Add standalone bson reader

NAMESPACE

@@ -15,11 +15,14 @@ export(gridfs)
 export(mongo)
 export(mongo_options)
 export(oid_to_timestamp)
+export(read_bson)
 export(ssl_options)
 importFrom(jsonlite,fromJSON)
 importFrom(jsonlite,toJSON)
 importFrom(jsonlite,validate)
 useDynLib(mongolite,R_bigint_as_char)
+useDynLib(mongolite,R_bson_reader_new)
+useDynLib(mongolite,R_bson_reader_read)
 useDynLib(mongolite,R_bson_to_json)
 useDynLib(mongolite,R_bson_to_list)
 useDynLib(mongolite,R_bson_to_raw)

R/reader.R

@@ -0,0 +1,41 @@
+#' Standalone BSON reader
+#'
+#' Utility to parse BSON files into R without using MongoDB. This is useful
+#' to read data from a `mongoexport` dump without needing local mongodb server.
+#' This reads all data in memory. To import a bson dump into your local mongodb
+#' server, use the [mongo$import][mongo] function instead.
+#'
+#' @export
+#' @useDynLib mongolite R_bson_reader_new R_bson_reader_read
+#' @param con either a path to a file, a url, or a a connection object
+#' @param as_json return data as json strings instead of R lists
+#' @param verbose print some output as we read
+#' @examples
+#' diamonds <- read_bson("http://jeroen.github.io/data/diamonds.bson")
+read_bson <- function(con, as_json = FALSE, verbose = TRUE){
+  if(length(con) && is.character(con)){
+    con <- if(grepl("^https?://", con)){
+      url(con)
+    } else {
+      file(normalizePath(con, mustWork = TRUE), raw = TRUE)
+    }
+  }
+  stopifnot(inherits(con, 'connection'))
+  open(con, 'rb')
+  on.exit(close(con))
+  reader <- .Call(R_bson_reader_new, con)
+  output <- new.env(parent = emptyenv())
+  i <- 0
+  one <- function(as_json = FALSE){
+    .Call(R_bson_reader_read, reader, as_json)
+  }
+  while(length(obj <- one(as_json))){
+    i <- i+1
+    if(isTRUE(verbose))
+      cat("\rRead", i, file = stderr())
+    output[[sprintf('%09d', i)]] <- obj
+  }
+  if(isTRUE(verbose))
+    cat("\rDone!\n", file = stderr())
+  unname(as.list(output, sorted = TRUE))
+}

src/mongolite.h

@@ -22,5 +22,5 @@ SEXP client2r(mongoc_client_t *client);
 SEXP gridfs2r(mongoc_gridfs_t *fs, SEXP prot);
 void mongolite_log_handler (mongoc_log_level_t log_level, const char *log_domain, const char *message, void *user_data);
 SEXP ConvertObject(bson_iter_t* iter, bson_iter_t* counter);
-SEXP bson2list(bson_t *b);
+SEXP bson2list(const bson_t *b);
 SEXP bson_to_str(const bson_t * b);

src/reader.c

@@ -71,3 +71,32 @@ SEXP R_mongo_restore(SEXP con, SEXP ptr_col, SEXP verb) {
   mongoc_bulk_operation_destroy (bulk);
   return Rf_ScalarInteger(count);
 }
+
+static void fin_bson_reader(SEXP ptr){
+  if(!R_ExternalPtrAddr(ptr)) return;
+  bson_reader_destroy(R_ExternalPtrAddr(ptr));
+  R_SetExternalPtrProtected(ptr, R_NilValue);
+  R_ClearExternalPtr(ptr);
+}
+
+SEXP R_bson_reader_new(SEXP con) {
+  bson_reader_t *reader = bson_reader_new_from_handle(con, bson_reader_feed, bson_reader_finalize);
+  SEXP ptr = PROTECT(R_MakeExternalPtr(reader, R_NilValue, con));
+  R_RegisterCFinalizerEx(ptr, fin_bson_reader, 1);
+  Rf_setAttrib(ptr, R_ClassSymbol, Rf_mkString("bson_reader"));
+  UNPROTECT(1);
+  return ptr;
+}
+
+SEXP R_bson_reader_read(SEXP ptr, SEXP as_json){
+  bson_reader_t *reader = R_ExternalPtrAddr(ptr);
+  if(!reader)
+    Rf_error("This reader has been destroyed.");
+  bool reached_eof = 0;
+  const bson_t *doc = bson_reader_read (reader, &reached_eof);
+  if(reached_eof)
+    return R_NilValue;
+  if(doc == NULL)
+    Rf_error("Failed to read all documents");
+  return Rf_asLogical(as_json) ? bson_to_str(doc) : bson2list(doc);
+}

src/utils.c

@@ -36,7 +36,7 @@ SEXP mkRaw(const unsigned char *buf, int len){
   return out;
 }
 
-SEXP bson2list(bson_t *b){
+SEXP bson2list(const bson_t *b){
   bson_iter_t iter1;
   bson_iter_t iter2;
   bson_iter_init(&iter1, b);