Add config file parsing

MANIFEST.in

@@ -1 +1 @@
-include README LICENSE requirements.txt
+include README.rst requirements.txt bwscanner/data/config.ini

bwscanner/configutil.py

@@ -0,0 +1,46 @@
+import os.path
+from ConfigParser import SafeConfigParser
+from pkg_resources import resource_string
+
+from bwscanner.logger import log
+
+
+def read_config(cfg_path):
+    log.debug('Reading config %s' % cfg_path)
+    if not config_exists(cfg_path):
+        copy_config(cfg_path)
+    parser = SafeConfigParser()
+    parser.read([cfg_path])
+    cfg_dict = dict(parser.items('default'))
+    int_keys = cfg_dict['int_keys'].split(' ')
+    bool_keys = cfg_dict['bool_keys'].split(' ')
+    for k in int_keys:
+        cfg_dict[k] = int(cfg_dict[k])
+    for i in bool_keys:
+        cfg_dict[k] = bool(cfg_dict[k])
+    bw_files = dict(parser.items('bw_files'))
+    cfg_bw_files = {}
+    for k, v in bw_files.items():
+        print(k, v)
+        if 'm' in k:
+            number = k.rstrip('m')
+            size = 1024 * int(number)
+            cfg_bw_files[size] = (k.upper(), v)
+    cfg_dict['bw_files'] = cfg_bw_files
+    return cfg_dict
+
+
+def config_exists(cfg_path):
+    return os.path.isfile(cfg_path)
+
+
+def copy_config(cfg_path, cfg_default_path=None):
+    # FIXME: obtain the path instead of the content
+    if cfg_default_path is None:
+        content = resource_string(__name__, 'data/config.ini')
+    else:
+        with open(cfg_default_path) as fp:
+            content = cfg_default_path.read()
+    log.debug("cfg_default_path %s" % cfg_default_path)
+    with open(cfg_path, 'w') as fp:
+        fp.write(content)

bwscanner/data/config.ini

@@ -0,0 +1,25 @@
+[default]
+
+data_dir = ~/.config/bwscanner
+measurement_dir = ~/.config/bwscanner/measurements
+tor_dir = ~/.config/bwscanner/tordata
+loglevel = info
+logfile = bwscanner.log
+baseurl = https://siv.sunet.se/bwauth/
+launch_tor = True
+circuit_build_timeout = 20
+partitions = 1
+current_partition = 1
+timeout = 120
+request_limit = 10
+int_keys = partitions current_partition timeout request_limit
+bool_keys = launch_tor
+
+[bw_files]
+
+64M = 6258de4f4d602be75a3458117b29d2c580c4bcb7ba5b9d2c4135c7603109f554
+32M = 5a5d66d7865f09498d776f20c9e9791b055a4fff357185f84fb4ecfca7da93f0
+16M = 6258de4f4d602be75a3458117b29d2c580c4bcb7ba5b9d2c4135c7603109f554
+8M = 738c5604295b9377f7636ce0c2c116f093bb50372f589a6c2332a3bb6bba096a
+4M = 4daaa42377d3c87577797d44a8fa569038e7a9d6a5d417a09d8ba41a69456164
+2M = 3e39b0bb92912cf1ad6c01fb7c9d592e814a691c61de1f649416f6bba2d15082

bwscanner/scanner.py

@@ -6,22 +6,30 @@
 from twisted.internet import reactor
 
 from bwscanner.attacher import connect_to_tor
+from bwscanner.configutil import read_config
 from bwscanner.logger import setup_logging, log
 from bwscanner.measurement import BwScan
 from bwscanner.aggregate import write_aggregate_data
 
 
 BWSCAN_VERSION = '0.0.1'
+APP_NAME = 'bwscanner'
+DATA_DIR = os.environ.get("BWSCANNER_DATADIR", click.get_app_dir(APP_NAME))
+CONFIG_FILE = 'config.ini'
+LOG_FILE = 'bwscanner.log'
 
 
 class ScanInstance(object):
     """
     Store the configuration and state for the CLI tool.
     """
-    def __init__(self, data_dir):
+    def __init__(self, data_dir, measurement_dir=None):
         self.data_dir = data_dir
-        self.measurement_dir = os.path.join(data_dir, 'measurements')
-        self.tor_dir = os.path.join(data_dir, 'tor_data')
+        if measurement_dir is None:
+            self.measurement_dir = os.path.join(data_dir, 'measurements')
+        else:
+            self.measurement_dir = measurement_dir
+        self.tor_state = None
 
     def __repr__(self):
         return '<BWScan %r>' % self.data_dir
@@ -30,18 +38,22 @@ def __repr__(self):
 pass_scan = click.make_pass_decorator(ScanInstance)
 
 
+# FIXME: check these errors
+# pylint: disable=unexpected-keyword-arg
+# pylint: disable=no-value-for-parameter
 @click.group()
 @click.option('--data-dir', type=click.Path(),
-              default=os.environ.get("BWSCANNER_DATADIR", click.get_app_dir('bwscanner')),
               help='Directory where bwscan should stores its measurements and '
               'other data.')
-@click.option('-l', '--loglevel', help='The logging level the scanner will use (default: info)',
-              default='info', type=click.Choice(['debug', 'info', 'warn', 'error', 'critical']))
-@click.option('-f', '--logfile', type=click.Path(), help='The file the log will be written to',
-              default=os.environ.get("BWSCANNER_LOGFILE", 'bwscanner.log'))
-@click.option('--launch-tor/--no-launch-tor', default=False,
+@click.option('-l', '--loglevel',
+              help='The logging level the scanner will use (default: info)',
+              type=click.Choice(
+                      ['debug', 'info', 'warn', 'error', 'critical']))
+@click.option('-f', '--logfile', type=click.Path(),
+              help='The file the log will be written to')
+@click.option('--launch-tor/--no-launch-tor',
               help='Launch Tor or try to connect to an existing Tor instance.')
-@click.option('--circuit-build-timeout', default=20,
+@click.option('--circuit-build-timeout',
               help='Option passed when launching Tor.')
 @click.version_option(BWSCAN_VERSION)
 @click.pass_context
@@ -51,9 +63,12 @@ def cli(ctx, data_dir, loglevel, logfile, launch_tor, circuit_build_timeout):
     bandwidth measurements can then be aggregate to create the bandwidth
     values used by the Tor bandwidth authorities when creating the Tor consensus.
     """
+    for k, v in ctx.default_map.items():
+        if ctx.params.get(k) is None:
+            ctx.params[k] = v
+
     # Create the data directory if it doesn't exist
-    data_dir = os.path.abspath(data_dir)
-    ctx.obj = ScanInstance(data_dir)
+    ctx.obj = ScanInstance(ctx.params.get('data_dir'))
 
     if not os.path.isdir(ctx.obj.measurement_dir):
         os.makedirs(ctx.obj.measurement_dir)
@@ -63,7 +78,8 @@ def cli(ctx, data_dir, loglevel, logfile, launch_tor, circuit_build_timeout):
                                        ctx.obj.tor_dir)
 
     # Set up the logger to only output log lines of level `loglevel` and above.
-    setup_logging(log_level=loglevel, log_name=logfile)
+    setup_logging(log_level=ctx.params.get('loglevel'),
+                  log_name=ctx.params.get('logfile'))
 
 
 @cli.command(short_help="Measure the Tor relays.")
@@ -85,7 +101,8 @@ def scan(scan, partitions, current_partition, timeout, request_limit):
 
     # XXX: check that each run is producing the same input set!
     scan_time = str(int(time.time()))
-    scan_data_dir = os.path.join(scan.measurement_dir, '{}.running'.format(scan_time))
+    scan_data_dir = os.path.join(scan.measurement_dir,
+                                 '{}.running'.format(scan_time))
     if not os.path.isdir(scan_data_dir):
         os.makedirs(scan_data_dir)
 
@@ -160,3 +177,8 @@ def aggregate(scan, scan_name, previous):
     scan.tor_state.addErrback(lambda failure: log.failure("Unexpected error"))
     scan.tor_state.addCallback(lambda _: reactor.stop())
     reactor.run()
+
+
+def start():
+    config = read_config(os.path.join(DATA_DIR, CONFIG_FILE))
+    return cli(default_map=config)

setup.py

@@ -39,9 +39,9 @@
       },
       python_requires=">=2.7",
       # data_files = [('path', ['filename'])]
-      data_files=[],
+      include_package_data=True,
       entry_points={
         "console_scripts": [
-            'bwscan = bwscanner.scanner:cli',
+            'bwscan = bwscanner.scanner:start',
         ]},
      )