Merge pull request #50 from ropensci-review-tools/ctbs

contributors dashboard plots

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: repometrics
 Title: Metrics for Your Code Repository
-Version: 0.1.2.007
+Version: 0.1.2.026
 Authors@R: 
     person("Mark", "Padgham", , "[email protected]", role = c("aut", "cre"),
            comment = c(ORCID = "0000-0003-2172-5265"))

R/quarto-dashboard.R

@@ -62,6 +62,12 @@ quarto_insert_pkg_name <- function (dir, pkg_name) {
     index_qmd <- brio::read_lines (f_index)
     i <- grep ("^title\\:", index_qmd)
     index_qmd [i] <- paste0 ("title: The {", pkg_name, "} package")
+    index_qmd <- gsub (
+        "the XXX repository",
+        paste ("the", pkg_name, "repository"),
+        index_qmd,
+        fixed = TRUE
+    )
     brio::write_lines (index_qmd, f_index)
 
     f_yaml <- fs::path (dir, "_quarto.yml")
@@ -82,7 +88,7 @@ check_dashboard_arg <- function (data) {
         "dependencies_downstream", "gh_repo_workflow", "gitlog",
         "issue_comments_from_gh_api", "issues_from_gh_api", "libyears",
         "prs_from_gh_api", "releases_from_gh_api", "repo_forks",
-        "repo_from_gh_api", "repo_stargazers"
+        "repo_from_gh_api", "repo_stargazers", "contributors"
     )
     checkmate::assert_names (names (data$cm), identical.to = nms)
 
@@ -123,7 +129,8 @@ check_dashboard_arg <- function (data) {
         "releases_from_gh_api" = 10L,
         "repo_forks" = 2L,
         "repo_from_gh_api" = 18L,
-        "repo_stargazers" = 2L
+        "repo_stargazers" = 2L,
+        "contributors" = 3L
     )
     if (!identical (ncols, ncols_expected)) {
         cli::cli_abort (paste0 (

R/repometrics-data.R

@@ -18,6 +18,7 @@ repometrics_data <- function (path, step_days = 1L, num_cores = -1L) {
         num_cores = num_cores
     )
     cm <- cm_data (path)
+    cm$contributors <- get_all_contribs (cm$contribs_from_log, cm$contribs_from_gh_api)
 
     list (pkgstats = pkgstats, cm = cm)
 }

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": "0.1.2.007",
+  "version": "0.1.2.026",
   "programmingLanguage": {
     "@type": "ComputerLanguage",
     "name": "R",

inst/extdata/quarto/fn-stats.qmd

@@ -1,5 +1,5 @@
 ---
-title: "Metrics from git commit history "
+title: "Git history"
 execute:
   echo: false
 format: 
@@ -276,7 +276,7 @@ loc <- tidyr::pivot_longer (
         "ndoc" ~ "doc"
     )) |>
     dplyr::group_by (date, dir, loc_measure) |>
-    dplyr::summarise (value = mean (value)) |>
+    dplyr::summarise (value = mean (value), .groups = "keep") |>
     dplyr::ungroup () |>
     dplyr::arrange (by = dplyr::desc (date))
 

inst/extdata/quarto/github.qmd

@@ -1,5 +1,5 @@
 ---
-title: "Metrics from GitHub "
+title: "Contributors"
 execute:
   echo: false
 format: 
@@ -14,3 +14,222 @@ library (repometrics)
 ```
 This page includes several graphs providing insight into the historical
 development of the package.
+
+```{r load-data}
+dat <- readRDS ("results.Rds")
+```
+```{r ctbs-git-data}
+ctbs <- dat$cm$contributors
+log <- dat$cm$gitlog
+log <- log [which (!log$aut_name == "GitHub"), ]
+
+# Insert gh handles in git log:
+log$gh_handle <- NA_character_
+
+# Match by log$aut_name:
+index <- match (log$aut_name, ctbs$name)
+log$gh_handle <- ctbs$gh_handle [index]
+# Match by log$aut_email:
+index <- which (is.na (log$gh_handle))
+index_email <- match (log$aut_email [index], ctbs$email)
+log$gh_handle [index] <- ctbs$gh_handle [index_email]
+
+is_test_env <- Sys.getenv ("REPOMETRICS_TESTS") == "true"
+if (is_test_env) {
+    log$gh_handle <- "mpadge"
+}
+
+
+log <- dplyr::mutate (log, date = as.Date (timestamp)) |>
+    dplyr::mutate (date = as.Date (gsub ("\\-[0-9]{2}$", "-01", date))) |>
+    dplyr::group_by (gh_handle, date) |>
+    dplyr::summarise (
+        nfiles_changed = sum (nfiles_changed),
+        lines_added = sum (lines_added) - sum (whitespace_added),
+        lines_removed = sum (lines_removed) - sum (whitespace_removed),
+        ncommits = dplyr::n (),
+        .groups = "keep"
+    ) |>
+    dplyr::ungroup () |>
+    dplyr::filter (!is.na (gh_handle))
+
+dates <- seq (min (log$date), max (log$date), by = "months")
+
+log_reg <- dplyr::group_by (log, gh_handle) |>
+    tidyr::complete (date = dates) |>
+    tidyr::fill () |>
+    dplyr::arrange (by = date) |>
+    dplyr::ungroup () |>
+    tidyr::replace_na (list (
+        nfiles_changed = 0L,
+        lines_added = 0L,
+        lines_removed = 0L,
+        ncommits = 0L
+    )) |>
+    dplyr::group_by (gh_handle) |>
+    dplyr::mutate (
+        nfiles_changed = cumsum (nfiles_changed),
+        lines_added = cumsum (lines_added),
+        lines_removed = cumsum (lines_removed),
+        ncommits = cumsum (ncommits),
+        date = as.Date (date)
+    ) |>
+    dplyr::ungroup ()
+
+log_reg <- tidyr::pivot_longer (
+    log_reg,
+    c ("nfiles_changed", "lines_added", "lines_removed", "ncommits")
+)
+```
+
+```{r ojs-in-log, echo = FALSE}
+ojs_define (log_in = log_reg)
+```
+```{ojs log-transpose-and-filter}
+log = {
+    return transpose(log_in).map(row => ({
+        ...row,
+        date: new Date(row.date)
+    }));
+}
+viewof varName = Inputs.radio(
+    new Map([
+        ["Nr. files changed", "nfiles_changed"],
+        ["Lines added", "lines_added"],
+        ["Lines removed", "lines_removed"],
+        ["Nr. commits", "ncommits"],
+    ]),
+    {
+        label: "Variable:",
+        value: "ncommits"
+    }
+)
+logFiltered = log.filter(function(row) {
+    return varName.includes(row.name)
+})
+```
+
+```{ojs count-contributors}
+function getMaxValues(log, varName) {
+    const ctbs = log.map((item) => item.gh_handle);
+    const uniqueCtbs = [... new Set(ctbs)];
+
+    const maxValues = {};
+    uniqueCtbs.forEach(handle => {
+        const logHandle = log.filter(item => item.gh_handle === handle);
+        if (logHandle.some(item => item.name === varName)) {
+            maxValues[handle] = Math.max(...logHandle
+                .filter(item => item.name === varName)
+                .map(item => item.value));
+        } else {
+            console.warn(`No entry found with name ${varName} for handle: ${handle}`);
+        }
+    })
+
+    const sortedMaxValues = Object.fromEntries(
+        Object.entries(maxValues).sort((a, b) => b[1] - a[1])
+    );
+
+    return sortedMaxValues;
+}
+maxValues = getMaxValues(log, varName);
+numUniqueCtbs = Object.keys(maxValues).length;
+```
+
+```{ojs num-contribs-selector}
+viewof numContribs = Inputs.range(
+    [1, numUniqueCtbs],
+    {
+        step: 1,
+        label: "Num. of Top Contributors",
+        value: 5
+    }
+)
+function reduceLogToCtbs(log, maxValues, numContribs) {
+    const topCtbs = Object.keys(maxValues).slice(0, numContribs);
+    const logReduced = log.filter(row => topCtbs.includes(row.gh_handle));
+    return logReduced;
+}
+logReduced = reduceLogToCtbs(logFiltered, maxValues, numContribs);
+```
+
+```{ojs log-scale-checkbox}
+viewof logScale = Inputs.radio(["true", "false"],
+    {
+        value: "true",
+        label: "Log Scale:"
+    }
+)
+logType = logScale === "true" ? "log" : "linear";
+```
+
+
+
+```{ojs log-plot}
+Plot.plot({
+    style: `
+        overflow: visible;
+    `,
+    marginLeft: 90,
+    marginBottom: 50,
+    x: { grid: true },
+    y: { grid: true, type: logType },
+    marks: [
+        Plot.ruleY([0]),
+        Plot.lineY(
+            logReduced,
+            {
+                x: "date",
+                y: "value",
+                stroke: "gh_handle",
+                fontSize: 18,
+                curve: "basis-open"
+            },
+        ),
+        Plot.text(
+            logReduced,
+            Plot.selectLast({
+                x: "date",
+                y: "value",
+                z: "gh_handle",
+                text: "gh_handle",
+                textAnchor: "start",
+                dx: 3,
+                fontSize: 18
+            })
+        ),
+        Plot.axisX(
+            {
+                fontSize: 14,
+                label: "",
+                labelArrow: false
+            }
+        ),
+        Plot.axisX(
+            {
+                fontSize: 20,
+                label: "Date",
+                labelAnchor: "center",
+                labelOffset: 40,
+                ticks: []
+            }
+        ),
+        Plot.axisY(
+            {
+                fontSize: 14,
+                label: "",
+                labelArrow: false
+            }
+        ),
+        Plot.axisY(
+            {
+                fontSize: 24,
+                label: "Value",
+                labelAnchor: "center",
+                labelOffset: 85,
+                ticks: []
+            }
+        ),
+    ]
+})
+```

inst/extdata/quarto/index.qmd

@@ -2,4 +2,4 @@
 title: "Historical Package Development"
 ---
 
-This site contains stuff.
+This is a dashboard for the XXX repository.

tests/testthat/helper-cm-data.R

@@ -55,6 +55,10 @@ mock_cm_data <- function () {
         do.call (i, list (path = path))
     })
     names (res) <- gsub ("^cm\\_data\\_", "", data_fns)
+    res$contributors <- get_all_contribs (
+        res$contribs_from_log,
+        res$contribs_from_gh_api
+    )
 
     fs::dir_delete (path)