import jsonlite, fixes to oa_combine to add dataset column for each u…

…nique

dataset combined
fixed some tests for new changes

DESCRIPTION

@@ -12,6 +12,7 @@ BugReports: https://github.com/sckott/openadds/issues
 LazyData: true
 Imports:
     httr (>= 1.2.1),
+    jsonlite (>= 1.0),
     readr (>= 1.0.0),
     dplyr (>= 0.5.0),
     tibble (>= 1.1),

R/combine.R

@@ -41,6 +41,10 @@ oa_combine <- function(...) {
   paths <- vapply(cb, attr, "", which = "path")
   readmes <- vapply(cb, attr, "", which = "readme")
 
+  dnames <- unlist(lapply(cb, function(z) {
+    rep(attr(z, "name"), length(z))
+  }))
+
   # unnest list of data.frames
   cbdat <- unlist(cb, recursive = FALSE)
 
@@ -61,6 +65,7 @@ oa_combine <- function(...) {
         }
         out[[i]] <- stats::setNames(tmp, nms)
       }
+      out[[i]]$dataset <- dnames[i]
     }
   }
   structure(dplyr::bind_rows(out),

R/oa_get.R

@@ -58,13 +58,9 @@ oa_get.character <- function(x, overwrite = FALSE, ...) {
   resp <- oa_GET(x, ...)
   structure(resp, class = "oa",
             id = x,
-            path = make_path(basename(x)),
+            path = make_path(x),
             readme = read_me(x),
             name = get_name(x))
-  # structure(resp, class = c("tbl_df", "data.frame", "oa"),
-  #           id = x,
-  #           path = make_path(basename(x)),
-  #           readme = read_me(x))
 }
 
 oa_GET <- function(url, ...){
@@ -81,16 +77,15 @@ oa_GET <- function(url, ...){
     ff <- res$request$output$path
   }
   switch(strextract(basename(ff), "\\zip|csv|geojson"),
-         csv = read_csv_(ff),
+         csv = list(read_csv_(ff)),
          zip = read_zip_(ff),
-         geojson = read_geojson_(ff)
+         geojson = list(read_geojson_(ff))
   )
 }
 
 get_name <- function(x) gsub("\\..+", "", basename(x))
 
 make_path <- function(x) {
-  # file.path(oa_cache_path(), x)
   xx <- grep("[A-Za-z]", strsplit(x, "/")[[1]], value = TRUE)
   xx <- xx[!grepl("http|openaddresses|runs", xx)]
   file.path(oa_cache_path(), paste0(xx, collapse = "_"))
@@ -108,7 +103,6 @@ read_zip_ <- function(fname) {
     file_type(exdir),
     csv = {
       files <- list.files(exdir, pattern = ".csv", full.names = TRUE, recursive = TRUE)
-      #if (length(files) > 1) stop('More than 1 csv file found', call. = FALSE)
       lapply(files, read_csv_)
     },
     shp = {
@@ -117,7 +111,6 @@ read_zip_ <- function(fname) {
     },
     geojson = {
       files <- list.files(exdir, pattern = ".geojson", full.names = TRUE, recursive = TRUE)
-      #if (length(files) > 1) stop('More than 1 csv file found', call. = FALSE)
       lapply(files, read_geojson_)
     }
   )
@@ -128,12 +121,6 @@ read_geojson_ <- function(x) {
 }
 
 read_shp_ <- function(x) {
-  # shpfile <- list.files(dir, pattern = ".shp", full.names = TRUE, recursive = TRUE)
-  # if (length(shpfile) != 1) {
-  #   shpfile2 <- grep("\\.shp$", shpfile, value = TRUE)
-  #   shpfile <- shpfile2[1]
-  #   message("Many shp files, using \n", shpfile2[1], "\n others available:\n", shpfile2[-1])
-  # }
   tmp <- maptools::readShapeSpatial(x)
   tibble::as_data_frame(tmp@data)
 }

tests/testthat/test-oa_combine.R

@@ -20,7 +20,7 @@ test_that("oa_combine works", {
 
   expect_gt(NROW(aa), NROW(out1))
   expect_gt(NROW(aa), NROW(out2))
-  expect_gt(NROW(out1), NROW(out2))
+  expect_gt(NROW(out1[[1]]), NROW(out2[[1]]))
 })
 
 test_that("oa_combine fails well", {

tests/testthat/test-oa_get.R

@@ -16,28 +16,28 @@ test_that("oa_get works", {
   # works for zip files
   aa <- oa_get(urls[3])
 
-  expect_is(aa, "data.frame")
   expect_is(aa, "oa")
+  expect_is(aa[[1]], "data.frame")
   expect_equal(attr(aa, "id"), urls[3])
 
   # works for csv files
   bb <- suppressWarnings(oa_get(urls[5]))
 
-  expect_is(bb, "data.frame")
   expect_is(bb, "oa")
+  expect_is(bb[[1]], "data.frame")
   expect_equal(attr(bb, "id"), urls[5])
 
   # works for geojson files
   cc <- oa_get(urls[2])
 
-  expect_is(cc, "data.frame")
   expect_is(cc, "oa")
+  expect_is(cc[[1]], "data.frame")
   expect_equal(attr(cc, "id"), urls[2])
   # geojson data exists
-  expect_true("geometry.type" %in% names(cc))
-  expect_true("geometry.coordinates" %in% names(cc))
-  expect_is(cc$geometry.coordinates[[1]], "array")
-  expect_type(cc$geometry.coordinates[[1]][,,1], "double")
+  expect_true("geometry.type" %in% names(cc[[1]]))
+  expect_true("geometry.coordinates" %in% names(cc[[1]]))
+  expect_is(cc[[1]]$geometry.coordinates[[1]], "array")
+  expect_type(cc[[1]]$geometry.coordinates[[1]][,,1], "double")
 })
 
 test_that("oa_get fails well", {