trappy.stats.grammar module

Grammar module allows the user to easily define relations between data events and perform basic logical and arithmetic operations on the data. The parser also handles super-indexing and variable forwarding.

class trappy.stats.grammar.Parser(data, pvars=None, window=(0, None), filters=None, **kwargs)

Bases: object

A parser class for solving simple data accesses and super-indexing data

Parameters:

• data (instance of trappy.ftrace.BareTrace or a child class (like trappy.ftrace.FTrace)) – Trace Object
• pvars (dict) – A dictionary of variables that need to be accessed from within the grammar
• method (str) – The method to be used for reindexing data This can be one of the standas pandas.DataFrame methods (eg. pad, bfill, nearest). The default is pad or use the last valid observation.
• limit (int) – The number of indices a value will be propagated when reindexing. The default is None
• fill (bool) – Whether to fill the NaNs in the data. The default value is True.
• window (tuple) – A window of time in which to apply the data accesses. By default the data accesses happen accross the whole trace. With the window parameter you can limit it to a window of time inside the trace. The first element of the tuple is the starting time and the second the ending time (set to None for end of trace).
• filters (dict) –

  Restrict the parsing to the rows that match the specified criteria. For Example:

  filters =
          {
              "pid": 3338,
              "cpu": [0, 2, 4],
          }
  

  will only consider rows whose pid column is 3338 and cpu is either 0, 2 or 4.

• Operators

  Operation	operator	Associativity
  Exponentiation	**	Left
  Unary	-	Right
  Multiply/Divide	*, /, //, %	Left
  Add/Subtract	+, -,	Left
  Comparison	>, <, >=, <=, ==, !=	Left
  Logical	&&, ||, |, &	Left

• Data Accessors

  Since the goal of the grammar is to provide an easy language to access and compare data from a trappy.trace.FTrace object. The parser provides a simple notation to access this data.

  Statically Defined Events

  import trappy
  from trappy.stats.grammar import Parser
  
  trace = trappy.FTrace("path/to/trace/file")
  parser = Parser(trace)
  parser.solve("trappy.thermal.Thermal:temp * 2")
  

  Aliasing

  import trappy
  from trappy.stats.grammar import Parser
  
  pvars = {"THERMAL": trappy.thermal.Thermal}
  trace = trappy.FTrace("path/to/trace/file")
  parser = Parser(trace, pvars=pvars)
  parser.solve("THERMAL:temp * 2")
  

  Using Event Name

  import trappy
  from trappy.stats.grammar import Parser
  trace = trappy.FTrace("path/to/trace/file")
  parser = Parser(trace)
  parser.solve("thermal:temp * 2")
  

  The event trappy.thermal.Thermal is aliased as thermal in the grammar

  Dynamic Events

  import trappy
  from trappy.stats.grammar import Parser
  
  # Register Dynamic Event
  cls = trappy.register_dynamic_ftrace("my_unique_word", "event_name")
  
  pvars = {"CUSTOM": cls}
  trace = trappy.FTrace("path/to/trace/file")
  parser = Parser(trace, pvars=pvars)
  parser.solve("CUSTOM:col * 2")
  

  See also

  trappy.dynamic.register_dynamic_ftrace

inspect(accessor)

A function to inspect the accessor for information

Parameters:accessor (str) – A data accessor of the format <event>:<column> Returns:A dictionary of information

ref(mask)

Reference super indexed data with a boolean mask

Parameters:mask (pandas.Series) – A boolean pandas.Series that can be used to reference the aggregated data in the parser Returns:aggregated_data[mask]

solve(expr)

Parses and solves the input expression

Parameters:expr (str) – The input expression Returns:The return type may vary depending on the expression. For example:

Vector

import trappy
from trappy.stats.grammar import Parser

trace = trappy.FTrace("path/to/trace/file")
parser = Parser(trace)
parser.solve("trappy.thermal.Thermal:temp * 2")

Scalar

import trappy
from trappy.stats.grammar import Parser

trace = trappy.FTrace("path/to/trace/file")
parser = Parser(trace)
parser.solve("numpy.mean(trappy.thermal.Thermal:temp)")

Vector Mask

import trappy
from trappy.stats.grammar import Parser

trace = trappy.FTrace("path/to/trace/file")
parser = Parser(trace)
parser.solve("trappy.thermal.Thermal:temp > 65000")

trappy.stats.grammar.eval_binary_op(tokens)

Evaluate Binary operators

Parameters:tokens (list) – The grammar tokens

trappy.stats.grammar.eval_unary_op(tokens)

Unary Op Evaluation

Parameters:tokens (list) – The grammar tokens

trappy.stats.grammar.get_parse_expression(parse_func, parse_var_id)

return a parse expression with for the input parseActions

trappy.stats.grammar.iterate_binary_ops(tokens)

An iterator for Binary Operation tokens

Parameters:tokens (list) – The grammar tokens

trappy.stats.grammar.parse_num(tokens)

Parser function for numerical data

Parameters:tokens (list) – The grammar tokens

trappy.stats.grammar.str_to_attr(cls_str)

Bring the attr specified into current scope

and return a handler

Parameters:cls_str (str) – A string representing the class Returns:A class object