feat: Automatically extract Posit versions (#31)

DESCRIPTION

@@ -1,7 +1,7 @@
 Type: Package
 Package: audit.base
 Title: Base package for Posit Checks
-Version: 0.6.16
+Version: 0.6.17
 Authors@R:
     person("Jumping", "Rivers", , "[email protected]", role = c("aut", "cre"))
 Description: Base package for sharing classes between posit audit
@@ -25,6 +25,7 @@ Imports:
     yaml
 Suggests: 
     jsonlite,
+    rvest,
     testthat (>= 3.0.0)
 Remotes:
     jumpingrivers/serverHeaders

NAMESPACE

@@ -20,7 +20,7 @@ export(init_r6_check)
 export(init_r6_checks)
 export(logger)
 export(print_colour_versions)
-export(update_software_csv)
+export(update_all_versions)
 import(R6)
 importFrom(dplyr,"%>%")
 importFrom(rlang,.data)

NEWS.md

@@ -1,3 +1,7 @@
+# audit.base 0.6.17 _2024-06-24_
+- chore: Software bump
+- feat: Automatically extract connect/workbench versions
+
 # audit.base 0.6.16 _2024-06-24_
 - chore: Use jrStyling
 

R/create_software_tibble.R

@@ -1,14 +1,3 @@
-#' Helper function for updating R/Py/Quarto versions
-#'
-#' Currently this function would only be called by the package author, as it
-#' updates inst/extdata/versions/software.csv
-#' @export
-update_software_csv = function() {
-  software = create_software_tibble()
-  readr::write_csv(software, file = "inst/extdata/versions/software.csv")
-  return(invisible(software))
-}
-
 # https://gitlab.com/jumpingrivers/services/de/spd/infrastructure-template/-/blob/5c584fced32a6fc8fd7b25b3ea78f6fb7a8bd7ca/template/ansible/scripts/versions.sh
 create_software_tibble = function() {
   r = get_latest_versions_from_posit("r")

R/get_posit_cves.R

@@ -0,0 +1,40 @@
+get_posit_remote_versions = function(type = c("connect", "workbench")) {
+  type = match.arg(type)
+  url = if (type == "connect") "https://docs.posit.co/connect/news/" 
+        else "https://docs.posit.co/ide/news/"
+  page = rvest::read_html(url)
+  sections = rvest::html_elements(page, "section")
+  v_tibbles = purrr::map_df(sections, extract_posit_cves)
+  all_v = get_all_remote_versions(page) |>
+    dplyr::filter(!.data$version %in% v_tibbles$version) %>%
+    dplyr::bind_rows(v_tibbles) %>%
+    dplyr::arrange(dplyr::desc(.data$version))
+  all_v
+}
+
+extract_posit_cves = function(section) {
+  posit_name = rvest::html_attrs(section)
+  posit_name = as.vector(posit_name["id"])
+  posit_version = stringr::str_extract(posit_name, "[0-9]{4}\\.[0-9]{1,2}\\.[0-9]{1,2}")
+
+  li = section %>%
+    rvest::html_elements("li") %>%
+    rvest::html_text2()
+
+  cves = stringr::str_extract(li, "^CVE-[0-9]{4}-[0-9]*")
+  cves = cves[!is.na(cves)]
+
+  if (length(cves) == 0L || is.na(posit_version)) {
+    tibble::tibble(version = character(0), cve = character(0))
+  } else {
+    tibble::tibble(version = posit_version, cve = cves)
+  }
+}
+
+get_all_remote_versions = function(page) {
+  versions = page %>%
+    rvest::html_nodes("h2") %>%
+    rvest::html_text()
+  v = stringr::str_extract(versions, "202[0-9]\\.[0-9]{2}\\.[0-9]{1,2}")
+  tibble::tibble(version = v[!is.na(v)], cve = "")
+}

R/posit_versions.R

@@ -12,8 +12,8 @@ get_posit_versions = function(type = c("connect", "workbench", "drivers")) {
   fname = system.file("extdata", "versions", paste0(type, ".csv"),
     mustWork = TRUE, package = "audit.base"
   )
-  versions = readr::read_csv(fname, comment = "#", col_types = c("c", "D", "c"))
-  versions = dplyr::arrange(versions, dplyr::desc(date))
+  versions = readr::read_csv(fname, comment = "#", col_types = c("c", "c"))
+  versions = dplyr::arrange(versions, dplyr::desc(.data$version))
   return(versions)
 }
 
@@ -45,13 +45,17 @@ audit_posit_version = function(posit_version, type = c("connect", "workbench", "
   return(invisible(NULL))
 }
 
+
+
 lookup_version = function(posit_version, type) {
   versions = get_posit_versions(type = type)
   version_as_date = version_to_date(posit_version)
-  if (is.na(version_as_date) || version_as_date < min(versions$date)) {
+  all_dates = unlist(purrr::map(versions$version, version_to_date))
+
+  if (is.na(version_as_date) || version_as_date < min(all_dates)) {
     # Older than DB
     row_number = NA_integer_
-  } else if (version_as_date > max(versions$date)) {
+  } else if (version_as_date > max(all_dates)) {
     # Newer than DB
     row_number = 1L
   } else {

R/update_all_versions.R

@@ -0,0 +1,20 @@
+#' Helper functions for updating R/Py/Quarto/Posit versions
+#'
+#' Currently this function would only be called by the package author, as it
+#' updates inst/extdata/versions/*.csv
+#' @export
+update_all_versions = function() {
+  readr::write_csv(
+    get_posit_remote_versions(type = "connect"),
+    file = "inst/extdata/versions/connect.csv"
+  )
+
+  readr::write_csv(
+    get_posit_remote_versions(type = "workbench"),
+    file = "inst/extdata/versions/workbench.csv"
+  )
+
+  software = create_software_tibble()
+  readr::write_csv(software, file = "inst/extdata/versions/software.csv")
+  return(invisible(NULL))
+}