calculate_features_from_versions: Calculate features from combinations of PMD alerts of a new...

In crotman/kludgenudger: support for Bruno Crotman's thesis about kludges in software development

Description

Reads or gets as input two versions of source codes, a new and an old, and returns the info about the alerts and the combination between the alerts. The main output are the features related to combinations of an alerts: an alert from the new version and an alert from the old version. All the combinations between an old alert and a new alert are considered. The main goal is to differentiate between Fixed, Old and New alerts. If a combination of alerts from the new and the old versions have many features that indicate they have the same role in the code, the alert in the new version may be considered an old alert.

Usage

calculate_features_from_versions(
  code_file_new = "",
  code_file_old = "",
  code_new = "",
  code_old = "",
  pmd_path,
  rule_path = "rulesets/java/quickstart.xml",
  blockrules_location = "data/blockrules/blockrules.xml",
  mostra_new = c(10, 43, 17, 15, 18, 16, 45, 44),
  mostra_old = c(10, 42, 41, 15, 16, 43),
  glue_string = "",
  optimize_feature_calculation = TRUE,
  alerts_new = NA,
  alerts_old = NA,
  ast_new = NA,
  ast_old = NA,
  diff = NA,
  only_categorised = FALSE
)

Arguments

code_file_new	file containing the source code of the new version. If this parameter is empty, the parameter code_new is used
code_file_old	file containing the source code of the old version. If this parameter is empty, the parameter code_old is used
code_new	source code of the new version. If this parameter is empty, the parameter code_file_new is used
code_old	source code of the old version. If this parameter is empty, the parameter code_file_old is used
pmd_path	complete path to pmd.bat including the name of the file
rule_path	path to the rules used for the PMD alerts
blockrules_location	path to the rules for the Abstract Syntax Tree
mostra_new	which nodes of the new version must appear?
mostra_old	which nodes of the old version must appear?
glue_string	template of a string that will be passed to str_glue and will be available in the returned info
optimize_feature_calculation	if true, the features are not calculated for all the combinations of new and old alerts if there is a perfect match

Value

a list with information about the comparison between old and new versions. The list contains the following items:

• versions_executed: a dataframe containing the PMD alerts generated for each version

• versions_crossed: a dataframe containing information about the combination of versions including a map between the lines of code in the new and in the old version

• graph_old_with_alert: a tidygraph containing the abstract syntax tree of the old version and information about PMD alerts

• graph_new_with_alert: a tidygraph containing the abstract syntax tree of the new version and information about PMD alerts

• graph_old_with_group: a tidygraph containing the abstract syntax tree of the old version and information about PMD alerts. Here, the nodes of the new and the old versions are categorized in groups

• graph_new_with_group: a tidygraph containing the abstract syntax tree of the new version and information about PMD alerts. Here, the nodes of the new and the old versions are categorized in groups

• graphs_from_alerts_old: a dataframe containing, for each alert of the old version, the path from the node related to the alert to the root node (compilation unit), in a tidygraph

• graphs_from_alerts_new: a dataframe containing, for each alert of the new version, the path from the node related to the alert to the root node (compilation unit), in a tidygraph

• features: a dataframe containing, for each combination of old and new alert, the features related to the combinations

• categorised_alerts: dataframe with old and new alerts categorised in fixed, open and new

crotman/kludgenudger documentation built on Oct. 19, 2021, 7:30 p.m.