update paper machine and papers

.github/workflows/update-publications.yaml

@@ -59,6 +59,40 @@ jobs:
           bib <- paste(c(bib_old, bib_new), collapse = "\n")
           
           write(bib, "_data/papers.bib")
+
+          ## filter latest version
+          df <- bib2df("_data/papers.bib") |>
+            dplyr::mutate(
+              version = dplyr::if_else(
+                grepl("wellcomeopenres", DOI),
+                sub("^.*\\.([1-9][0-9]*)$", "\\1", DOI),
+                "1"
+              ),
+              base_doi = dplyr::if_else(
+                grepl("wellcomeopenres", DOI),
+                sub("\\.([1-9][0-9]*)$", "", DOI),
+                DOI
+              )
+            ) |>
+            dplyr::group_by(base_doi) |>
+            dplyr::filter(version == max(version)) |>
+            dplyr::ungroup() |>
+            dplyr::select(-base_doi, -version)
+
+          ## filter preprints
+          df <- df |>
+            dplyr::group_by(TITLE) |>
+            dplyr::mutate(n = dplyr::n()) |>
+            dplyr::mutate(id = seq_len(dplyr::n())) |>
+            dplyr::mutate(n_pp = sum(JOURNAL == "medRxiv")) |>
+            dplyr::filter(
+              !((n_pp < n) & (JOURNAL == "medRxiv"))
+            ) |>
+            dplyr::filter(
+              !(n > 1 & id < max(id))
+            ) |>
+            dplyr::ungroup()
+
         shell: Rscript {0}
 
       - uses: peter-evans/create-pull-request@v6

_automation/get_new_papers.R

@@ -1,37 +1,55 @@
-`%>%` <- magrittr::`%>%`
-
 get_new_papers_orcid <- function(orcid, from_date) {
-
-  rcrossref::cr_works(filter = list(orcid = orcid,
-                                    from_deposit_date = from_date,
-                                    type = "journal-article")) %>%
+  refs <- rcrossref::cr_works(filter = list(
+    orcid = orcid,
+    from_deposit_date = from_date
+  )) |>
     purrr::pluck("data")
-
+
+  if (nrow(refs) > 0) {
+    refs <- refs |>
+      dplyr::mutate(
+        container.title = dplyr::if_else(
+          publisher == "Cold Spring Harbor Laboratory",
+          "medRxiv",
+          container.title
+        ),
+        type = dplyr::if_else(
+          publisher == "Cold Spring Harbor Laboratory",
+          "journal-article",
+          type
+        )
+      ) |>
+      dplyr::filter(type == "journal-article")
+    existing_papers <- dois_from_bib("_data/papers.bib")
+    refs <- refs |>
+      dplyr::filter(!(doi %in% existing_papers))
+  }
+  return(refs)
 }
 
 get_new_papers_team <- function(from_date) {
 
-  recent_papers <- fs::dir_ls("_data/team", regexp = "\\w+\\-\\w+\\.yml") %>%
-    purrr::map(yaml::read_yaml) %>%
-    purrr::keep(function(x) x[["current-member"]]) %>%
-    purrr::map("orcid") %>%
-    purrr::keep(~ !is.null(.x)) %>%
-    purrr::map_dfr(get_new_papers_orcid, from_date = from_date) %>%
+  recent_papers <- fs::dir_ls("_data/team", regexp = "\\w+\\-\\w+\\.yml") |>
+    purrr::map(yaml::read_yaml) |>
+    purrr::keep(function(x) x[["current-member"]]) |>
+    purrr::map("orcid") |>
+    purrr::keep(~ !is.null(.x)) |>
+    purrr::map_dfr(get_new_papers_orcid, from_date = from_date) |>
     dplyr::distinct()
 
   return(recent_papers)
 
 }
 
 create_bibentries <- function(papers) {
-  papers %>%
+  papers |>
     dplyr::mutate(
       author = purrr::map(author, ~ dplyr::filter(.x, !is.na(given) &!is.na(family))),
       author = purrr::map(author, ~ paste(.x$given, .x$family)),
       author = purrr::map(author, as.person),
       year = format(lubridate::parse_date_time(issued, c("Y-m-d", "Y-m", "Y")), "%Y")
-    ) %>%
-    dplyr::rowwise() %>%
+    ) |>
+    dplyr::rowwise() |>
     dplyr::mutate(
       bib = list(bibentry(
         "Article", 
@@ -42,7 +60,13 @@ create_bibentries <- function(papers) {
         doi = doi,
         key = ids::adjective_animal()
       ))
-    ) %>%
+    ) |>
     dplyr::pull(bib)
 
 }
+
+dois_from_bib <- function(bib) {
+  bib |>
+    bibtex::read.bib() |>
+    purrr::map_chr(~ .x$doi)
+}