Merge lp:~joetalbott/utah/parser_update into lp:utah

[Joe Talbott]

This branch adds support for parsing UTAH client logs stored in string.

In addition there is improved error handling and I removed a hardcoded path.

[Javier Collado (javier.collado)]

Looks fine to me.

I've seen http support has been added to download a remote file. Note that you
can use utah.url.url_argument to add support for launchpad url's as well:

from utah.url import url_argument

# bzr export log_file_url
log_file = url_argument(log_file_url)

review: Approve

[Joe Talbott (joetalbott)]

This code is to be packaged as a separate module and will be run from the dashboard. Once we get pieces of utah packaged individually I'll use the utah.url module. For the time being I'm going to merge.

=== modified file 'utah/parser.py'
--- utah/parser.py	2012-11-14 22:10:40 +0000
+++ utah/parser.py	2012-11-21 15:03:20 +0000
@@ -1,5 +1,8 @@
+import argparse
+import jsonschema
+import logging
 import os
-import jsonschema
+import urllib2
 import yaml
 
 class UTAHParser(object):
@@ -67,25 +70,31 @@
         },
     }
 
-    def parse(self, logdata):
+    def parse(self, logdata, as_string=False):
         """
         Parse utah client results.
 
         logdata should be either a filename or a handle to a file
         object.
 
-        Returns the parsed yaml data.
+        Returns the parsed yaml data or None.
 
         """
 
-        if isinstance(logdata, str):
+        if as_string:
+            return self._parse_string(logdata)
+        elif isinstance(logdata, str):
             return self._parse_logfile(logdata)
         else:
             return self._parse_stream(logdata)
 
     def _parse_stream(self, stream):
         """ Parse client output from stream. """
-        data = yaml.load(stream)
+        try:
+            data = yaml.load(stream)
+        except yaml.YAMLError as e:
+            logging.error(e)
+            return None
 
         jsonschema.validate(data, self.CLIENT_OUTPUT_SCHEMA)
 
@@ -96,13 +105,36 @@
 
         data = None
 
-        with open(logfile, 'r') as fp:
+        if logfile.startswith('http'):
+            fp = urllib2.urlopen(logfile)
             data = self._parse_stream(fp)
-
-        return data
+        else:
+            with open(logfile, 'r') as fp:
+                data = self._parse_stream(fp)
+
+        return data
+
+    def _parse_string(self, log_data):
+        """ Parse client data from a string. """
+        try:
+            data = yaml.load(log_data)
+        except yaml.YAMLError as e:
+            logging.error(e)
+            return None
+
+        jsonschema.validate(data, self.CLIENT_OUTPUT_SCHEMA)
+
+        return data
+
 
 if __name__ == "__main__":
+    arg_parser = argparse.ArgumentParser(description='phoenix bootstrapper')
+    arg_parser.add_argument('logfile', metavar='LOGFILE', type=str,
+            help='log file to parse')
+
+    args = arg_parser.parse_args()
+
     parser = UTAHParser()
-    data = parser.parse("/tmp/utah.yaml")
+    data = parser.parse(args.logfile)
     print(data)