PR for feature #58 (access private datasets and be able to upsert and full replace datasets)

NAMESPACE

@@ -6,6 +6,7 @@ export(ls.socrata)
 export(read.socrata)
 export(read.socrataGEO)
 export(validateUrl)
+export(write.socrata)
 importFrom(geojsonio,geojson_read)
 importFrom(httr,GET)
 importFrom(httr,add_headers)

R/errorHandling.R

@@ -6,11 +6,18 @@
 # but one that is not compatible with RSocrata.
 # See \url{https://github.com/Chicago/RSocrata/issues/16}
 #
-# @param url - SOPA url
+# @param url - SODA url
+# @param optional email - The email to the Socrata account with read access to the dataset
+# @param optional password - The password associated with the email to the Socrata account
 #' @importFrom httr stop_for_status GET add_headers
-errorHandling <- function(url = "", app_token = NULL) {
-  rsp <- httr::GET(url, httr::add_headers("X-App-Token" = app_token))
-
+errorHandling <- function(url = "", app_token = NULL, email = NULL, password = NULL) {
+
+  if(is.null(email) && is.null(password)){
+    rsp <- httr::GET(url, httr::add_headers("X-App-Token" = app_token))
+  } else { # email and password are not NULL
+    rsp <- httr::GET(url, httr::add_headers("X-App-Token" = app_token), httr::authenticate(email, password))
+  } 
+
   if (rsp$status_code == 200) {
     invisible("OK. Your request was successful.")
 

R/metadata.R

@@ -5,6 +5,8 @@
 #' \code{http://DOMAIN/api/views/FOUR-FOUR/columns.json}, which is used here. 
 #' 
 #' @param url - A Socrata resource URL, or a Socrata "human-friendly" URL!
+#' @param optional email - The email to the Socrata account with read access to the dataset
+#' @param optional password - The password associated with the email to the Socrata account
 #' 
 #' @source \url{http://stackoverflow.com/a/29782941}
 #'
@@ -24,13 +26,13 @@
 #' @author John Malc \email{cincenko@@outlook.com} 
 #'  
 #' @export
-getMetadata <- function(url = "") {
+getMetadata <- function(url = "", email = NULL, password = NULL) {
 
   urlParsedBase <- httr::parse_url(url)
   mimeType <- mime::guess_type(urlParsedBase$path)
 
   # use function below to get them using =COUNT(*) SODA query
-  gQRC <- getQueryRowCount(urlParsedBase, mimeType) 
+  gQRC <- getQueryRowCount(urlParsedBase, mimeType, email, password) 
 
   # create URL for metadata data frame
   fourByFour <- substr(basename(urlParsedBase$path), 1, 9)
@@ -39,8 +41,15 @@ getMetadata <- function(url = "") {
 
   # execute it
   URL <- httr::build_url(urlParsed)
-  df <- jsonlite::fromJSON(URL)
 
+  if(is.null(email) && is.null(password)){
+    df <- jsonlite::fromJSON(URL)
+  } else { # email and password are not NULL
+    response <- httr::GET(URL, httr::authenticate(email, password))
+    response_content <- httr::content(response, as="text")
+    df <- jsonlite::fromJSON(response_content)
+  } 
+
   # number of rows can be sometimes "cached". If yes, then below we calculate the maximum number of 
   # rows from all non-null and null fields. 
   # If not, then it uses "getQueryRowCount" fnct with SODA =COUNT(*) SODA query.
@@ -62,7 +71,7 @@ getMetadata <- function(url = "") {
 # @author Gene Leynes \email{gleynes@@gmail.com}
 #
 #' @importFrom httr GET build_url content
-getQueryRowCount <- function(urlParsed, mimeType) {
+getQueryRowCount <- function(urlParsed, mimeType, email = NULL, password = NULL) {
   ## Construct the count query based on the URL,
   if (is.null(urlParsed[['query']])) {
     ## If there is no query at all, create a simple count
@@ -81,7 +90,7 @@ getQueryRowCount <- function(urlParsed, mimeType) {
                    urlParsed[[c('path')]], cntQueryText)
 
   ## Execute the query to count the rows
-  totalRowsResult <- errorHandling(cntUrl, app_token = NULL)
+  totalRowsResult <- errorHandling(cntUrl, app_token = NULL, email, password)
 
   ## Parsing the result depends on the mime type
   if (mimeType == "application/json") {

R/returnData.R

@@ -82,6 +82,8 @@ getContentAsDataFrame <- function(response) {
 #' @param limit - defaults to the max of 50000. See \url{http://dev.socrata.com/docs/paging.html}.
 #' @param domain - A Socrata domain, e.g \url{http://data.cityofchicago.org} 
 #' @param fourByFour - a unique 4x4 identifier, e.g. "ydr8-5enu". See more \code{\link{isFourByFour}}
+#' @param optional email - The email to the Socrata account with read access to the dataset
+#' @param optional password - The password associated with the email to the Socrata account
 #' 
 #' @author Hugh J. Devlin, Ph. D. \email{Hugh.Devlin@@cityofchicago.org}
 #' 
@@ -96,7 +98,7 @@ getContentAsDataFrame <- function(response) {
 #' 
 #' @export
 read.socrata <- function(url = NULL, app_token = NULL, limit = 50000, domain = NULL, fourByFour = NULL, 
-                         query = NULL) {
+                         query = NULL, email = NULL, password = NULL) {
 
   if (is.null(url) == TRUE) {
     buildUrl <- paste0(domain, "/resource/", fourByFour, ".json")
@@ -107,16 +109,16 @@ read.socrata <- function(url = NULL, app_token = NULL, limit = 50000, domain = N
   validUrl <- validateUrl(url) 
   parsedUrl <- httr::parse_url(validUrl)
 
-  response <- errorHandling(validUrl, app_token)
+  response <- errorHandling(validUrl, app_token, email, password)
   results <- getContentAsDataFrame(response)
   dataTypes <- getSodaTypes(response)
 
-  rowCount <- as.numeric(getMetadata(cleanQuest(validUrl))[1])
+  rowCount <- as.numeric(getMetadata(cleanQuest(validUrl), email, password)[1])
 
   ## More to come? Loop over pages implicitly
   while (nrow(results) < rowCount) { 
     query_url <- paste0(validUrl, ifelse(is.null(parsedUrl$query), "?", "&"), "$offset=", nrow(results), "&$limit=", limit)
-    response <- errorHandling(query_url, app_token)
+    response <- errorHandling(query_url, app_token, email, password)
     page <- getContentAsDataFrame(response)
     results <- plyr::rbind.fill(results, page) # accumulate data
   }	

R/writeData.R

@@ -0,0 +1,74 @@
+#' Methods required for uploading data to Socrata
+
+#' Wrap httr GET in some diagnostics
+#' 
+#' In case of failure, report error details from Socrata.
+#' 
+#' @param url - Socrata Open Data Application Program Interface (SODA) endpoint (JSON only for now)
+#' @param json_data_to_upload - JSON encoded data to update your SODA endpoint with
+#' @param http_verb - PUT or POST depending on update mode
+#' @param email - email associated with Socrata account (will need write access to dataset)
+#' @param password - password associated with Socrata account (will need write access to dataset)
+#' @param app_token - optional app_token associated with Socrata account
+#' @return httr a response object
+#' @importFrom httr GET
+#' 
+#' @noRd
+checkUpdateResponse <- function(json_data_to_upload, url, http_verb, email, password, app_token = NULL) {
+  if(http_verb == "POST"){
+    response <- httr::POST(url,
+      body = json_data_to_upload,
+      httr::authenticate(email, password),
+      httr::add_headers("X-App-Token" = app_token,
+                  "Content-Type" = "application/json")) #, verbose())
+  } else if(http_verb == "PUT"){
+    response <- httr::PUT(url,
+      body = json_data_to_upload,
+      httr::authenticate(email, password),
+      httr::add_headers("X-App-Token" = app_token,
+                  "Content-Type" = "application/json")) # , verbose())
+  }
+
+  # TODO: error handling
+  # errorHandling(response)
+
+  return(response)
+}
+
+
+#' @description Method for updating Socrata datasets 
+#'
+#' @param dataframe - dataframe to upload to Socrata
+#' @param dataset_json_endpoint - Socrata Open Data Application Program Interface (SODA) endpoint (JSON only for now)
+#' @param update_mode - "UPSERT" or "REPLACE"; consult http://dev.socrata.com/publishers/getting-started.html 
+#' @param email - The email to the Socrata account with read access to the dataset
+#' @param password - The password associated with the email to the Socrata account
+#' @param app_token - a (non-required) string; SODA API token can be used to query the data 
+#' portal \url{http://dev.socrata.com/consumers/getting-started.html}
+#' 
+#' @author Mark Silverberg \email{mark.silverberg@@socrata.com}
+#' 
+#' @importFrom httr parse_url build_url
+#' @importFrom plyr rbind.fill
+#' 
+#' @export
+write.socrata <- function(dataframe, dataset_json_endpoint, update_mode, email, password, app_token = NULL) {
+
+  # translate update_mode to http_verbs
+  if(update_mode == "UPSERT"){
+    http_verb <- "POST"
+  } else if(update_mode == "REPLACE") {
+    http_verb <- "PUT"
+  } else {
+    stop("update_mode must be UPSERT or REPLACE")
+  }
+
+  # convert dataframe to JSON
+  dataframe_as_json_string <- jsonlite::toJSON(dataframe)
+
+  # do the actual upload
+  response <- checkUpdateResponse(dataframe_as_json_string, dataset_json_endpoint, http_verb, email, password, app_token)
+
+  return(response)
+
+}

man/getMetadata.Rd

@@ -7,10 +7,14 @@
 \url{http://stackoverflow.com/a/29782941}
 }
 \usage{
-getMetadata(url = "")
+getMetadata(url = "", email = NULL, password = NULL)
 }
 \arguments{
 \item{url}{- A Socrata resource URL, or a Socrata "human-friendly" URL!}
+
+\item{optional}{email - The email to the Socrata account with read access to the dataset}
+
+\item{optional}{password - The password associated with the email to the Socrata account}
 }
 \value{
 a list (!) containing a number of rows & columns and a data frame of metadata

man/read.socrata.Rd

@@ -5,7 +5,7 @@
 \title{Get a full Socrata data set as an R data frame}
 \usage{
 read.socrata(url = NULL, app_token = NULL, limit = 50000, domain = NULL,
-  fourByFour = NULL, query = NULL)
+  fourByFour = NULL, query = NULL, email = NULL, password = NULL)
 }
 \arguments{
 \item{url}{- A Socrata resource URL, or a Socrata "human-friendly" URL,
@@ -26,15 +26,21 @@ portal \url{http://dev.socrata.com/consumers/getting-started.html}}
 
 \item{query}{- Based on query language called the "Socrata Query Language" ("SoQL"), see
 \url{http://dev.socrata.com/docs/queries.html}.}
+
+\item{optional}{email - The email to the Socrata account with read access to the dataset}
+
+\item{optional}{password - The password associated with the email to the Socrata account}
 }
 \description{
 Manages throttling and POSIX date-time conversions. We support only .json suffix.
 }
 \examples{
+\dontrun{
 df_1 <- read.socrata(url = "http://soda.demo.socrata.com/resource/4334-bgaj.csv")
 df_2 <- read.socrata(domain = "http://data.cityofchicago.org/", fourByFour = "ydr8-5enu")
 df_3 <- read.socrata(url = "http://data.cityofchicago.org/resource/ydr8-5enu.json")
 }
+}
 \author{
 Hugh J. Devlin, Ph. D. \email{Hugh.Devlin@cityofchicago.org}
 }

man/write.socrata.Rd

@@ -0,0 +1,29 @@
+% Generated by roxygen2 (4.1.1): do not edit by hand
+% Please edit documentation in R/writeData.R
+\name{write.socrata}
+\alias{write.socrata}
+\usage{
+write.socrata(dataframe, dataset_json_endpoint, update_mode, email, password,
+  app_token = NULL)
+}
+\arguments{
+\item{dataframe}{- dataframe to upload to Socrata}
+
+\item{dataset_json_endpoint}{- Socrata Open Data Application Program Interface (SODA) endpoint (JSON only for now)}
+
+\item{update_mode}{- "UPSERT" or "REPLACE"; consult http://dev.socrata.com/publishers/getting-started.html}
+
+\item{email}{- The email to the Socrata account with read access to the dataset}
+
+\item{password}{- The password associated with the email to the Socrata account}
+
+\item{app_token}{- a (non-required) string; SODA API token can be used to query the data
+portal \url{http://dev.socrata.com/consumers/getting-started.html}}
+}
+\description{
+Method for updating Socrata datasets
+}
+\author{
+Mark Silverberg \email{mark.silverberg@socrata.com}
+}
+

tests/testthat/test-readPrivateDataset.R

@@ -0,0 +1,26 @@
+context("Test reading private Socrata dataset with email and password")
+
+privateResourceToReadCsvUrl <- "https://soda.demo.socrata.com/resource/a9g2-feh2.csv"
+privateResourceToReadJsonUrl <- "https://soda.demo.socrata.com/resource/a9g2-feh2.json"
+socrataEmail <- Sys.getenv("SOCRATA_EMAIL", "")
+socrataPassword <- Sys.getenv("SOCRATA_PASSWORD", "")
+
+test_that("read Socrata CSV that requires a login", {
+	# should error when no email and password are sent with the request
+  expect_error(read.socrata(url = privateResourceToReadCsvUrl))
+  # try again, this time with email and password in the request
+  df <- read.socrata(url = privateResourceToReadCsvUrl, email = socrataEmail, password = socrataPassword)
+  # tests
+  expect_equal(2, ncol(df), label="columns")
+  expect_equal(3, nrow(df), label="rows")
+})
+
+test_that("read Socrata JSON that requires a login", {
+	# should error when no email and password are sent with the request
+  expect_error(read.socrata(url = privateResourceToReadJsonUrl))
+  # try again, this time with email and password in the request
+  df <- read.socrata(url = privateResourceToReadJsonUrl, email = socrataEmail, password = socrataPassword)
+  # tests
+  expect_equal(2, ncol(df), label="columns")
+  expect_equal(3, nrow(df), label="rows")
+})

tests/testthat/test-writeData.R

@@ -0,0 +1,44 @@
+context("write Socrata datasets")
+
+socrataEmail <- Sys.getenv("SOCRATA_EMAIL", "")
+socrataPassword <- Sys.getenv("SOCRATA_PASSWORD", "")
+
+test_that("add a row to a dataset", {
+	datasetToAddToUrl <- "https://soda.demo.socrata.com/resource/xh6g-yugi.json"
+
+	# populate df_in with two columns, each with a random number
+	x <- sample(-1000:1000, 1)
+	y <- sample(-1000:1000, 1)
+	df_in <- data.frame(x,y)
+
+	# write to dataset
+	write.socrata(df_in,datasetToAddToUrl,"UPSERT",socrataEmail,socrataPassword)
+
+	# read from dataset and store last (most recent) row for comparisons / tests
+	df_out <- read.socrata(url = datasetToAddToUrl, email = socrataEmail, password = socrataPassword)
+	df_out_last_row <- tail(df_out, n=1)
+
+	expect_equal(df_in$x, as.numeric(df_out_last_row$x), label = "x value")
+	expect_equal(df_in$y, as.numeric(df_out_last_row$y), label = "y value")
+})
+
+
+test_that("fully replace a dataset", {
+	datasetToReplaceUrl <- "https://soda.demo.socrata.com/resource/kc76-ybeq.json"
+
+	# populate df_in with two columns of random numbers
+	x <- sample(-1000:1000, 5)
+	y <- sample(-1000:1000, 5)
+	df_in <- data.frame(x,y)
+
+	# write to dataset
+	write.socrata(df_in,datasetToReplaceUrl,"REPLACE",socrataEmail,socrataPassword)
+
+	# read from dataset for comparisons / tests
+	df_out <- read.socrata(url = datasetToReplaceUrl, email = socrataEmail, password = socrataPassword)
+
+	expect_equal(ncol(df_in), ncol(df_out), label="columns")
+	expect_equal(nrow(df_in), nrow(df_out), label="rows")
+	expect_equal(df_in$x, as.numeric(df_out$x), label = "x values")
+	expect_equal(df_in$y, as.numeric(df_out$y), label = "y values")
+})