From 5a9f81876ba0b5830a1652eb85cdf971f9e39bf7 Mon Sep 17 00:00:00 2001
From: Raffaele Mancuso <raffaele.mancuso4@unibo.it>
Date: Tue, 4 Feb 2025 10:30:04 +0100
Subject: [PATCH] Implement timeout in httr

Fixes #312
---
 R/oa_fetch.R      | 24 +++++++++++++++---------
 man/oa_fetch.Rd   |  7 ++++++-
 man/oa_request.Rd |  7 ++++++-
 3 files changed, 27 insertions(+), 11 deletions(-)

diff --git a/R/oa_fetch.R b/R/oa_fetch.R
index 050fbe2..a16a680 100644
--- a/R/oa_fetch.R
+++ b/R/oa_fetch.R
@@ -83,7 +83,8 @@ oa_fetch <- function(entity = if (is.null(identifier)) NULL else id_type(shorten
                      count_only = FALSE,
                      mailto = oa_email(),
                      api_key = oa_apikey(),
-                     verbose = FALSE) {
+                     verbose = FALSE,
+                     timeout = 30) {
   output <- match.arg(output)
   entity <- match.arg(entity, oa_entities())
 
@@ -142,7 +143,8 @@ oa_fetch <- function(entity = if (is.null(identifier)) NULL else id_type(shorten
       mailto = mailto,
       api_key = api_key,
       parse = output != "raw",
-      verbose = verbose
+      verbose = verbose,
+      timeout = timeout
     )
   }
 
@@ -197,6 +199,9 @@ oa_fetch <- function(entity = if (is.null(identifier)) NULL else id_type(shorten
 #' If FALSE, returns the raw JSON response as string.
 #' @param verbose Logical.
 #' If TRUE, print information about the querying process. Defaults to TRUE.
+#' @param timeout Numeric.
+#' Number of seconds to wait for a response until giving up. Can not be less than 1 ms.
+#' Defaults to 30.
 #'
 #' @return a data.frame or a list of bibliographic records.
 #'
@@ -312,7 +317,8 @@ oa_request <- function(query_url,
                        mailto = oa_email(),
                        api_key = oa_apikey(),
                        parse = TRUE,
-                       verbose = FALSE) {
+                       verbose = FALSE,
+                       timeout = 30) {
   # https://httr.r-lib.org/articles/api-packages.html#set-a-user-agent
   ua <- httr::user_agent("https://github.com/ropensci/openalexR/")
 
@@ -335,7 +341,7 @@ oa_request <- function(query_url,
   }
 
   # first, download info about n. of items returned by the query
-  res <- api_request(query_url, ua, query = query_ls, api_key = api_key, parse = FALSE)
+  res <- api_request(query_url, ua, query = query_ls, api_key = api_key, parse = FALSE, timeout = timeout)
   res_parsed <- jsonlite::fromJSON(res, simplifyVector = FALSE)
   res_meta <- res_parsed$meta
   if (parse) {
@@ -364,7 +370,7 @@ oa_request <- function(query_url,
       if (verbose) cat("=")
       Sys.sleep(1 / 10)
       query_ls[[paging]] <- next_page
-      res <- api_request(query_url, ua, query = query_ls)
+      res <- api_request(query_url, ua, query = query_ls, timeout = timeout)
       data <- c(data, res[[result_name]])
       i <- i + 1
       next_page <- get_next_page("cursor", i, res)
@@ -412,10 +418,10 @@ oa_request <- function(query_url,
     query_ls[[paging]] <- next_page
 
     if (parse) {
-      res <- api_request(query_url, ua, query = query_ls, parse = TRUE)
+      res <- api_request(query_url, ua, query = query_ls, parse = TRUE, timeout = timeout)
       if (!is.null(res[[result_name]])) data[[i]] <- res[[result_name]]
     } else {
-      raw <- api_request(query_url, ua, query = query_ls, parse = FALSE)
+      raw <- api_request(query_url, ua, query = query_ls, parse = FALSE, timeout = timeout)
       data[[i]] <- raw
     }
   }
@@ -707,8 +713,8 @@ oa_random <- function(entity = oa_entities(),
   final_res
 }
 
-api_request <- function(query_url, ua, query, api_key = oa_apikey(), parse = TRUE) {
-  res <- httr::GET(query_url, ua, query = query, httr::add_headers(api_key = api_key))
+api_request <- function(query_url, ua, query, api_key = oa_apikey(), parse = TRUE, timeout = 30) {
+  res <- httr::GET(query_url, ua, query = query, httr::add_headers(api_key = api_key), httr::timeout(timeout))
 
   empty_res <- if (parse) list() else "{}"
 
diff --git a/man/oa_fetch.Rd b/man/oa_fetch.Rd
index c261ff3..e64599e 100644
--- a/man/oa_fetch.Rd
+++ b/man/oa_fetch.Rd
@@ -20,7 +20,8 @@ oa_fetch(
   count_only = FALSE,
   mailto = oa_email(),
   api_key = oa_apikey(),
-  verbose = FALSE
+  verbose = FALSE,
+  timeout = 30
 )
 }
 \arguments{
@@ -110,6 +111,10 @@ Your OpenAlex Premium API key, if available.}
 Default to \code{verbose = FALSE}.
 To shorten the printed query URL, set the environment variable openalexR.print
 to the number of characters to print: \code{Sys.setenv(openalexR.print = 70)}.}
+
+\item{timeout}{Numeric.
+Number of seconds to wait for a response until giving up. Can not be less than 1 ms.
+Defaults to 30.}
 }
 \value{
 A data.frame or a list. Result of the query.
diff --git a/man/oa_request.Rd b/man/oa_request.Rd
index 9de2c86..effea5d 100644
--- a/man/oa_request.Rd
+++ b/man/oa_request.Rd
@@ -13,7 +13,8 @@ oa_request(
   mailto = oa_email(),
   api_key = oa_apikey(),
   parse = TRUE,
-  verbose = FALSE
+  verbose = FALSE,
+  timeout = 30
 )
 }
 \arguments{
@@ -50,6 +51,10 @@ If FALSE, returns the raw JSON response as string.}
 
 \item{verbose}{Logical.
 If TRUE, print information about the querying process. Defaults to TRUE.}
+
+\item{timeout}{Numeric.
+Number of seconds to wait for a response until giving up. Can not be less than 1 ms.
+Defaults to 30.}
 }
 \value{
 a data.frame or a list of bibliographic records.