diff --git a/frc971/analysis/logentry.py b/frc971/analysis/logentry.py
new file mode 100644
index 0000000..5b9a2ed
--- /dev/null
+++ b/frc971/analysis/logentry.py
@@ -0,0 +1,265 @@
+#!/usr/bin/python3
+
+import re
+import sys
+
+"""
+A regular expression to match the envelope part of the log entry.
+Parsing of the JSON msg is handled elsewhere.
+"""
+LOG_RE = re.compile("""
+  (.*?)              # 1 name
+  \((\d+)\)          # 2 pid
+  \((\d+)\)          # 3 message_index
+  :\s
+  (\w+?)             # 4 level
+  \s+at\s+
+  (\d+\.\d+)s        # 5 time
+  :\s
+  ([A-Za-z0-9_./-]+) # 6 filename
+  :\s
+  (\d+)              # 7 linenumber
+  :\s
+  (.*)               # 8 msg
+  """, re.VERBOSE)
+
+class LogEntry:
+  """
+  This class provides a way to parse log entries.
+  The header portion of the log entry is parsed eagerly.
+  The structured portion of a log entry is parsed on demand.
+  """
+
+  def __init__(self, line):
+    """Populates a LogEntry from a line."""
+    m = LOG_RE.match(line)
+    if m is None:
+        print("LOG_RE failed on", line)
+        sys.exit(1)
+    self.name = m.group(1)
+    self.pid_index = int(m.group(2))
+    self.msg_index = int(m.group(3))
+    self.level = m.group(4)
+    self.time = float(m.group(5))
+    self.filename = m.group(6)
+    self.linenumber = m.group(7)
+    self.msg = m.group(8)
+
+  def __str__(self):
+    """Formats the data cleanly."""
+    return '%s(%d)(%d): %s at %fs: %s: %d: %s' % (
+        self.name, self.pid, self.msg_index, self.level, self.time, self.filename, self.linenumber, self.msg)
+
+  def ParseStruct(self):
+    """Parses the message as a structure.
+
+    Returns:
+      struct_name, struct_type, json dict.
+    """
+    struct_name_index = self.msg.find(':')
+    struct_name = self.msg[0:struct_name_index]
+
+    struct_body = self.msg[struct_name_index+2:]
+    tokens = []
+    this_token = ''
+    # For the various deliminators, append what we have found so far to the
+    # list and the token.
+    for char in struct_body:
+      if char == '{':
+        if this_token:
+          tokens.append(this_token)
+          this_token = ''
+        tokens.append('{')
+      elif char == '}':
+        if this_token:
+          tokens.append(this_token)
+          this_token = ''
+        tokens.append('}')
+      elif char == '[':
+        if this_token:
+          tokens.append(this_token)
+          this_token = ''
+        tokens.append('[')
+      elif char == ']':
+        if this_token:
+          tokens.append(this_token)
+          this_token = ''
+        tokens.append(']')
+      elif char == ':':
+        if this_token:
+          tokens.append(this_token)
+          this_token = ''
+        tokens.append(':')
+      elif char == ',':
+        if this_token:
+          tokens.append(this_token)
+          this_token = ''
+        tokens.append(',')
+      elif char == ' ':
+        if this_token:
+          tokens.append(this_token)
+          this_token = ''
+      else:
+        this_token += char
+    if this_token:
+      tokens.append(this_token)
+
+    struct_type = tokens[0]
+    json = dict()
+    # Now that we have tokens, parse them.
+    self.JsonizeTokens(json, tokens, 1)
+
+    return (struct_name, struct_type, json)
+
+  def JsonizeTokens(self, json, tokens, token_index):
+    """Creates a json-like dictionary from the provided tokens.
+
+    Args:
+      json: dict, The dict to stick the elements in.
+      tokens: list of strings, The list with all the tokens in it.
+      token_index: int, Where to start in the token list.
+
+    Returns:
+      int, The last token used.
+    """
+    # Check that the message starts with a {
+    if tokens[token_index] != '{':
+      print(tokens)
+      print('Expected { at beginning, found', tokens[token_index])
+      return None
+
+    # Eat the {
+    token_index += 1
+
+    # States and state variable for parsing elements.
+    STATE_INIT = 'init'
+    STATE_HAS_NAME = 'name'
+    STATE_HAS_COLON = 'colon'
+    STATE_EXPECTING_SUBMSG = 'submsg'
+    STATE_EXPECTING_COMMA = 'comma'
+    parser_state = STATE_INIT
+
+    while token_index < len(tokens):
+      if tokens[token_index] == '}':
+        # Finish if there is a }
+        return token_index + 1
+      elif tokens[token_index] == '{':
+        if parser_state != STATE_EXPECTING_SUBMSG:
+          print(tokens)
+          print(parser_state)
+          print('Bad input, was not expecting {')
+          return None
+        # Found a submessage, parse it.
+        sub_json = dict()
+        token_index = self.JsonizeTokens(sub_json, tokens, token_index)
+        json[token_name] = sub_json
+        parser_state = STATE_EXPECTING_COMMA
+      else:
+        if parser_state == STATE_INIT:
+          # This token is the name.
+          token_name = tokens[token_index]
+          parser_state = STATE_HAS_NAME
+        elif parser_state == STATE_HAS_NAME:
+          if tokens[token_index] != ':':
+            print(tokens)
+            print(parser_state)
+            print('Bad input, found', tokens[token_index], 'expected :')
+            return None
+          # After a name, comes a :
+          parser_state = STATE_HAS_COLON
+        elif parser_state == STATE_HAS_COLON:
+          # After the colon, figure out what is next.
+          if tokens[token_index] == '[':
+            # Found a sub-array!
+            sub_array = []
+            token_index = self.__JsonizeTokenArray(sub_array, tokens, token_index)
+            json[token_name] = sub_array
+            parser_state = STATE_EXPECTING_COMMA
+          elif tokens[token_index + 1] == '{':
+            # Found a sub-message, trigger parsing it.
+            parser_state = STATE_EXPECTING_SUBMSG
+          else:
+            # This is just an element, move on.
+            json[token_name] = tokens[token_index]
+            parser_state = STATE_EXPECTING_COMMA
+        elif parser_state == STATE_EXPECTING_COMMA:
+          # Complain if there isn't a comma here.
+          if tokens[token_index] != ',':
+            print(tokens)
+            print(parser_state)
+            print('Bad input, found', tokens[token_index], 'expected ,')
+            return None
+          parser_state = STATE_INIT
+        else:
+          print('Bad parser state')
+          return None
+        token_index += 1
+
+    print('Unexpected end')
+    return None
+
+  def __JsonizeTokenArray(self, sub_array, tokens, token_index):
+    """Parses an array from the provided tokens.
+
+    Args:
+      sub_array: list, The list to stick the elements in.
+      tokens: list of strings, The list with all the tokens in it.
+      token_index: int, Where to start in the token list.
+
+    Returns:
+      int, The last token used.
+    """
+    # Make sure the data starts with a '['
+    if tokens[token_index] != '[':
+      print(tokens)
+      print('Expected [ at beginning, found', tokens[token_index + 1])
+      return None
+
+    # Eat the '['
+    token_index += 1
+
+    # Loop through the tokens.
+    while token_index < len(tokens):
+      if tokens[token_index + 1] == ',':
+        # Next item is a comma, so we should just add the element.
+        sub_array.append(tokens[token_index])
+        token_index += 2
+      elif tokens[token_index + 1] == ']':
+        # Next item is a ']', so we should just add the element and finish.
+        sub_array.append(tokens[token_index])
+        token_index += 1
+        return token_index
+      else:
+        # Otherwise, it must be a sub-message.
+        sub_json = dict()
+        token_index = self.JsonizeTokens(sub_json, tokens, token_index + 1)
+        sub_array.append(sub_json)
+        if tokens[token_index] == ',':
+          # Handle there either being another data element.
+          token_index += 1
+        elif tokens[token_index] == ']':
+          # Handle the end of the array.
+          return token_index
+        else:
+          print('Unexpected ', tokens[token_index])
+          return None
+
+    print('Unexpected end')
+    return None
+
+
+if __name__ == '__main__':
+  def ParseLine(line):
+    return LogEntry(line)
+
+  print('motor_writer(2240)(07421): DEBUG   at 0000000819.99620s: ../../frc971/output/motor_writer.cc: 105: sending: .aos.controls.OutputCheck{pwm_value:221, pulse_length:2.233333}')
+  line = ParseLine('motor_writer(2240)(07421): DEBUG   at 0000000819.99620s: ../../frc971/output/motor_writer.cc: 105: sending: .aos.controls.OutputCheck{pwm_value:221, pulse_length:2.233333}')
+  if '.aos.controls.OutputCheck' in line.msg:
+    print(line)
+    print(line.ParseStruct())
+
+  line = ParseLine('claw(2263)(19404): DEBUG   at 0000000820.00000s: ../../aos/common/controls/control_loop-tmpl.h: 104: position: .frc971.control_loops.ClawGroup.Position{top:.frc971.control_loops.HalfClawPosition{position:1.672153, front:.frc971.HallEffectStruct{current:f, posedge_count:0, negedge_count:52}, calibration:.frc971.HallEffectStruct{current:f, posedge_count:6, negedge_count:13}, back:.frc971.HallEffectStruct{current:f, posedge_count:0, negedge_count:62}, posedge_value:0.642681, negedge_value:0.922207}, bottom:.frc971.control_loops.HalfClawPosition{position:1.353539, front:.frc971.HallEffectStruct{current:f, posedge_count:2, negedge_count:150}, calibration:.frc971.HallEffectStruct{current:f, posedge_count:8, negedge_count:18}, back:.frc971.HallEffectStruct{current:f, posedge_count:0, negedge_count:6}, posedge_value:0.434514, negedge_value:0.759491}}')
+  print(line.ParseStruct())
+
+  line = ParseLine('joystick_proxy(2255)(39560): DEBUG   at 0000000820.00730s: ../../aos/prime/input/joystick_input.cc: 61: sending: .aos.RobotState{joysticks:[.aos.Joystick{buttons:0, axis:[0.000000, 1.000000, 1.000000, 0.000000]}, .aos.Joystick{buttons:0, axis:[-0.401575, 1.000000, -1.007874, 0.000000]}, .aos.Joystick{buttons:0, axis:[0.007874, 0.000000, 1.000000, -1.007874]}, .aos.Joystick{buttons:0, axis:[0.000000, 0.000000, 0.000000, 0.000000]}], test_mode:f, fms_attached:f, enabled:T, autonomous:f, team_id:971, fake:f}')
+  print(line.ParseStruct())