✨ generate_badge_tables type functions now generate badges statically
#117
Conversation
jdhoffa
added
feature
|
|
||
| format_lifecycle <- function(lifecycle_badge) { | ||
| if (is.na(lifecycle_badge)) { | ||
| return("No lifecycle badge found.") | ||
| } | ||
|
|
||
| desc <- "![Lifecycle]" | ||
| link <- "https://lifecycle.r-lib.org/articles/stages.html" | ||
|
|
||
| glue::glue("[{desc}({lifecycle_badge})]({link})") | ||
| } | ||
|
|
||
| table_lifecycle <- function(repo_path) { | ||
| readme <- get_gh_text_file(repo_path, file_path = "README.md") | ||
|
|
||
| if (is.null(readme)) { | ||
| return(NA_character_) | ||
| } | ||
|
|
||
| pattern <- "https://img.shields.io/badge/lifecycle-\\S+.svg" | ||
|
|
||
| lifecycle_badge <- readme[grepl(pattern, readme)][[1L]] # nolint: regex_subset_linter | ||
| lifecycle_badge <- gsub(".*(https[^)]*\\.svg).*", "\\1", lifecycle_badge) | ||
|
|
||
| if (length(lifecycle_badge) == 0L) { | ||
| return(NA_character_) | ||
| } | ||
|
|
||
| return(lifecycle_badge) | ||
| } |
There was a problem hiding this comment.
Note to reviewer:
These changes have very little to do with this PR.
All I have done is copy and paste two functions from the generate_package_table.R file (which I deleted).
Those were the only two functions that were used anywhere in the package, and since they were used by this function, I felt they should live here.
| banks_repos <- list( | ||
| c( | ||
| path = "RMI-PACTA/r2dii.data", | ||
| lifecycle = "stable", | ||
| ci_check = "R.yml" | ||
| ), | ||
| c( | ||
| path = "RMI-PACTA/r2dii.match", | ||
| lifecycle = "stable", | ||
| ci_check = "R.yml" | ||
| ), | ||
| c( | ||
| path = "RMI-PACTA/r2dii.analysis", | ||
| lifecycle = "stable", | ||
| ci_check = "R.yml" | ||
| ), | ||
| c( | ||
| path = "RMI-PACTA/r2dii.plot", | ||
| lifecycle = "experimental", | ||
| ci_check = "R.yml" | ||
| ) |
There was a problem hiding this comment.
generate_package_table function.
There was a problem hiding this comment.
nit: doesn't table_lifecycle() aim to scrape the lifecycle status from the repo's README, in which case (assuming it works reliably) in wouldn't be necessary to maintain the lifecycle here?
|
|
||
| test_that("handles empty input", { | ||
| empty_input <- character(0) | ||
| result <- generate_package_table(empty_input) | ||
|
|
||
| expect_true(tibble::is_tibble(result) && nrow(result) == 0) | ||
| }) |
There was a problem hiding this comment.
There is no need for the function to handle empty input...
|
Supersedes #119 |
R/format_badges.R
Outdated
| format_badge_url <- function(display, path) { | ||
| glue::glue("[{display}]({path})") | ||
| } | ||
|
|
||
| format_name_badge <- function(repo_path) { |
There was a problem hiding this comment.
nit: the naming convention used is confusing to me. I kinda expected format_badge_url() to return a "formatted" badge URL (whatever that would be 🤣), but it returns markdown for a linked badge.
There was a problem hiding this comment.
I renamed it build_markdown_link
| "" # nolint: line_length_linter | ||
| ), | ||
| path = glue::glue( | ||
| "https://github.com/{repo_path}/actions/workflows/{ci_check}?query=branch%3Amain" # nolint: line_length_linter |
There was a problem hiding this comment.
I think all of the relevant repos use main not master, so this is not a problem?
Note to reviewer
Sorry in advance, I know this is a big PR. I am overhauling how most of these functions work entirely.
The simplest way to review that the behavior of this PR functions as expected would be to inspect the rendered README.md
Before
A nightly knit action would reknit the
README.Rmdand dynamically scrape different files (e.g.DESCRIPTION,CODEOWNERS, etc.) within each GitHub repo, and re-render theREADME.mdwith the latest badges.This was an over-engineered solution (and the GH action was a pain in the ass to approve)
After this PR
Now, we leverage the dynamic updating of the badge itself at it's source link, and simply use the functions here to format the relevant
.mdURL.latest_shacolumn entirely, this added very little actual valuelifecycleandci_checkargument inputs need to manually be kept up to date with whatever they should point to in the source repo, there is not automated way to check if these are valid (now).nightly_buildaction entirely (finally).✅ While I was there, I added some missing repos to the situation report, and updated relevant
lifecycleandci_checkvalues as appropriate.Closes #51 #47 #35 #34 #10
Supersedes #33