feat: Avoid roundtrip to and from HTML for the entire file (#826)

R/auto.R

@@ -248,8 +248,11 @@ merge_dev_news <- function(fledgeling, new_version) {
 
   n_dev <- rle(dev_idx)$lengths[[1]]
 
+  raw <- write_news_sections(fledgeling[["news"]][seq_len(n_dev), ])
+  versions <- parse_news_md(raw)
+
   news <- regroup_news(
-    unlist(fledgeling[["news"]]$news[seq_len(n_dev)], recursive = FALSE)
+    unlist(news_from_versions(versions), recursive = FALSE)
   )
 
   raw <- format_news_subsections(news, header_level = 2)

R/fledgling.R

@@ -44,9 +44,9 @@ read_news <- function(news_lines = NULL) {
     }
   }
 
-  news <- parse_news_md(news_lines, strict = TRUE)
+  versions <- parse_news_md(news_lines)
 
-  if (is.null(news)) {
+  if (is.null(versions)) {
     return(
       list(
         section_df = NULL,
@@ -57,7 +57,7 @@ read_news <- function(news_lines = NULL) {
   }
 
   # NEWS content under no version
-  if (length(news) == 1 && !nzchar(names(news))) {
+  if (length(versions) == 1 && !nzchar(names(versions))) {
     cli::cli_abort("All {.file NEWS.md} content must be under version headers.")
   }
 
@@ -77,9 +77,9 @@ read_news <- function(news_lines = NULL) {
     section_start
   }
 
-  duplicate_version_names_present <- anyDuplicated(names(news))
+  duplicate_version_names_present <- anyDuplicated(names(versions))
   if (duplicate_version_names_present) {
-    duplicated_version_names <- toString(names(news)[duplicated(names(news))])
+    duplicated_version_names <- toString(names(versions)[duplicated(names(versions))])
     cli::cli_abort(
       c(
         "Can't deal with duplicate version names: {duplicated_version_names}.",
@@ -88,25 +88,20 @@ read_news <- function(news_lines = NULL) {
     )
   }
 
-  starts <- purrr::map_int(names(news), get_section_start, news_lines)
+  starts <- purrr::map_int(names(versions), get_section_start, news_lines)
   ends <- c(starts[seq_along(starts[-1]) + 1] - 1L, length(news_lines))
 
   section_df <- tibble::tibble(
     start = starts,
     end = ends,
     h2 = grepl("##", news_lines[starts]), # TODO does not account for all syntaxes,
     raw = map2_chr(starts, ends, ~ paste(news_lines[seq2(.x, .y)], collapse = "\n")),
-    news = news_collection_fix_name_and_level(news)
+    versions = versions,
+    section_state = "keep",
+    title = names(versions),
+    parse_versions(names(versions))[, c("version", "date", "nickname")],
   )
 
-  section_df$section_state <- "keep"
-
-  section_df$title <- names(news)
-
-  parsed_titles <- parse_versions(names(news))[, c("version", "date", "nickname")]
-
-  section_df <- vctrs::vec_cbind(section_df, parsed_titles)
-
   # create, update or re-use preamble
   is_preamble_absent <- (section_df[["start"]][[1]] == 1)
   if (is_preamble_absent) {
@@ -128,8 +123,21 @@ read_news <- function(news_lines = NULL) {
   )
 }
 
-news_collection_fix_name_and_level <- function(news_collection) {
-  news_wrapped <- unname(split(news_collection, seq_len(length(news_collection))))
+parse_news_md <- function(news) {
+  news <- protect_hashtag(news)
+  versions <- versions_from_news(news)
+  if (is.null(versions)) {
+    return(NULL)
+  }
+
+  check_top_level_headers(versions)
+  rlang::set_names(unclass(versions), news_collection_get_section_name(versions))
+}
+
+news_from_versions <- function(news_collection) {
+  news_treated <- news_collection_treat_section(news_collection)
+
+  news_wrapped <- unname(split(news_treated, seq_len(length(news_treated))))
 
   map(news_wrapped, news_fix_name_and_level)
 }

R/pandoc.R

@@ -1,6 +1,4 @@
-parse_news_md <- function(news = brio::read_lines(news_path()), strict = FALSE) {
-  news <- protect_hashtag(news)
-
+versions_from_news <- function(news) {
   temp_file <- withr::local_tempfile(fileext = ".md")
   brio::write_lines(news, temp_file)
 
@@ -31,17 +29,9 @@ parse_news_md <- function(news = brio::read_lines(news_path()), strict = FALSE)
   }
   if (length(versions) == 0) {
     cli::cli_abort("Empty {.file NEWS.md}")
-
-    contents <- markdownify(html)
-    return(list(contents))
-  }
-
-  if (strict) {
-    check_top_level_headers(versions)
   }
 
-  out <- news_collection_treat_section(versions)
-  out
+  versions
 }
 
 news_collection_treat_section <- function(news_collection) {
@@ -76,6 +66,10 @@ news_treat_section <- function(section) {
   )
 }
 
+news_collection_get_section_name <- function(news_collection) {
+  purrr::map_chr(news_collection, news_get_section_name)
+}
+
 news_get_section_name <- function(section) {
   xml2::xml_text(xml2::xml_child(section))
 }

R/update-news.R

@@ -19,7 +19,47 @@ update_news_impl <- function(commits,
   }
 
   fledgeling <- fledgeling %||% read_fledgling()
+  add_news_to_fledgeling(fledgeling, news_lines, which, news_items)
+}
+
+add_news_to_fledgeling_samedev <- function(fledgeling, news_lines) {
+  # Append and regroup
+  initializing <- is.null(fledgeling[["news"]])
+
+  if (initializing) {
+    fledgeling[["news"]] <- tibble::tibble(
+      start = 3,
+      h2 = FALSE,
+      version = fledgeling[["version"]],
+      date = "",
+      nickname = "",
+      original = "",
+      news = list(parse_news_lines(news_lines)),
+      raw = news_lines,
+      section_state = "new"
+    )
+  } else {
+    old_news <- news_from_versions(fledgeling[["news"]]$versions[1])[[1]]
+    combined <- c(parse_news_lines(news_lines), old_news)
+    combined <- purrr::discard(combined, purrr::is_empty)
+    regrouped <- regroup_news(combined)
+    fledgeling[["news"]]$raw[[1]] <- format_news_subsections(regrouped, header_level = 2)
+    fledgeling[["news"]][1, ]$section_state <- "new"
+  }
+
+  if (fledge_chatty()) {
+    cli_alert("Added items to {.file {news_path()}}.")
+  }
+
+  fledgeling
+}
 
+add_news_to_fledgeling <- function(
+  fledgeling,
+  news_lines,
+  which,
+  news_items
+) {
   # isTRUE() as NEWS.md can be empty
   dev_header_present <- isTRUE(
     grepl(
@@ -35,100 +75,72 @@ update_news_impl <- function(commits,
       which <- "dev"
     }
   }
-  initializing <- is.null(fledgeling[["news"]])
 
   if (which == "samedev") {
     if (!dev_header_present) {
       cli::cli_abort("Can't find a development version header in {.file NEWS.md}.")
     }
 
-    # Append and regroup
-
-    if (initializing) {
-      fledgeling[["news"]] <- tibble::tibble(
-        start = 3,
-        h2 = FALSE,
-        version = fledgeling[["version"]],
-        date = "",
-        nickname = "",
-        original = "",
-        news = list(parse_news_md(news_lines)),
-        raw = "",
-        section_state = "new"
-      )
-    } else {
-      combined <- c(
-        parse_news_md(news_lines),
-        fledgeling[["news"]]$news[[1]]
-      )
-      combined <- purrr::discard(combined, purrr::is_empty)
-      regrouped <- regroup_news(combined)
-      fledgeling[["news"]]$raw[[1]] <- format_news_subsections(regrouped, header_level = 2)
-      fledgeling[["news"]][1, ]$section_state <- "new"
-    }
+    return(add_news_to_fledgeling_samedev(fledgeling, news_lines))
+  }
 
-    if (fledge_chatty()) {
-      cli_alert("Added items to {.file {news_path()}}.")
-    }
-  } else {
-    current_version <- fledgeling[["version"]]
+  initializing <- is.null(fledgeling[["news"]])
 
-    new_version <- fledge_guess_version(current_version, which)
-    fledgeling[["version"]] <- new_version
+  current_version <- fledgeling[["version"]]
 
-    # In the galley test for the demo vignette, for some reason, `is.na(get_date())`
-    if (!is.null(fledgeling[["date"]]) && !is.na(get_date())) {
-      fledgeling[["date"]] <- as.character(get_date())
-    }
+  new_version <- fledge_guess_version(current_version, which)
+  fledgeling[["version"]] <- new_version
 
-    if (initializing) {
-      no_actual_commit <- (nrow(news_items) == 1) &&
-        (news_items[["description"]] == same_as_previous())
+  # In the galley test for the demo vignette, for some reason, `is.na(get_date())`
+  if (!is.null(fledgeling[["date"]]) && !is.na(get_date())) {
+    fledgeling[["date"]] <- as.character(get_date())
+  }
 
-      if (no_actual_commit) {
-        news_lines <- sprintf("## Uncategorized\n\n- %s", added_changelog())
-      }
-    }
+  if (initializing) {
+    no_actual_commit <- (nrow(news_items) == 1) &&
+      (news_items[["description"]] == same_as_previous())
 
-    if (dev_header_present) {
-      combined <- c(
-        parse_news_md(news_lines),
-        fledgeling[["news"]]$news[[1]]
-      )
-      combined <- purrr::discard(combined, purrr::is_empty)
-      news <- regroup_news(combined)
-      fledgeling[["news"]] <- fledgeling[["news"]][-1, ]
-    } else {
-      news <- parse_news_md(news_lines)
+    if (no_actual_commit) {
+      news_lines <- sprintf("## Uncategorized\n\n- %s", added_changelog())
     }
+  }
 
-    raw <- format_news_subsections(news, header_level = 2)
+  if (dev_header_present) {
+    old_news <- news_from_versions(fledgeling[["news"]]$versions[1])[[1]]
+    combined <- c(parse_news_lines(news_lines), old_news)
+    combined <- purrr::discard(combined, purrr::is_empty)
+    news <- regroup_news(combined)
+    fledgeling[["news"]] <- fledgeling[["news"]][-1, ]
+  } else {
+    news <- parse_news_lines(news_lines)
+  }
 
-    section_df <- tibble::tibble(
-      start = 3,
-      end = NA,
-      h2 = fledgeling[["news"]][["h2"]][1] %||% FALSE,
-      version = new_version,
-      date = maybe_date(fledgeling[["news"]]),
-      nickname = NA,
-      news = list(news),
-      raw = raw,
-      title = "",
-      section_state = "new"
-    )
+  raw <- format_news_subsections(news, header_level = 2)
+
+  section_df <- tibble::tibble(
+    start = 3,
+    end = NA,
+    h2 = fledgeling[["news"]][["h2"]][1] %||% FALSE,
+    version = new_version,
+    date = maybe_date(fledgeling[["news"]]),
+    nickname = NA,
+    news = list(news),
+    raw = raw,
+    title = "",
+    section_state = "new"
+  )
 
-    fledgeling[["news"]] <- vctrs::vec_rbind(
-      section_df,
-      fledgeling[["news"]]
-    )
+  fledgeling[["news"]] <- vctrs::vec_rbind(
+    section_df,
+    fledgeling[["news"]]
+  )
 
-    if (fledge_chatty()) {
-      cli_h2("Updating Version")
+  if (fledge_chatty()) {
+    cli_h2("Updating Version")
 
-      cli_alert_success("Package version bumped to {.field {new_version}}.")
+    cli_alert_success("Package version bumped to {.field {new_version}}.")
 
-      cli_alert("Added header to {.file {news_path()}}.")
-    }
+    cli_alert("Added header to {.file {news_path()}}.")
   }
 
   fledgeling
@@ -307,3 +319,9 @@ get_date <- function() {
   author_time <- parsedate::parse_iso_8601(Sys.getenv("GIT_COMMITTER_DATE"))
   as.Date(author_time)
 }
+
+parse_news_lines <- function(news) {
+  news <- protect_hashtag(news)
+  versions <- versions_from_news(news)
+  news_collection_treat_section(versions)
+}