cm data

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: repometrics
 Title: Metrics for Your Code Repository
-Version: 0.1.1.069
+Version: 0.1.1.077
 Authors@R: 
     person("Mark", "Padgham", , "[email protected]", role = c("aut", "cre"),
            comment = c(ORCID = "0000-0003-2172-5265"))

R/chaoss-internal-change-req.R

@@ -15,9 +15,18 @@
 chaoss_internal_change_req <- function (path, end_date = Sys.Date ()) {
 
     log <- git_log_in_period (path, end_date, get_repometrics_period ())
-    res <- NA_integer_
-    if (nrow (log) > 0L) {
-        res <- length (which (log$merge)) / nrow (log)
-    }
+    prs <- cm_data_prs_from_gh_api (path)
+    prs <- prs [which (prs$merged), ]
+    closed_dates <- as.Date (prs$closed_at)
+    start_date <- end_date - get_repometrics_period ()
+    index <- which (closed_dates >= start_date & closed_dates <= end_date)
+    prs <- prs [index, ]
+
+    num_commits <- vapply (prs$commit_oids, function (i) {
+        length (strsplit (i, ",") [[1]])
+    }, integer (1L), USE.NAMES = FALSE)
+
+    res <- sum (num_commits) / nrow (log)
+
     return (res)
 }

R/chaoss-internal-num-ctb.R

@@ -2,16 +2,7 @@ chaoss_internal_num_contributors <- function (path, end_date = Sys.Date ()) {
 
     log <- git_log_in_period (path, end_date, get_repometrics_period ())
 
-    auths_un <- unique (log$author)
-
-    # separate handles from emails:
-    emails <- regmatches (auths_un, gregexpr ("<.*>", auths_un))
-    emails <- vapply (emails, function (i) {
-        ifelse (length (i) == 0L, "", gsub ("<|>", "", i))
-    }, character (1L))
-    handles <- gsub ("<.*$", "", auths_un)
-
-    # Remove any duplicates of either, but excluding non-entries:
+    # Remove any duplicates of either names or emails, but excluding non-entries:
     rm_dup_rows <- function (x) {
         x <- gsub ("\\s+", "", x)
         index <- seq_along (x)
@@ -21,15 +12,13 @@ chaoss_internal_num_contributors <- function (path, end_date = Sys.Date ()) {
         }
         return (index)
     }
-    index1 <- rm_dup_rows (handles)
-    index2 <- rm_dup_rows (emails)
+    index1 <- rm_dup_rows (log$aut_name)
+    index2 <- rm_dup_rows (log$aut_email)
 
     # Then extract only instances where neither handles nor emails are
     # duplicated:
     index_table <- table (c (index1, index2))
     index <- as.integer (names (index_table) [which (index_table == 2L)])
 
-    auths_un <- auths_un [index]
-
-    return (length (auths_un))
+    return (length (index))
 }

R/cm-data-dependencies.R

@@ -27,17 +27,19 @@ cm_data_dependencies <- function (path) {
 cm_data_libyears <- function (path) {
 
     deps <- cm_data_dependencies (path)
-    cran_db <- data.frame (tools::CRAN_package_db ())
+    cran_db <- data.frame (cran_pkg_db ())
     index <- match (deps$name, cran_db$Package)
     deps$cran_version <- cran_db$Version [index]
     deps$published <- as.Date (cran_db$Published [index])
     deps <- deps [which (!is.na (deps$published)), ]
 
-    rel <- releases_from_gh_api (path, latest_only = TRUE)
+    rel <- cm_data_releases_from_gh_api (path, latest_only = TRUE)
     rel_date <- as.Date (strftime (rel$published_at, format = "%Y-%m-%d"))
 
     dt <- difftime (deps$published, rel_date, units = "days")
     dt <- as.numeric (dt) / 365.25 # In years
 
     c (mean = mean (dt), median = stats::median (dt))
 }
+
+cran_pkg_db <- memoise::memoise (tools::CRAN_package_db)

R/cm-data-gh-contribs.R

@@ -1,4 +1,6 @@
-contribs_from_log <- function (log) {
+cm_data_contribs_from_log <- function (path) {
+
+    log <- cm_data_gitlog (path)
 
     gh_handle <- unique (log$aut_name)
     gh_email <- log$aut_email [match (gh_handle, log$aut_name)]
@@ -27,7 +29,7 @@
     ) [index, ]
 }
 
-contribs_from_gh_api <- function (path, n_per_page = 100) {
+cm_data_contribs_from_gh_api_internal <- function (path, n_per_page = 100) {
 
     is_test_env <- Sys.getenv ("REPOMETRICS_TESTS") == "true"
 
@@ -80,6 +82,8 @@
 
     return (ctbs)
 }
+cm_data_contribs_from_gh_api <-
+    memoise::memoise (cm_data_contribs_from_gh_api_internal)
 
 user_from_gh_api <- function (user) {
 

R/cm-data-gh-issues.R

@@ -1,4 +1,4 @@
-issues_from_gh_api <- function (path, n_per_page = 100) {
+cm_data_issues_from_gh_api_internal <- function (path, n_per_page = 100) {
 
     is_test_env <- Sys.getenv ("REPOMETRICS_TESTS") == "true"
 
@@ -69,6 +69,8 @@ issues_from_gh_api <- function (path, n_per_page = 100) {
 
     return (issues)
 }
+cm_data_issues_from_gh_api <-
+    memoise::memoise (cm_data_issues_from_gh_api_internal)
 
 get_issue_reactions <- function (body) {
 
@@ -87,7 +89,7 @@ get_issue_reactions <- function (body) {
     return (reaction_counts)
 }
 
-issue_comments_from_gh_api <- function (path, n_per_page = 100) {
+cm_data_issue_comments_from_gh_api_internal <- function (path, n_per_page = 100) {
 
     is_test_env <- Sys.getenv ("REPOMETRICS_TESTS") == "true"
 
@@ -138,3 +140,5 @@ issue_comments_from_gh_api <- function (path, n_per_page = 100) {
         issue_body = issue_body
     )
 }
+cm_data_issue_comments_from_gh_api <-
+    memoise::memoise (cm_data_issue_comments_from_gh_api_internal)

R/cm-data-gh-prs.R

@@ -107,7 +107,7 @@ gh_prs_qry <- function (org = "ropensci-review-tools",
     return (q)
 }
 
-prs_from_gh_api <- function (path, n_per_page = 30L) {
+cm_data_prs_from_gh_api_internal <- function (path, n_per_page = 30L) {
 
     is_test_env <- Sys.getenv ("REPOMETRICS_TESTS") == "true"
     if (is_test_env) {
@@ -158,7 +158,7 @@ prs_from_gh_api <- function (path, n_per_page = 30L) {
     reviews <- lapply (pr_data, function (i) {
         login <- vapply (i$reviews$nodes, function (j) j$author$login, character (1L))
         state <- vapply (i$reviews$nodes, function (j) j$state, character (1L))
-        submitted_at <- vapply (i$reviews$nodes, function (j) j$submittedAt, character (1L))
+        submitted_at <- vapply (i$reviews$nodes, function (j) null2na_char (j$submittedAt), character (1L))
         body <- vapply (i$reviews$nodes, function (j) j$body, character (1L))
         data.frame (
             login = login,
@@ -193,3 +193,4 @@ prs_from_gh_api <- function (path, n_per_page = 30L) {
         reviews = I (reviews)
     )
 }
+cm_data_prs_from_gh_api <- memoise::memoise (cm_data_prs_from_gh_api_internal)

R/cm-data-gh-releases.R

@@ -1,4 +1,4 @@
-releases_from_gh_api <- function (path, n_per_page = 100L, latest_only = FALSE) {
+cm_data_releases_from_gh_api_internal <- function (path, n_per_page = 100L, latest_only = FALSE) {
 
     checkmate::assert_integerish (n_per_page)
     checkmate::assert_logical (latest_only)
@@ -48,3 +48,4 @@ releases_from_gh_api <- function (path, n_per_page = 100L, latest_only = FALSE)
         published_at = vapply (body, function (i) i$published_at, character (1L))
     )
 }
+cm_data_releases_from_gh_api <- memoise::memoise (cm_data_releases_from_gh_api_internal)

R/cm-data-gh-repo.R

@@ -1,4 +1,4 @@
-repo_from_gh_api <- function (path) {
+cm_data_repo_from_gh_api_internal <- function (path) {
 
     or <- org_repo_from_path (path)
 
@@ -36,3 +36,4 @@ repo_from_gh_api <- function (path) {
         default_branch = null2na_char (body$default_branch)
     )
 }
+cm_data_repo_from_gh_api <- memoise::memoise (cm_data_repo_from_gh_api_internal)

R/cm-data-git.R

@@ -50,3 +50,26 @@
     )
 }
 cm_data_gitlog <- memoise::memoise (cm_data_gitlog_internal)
+
+git_log_in_period <- function (path, end_date = Sys.Date (), period = 90) {
+
+    checkmate::assert_character (path, len = 1L)
+    checkmate::assert_directory (path)
+    checkmate::assert_date (end_date)
+
+    log <- cm_data_gitlog (path)
+
+    if (nrow (log) == 0) {
+        return (log)
+    }
+    dates <- as.Date (log$time)
+    today_minus_period <- as.Date (end_date - period)
+    index <- which (dates >= today_minus_period)
+    log <- log [index, ]
+
+    if (dates [1] > end_date) {
+        log <- log [which (dates <= end_date), ]
+    }
+
+    return (log)
+}

R/cm-data.R

@@ -0,0 +1,35 @@
+cm_data <- function (path) {
+
+    data_fns <- get_cm_data_fns ()
+
+    if (all_cm_data_fns_memoised (data_fns, path)) {
+        res <- lapply (data_fns, function (i) {
+            do.call (i, list (path))
+        })
+    } else {
+        res <- pbapply::pblapply (data_fns, function (i) {
+            do.call (i, list (path))
+        })
+    }
+    names (res) <- gsub ("^cm\\_data\\_", "", data_fns)
+
+    return (res)
+}
+
+get_cm_data_fns <- function () {
+
+    pkg_fns <- ls (envir = asNamespace ("repometrics"))
+    data_fns <- grep ("^cm\\_data\\_", pkg_fns, value = TRUE)
+    data_fns [which (!grepl ("\\_internal$", data_fns))]
+}
+
+all_cm_data_fns_memoised <- function (data_fns, path) {
+    is_memoised <- vapply (data_fns, function (i) {
+        tryCatch (
+            memoise::has_cache (get (i)) (path),
+            error = function (e) FALSE
+        )
+    }, logical (1L))
+
+    length (which (is_memoised)) > (length (data_fns) / 2)
+}

R/cm-metrics-num-commits.R

@@ -0,0 +1,6 @@
+cm_metrics_num_commits <- function (path, end_date = Sys.Date ()) {
+
+    log <- git_log_in_period (path, end_date, get_repometrics_period ())
+
+    return (nrow (log))
+}

codemeta.json

@@ -8,7 +8,7 @@
   "codeRepository": "https://github.com/ropensci-review-tools/repometrics",
   "issueTracker": "https://github.com/ropensci-review-tools/repometrics/issues",
   "license": "https://spdx.org/licenses/GPL-3.0",
-  "version": "0.1.1.069",
+  "version": "0.1.1.077",
   "programmingLanguage": {
     "@type": "ComputerLanguage",
     "name": "R",

tests/testthat/gh_libyears/ghrepos/repo/releases-4aabf7.json

@@ -0,0 +1,56 @@
+[
+    {
+        "url": "ghrepos/repo//releases/158953460",
+        "assets_url": "ghrepos/repo//releases/158953460/assets",
+        "upload_url": "https://uploads.github.com/repos/repo//releases/158953460/assets{?name,label}",
+        "html_url": "https://github.com/repo//releases/tag/v1.0.5",
+        "id": 158953460,
+        "author": {
+            "login": "mpadge",
+            "id": 6697851,
+            "node_id": "MDQ6VXNlcjY2OTc4NTE=",
+            "avatar_url": "https://avatars.githubusercontent.com/u/6697851?v=4",
+            "gravatar_id": "",
+            "url": "ghusers/mpadge",
+            "html_url": "https://github.com/mpadge",
+            "followers_url": "ghusers/mpadge/followers",
+            "following_url": "ghusers/mpadge/following{/other_user}",
+            "gists_url": "ghusers/mpadge/gists{/gist_id}",
+            "starred_url": "ghusers/mpadge/starred{/owner}{/repo}",
+            "subscriptions_url": "ghusers/mpadge/subscriptions",
+            "organizations_url": "ghusers/mpadge/orgs",
+            "repos_url": "ghusers/mpadge/repos",
+            "events_url": "ghusers/mpadge/events{/privacy}",
+            "received_events_url": "ghusers/mpadge/received_events",
+            "type": "User",
+            "user_view_type": "public",
+            "site_admin": false
+        },
+        "node_id": "RE_kwDOAwboEs4JeW_0",
+        "tag_name": "v1.0.5",
+        "target_commitish": "main",
+        "name": "goodpractice 1.0.5",
+        "draft": false,
+        "prerelease": false,
+        "created_at": "2024-06-05T09:31:17Z",
+        "published_at": "2024-06-05T09:34:42Z",
+        "assets": [
+
+        ],
+        "tarball_url": "ghrepos/repo//tarball/v1.0.5",
+        "zipball_url": "ghrepos/repo//zipball/v1.0.5",
+        "body": "* New maintainer: rOpenSci\r\n* Package reinstated on CRAN, after archiving of previous version.\r\n* CRAN fixes - skipping failing test and adding \\alias{goodpractice} to package Rd\r\n* Adding docs.ropensci site to DESCRIPTION",
+        "reactions": {
+            "url": "ghrepos/repo//releases/158953460/reactions",
+            "total_count": 1,
+            "+1": 0,
+            "-1": 0,
+            "laugh": 0,
+            "hooray": 0,
+            "confused": 0,
+            "heart": 1,
+            "rocket": 0,
+            "eyes": 0
+        }
+    }
+]

tests/testthat/test-chaoss-metrics-internal.R

@@ -4,7 +4,7 @@ test_that ("chaoss internal num_commits", {
 
     path <- generate_test_pkg ()
 
-    n <- chaoss_internal_num_commits (path, end_date = end_date)
+    n <- cm_metrics_num_commits (path, end_date = end_date)
     expect_equal (n, 4L)
 
     n <- chaoss_internal_num_contributors (path, end_date = end_date)
@@ -39,15 +39,3 @@ test_that ("chaoss has CI internal", {
 
     fs::dir_delete (path)
 })
-
-test_that ("chaoss internal change requests", {
-
-    path <- generate_test_pkg ()
-
-    x <- chaoss_internal_change_req (path, end_date = end_date)
-    expect_equal (x, 0)
-    x <- chaoss_internal_change_req (path, end_date = Sys.Date ())
-    expect_equal (x, NA_integer_) # no commits, so NA returned
-
-    fs::dir_delete (path)
-})