From 14b7ebeff576bf3f3cbf7b115900a6e825f76c15 Mon Sep 17 00:00:00 2001
From: Carl Boettiger <cboettig@gmail.com>
Date: Thu, 15 Feb 2018 21:13:05 +0000
Subject: [PATCH] Use built-in base uri mechanism from jsonld

Much much faster (and simpler) than manual expansion with `@base`.  Follow-up on #5
---
 R/rdf.R                               | 12 +++++++
 R/rdf_parse.R                         | 25 +++-----------
 inst/examples/profile_performance.R   | 39 ++++++++++++++++++++++
 tests/testthat/test-parse-serialize.R | 47 +++++++++++++++++++++++++++
 4 files changed, 103 insertions(+), 20 deletions(-)
 create mode 100644 inst/examples/profile_performance.R

diff --git a/R/rdf.R b/R/rdf.R
index d67df76..309bb04 100644
--- a/R/rdf.R
+++ b/R/rdf.R
@@ -25,6 +25,18 @@
 #' Typical use will be simply to initialize a container to which
 #' the user would manually add triples using \code{\link{rdf_add}}.
 #'
+#' Overview of configuration options
+#' rdflib_storage:
+#'   - NULL or "memory" for in memory storage. (default)
+#'   - "BDB" for disk-based storage in Berkeley Database
+#' rdflib_print_format: 
+#'   - NULL or "nquads" (default)
+#'   - any valid serializer name: e.g. "rdfxml", "jsonld", "turtle",  "ntriples"
+#' rdflib_base_uri:
+#'   - Default base URI to use (when serializing JSON-LD only at this time)
+#'     default is "localhost://"
+#'
+#'
 #'
 #' @importClassesFrom redland World Model Storage
 #' @importMethodsFrom redland freeWorld freeModel freeStorage
diff --git a/R/rdf_parse.R b/R/rdf_parse.R
index 30d6904..7ab2131 100644
--- a/R/rdf_parse.R
+++ b/R/rdf_parse.R
@@ -37,8 +37,11 @@ rdf_parse <- function(doc,
   ## be a read-only task!
   if(format == "jsonld"){
     tmp <- tempfile()
-    tmp <- add_base_uri(doc, tmp)
-    x <- jsonld::jsonld_to_rdf(tmp)
+    #tmp <- add_base_uri(doc, tmp)
+    x <- jsonld::jsonld_to_rdf(doc, 
+                               options = 
+           list(base = getOption("rdflib_base_uri", "localhost://"),
+                format = "application/nquads"))
     writeLines(x, tmp)
     format <- "nquads"
     doc <- tmp
@@ -56,24 +59,6 @@ rdf_parse <- function(doc,
   rdf
 }
 
-# Whenever we convert JSON-LD to RDF we should set a @base if not set.
-# https://json-ld.org/playground does this (with it's own url) 
-# but jsonld R package does not.
-# For details, see https://github.com/cboettig/rdflib/issues/5
-#
-#' @importFrom jsonld jsonld_expand jsonld_compact
-add_base_uri <- function(doc, tmp = tempfile()){
-  
-  ## Cannot assume it has context, may already be expanded 
-  ## (e.g. from rdf_serialize).  Expanding will also make 
-  ## any preset @base context take precedence 
-  expanded <- jsonld::jsonld_expand(doc)
-  base <- getOption("rdflib_base_uri", "localhost://")
-  context <- paste0('{"@base": "', base, '"}')
-  compact <- jsonld::jsonld_compact(expanded, context)
-  writeLines(compact, tmp)
-  tmp
-}
 
 
 # rdf functions like working with local files
diff --git a/inst/examples/profile_performance.R b/inst/examples/profile_performance.R
new file mode 100644
index 0000000..ed9207d
--- /dev/null
+++ b/inst/examples/profile_performance.R
@@ -0,0 +1,39 @@
+
+library(nycflights13)
+library(tidyverse)
+library(rdflib)
+source(system.file("examples/as_rdf.R", package="rdflib"))
+
+
+## Tidyverse Style
+df <- flights %>% 
+  left_join(airlines) %>%
+  left_join(planes, by="tailnum") %>% 
+  select(carrier, manufacturer, model)
+
+## JSON-LD approach -- 5 sec, 
+## ~ all rdf_parse
+## ~ all of which is jsonld_expand()
+profvis::profvis(
+  x2 <- as_rdf.list(airports)
+)
+
+
+## rdf_add approach 12 sec
+## ~ all of which is rdf_add
+## ~ all of which is calls to new / initialize S4 method
+#3 ~ none of which is calls to the low level librdf_* C calls
+profvis::profvis(
+x2 <- as_rdf(airports,  "faa", "x:")
+)
+
+
+
+x1 <- as_rdf(airlines, "carrier", "x:")
+x3 <- as_rdf(planes, "tailnum", "x:")
+
+system.time(
+  x4 <- as_rdf(flights, NULL, "x:")
+)
+
+rdf <- c(x1,x2,x3)
diff --git a/tests/testthat/test-parse-serialize.R b/tests/testthat/test-parse-serialize.R
index 178e2a5..aa668b3 100644
--- a/tests/testthat/test-parse-serialize.R
+++ b/tests/testthat/test-parse-serialize.R
@@ -82,9 +82,56 @@ testthat::test_that("we can serialize turtle with a baseUri", {
   rdf_free(rdf)
 })
 
+
+## JSON-LD tests with default base uri
+
+  testthat::test_that("@id is not a URI, we should get localhost", {
+    ex <- '{
+      "@context": "http://schema.org/",
+      "@id": "person_id",
+      "name": "Jane Doe"
+    }'
+    
+    rdf <- rdf_parse(ex, "jsonld")
+    testthat::expect_output(cat(format(rdf, "nquads")), "localhost")
+    rdf_free(rdf)
+  })
+
+  testthat::test_that("@id is a URI, we should not get localhost", {
+    ex <- '{
+      "@context": "http://schema.org/",
+      "@id": "uri:person_id",
+      "name": "Jane Doe"
+    }'
+    rdf <- rdf_parse(ex, "jsonld")
+    testthat::expect_false(grepl("localhost", format(rdf, "nquads")))
+    rdf_free(rdf)
+  })  
+  
+  testthat::test_that("we can alter the base URI", {
+    ex <- '{
+      "@id": "person_id",
+      "schema:name": "Jane Doe"
+    }'
+    options(rdflib_base_uri = "http://example.com/")
+    rdf <- rdf_parse(ex, "jsonld")
+    testthat::expect_output(cat(format(rdf, "nquads")), "http://example.com")
+    rdf_free(rdf)
+    
+    
+    options(rdflib_base_uri = "")
+    rdf <- rdf_parse(ex, "jsonld")
+    testthat::expect_silent(cat(format(rdf, "nquads")))
+    rdf_free(rdf)
+    
+    options(rdflib_base_uri = NULL)
+  })
+  
   
   
   
+
+  
   
 testthat::test_that("we can parse into an existing rdf model", {
     rdf1 <- rdf_parse(system.file("extdata/ex.xml", package = "rdflib"))