Add support for ec2 specific contextual fields

README.md

@@ -24,9 +24,10 @@ Note that the following documentation relates to the code currently in the maste
 * Handler agnosticism
 * Differentiate user events and system logs
 * Auto-enrich with useful data (hostname, pid, etc..)
+* Enrich with AWS EC2 metadata (instance-id, instance-type, region and ipv4) if available
 * Easily provide contexual data
-* Dynamic severity levels
 * Context binding to prevent repetition
+* Dynamic severity levels
 * Retroactive logging (WIP)
 * Assist in user tracing (via auto-provided context ids)
 * Zero logging exceptions. There should be zero logging exceptions causing the app to crash but rather Wryte should log errors whenever logging errors occur.
@@ -260,7 +261,7 @@ wryter.critical('Logging nested dicts', {'key': 'value', 'nested': { 'key1': 'va
 
 ```
 
-#### Reserved Fields
+#### Reserved Contextual Fields
 
 The following are fields reserved to Wryte:
 
@@ -272,9 +273,18 @@ The following are fields reserved to Wryte:
 * `level`
 * `type`
 * `name`
+* `ec2_instance_id` (optional)
+* `ec2_instance_type` (optional)
+* `ec2_region` (optional)
+* `ec2_ipv4` (optional)
 
 Note that even if the fields are provided in a log message, they are bound to be overriden by Wryte.
 
+#### Enriching with AWS EC2 instance context
+
+The optional fields mentioned above are added to the context if the `WRYTE_EC2_ENABLED` env var is set.
+
+Note that if the data is unattainable for any reason, Wryte will spit out an error message stating so when the logger is instantiated.
 
 #### Binding contextual information to a logger
 

tests/test_wryte.py

@@ -1,3 +1,4 @@
+import os
 import sys
 import uuid
 import shlex
@@ -193,3 +194,8 @@ def test_set_level_from_log(self):
 
     def test_cli(self):
         _invoke('main info My Message x=y')
+
+    def test_aws(self):
+        os.environ['WRYTE_EC2_ENABLED'] = 'true'
+        w = Wryte(name=str(uuid.uuid4()))
+        w.info('Message')

wryte.py

@@ -20,6 +20,13 @@
 import logging
 from datetime import datetime
 
+try:
+    # Python 2
+    import urllib2 as urllib
+except ImportError:
+    # Python 3
+    import urllib.request as urllib
+
 try:
     import colorama
     from colorama import Fore, Style
@@ -156,7 +163,8 @@ def __init__(self,
                  bare=False,
                  jsonify=False,
                  color=True,
-                 simple=False):
+                 simple=False,
+                 enable_ec2=False):
         """Instantiate a logger instance.
 
         Either a JSON or a Console handler will be added to the logger
@@ -177,8 +185,8 @@ def __init__(self,
         self.color = color
         self.simple = simple
 
-        self._log = self._get_base(self.logger_name, hostname)
         self.logger = self._logger(self.logger_name)
+        self._log = self._get_base(self.logger_name, hostname, enable_ec2)
 
         if not bare:
             self._configure_handlers(level, jsonify)
@@ -190,20 +198,44 @@ def _logger(name):
         logger = logging.getLogger(name)
         return logger
 
-    @staticmethod
-    def _get_base(name, hostname):
+    def _get_base(self, name, hostname, enable_ec2=False):
         """Generate base fields for each log message.
 
         This is evaluated once when the logger's instance is instantiated.
         It is then later copied by each log message.
         """
-        return {
+        def fetch_ec2(attribute):
+            try:
+                return urllib.urlopen(
+                    'http://169.254.169.254/latest/meta-data/{0}'.format(
+                        attribute)).read().decode()
+            # Yuch. But shouldn't take a risk that any exception will raise
+            except Exception:
+                return None
+
+        # TODO: Document that these are only generated once.
+        base = {
             'name': name,
             'hostname': hostname or socket.gethostname(),
             'pid': os.getpid(),
-            'type': 'log'
+            'type': 'log',
         }
 
+        if self._env('EC2_ENABLED') or enable_ec2:
+            # To test that ec2 data is actually attainable
+            instance_id = fetch_ec2('instance-id')
+            if instance_id:
+                base['ec2_instance_id'] = instance_id
+                base['ec2_instance_type'] = fetch_ec2('instance-type')
+                base['ec2_region'] = fetch_ec2('placement/availability-zone')
+                base['ec2_ipv4'] = fetch_ec2('local-ipv4')
+            else:
+                self.logger.error(
+                    'WRYTE EC2 env var set but EC2 metadata endpoint is '
+                    'unavailable or the data could not be retrieved.')
+
+        return base
+
     @staticmethod
     def _get_timestamp():
         # `now()` needs to compensate for timezones, and so it takes much