Update embedded curl to 5.2.3 (#728)

* Update embedded curl to 5.2.3

* Patch curl to compile on FreeBSD

src/library/curl/DESCRIPTION

@@ -1,9 +1,9 @@
 Package: curl
 Type: Package
 Title: A Modern and Flexible Web Client for R
-Version: 5.2.0
+Version: 5.2.3
 Authors@R: c(
-    person("Jeroen", "Ooms", role = c("aut", "cre"), email = "jeroen@berkeley.edu",
+    person("Jeroen", "Ooms", role = c("aut", "cre"), email = "jeroenooms@gmail.com",
       comment = c(ORCID = "0000-0002-4035-0289")),
     person("Hadley", "Wickham", , "hadley@rstudio.com", role = "ctb"),
     person("RStudio", role = "cph")
@@ -23,17 +23,17 @@ SystemRequirements: libcurl: libcurl-devel (rpm) or
 URL: https://jeroen.r-universe.dev/curl https://curl.se/libcurl/
 BugReports: https://github.com/jeroen/curl/issues
 Suggests: spelling, testthat (>= 1.0.0), knitr, jsonlite, later,
-        rmarkdown, magrittr, httpuv (>= 1.4.4), webutils
+        rmarkdown, httpuv (>= 1.4.4), webutils
 VignetteBuilder: knitr
 Depends: R (>= 3.0.0)
-RoxygenNote: 7.2.3
+RoxygenNote: 7.3.0
 Encoding: UTF-8
 Language: en-US
 NeedsCompilation: yes
-Packaged: 2023-12-07 23:07:08 UTC; jeroen
+Packaged: 2024-09-19 15:43:51 UTC; jeroen
 Author: Jeroen Ooms [aut, cre] (<https://orcid.org/0000-0002-4035-0289>),
   Hadley Wickham [ctb],
   RStudio [cph]
-Maintainer: Jeroen Ooms <jeroen@berkeley.edu>
+Maintainer: Jeroen Ooms <jeroenooms@gmail.com>
 Repository: CRAN
-Date/Publication: 2023-12-08 07:30:02 UTC
+Date/Publication: 2024-09-20 11:50:24 UTC

src/library/curl/NAMESPACE

@@ -64,7 +64,6 @@ useDynLib(curl,R_handle_getcustom)
 useDynLib(curl,R_handle_getheaders)
 useDynLib(curl,R_handle_reset)
 useDynLib(curl,R_handle_setform)
-useDynLib(curl,R_handle_setheaders)
 useDynLib(curl,R_handle_setopt)
 useDynLib(curl,R_multi_add)
 useDynLib(curl,R_multi_cancel)

src/library/curl/NEWS

@@ -1,3 +1,21 @@
+5.2.3
+ - Remove some CMD check verbosity per new CRAN rules
+ - New maintainer email address
+
+5.2.2
+ - Disable a compression test for libcurl 8.7.1 which has a bug for deflate
+ - Change a unit test to work around firewall problems on CRANs WinBuilder
+
+5.2.1
+ - In handle_setheader(), setting a header to a non-empty whitespace string will
+   now send a blank header. Using an empty string will still remove the header
+   altogether (this has not changed).
+ - Internally simplify handle_setheader() as a wrapper for handle_setopt(httpheader)
+ - Fix an issue where curl_fetch_stream() might not return all bytes received
+   by the C layer if the server closed the connection mid-response.
+ - No longer enable CURLOPT_UNRESTRICTED_AUTH by default
+ - Fix code to build with -DR_NO_REMAP (add all the Rf_ prefixes)
+
 5.2.0
  - The CURL_CA_BUNDLE envvar is now also used on non-Windows.
  - curl_echo() now uses a random available port to run httpuv

src/library/curl/R/email.R

@@ -19,6 +19,13 @@
 #' By default, the port will be 25, unless \code{smtps://} is specified--then
 #' the default will be 465 instead.
 #'
+#' For internet SMTP servers you probably need to pass a
+#' \href{https://curl.se/libcurl/c/CURLOPT_USERNAME.html}{username} and
+#' \href{https://curl.se/libcurl/c/CURLOPT_PASSWORD.html}{passwords} option.
+#' For some servers you also need to pass a string with
+#' \href{https://curl.se/libcurl/c/CURLOPT_LOGIN_OPTIONS.html}{login_options}
+#' for example \code{login_options="AUTH=NTLM"}.
+#'
 #' @section Encrypting connections via SMTPS or STARTTLS:
 #'
 #' There are two different ways in which SMTP can be encrypted: SMTPS servers
@@ -51,8 +58,8 @@
 #' for details. Default will try to SSL, proceed as normal otherwise.
 #' @param verbose print output
 #' @param ... other options passed to \code{\link{handle_setopt}}. In most cases
-#' you will need to set a \code{username} and \code{password} to authenticate
-#' with the SMTP server.
+#' you will need to set a \code{username} and \code{password} or \code{login_options}
+#' to authenticate with the SMTP server, see details.
 #' @examples \dontrun{# Set sender and recipients (email addresses only)
 #' recipients <- readline("Enter your email address to receive test: ")
 #' sender <- 'test@noreply.com'

src/library/curl/R/handle.R

@@ -69,20 +69,21 @@ handle_setopt <- function(handle, ..., .list = list()){
 }
 
 #' @export
-#' @useDynLib curl R_handle_setheaders
 #' @rdname handle
 handle_setheaders <- function(handle, ..., .list = list()){
-  stopifnot(inherits(handle, "curl_handle"))
-  opts <- c(list(...), .list)
-  if(!all(vapply(opts, is.character, logical(1)))){
+  x <- c(list(...), .list)
+  if(!all(vapply(x, is.character, logical(1)))){
     stop("All headers must be strings.")
   }
-  opts$Expect = ""
-  names <- names(opts)
-  values <- as.character(unlist(opts))
-  vec <- paste0(names, ": ", values)
-  .Call(R_handle_setheaders, handle, vec)
-  invisible(handle)
+  handle_setopt(handle, httpheader = format_request_headers(x))
+}
+
+format_request_headers <- function(x){
+  x$Expect = ""
+  names <- names(x)
+  values <- as.character(unlist(x))
+  postfix <- ifelse(grepl("^\\s+$", values), ";", paste(":", values))
+  paste0(names, postfix)
 }
 
 #' @useDynLib curl R_handle_getheaders

src/library/curl/configure

@@ -64,21 +64,21 @@ if [ $? -ne 0 ]; then
   exit 1
 fi
 
-# Hack for MacOS 11 which has newer headers than the libcurl lib
-# if [ `uname` = "Darwin" ]; then
-#   MACVERSION=`sw_vers -productVersion` || true
-#   case "$MACVERSION" in
-#     "11"*)
-#       PKG_CFLAGS="$PKG_CFLAGS -DDISABLE_CURL_EASY_OPTION";;
-#   esac
-# fi
+# Disable purling on oldrel due to https://github.com/yihui/knitr/issues/2338
+if [ `uname` = "Darwin" ] && [ "${R_VERSION}" = "4.3.3" ]; then
+  for file in vignettes/*.Rmd; do
+    sed -i '' '/```/,$d' $file || true
+    touch inst/doc/*  || true
+  done
+fi
 
 # Write to Makevars
 sed -e "s|@cflags@|$PKG_CFLAGS|" -e "s|@libs@|$PKG_LIBS|" src/Makevars.in > src/Makevars
 
 # Extract curlopt symbols
 echo '#include <curl/curl.h>' | ${CC} ${CPPFLAGS} ${PKG_CFLAGS} ${CFLAGS} -E -xc - \
-  | grep "^[ \t]*CURLOPT_.*," | sed s/,// > tools/option_table.txt
+  | grep "^[ \t]*CURLOPT_.*," | sed s/,// | sed 's/__attribute__[(][(].*[)][)] =/=/' \
+  > tools/option_table.txt
 
 # Success
 exit 0

src/library/curl/src/Makevars.in

@@ -1,5 +1,5 @@
 PKG_CFLAGS=$(C_VISIBILITY)
-PKG_CPPFLAGS=@cflags@ -DSTRICT_R_HEADERS
+PKG_CPPFLAGS=@cflags@ -DSTRICT_R_HEADERS -DR_NO_REMAP
 PKG_LIBS=@libs@
 
 all: clean

src/library/curl/src/Makevars.win

@@ -7,7 +7,7 @@ PKG_LIBS = \
 	-lwinhttp -lcurl -lnghttp2 -lssh2 -lz -lssl -lcrypto -pthread -lgdi32 -lws2_32 -lcrypt32 -lbcrypt -lwldap32
 
 PKG_CPPFLAGS= \
-	-I$(RWINLIB)/include -DCURL_STATICLIB -DSTRICT_R_HEADERS
+	-I$(RWINLIB)/include -DCURL_STATICLIB -DSTRICT_R_HEADERS -DR_NO_REMAP
 
 all: clean winlibs
 

src/library/curl/src/callbacks.c

@@ -4,11 +4,11 @@ int R_curl_callback_progress(SEXP fun,
                              double dltotal, double dlnow,
                              double ultotal, double ulnow) {
 
-  SEXP down = PROTECT(allocVector(REALSXP, 2));
+  SEXP down = PROTECT(Rf_allocVector(REALSXP, 2));
   REAL(down)[0] = dltotal;
   REAL(down)[1] = dlnow;
 
-  SEXP up = PROTECT(allocVector(REALSXP, 2));
+  SEXP up = PROTECT(Rf_allocVector(REALSXP, 2));
   REAL(up)[0] = ultotal;
   REAL(up)[1] = ulnow;
 
@@ -21,19 +21,19 @@ int R_curl_callback_progress(SEXP fun,
     return CURL_READFUNC_ABORT;
   }
 
-  if (TYPEOF(res) != LGLSXP || length(res) != 1) {
+  if (TYPEOF(res) != LGLSXP || Rf_length(res) != 1) {
     UNPROTECT(4);
     Rf_warning("progress callback must return boolean");
     return 0;
   }
 
-  int out = asLogical(res);
+  int out = Rf_asLogical(res);
   UNPROTECT(4);
   return !out;
 }
 
 size_t R_curl_callback_read(char *buffer, size_t size, size_t nitems, SEXP fun) {
-  SEXP nbytes = PROTECT(ScalarInteger(size * nitems));
+  SEXP nbytes = PROTECT(Rf_ScalarInteger(size * nitems));
   SEXP call = PROTECT(Rf_lang2(fun, nbytes));
 
   int ok;
@@ -50,7 +50,7 @@ size_t R_curl_callback_read(char *buffer, size_t size, size_t nitems, SEXP fun)
     return CURL_READFUNC_ABORT;
   }
 
-  size_t bytes_read = length(res);
+  size_t bytes_read = Rf_length(res);
   memcpy(buffer, RAW(res), bytes_read);
 
   UNPROTECT(3);
@@ -59,7 +59,7 @@ size_t R_curl_callback_read(char *buffer, size_t size, size_t nitems, SEXP fun)
 
 /* origin is always SEEK_SET in libcurl, not really useful to pass on */
 int R_curl_callback_seek(SEXP fun, curl_off_t offset, int origin){
-  SEXP soffset = PROTECT(ScalarReal(offset));
+  SEXP soffset = PROTECT(Rf_ScalarReal(offset));
   SEXP call = PROTECT(Rf_lang2(fun, soffset));
   int ok;
   R_tryEval(call, R_GlobalEnv, &ok);
@@ -71,8 +71,8 @@ int R_curl_callback_debug(CURL *handle, curl_infotype type_, char *data,
                           size_t size, SEXP fun) {
 
   /* wrap type and msg into R types */
-  SEXP type = PROTECT(ScalarInteger(type_));
-  SEXP msg = PROTECT(allocVector(RAWSXP, size));
+  SEXP type = PROTECT(Rf_ScalarInteger(type_));
+  SEXP msg = PROTECT(Rf_allocVector(RAWSXP, size));
   memcpy(RAW(msg), data, size);
 
   /* call the R function */

src/library/curl/src/curl.c

@@ -77,7 +77,7 @@ static size_t push(void *contents, size_t sz, size_t nmemb, void *ctx) {
     //Rprintf("Resizing buffer to %d.\n", newlimit);
     void *newbuf = realloc(req->buf, newlimit);
     if(!newbuf)
-      error("Failure in realloc. Out of memory?");
+      Rf_error("Failure in realloc. Out of memory?");
     req->buf = newbuf;
     req->limit = newlimit;
   }
@@ -130,6 +130,16 @@ static size_t rcurl_read(void *target, size_t sz, size_t ni, Rconnection con) {
 
   /* append data to the target buffer */
   size_t total_size = pop(target, req_size, req);
+
+  if (total_size > 0 && (!con->blocking || req->partial)) {
+      // If we can return data without waiting, and the connection is
+      // non-blocking (or using curl_fetch_stream()), do so.
+      // This ensures that bytes we already received get flushed
+      // to the target buffer before a connection error.
+      con->incomplete = req->has_more || req->size;
+      return total_size;
+  }
+
   while((req_size > total_size) && req->has_more) {
     /* wait for activity, timeout or "nothing" */
 #ifdef HAS_MULTI_WAIT
@@ -246,14 +256,14 @@ static Rboolean rcurl_open(Rconnection con) {
 }
 
 SEXP R_curl_connection(SEXP url, SEXP ptr, SEXP partial) {
-  if(!isString(url))
-    error("Argument 'url' must be string.");
+  if(!Rf_isString(url))
+    Rf_error("Argument 'url' must be string.");
 
   /* create the R connection object, mimicking base::url() */
   Rconnection con;
 
   /* R wants description in native encoding, but we use UTF-8 URL below */
-  SEXP rc = PROTECT(R_new_custom_connection(translateChar(STRING_ELT(url, 0)), "r", "curl", &con));
+  SEXP rc = PROTECT(R_new_custom_connection(Rf_translateChar(STRING_ELT(url, 0)), "r", "curl", &con));
 
   /* setup curl. These are the parts that are recycable. */
   request *req = malloc(sizeof(request));
@@ -262,12 +272,12 @@ SEXP R_curl_connection(SEXP url, SEXP ptr, SEXP partial) {
   req->limit = CURL_MAX_WRITE_SIZE;
   req->buf = malloc(req->limit);
   req->manager = curl_multi_init();
-  req->partial = asLogical(partial); //only for curl_fetch_stream()
+  req->partial = Rf_asLogical(partial); //only for curl_fetch_stream()
   req->used = 0;
 
   /* allocate url string */
-  req->url = malloc(strlen(translateCharUTF8(asChar(url))) + 1);
-  strcpy(req->url, translateCharUTF8(asChar(url)));
+  req->url = malloc(strlen(Rf_translateCharUTF8(Rf_asChar(url))) + 1);
+  strcpy(req->url, Rf_translateCharUTF8(Rf_asChar(url)));
 
   /* set connection properties */
   con->incomplete = FALSE;

src/library/curl/src/download.c

@@ -4,35 +4,35 @@
 #include "curl-common.h"
 
 SEXP R_download_curl(SEXP url, SEXP destfile, SEXP quiet, SEXP mode, SEXP ptr, SEXP nonblocking) {
-  if(!isString(url))
-    error("Argument 'url' must be string.");
+  if(!Rf_isString(url))
+    Rf_error("Argument 'url' must be string.");
 
-  if(!isString(destfile))
-    error("Argument 'destfile' must be string.");
+  if(!Rf_isString(destfile))
+    Rf_error("Argument 'destfile' must be string.");
 
-  if(!isLogical(quiet))
-    error("Argument 'quiet' must be TRUE/FALSE.");
+  if(!Rf_isLogical(quiet))
+    Rf_error("Argument 'quiet' must be TRUE/FALSE.");
 
-  if(!isString(mode))
-    error("Argument 'mode' must be string.");
+  if(!Rf_isString(mode))
+    Rf_error("Argument 'mode' must be string.");
 
   /* get the handle */
   CURL *handle = get_handle(ptr);
   reset_errbuf(get_ref(ptr));
 
   /* open file */
-  FILE *dest = fopen(CHAR(asChar(destfile)), CHAR(asChar(mode)));
+  FILE *dest = fopen(CHAR(Rf_asChar(destfile)), CHAR(Rf_asChar(mode)));
   if(!dest)
-    error("Failed to open file %s.", CHAR(asChar(destfile)));
+    Rf_error("Failed to open file %s.", CHAR(Rf_asChar(destfile)));
 
   /* set options */
-  curl_easy_setopt(handle, CURLOPT_URL, translateCharUTF8(asChar(url)));
-  curl_easy_setopt(handle, CURLOPT_NOPROGRESS, asLogical(quiet));
+  curl_easy_setopt(handle, CURLOPT_URL, Rf_translateCharUTF8(Rf_asChar(url)));
+  curl_easy_setopt(handle, CURLOPT_NOPROGRESS, Rf_asLogical(quiet));
   curl_easy_setopt(handle, CURLOPT_WRITEFUNCTION, push_disk);
   curl_easy_setopt(handle, CURLOPT_WRITEDATA, dest);
 
   /* perform blocking request */
-  CURLcode status = asLogical(nonblocking) ?
+  CURLcode status = Rf_asLogical(nonblocking) ?
     curl_perform_with_interrupt(handle) : curl_easy_perform(handle);
 
   /* cleanup */
@@ -47,5 +47,5 @@ SEXP R_download_curl(SEXP url, SEXP destfile, SEXP quiet, SEXP mode, SEXP ptr, S
 
   /* check for success */
   stop_for_status(handle);
-  return ScalarInteger(0);
+  return Rf_ScalarInteger(0);
 }

src/library/curl/src/escape.c

@@ -3,17 +3,17 @@
 
 SEXP R_curl_escape(SEXP url, SEXP unescape_) {
   if (!Rf_isString(url))
-    error("`url` must be a character vector.");
+    Rf_error("`url` must be a character vector.");
   CURL *handle = curl_easy_init();
   int n = Rf_length(url);
-  SEXP output = PROTECT(allocVector(STRSXP, n));
+  SEXP output = PROTECT(Rf_allocVector(STRSXP, n));
 
   for (int i = 0; i < n; i++) {
     const char *input = CHAR(STRING_ELT(url, i));
-    char *out = asLogical(unescape_) ?
+    char *out = Rf_asLogical(unescape_) ?
       curl_easy_unescape(handle, input, 0, NULL) : curl_easy_escape(handle, input, 0);
     if(out != NULL){
-      SET_STRING_ELT(output, i, mkCharCE(out, CE_UTF8));
+      SET_STRING_ELT(output, i, Rf_mkCharCE(out, CE_UTF8));
       curl_free(out);
     }
   }