build-cran-binaries/local/dedupe-audit-issue.R

# One-off maintenance: collapse the duplicate arch subsections that accumulated
# in the "Missing package binaries for latest version (<family>)" issues.
#
# A bug in weekly-missing-binaries-audit.R matched the existing "### <arch>"
# subsection by its bare header while writing headers with a
# " (N missing, M to rebuild)" suffix, so every audit run appended a fresh block
# instead of replacing it. This script rewrites each "## <platform>" section to
# keep only the *last* (freshest) block per arch. The audit fix prevents further
# accumulation; this cleans up what is already there.
#
# Env: FORGEJO_TOKEN (required). DRY_RUN=1 to preview counts without patching.

library(httr2, quietly = TRUE)

forgejo_base <- "https://git.devxy.io/api/v1"
repo <- "devxy/build-cran-binaries"
token <- Sys.getenv("FORGEJO_TOKEN")
dry_run <- nchar(Sys.getenv("DRY_RUN")) > 0

if (nchar(token) == 0) {
  stop("FORGEJO_TOKEN env var is not set")
}

issue_titles <- c(
  "Missing package binaries for latest version (Alpine)",
  "Missing package binaries for latest version (Ubuntu)",
  "Missing package binaries for latest version (Red Hat)"
)

# Collapse one "## <platform>" block: keep only the last block per arch,
# emitted in order of first appearance. `pl[1]` is the "## <platform>" header.
dedupe_platform <- function(pl) {
  sub_hdr <- which(grepl("^### ", pl))
  if (length(sub_hdr) == 0) {
    return(pl)
  }
  preamble <- pl[seq_len(sub_hdr[1] - 1)]
  sub_end <- c(sub_hdr[-1] - 1, length(pl))
  blocks <- lapply(seq_along(sub_hdr), function(k) {
    pl[seq(sub_hdr[k], sub_end[k])]
  })
  arches <- vapply(
    blocks,
    function(b) sub("^### (\\S+).*", "\\1", b[1]),
    character(1)
  )
  # Index of the last block for each arch, kept in first-appearance order.
  last_idx <- vapply(
    unique(arches),
    function(a) max(which(arches == a)),
    integer(1)
  )
  keep <- sort(last_idx)
  out <- preamble
  for (i in keep) {
    out <- c(out, blocks[[i]])
  }
  out
}

process_issue <- function(title) {
  search_url <- sprintf(
    "%s/repos/%s/issues?type=issues&state=open&q=%s&limit=50",
    forgejo_base,
    repo,
    utils::URLencode(title, reserved = TRUE)
  )
  resp <- request(search_url) |>
    req_headers(Authorization = paste("token", token)) |>
    req_perform()
  issues <- resp_body_json(resp, simplifyVector = FALSE)
  match_idx <- which(vapply(issues, function(x) x$title, character(1)) == title)
  if (length(match_idx) == 0) {
    cat(sprintf("[skip] No issue found: %s\n", title))
    return(invisible())
  }

  issue_number <- issues[[match_idx[1]]]$number
  body <- issues[[match_idx[1]]]$body
  if (is.null(body) || nchar(body) == 0) {
    cat(sprintf("[skip] Empty body: #%d %s\n", issue_number, title))
    return(invisible())
  }

  lines <- strsplit(body, "\n", fixed = TRUE)[[1]]
  before <- sum(grepl("^### ", lines))

  # Split off the "## Excluded packages" footer so it is preserved verbatim.
  excl_idx <- which(lines == "## Excluded packages")
  footer <- character(0)
  if (length(excl_idx) > 0) {
    pre_dash <- which(lines == "---" & seq_along(lines) < excl_idx[1])
    cut <- if (length(pre_dash) > 0) pre_dash[length(pre_dash)] else excl_idx[1]
    footer <- lines[seq(cut, length(lines))]
    lines <- lines[seq_len(cut - 1)]
  }

  # Platform headers ("## <platform>"); everything before the first is preamble.
  plat_idx <- which(grepl("^## ", lines))
  if (length(plat_idx) == 0) {
    cat(sprintf("[skip] No platform sections: #%d %s\n", issue_number, title))
    return(invisible())
  }
  top <- lines[seq_len(plat_idx[1] - 1)]
  plat_end <- c(plat_idx[-1] - 1, length(lines))

  new_lines <- top
  for (j in seq_along(plat_idx)) {
    pl <- lines[seq(plat_idx[j], plat_end[j])]
    new_lines <- c(new_lines, dedupe_platform(pl))
  }
  if (length(footer) > 0) {
    new_lines <- c(new_lines, footer)
  }

  after <- sum(grepl("^### ", new_lines))
  cat(sprintf(
    "#%d %s: %d -> %d arch subsections%s\n",
    issue_number,
    title,
    before,
    after,
    if (dry_run) " (dry run, not patched)" else ""
  ))

  if (dry_run) {
    return(invisible())
  }

  patch_url <- sprintf(
    "%s/repos/%s/issues/%d",
    forgejo_base,
    repo,
    issue_number
  )
  request(patch_url) |>
    req_headers(
      Authorization = paste("token", token),
      `Content-Type` = "application/json"
    ) |>
    req_body_json(list(body = paste(new_lines, collapse = "\n"))) |>
    req_method("PATCH") |>
    req_perform()
  cat(sprintf("  patched #%d\n", issue_number))
}

for (t in issue_titles) {
  process_issue(t)
}
cat("Done.\n")