SENTRY_DSN environment variable.")
@@ -134,14 +189,26 @@ def main():
     if not opts.verbose:
         sys.stdout = StringIO()
 
-    print("Using DSN configuration:")
-    print(" ", dsn)
-    print()
-
-    traceback = ''.join([line for line in fileinput.input(args)])
-
     client = Client(dsn, include_paths=['raven'], string_max_length=100000)
-    send_message(client, opts.__dict__, traceback=traceback)
+
+    if opts.watch:
+        if len(args) != 1:
+            print('Error: In --watch mode you must provide exactly one file '
+                  'argument')
+            sys.exit(1)
+        tail = subprocess.Popen(['tail', '--follow=name', '--retry', args[0]], stdout=subprocess.PIPE)
+        group_log(client, opts.__dict__, tail.stdout, tail)
+
+    elif opts.tail:
+        if len(args) > 0:
+            print('Error: In --tail mode, no file arguments should be provided'
+                  ' -- pipe to STDIN instead')
+            sys.exit(1)
+        group_log(client, opts.__dict__, sys.stdin)
+
+    else:
+        log = ''.join([line for line in fileinput.input(args)])
+        send_message(client, opts.__dict__, log=log)
 
 
 if __name__ == '__main__':

setup.py

@@ -1,11 +1,11 @@
 import sys
-from setuptools.command.test import test as TestCommand
+import setuptools.command.test
 from setuptools import setup
 
 
-class PyTest(TestCommand):
+class PyTest(setuptools.command.test.test):
     def finalize_options(self):
-        TestCommand.finalize_options(self)
+        setuptools.command.test.test.finalize_options(self)
         self.test_args = []
         self.test_suite = True
 
@@ -17,20 +17,20 @@ def run_tests(self):
 
 setup(
     name='raven-harakiri',
-    version='0.1.2',
+    version='0.1.8',
     py_modules=['raven_harakiri'],
-    url='http://github.com/futurecolors/raven-harakiri',
+    url='http://github.com/LexMachinaInc/raven-harakiri',
     license='MIT',
-    author='Ilya Baryshev',
-    author_email='baryshev@gmail.com',
+    author='Sean Davis',
+    author_email='sdavis@lexmachina.com',
     description='Send UWSGI harakiri logs to sentry',
     long_description=open('README.rst').read(),
     entry_points={
         'console_scripts': [
             'raven-harakiri = raven_harakiri:main',
         ],
     },
-    install_requires=['raven>=3.4'],
+    install_requires=['raven>=3.4', 'six'],
     tests_require=['pytest'],
     cmdclass={'test': PyTest},
     classifiers=[

test_harakiri.py

@@ -61,7 +61,7 @@
 /lib/libpthread.so.0(+0x68ca) [0x7f2c44b4a8ca]
 /lib/libc.so.6(clone+0x6d) [0x7f2c43413b6d]"""
 
-from raven_harakiri import convert_traceback
+from raven_harakiri import convert_traceback, extract_http
 
 
 def test_convert():
@@ -78,3 +78,17 @@ def test_convert():
         'pre_context': [],
         'vars': {}
     }
+
+
+example_http = \
+"""                                                                                                                                                                        Mon Jan  2 14:43:09 2017 - HARAKIRI !!! worker 1 status !!!
+Mon Jan  2 14:43:09 2017 - HARAKIRI [core 0] 127.0.0.1 - GET /trigger_harakiri/ since 1483368158 
+Mon Jan  2 14:43:09 2017 - HARAKIRI !!! end of worker 1 status !!!
+"""
+
+
+def test_extract_http():
+    method, url, remote_addr = extract_http(example_http)
+    assert method == 'GET'
+    assert url == '/trigger_harakiri/'
+    assert remote_addr == '127.0.0.1'