Merge pull request #235 from ropensci/truncation-warning

Truncation warning

R/oa2df.R

@@ -226,7 +226,7 @@ works2df <- function(data, abstract = TRUE, verbose = TRUE,
             first_inst <- empty_inst
           }
           first_inst <- prepend(first_inst, "institution")
-          aff_raw <- list(au_affiliation_raw = l$raw_affiliation_string[1])
+          aff_raw <- list(au_affiliation_raw = replace_w_na(l$raw_affiliation_string[1]))
           l_author <- if (length(l$author) > 0) {
             prepend(replace_w_na(l$author), "au")
           } else {
@@ -318,8 +318,8 @@ abstract_build <- function(ab) {
 #' query_author <- oa_query(
 #'   identifier = NULL,
 #'   entity = "authors",
-#'   last_known_institution.id = "I71267560",
-#'   works_count = ">99"
+#'   last_known_institutions.id = "I71267560",
+#'   works_count = ">500"
 #' )
 #'
 #' res <- oa_request(
@@ -370,14 +370,13 @@ authors2df <- function(data, verbose = TRUE,
       fields$type,
       SIMPLIFY = FALSE
     )
-    sub_affiliation <- item$last_known_institution
-    if (!is.null(sub_affiliation)) {
+    sub_affiliation <- item$last_known_institutions
+    if (!is.null(sub_affiliation) && length(sub_affiliation)) {
+      sub_affiliation <- sub_affiliation[[1]]
       if (is.na(sub_affiliation[[1]])) {
         sub_affiliation <- empty_inst
       }
-      if (length(sub_affiliation$lineage) > 1) {
-        sub_affiliation$lineage <- paste(sub_affiliation$lineage, collapse = ", ")
-      }
+      sub_affiliation$lineage <- paste(sub_affiliation$lineage, collapse = ", ")
       sub_affiliation <- prepend(sub_affiliation, "affiliation")
     }
     sub_affiliation <- replace_w_na(sub_affiliation)

R/oa_fetch.R

@@ -141,19 +141,20 @@ oa_fetch <- function(entity = if (is.null(identifier)) NULL else id_type(shorten
   if (length(final_res[[1]]) == 0) { # || is.null(final_res[[1]][[1]]$id)
     return(NULL)
   }
+  final_res <- unlist(final_res, recursive = FALSE)
 
   if (output == "list") {
-    unlist(final_res, recursive = FALSE)
-  } else {
-    # Flatten out the initial chunking of 50 at a time
-    final_res <- list(unlist(final_res, recursive = FALSE))
-    do.call(rbind, lapply(
-      final_res, oa2df,
-      entity = entity, options = options, abstract = abstract,
-      count_only = count_only, group_by = group_by,
-      verbose = verbose
-    ))
+    return(final_res)
   }
+
+  # Flatten out the initial chunking of 50 at a time
+  final_res <- list(final_res)
+  do.call(rbind, lapply(
+    final_res, oa2df,
+    entity = entity, options = options, abstract = abstract,
+    count_only = count_only, group_by = group_by,
+    verbose = verbose
+  ))
 }
 
 #' Get bibliographic records from OpenAlex database
@@ -397,7 +398,38 @@ oa_request <- function(query_url,
     if (!is.null(res[[result_name]])) data[[i]] <- res[[result_name]]
   }
 
-  unlist(data, recursive = FALSE)
+  data <- unlist(data, recursive = FALSE)
+
+  if (grepl("filter", query_url) && grepl("works", query_url)) {
+    truncated <- unlist(truncated_authors(data))
+    if (length(truncated)) {
+      truncated <- shorten_oaid(truncated)
+      warning(
+        "\nThe following work(s) have truncated lists of authors: ",
+        paste(truncated, collapse = ", "),
+        ".\nQuery each work separately by its identifier to get full list of authors.\n",
+        "For example:\n  ",
+        paste0(
+          "lapply(c(\"",
+          paste(utils::head(truncated, 2), collapse = "\", \""),
+          "\"), \\(x) oa_fetch(identifier = x))"
+        ),
+        "\nDetails at https://docs.openalex.org/api-entities/authors/limitations."
+      )
+    }
+  }
+
+  data
+}
+
+truncated_authors <- function(list_result) {
+  lapply(
+    list_result,
+    function(x){
+      trunc <- x$is_authors_truncated
+      if (!is.null(trunc) && trunc) x$id else NULL
+    }
+  )
 }
 
 get_next_page <- function(paging, i, res = NULL) {

README.Rmd

@@ -189,7 +189,7 @@ Let's first check how many records match the query, then download the entire col
 ```{r}
 my_arguments <- list(
   entity = "authors",
-  last_known_institution.id = "I71267560",
+  last_known_institutions.id = "I71267560",
   works_count = ">499"
 )
 

vignettes/articles/A_Brief_Introduction_to_openalexR.Rmd

@@ -209,7 +209,7 @@ Let's first check how many records match the query, then set `count_only = FALSE
 ```{r}
 my_arguments <- list(
   entity = "authors",
-  last_known_institution.id = "I71267560",
+  last_known_institutions.id = "I71267560",
   works_count = ">499"
   )
 

vignettes/articles/Filters.Rmd

@@ -73,10 +73,10 @@ See the [original documentation](https://docs.openalex.org/how-to-use-the-api/ge
 |-----------------------------------------|--------------------------------------------------------------------------------------------|
 | `works_count`                           | \">99\"                                                                                    |
 | `cited_by_count`                        | \">1000\"                                                                                  |
-| `last_known_institution.id`             | \"I4200000001\"                                                                            |
-| `last_known_institution.ror`            | \"02nr0ka47\"                                                                              |
-| `last_known_institution.country_code`   | \"CA\"                                                                                     |
-| `last_known_institution.type`           | \"nonprofit\"                                                                              |
+| `last_known_institutions.id`            | \"I4200000001\"                                                                            |
+| `last_known_institutions.ror`           | \"02nr0ka47\"                                                                              |
+| `last_known_institutions.country_code`  | \"CA\"                                                                                     |
+| `last_known_institutions.type`          | \"nonprofit\"                                                                              |
 | `openalex`                              | \"A2208157607\"                                                                            |
 | `orcid`                                 | \"0000-0001-6187-6610\"                                                                    |
 | `mag`                                   | \"2208157607\"                                                                             |