Add Commit Networks

README.md

@@ -234,6 +234,11 @@ There are four types of networks that can be built using this library: author ne
      * The vertices in an artifact network denote any kind of artifact, e.g., source-code artifact (such as features or files) or communication artifact (such as mail threads or issues). All artifact-type vertices are uniquely identifiable by their name. There are only unipartite edges among artifacts in this type of network.
      * The relations (i.e., the edges' meaning and source) can be configured using the [`NetworkConf`](#networkconf) attribute `artifact.relation`. The relation also describes which kinds of artifacts are represented as vertices in the network. (For example, if "mail" is selected as `artifact.relation`, only mail-thread vertices are included in the network.)
 
+- Commit networks
+     * The vertices in a commit network denote any commits in the data. All vertices
+     are uniquely identifyable by the hash of the commit. There are only unipartite edges among commits in this type of network.
+     * The relations (i.e., the edges meaning and source) can be configured using the [`networkConf`](#networkconf) attribute `commit.relation`. The relation also describes the type of data used for network construction (`cochange` uses commit data, `commit.interaction` uses commit interaction data).
+
 - Bipartite networks
      * The vertices in a bipartite network denote both authors and artifacts. There are only bipartite edges from authors to artifacts in this type of network.
      * The relations (i.e., the edges' meaning and source) can be configured using the [`NetworkConf`](#networkconf) attribute `artifact.relation`.
@@ -249,6 +254,7 @@ Relations determine which information is used to construct edges among the verti
 - `cochange`
     * For author networks (configured via `author.relation` in the [`NetworkConf`](#networkconf)), authors who change the same source-code artifact are connected with an edge.
     * For artifact networks (configured via `artifact.relation` in the [`NetworkConf`](#networkconf)), source-code artifacts that are concurrently changed in the same commit are connected with an edge.
+    * For commit networks (configured vie `commit.relation` in the [`NetworkConf`](#networkconf)), commits are connected if they change the same artifact.
     * For bipartite networks (configured via `artifact.relation` in the [`NetworkConf`](#networkconf)), authors get linked to all source-code artifacts they have changed in their respective commits.
 
 - `mail`
@@ -269,6 +275,7 @@ Relations determine which information is used to construct edges among the verti
 - `commit.interaction`
     * For author networks (configured via `author.relation` in the [`NetworkConf`](#networkconf)), authors who contribute to interacting commits are connected with an edge.
     * For artifact networks (configured via `artifact.relation` in the [`NetworkConf`](#networkconf)), artifacts are connected when there is an interaction between two commits that occur in the artifacts.
+    * For commit networks (configured via `commit.relation` in the [`NetworkConf`](#networkconf)), commits are connected when they interact in the commit interaction data.
     * This relation does not apply for bipartite networks.
 
 #### Edge-construction algorithms for author networks

showcase.R

@@ -239,8 +239,8 @@ sample.pull.requests = add.vertex.attribute.author.issue.count(my.networks, x.da
 ## add vertex attributes for the project-level network
 x.net.as.list = list("1970-01-01 00:00:00-2030-01-01 00:00:00" = x$get.author.network())
 sample.entire = add.vertex.attribute.author.commit.count(x.net.as.list, x.data, aggregation.level = "complete")
-## add vertex attributes to commit network
-add.vertex.attribute.commit.network(x$get.commit.network(), x.data, "author.name", "NO_AUTHOR")
+## add vertex attributes to commit network. Default value 'NO_AUTHOR' is used if vertex is not in commit data
+add.vertex.attribute.commit.network(x$get.commit.network(), x.data, attr.name = "author.name", default.value = "NO_AUTHOR")
 
 
 ## / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /

tests/test-data.R

@@ -564,15 +564,15 @@ test_that("Compare two ProjectData Objects with commit.interactions", {
     proj.data.two$set.commits(create.empty.commits.list())
 
     ## create empty data frame of correct size
-    commit.interactions.data.expected = data.frame(matrix(nrow = 4, ncol = 8))
+    commit.interactions.data.expected = data.frame(matrix(nrow = 4, ncol = 9))
     ## assure that the correct type is used
-    for(i in seq_len(8)) {
+    for(i in seq_len(9)) {
         commit.interactions.data.expected[[i]] = as.character(commit.interactions.data.expected[[i]])
     }
     ## set everything except for authors as expected
     colnames(commit.interactions.data.expected) = c("commit.hash", "base.hash", "func", "file",
-                                                    "base.func", "base.file", "base.author",
-                                                    "interacting.author")
+                                                    "base.func", "base.file","artifact.type",
+                                                    "base.author", "interacting.author")
     commit.interactions.data.expected[["commit.hash"]] =
                                                         c("0a1a5c523d835459c42f33e863623138555e2526",
                                                         "418d1dc4929ad1df251d2aeb833dd45757b04a6f",
@@ -588,6 +588,8 @@ test_that("Compare two ProjectData Objects with commit.interactions", {
     commit.interactions.data.expected[["base.func"]] = c("test2.c::test2", "test2.c::test2",
                                                          "test3.c::test_function", "test2.c::test2")
     commit.interactions.data.expected[["base.file"]] = c("test2.c", "test2.c", "test3.c", "test2.c")
+    commit.interactions.data.expected[["artifact.type"]] = c("CommitInteraction", "CommitInteraction",
+                                                             "CommitInteraction", "CommitInteraction")
 
     expect_equal(proj.data.two$get.commit.interactions(), commit.interactions.data.expected)
 

tests/test-networks-commit.R

@@ -83,6 +83,9 @@ patrick::with_parameters_test_that("Network construction with commit-interaction
         )
     network = igraph::graph.data.frame(edges, directed = test.directed, vertices = vertices)
     expect_true(igraph::identical_graphs(network.built, network))
+
+    network.new.attr = add.vertex.attribute.commit.network(network.built, proj.data,  "deleted.lines", "NO_DATA")
+    expect_identical(igraph::V(network.new.attr)$deleted.lines, c("0", "0","0", "NO_DATA", "0", "NO_DATA"))
 }, patrick::cases(
     "directed: FALSE" = list(test.directed = FALSE),
     "directed: TRUE" = list(test.directed = TRUE)

tests/test-read.R

@@ -505,15 +505,15 @@ test_that("Read the commit-interactions data.", {
     commit.interactions.data.read = read.commit.interactions(proj.conf$get.value("datapath"))
     ## build the expected data.frame
 
-    commit.interactions.data.expected = data.frame(matrix(nrow = 4, ncol = 8))
+    commit.interactions.data.expected = data.frame(matrix(nrow = 4, ncol = 9))
     ## assure that the correct type is used
-    for(i in seq_len(8)) {
+    for(i in seq_len(ncol(commit.interactions.data.expected))) {
         commit.interactions.data.expected[[i]] = as.character(commit.interactions.data.expected[[i]])
     }
     ## set everything except for authors as expected
     colnames(commit.interactions.data.expected) = c("func", "commit.hash", "file", "base.hash",
                                                     "base.func", "base.file", "base.author",
-                                                    "interacting.author")
+                                                    "interacting.author", "artifact.type")
     commit.interactions.data.expected[["commit.hash"]] =
                                                         c("5a5ec9675e98187e1e92561e1888aa6f04faa338",
                                                           "0a1a5c523d835459c42f33e863623138555e2526",
@@ -529,6 +529,8 @@ test_that("Read the commit-interactions data.", {
     commit.interactions.data.expected[["base.func"]] = c("test3.c::test_function", "test2.c::test2",
                                                          "test2.c::test2", "test2.c::test2")
     commit.interactions.data.expected[["base.file"]] = c("test3.c", "test2.c", "test2.c", "test2.c")
+    commit.interactions.data.expected[["artifact.type"]] = c("CommitInteraction", "CommitInteraction",
+                                                             "CommitInteraction", "CommitInteraction")
     ## check the results
     expect_identical(commit.interactions.data.read, commit.interactions.data.expected,
                      info = "commit interaction data.")
@@ -543,11 +545,11 @@ test_that("Read the empty commit-interactions data.", {
     commit.interactions.data.read = read.commit.interactions("./codeface-data/results/testing/
                                                              test_empty_proximity/proximity")
     ## build the expected data.frame
-    commit.interactions.data.expected = data.frame(matrix(nrow = 0, ncol = 8))
+    commit.interactions.data.expected = data.frame(matrix(nrow = 0, ncol = 9))
     colnames(commit.interactions.data.expected) = c("func", "commit.hash", "file",
                                                     "base.hash", "base.func", "base.file",
-                                                    "base.author", "interacting.author")
-    for(i in seq_len(8)) {
+                                                    "base.author", "interacting.author", "artifact.type")
+    for(i in seq_len(ncol(commit.interactions.data.expected))) {
         commit.interactions.data.expected[[i]] = as.character(commit.interactions.data.expected[[i]])
     }
     ## check the results

util-data.R

@@ -415,7 +415,10 @@ ProjectData = R6::R6Class("ProjectData",
         #'
         #' This method should be called whenever the field \code{commit.interactions} is changed.
         update.commit.interactions = function() {
-            if (self$is.data.source.cached("commit.interactions")) {
+            stacktrace = get.stacktrace(sys.calls())
+            caller = get.second.last.element(stacktrace)
+            if (self$is.data.source.cached("commit.interactions") &&
+                (is.na(caller)|| paste(caller, collapse = " ") != "self$set.commits(commit.data)")) {
                 if (!self$is.data.source.cached("commits.unfiltered")) {
                     self$get.commits()
                 }
@@ -2143,8 +2146,6 @@ ProjectData = R6::R6Class("ProjectData",
             return(mylist)
         },
 
-        ## * * processed data ----------------------------------------------
-
         #' Group the commits of the given \code{data.source} by the given \code{group.column}.
         #' For each group, the column \code{"hash"} is duplicated and prepended to each
         #' group's data as first column (see below for details).
@@ -2162,12 +2163,11 @@ ProjectData = R6::R6Class("ProjectData",
         #'         as first column (with name \code{"data.vertices"})
         #'
         #' @seealso ProjectData$group.data.by.column
-        group.commits.by.data.column = function(data.source = c("commits", "mails", "issues"),
-                                                group.column = "artifact") {
+        group.commits.by.data.column = function(group.column = "artifact") {
             logging::loginfo("Grouping commits by data column.")
 
             ## store the commits per group that is determined by 'group.column'
-            mylist = self$group.data.by.column(data.source, group.column, "hash")
+            mylist = self$group.data.by.column("commits", group.column, "hash")
 
             return(mylist)
         },

util-networks-covariates.R

@@ -149,8 +149,8 @@ add.vertex.attribute = function(net.to.range.list, attr.name, default.value, com
 #' @param network the commit network
 #' @param project.data the project data from which to extract the values
 #' @param attr.name the name of the attribute
-#' @param default.value the dafault value of the attribute
-#'                      if it does not occur in the commit data
+#' @param default.value the default value that is used if the current hash
+#'                      is not contained in the commit data at all
 #'
 #' @return a network with new vertex attribute
 add.vertex.attribute.commit.network = function(network, project.data,
@@ -174,7 +174,6 @@ add.vertex.attribute.commit.network = function(network, project.data,
         attribute.values = c(attribute.values, value)
     }
     net.with.attr = igraph::set.vertex.attribute(network, attr.name, value = attribute.values)
-
 }