coder-chenzhi / plover

PLOVER

PLOVER is short for Pre-Logging OVERhead.

Main directory structure

|-- preliminary_study: the code and data for the preliminary study
|    |-- guard_addition_collector.py: the code used to collect supplementary logging guards
|    |-- raw_data: the collected raw data for the five studied Apache projects
|    |-- Summary.xlsx: the analysis result on the raw data
|-- detector: the code and data for the static analysis
|    |-- ImplementationDetails: implementation details
|    |-- core: core of our static analysis
|    |-- guards-bench-extractor: the code used to collect benchmark of guards
|    |-- log-entries-extractor: the code used to collect entry-points of static analysis
|    |-- commons: utility functions of the static analysis
|    |-- conf: configurations used to run the static analysis
|    |-- docs: results of the static analysis
|-- prioritization: : the code and data for the performance-impact-based prioritization
|    |-- pre-processing: pre-processing code to prepare the data used to train LtR
|    |-- lrt: code, data and the results of LtR

How to run the detector

• install JDK and Maven
• run mvn install to install the detector
• if you run the detector on the stuided five projects
  • change the paths in RunConfig.java in detector/commons to the right paths, including the paths to files containning the signatures of entry-points, the paths to files containning classpaths (used by Soot), the root path to the source code of projects, the paths to files containning the signatures of user-defined guards methods (can be empty file), and the paths to files containning the signatures of user-defined logging methods (can be empty file).
• if you run the detector on new projects
  • you also need to add new configuration files to detector/conf. Please follow the naming convention of existing configuration files. The benchmark of guards can be extracted by running guards-bench-extractor, and the entry-points can be extracted by running log-entries-extractor.
• run MainEntry.java in detector/core

How to run the prioritation

• install Jupyter Notebook (recommend to install anaconda to ease the management of dependencies)
• if you run the detector on the stuided five projects
  • just run the code in ltr sequently
• if you run the detector on new projects
  • you also need to run the code in pre-processing sequently