Skip to contents

Pulling stars, issues, and pull requests

gh-stats.Rmd

collect_git_commits() historically returned one tibble: commit metadata across the user’s non-fork repos. As of v0.1.0 it accepts an include argument that opts in to richer GitHub-level activity data: per-repo stars, authored issues, and authored prs. This vignette walks through each new type, what the schema looks like, and a few patterns for turning the data into something you can embed in your profile README.

Authentication

Every call is routed through gh::gh(), which reads a personal access token from the GITHUB_PAT environment variable. Without a token you are held to 60 requests per hour, which is not enough for any non-trivial account; with a fine-grained PAT (read-only Metadata, Contents, and Issues) the limit is 5,000 requests per hour. The /search/issues endpoint used by include = c("issues", "prs") has its own per-query ceiling of 1,000 results.

# In your .Renviron or a session:
# GITHUB_PAT=ghp_xxx
Sys.getenv("GITHUB_PAT") != ""

Commits (the default)

Calling the function with no include argument preserves the original behavior: a single tibble of commits, with derived date, hour, weekday, and year columns ready for the plotting helpers.

library(ghreadme)

commits <- collect_git_commits(
  user   = "your-username",
  emails = c("you@example.com", "you@work.example.com")
)

head(commits)

Stars received on your repos

include = "stars" returns one row per non-fork repo you own, with the star, fork, and watcher counts plus the language and description. Rows are sorted in descending order of stars.

stats <- collect_git_commits(
  user    = "your-username",
  include = "stars"
)
stats$stars

A common use is a “top-starred repos” bar chart in the README:

library(ggplot2)

stats$stars |>
  head(10) |>
  ggplot(aes(
    x = stargazers_count,
    y = reorder(repo, stargazers_count)
  )) +
  geom_col(fill = "#1E5F8C") +
  labs(
    x = "Stars",
    y = NULL,
    title = "Top 10 repos by stars"
  ) +
  theme_minimal()

Issues you’ve authored across GitHub

include = "issues" hits the /search/issues endpoint with author:<user> type:issue, so it returns issues anywhere on GitHub that you opened (not just on your own repos). Each row carries repo (owner/name), number, title, state, created_at, closed_at, and the issue url.

stats <- collect_git_commits(
  user    = "your-username",
  include = "issues"
)
stats$issues

A small “issues opened per year, by state” summary:

library(dplyr)

stats$issues |>
  mutate(year = as.integer(format(created_at, "%Y"))) |>
  count(year, state) |>
  tidyr::pivot_wider(names_from = state, values_from = n, values_fill = 0)

Pull requests you’ve authored

include = "prs" shares the schema of issues but filters the search with type:pr. The same state field (open or closed) applies.

stats <- collect_git_commits(
  user    = "your-username",
  include = "prs"
)
stats$prs

A “merge rate” approximation, treating closed PRs as merged (the search endpoint does not distinguish merged from closed-without-merging; pull the detail endpoint per PR if you need that):

stats$prs |>
  count(state) |>
  mutate(pct = n / sum(n))

Pulling everything in one call

Passing multiple values to include switches the return type from a single tibble to a named list, with one element per requested type.

stats <- collect_git_commits(
  user    = "your-username",
  emails  = "you@example.com",
  include = c("commits", "stars", "issues", "prs")
)

names(stats)
#> [1] "commits" "stars" "issues" "prs"

A one-shot “profile stats” summary suitable for the top of a profile README:

tibble::tibble(
  metric = c("Public repos", "Total stars", "Commits", "Issues opened",
             "Pull requests opened"),
  value  = c(
    nrow(stats$stars),
    sum(stats$stars$stargazers_count),
    nrow(stats$commits),
    nrow(stats$issues),
    nrow(stats$prs)
  )
) |>
  knitr::kable()

Date bounds

since and until apply to all four types: for commits they are passed to the API directly; for issues and pull requests they are translated to created:>=YYYY-MM-DD / created:<=YYYY-MM-DD qualifiers on the search query.

this_year <- collect_git_commits(
  user    = "your-username",
  emails  = "you@example.com",
  since   = "2025-01-01",
  include = c("commits", "issues", "prs")
)

Embedding in your README

The simplest pattern is to call collect_git_commits() once at the top of README.Rmd (with include set to whatever you want to show), then drop small tables or plots into the body. Because the call hits the API every knit, pair it with the scheduled rebuild from use_profile_readme_action() so the numbers stay fresh.

# In README.Rmd:
stats <- collect_git_commits(
  user    = "your-username",
  emails  = "you@example.com",
  include = c("commits", "stars", "prs")
)

# Then drop a chart or table into the README body