Add ADA parser as fallback

.Rbuildignore

@@ -20,3 +20,4 @@
 bugs
 ^\.github$
 ^configure.log$
+^src/ada

.gitignore

@@ -12,3 +12,4 @@ tools/option_table.txt
 inst/doc
 windows
 configure.log
+src/ada*

R/parser.R

@@ -8,5 +8,12 @@
 #' @useDynLib curl R_parse_url
 parse_url <- function(url){
   stopifnot(is.character(url))
-  .Call(R_parse_url, url)
+  result <- .Call(R_parse_url, url)
+  if(length(result$scheme))
+    result$scheme <- sub("\\:$", "", result$scheme)
+  if(length( result$query))
+    result$query <- sub("^\\?", "", result$query)
+  if(length(result$fragment))
+    result$fragment <- sub("^\\#", "", result$fragment)
+  result
 }

cleanup

@@ -1,2 +1,3 @@
 #!/bin/sh
 rm -f src/Makevars configure.log
+rm -f src/ada*

configure

@@ -64,8 +64,11 @@ if [ $? -ne 0 ]; then
   exit 1
 fi
 
-# On MacOS the runtime version can be different from SDK version
-if [ `uname` = "Darwin" ] && ${R_HOME}/bin/Rscript tools/testversion.R; then
+# On curl < 7.62 (RHEL-8) enable the fallback
+if [ "$TEST_ADA_PARSER" ] || ! ${R_HOME}/bin/Rscript tools/testversion.R "7.62"; then
+  tar xf tools/ada.tar.gz -C src
+  PKG_CFLAGS="$PKG_CFLAGS -DUSE_ADA_PARSER"
+elif [ `uname` = "Darwin" ] && ${R_HOME}/bin/Rscript tools/testversion.R "7.80"; then
   PKG_CFLAGS="$PKG_CFLAGS -DENABLE_ALL_FEATURES"
 fi
 

src/urlparser.c

@@ -1,6 +1,51 @@
 #include "curl-common.h"
 
-#ifdef HAS_CURL_PARSER
+static SEXP make_url_names(void){
+  SEXP names = PROTECT(Rf_allocVector(STRSXP, 9));
+  SET_STRING_ELT(names, 0, Rf_mkChar("url"));
+  SET_STRING_ELT(names, 1, Rf_mkChar("scheme"));
+  SET_STRING_ELT(names, 2, Rf_mkChar("host"));
+  SET_STRING_ELT(names, 3, Rf_mkChar("port"));
+  SET_STRING_ELT(names, 4, Rf_mkChar("path"));
+  SET_STRING_ELT(names, 5, Rf_mkChar("query"));
+  SET_STRING_ELT(names, 6, Rf_mkChar("fragment"));
+  SET_STRING_ELT(names, 7, Rf_mkChar("user"));
+  SET_STRING_ELT(names, 8, Rf_mkChar("password"));
+  UNPROTECT(1);
+  return names;
+}
+
+#ifdef USE_ADA_PARSER
+#include "ada_c.h"
+static SEXP get_ada_field(ada_string val){
+  if(val.length == 0)
+    return R_NilValue;
+  return Rf_ScalarString(Rf_mkCharLenCE(val.data, val.length, CE_UTF8));
+}
+
+SEXP R_parse_url(SEXP url) {
+  ada_url *result = ada_parse(CHAR(STRING_ELT(url, 0)), Rf_length(STRING_ELT(url, 0)));
+  if(!ada_is_valid(result)){
+    ada_free(result);
+    Rf_error("ADA failed to parse URL");
+  }
+  SEXP out = PROTECT(Rf_allocVector(VECSXP, 9));
+  SET_VECTOR_ELT(out, 0, get_ada_field(ada_get_href(result)));
+  SET_VECTOR_ELT(out, 1, get_ada_field(ada_get_protocol(result)));
+  SET_VECTOR_ELT(out, 2, get_ada_field(ada_get_hostname(result)));
+  SET_VECTOR_ELT(out, 3, get_ada_field(ada_get_port(result)));
+  SET_VECTOR_ELT(out, 4, get_ada_field(ada_get_pathname(result)));
+  SET_VECTOR_ELT(out, 5, get_ada_field(ada_get_search(result)));
+  SET_VECTOR_ELT(out, 6, get_ada_field(ada_get_hash(result)));
+  SET_VECTOR_ELT(out, 7, get_ada_field(ada_get_username(result)));
+  SET_VECTOR_ELT(out, 8, get_ada_field(ada_get_password(result)));
+
+  Rf_setAttrib(out, R_NamesSymbol, make_url_names());
+  UNPROTECT(1);
+  return out;
+}
+
+#elif defined(HAS_CURL_PARSER)
 static void fail_if(CURLUcode err){
   if(err != CURLUE_OK)
 #ifdef HAS_CURL_PARSER_STRERROR
@@ -24,26 +69,10 @@ static SEXP get_field(CURLU *h, CURLUPart part, CURLUcode field_missing){
   return field;
 }
 
-static SEXP make_url_names(void){
-  SEXP names = PROTECT(Rf_allocVector(STRSXP, 10));
-  SET_STRING_ELT(names, 0, Rf_mkChar("url"));
-  SET_STRING_ELT(names, 1, Rf_mkChar("scheme"));
-  SET_STRING_ELT(names, 2, Rf_mkChar("host"));
-  SET_STRING_ELT(names, 3, Rf_mkChar("port"));
-  SET_STRING_ELT(names, 4, Rf_mkChar("path"));
-  SET_STRING_ELT(names, 5, Rf_mkChar("query"));
-  SET_STRING_ELT(names, 6, Rf_mkChar("fragment"));
-  SET_STRING_ELT(names, 7, Rf_mkChar("user"));
-  SET_STRING_ELT(names, 8, Rf_mkChar("password"));
-  SET_STRING_ELT(names, 9, Rf_mkChar("options"));
-  UNPROTECT(1);
-  return names;
-}
-
 SEXP R_parse_url(SEXP url) {
   CURLU *h = curl_url();
   fail_if(curl_url_set(h, CURLUPART_URL, CHAR(STRING_ELT(url, 0)), 0));
-  SEXP out = PROTECT(Rf_allocVector(VECSXP, 10));
+  SEXP out = PROTECT(Rf_allocVector(VECSXP, 9));
   SET_VECTOR_ELT(out, 0, get_field(h, CURLUPART_URL, CURLUE_OK));
   SET_VECTOR_ELT(out, 1, get_field(h, CURLUPART_SCHEME, CURLUE_NO_SCHEME));
   SET_VECTOR_ELT(out, 2, get_field(h, CURLUPART_HOST, CURLUE_NO_HOST));
@@ -53,7 +82,6 @@ SEXP R_parse_url(SEXP url) {
   SET_VECTOR_ELT(out, 6, get_field(h, CURLUPART_FRAGMENT, CURLUE_NO_FRAGMENT));
   SET_VECTOR_ELT(out, 7, get_field(h, CURLUPART_USER, CURLUE_NO_USER));
   SET_VECTOR_ELT(out, 8, get_field(h, CURLUPART_PASSWORD, CURLUE_NO_PASSWORD));
-  SET_VECTOR_ELT(out, 9, get_field(h, CURLUPART_OPTIONS, CURLUE_NO_OPTIONS));
   curl_url_cleanup(h);
   Rf_setAttrib(out, R_NamesSymbol, make_url_names());
   UNPROTECT(1);

tools/testversion.R

@@ -1,3 +1,3 @@
 ver <- libcurlVersion()
 cat("Curl runtime version", ver, "\n")
-q('no', status = (numeric_version(ver) < "7.80"))
+q('no', status = (numeric_version(ver) < commandArgs(TRUE)))