commit 707ed17993344282ce35926dcb3b908c59ed82e9
parent 26e1a453d9ddf0575885050a5133c4a241b161a5
Author: eamoncaddigan <eamon.caddigan@gmail.com>
Date: Tue, 11 Aug 2015 16:39:57 -0400
Have sysdata with countries and languages, and tidied up some code.
Diffstat:
8 files changed, 76 insertions(+), 48 deletions(-)
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -7,5 +7,7 @@ Depends: R (>= 3.2.1)
Imports:
httr (>= 1.0.0),
jsonlite
+Suggests:
+ devtools
License: GPL-3
LazyData: true
diff --git a/R/guessGender.R b/R/guessGender.R
@@ -2,10 +2,11 @@
# Helper functions --------------------------------------------------------
-#' Check country and language code.
+#' Check country and language code.
#'
-#' Makes sure that no more than one of countryCode or languageCode is *not* NA
-#' (i.e., they can both be NA, or one can be NA).
+#' Makes sure that no more than one of countryCode or languageCode is *not* NA
+#' (i.e., they can both be NA, or one can be NA).
+#' @keywords internal
checkLanguageCountryCodes <- function(countryCode, languageCode) {
# TODO: Check code validity
if (sum(is.na(c(countryCode, languageCode))) < 1) {
@@ -30,17 +31,14 @@ getListElement <- function(listName, elementName) {
# API functions -----------------------------------------------------------
-#' Look up a vector of names.
+#' Look up a vector of names on genderize.io.
#'
#' This function actually implements the genderize.io API. Can only query 10
#' names at a time.
-#' @param nameVector A vector containing one or more names to look up.
-#' @param countryCode An optional ISO 3166-1 alpha-2 country code.
-#' @param languageCode An optional ISO 639-1 language code. Only one of
-#' countryCode or languageCode can be specified.
-#' @param apiKey An optional API key for genderize.io.
+#' @inheritParams guessGender
#' @keywords internal
-lookupNameVector <- function(nameVector, countryCode = NA, languageCode = NA, apiKey = NA) {
+lookupNameVectorGenderize <- function(nameVector,
+ countryCode = NA, languageCode = NA, apiKey = NA) {
# Make sure that no more than 10 names were passed
if (length(nameVector) > 10) {
stop("This only accepts 10 or fewer names")
@@ -49,16 +47,16 @@ lookupNameVector <- function(nameVector, countryCode = NA, languageCode = NA, ap
# Construct the query
query <- paste("name[", seq_along(nameVector), "]=", nameVector,
- sep="",
- collapse="&")
+ sep = "",
+ collapse = "&")
if (!is.na(countryCode)) {
- query <- paste(query, "&country_id=", countryCode, sep="")
+ query <- paste(query, "&country_id=", countryCode, sep = "")
}
if (!is.na(languageCode)) {
- query <- paste(query, "&language_id=", languageCode, sep="")
+ query <- paste(query, "&language_id=", languageCode, sep = "")
}
if (!is.na(apiKey)) {
- query <- paste(query, "&apikey=", apiKey, sep="")
+ query <- paste(query, "&apikey=", apiKey, sep = "")
}
# Run it!
@@ -66,7 +64,7 @@ lookupNameVector <- function(nameVector, countryCode = NA, languageCode = NA, ap
queryResult <- httr::GET("https://api.genderize.io", query = query,
httr::config(ssl_verifypeer = FALSE))
if (httr::status_code(queryResult) == 200) {
- responseFromJSON <- jsonlite::fromJSON(httr::content(queryResult, as="text"))
+ responseFromJSON <- jsonlite::fromJSON(httr::content(queryResult, as = "text"))
# Make sure this is a data.frame with the correct columns. I bet fromJSON
# can do this for me but I don't know how. This code works whether fromJSON
# returned a list (the response to one name) or a data.frame (the response
@@ -106,8 +104,9 @@ lookupNameVector <- function(nameVector, countryCode = NA, languageCode = NA, ap
#' @param apiKey An optional API key for genderize.io.
#' @export
#' @examples
-#' guessGender(c("Eamon", "Sean"), countryCode = "US")
-guessGender <- function(nameVector, countryCode = NA, languageCode = NA, apiKey = NA) {
+#' guessGender(c("Natalie", "Liam", "Eamon"), countryCode = "US")
+guessGender <- function(nameVector,
+ countryCode = NA, languageCode = NA, apiKey = NA) {
checkLanguageCountryCodes(countryCode, languageCode)
# genderize.io only handles 10 names at a time. Create a list of vectors, each
@@ -122,7 +121,7 @@ guessGender <- function(nameVector, countryCode = NA, languageCode = NA, apiKey
# Run the queries
responseList <- list()
for (i in seq_along(queryList)) {
- responseDF <- lookupNameVector(queryList[[i]], countryCode, apiKey)
+ responseDF <- lookupNameVectorGenderize(queryList[[i]], countryCode, apiKey)
if (is.null(responseDF)) {
break
} else {
diff --git a/R/sysdata.rda b/R/sysdata.rda
Binary files differ.
diff --git a/data-raw/getCountryAndLanguageCodes.R b/data-raw/getCountryAndLanguageCodes.R
@@ -1,2 +1,28 @@
-#https://api.genderize.io/languages
-#https://api.genderize.io/countries
+
+# Read genderize.io's lists of supported country and language codes -------
+
+queryResult <- httr::GET("https://api.genderize.io/languages",
+ httr::config(ssl_verifypeer = FALSE))
+if (httr::status_code(queryResult) == 200) {
+ responseFromJSON <- jsonlite::fromJSON(httr::content(queryResult, as="text"))
+ genderizeLanguages <- responseFromJSON[["languages"]]
+ # Don't know why they return an empty string.
+ genderizeLanguages <- genderizeLanguages[nchar(genderizeLanguages) > 0]
+} else {
+ stop("Couldn't load language list")
+}
+
+queryResult <- httr::GET("https://api.genderize.io/countries",
+ httr::config(ssl_verifypeer = FALSE))
+if (httr::status_code(queryResult) == 200) {
+ responseFromJSON <- jsonlite::fromJSON(httr::content(queryResult, as="text"))
+ genderizeCountries <- responseFromJSON[["countries"]]
+} else {
+ stop("Couldn't load country list")
+}
+
+
+# Save the lists ----------------------------------------------------------
+
+devtools::use_data(genderizeLanguages, genderizeCountries,
+ internal = TRUE, overwrite = TRUE)
diff --git a/man/checkLanguageCountryCodes.Rd b/man/checkLanguageCountryCodes.Rd
@@ -8,6 +8,7 @@ checkLanguageCountryCodes(countryCode, languageCode)
}
\description{
Makes sure that no more than one of countryCode or languageCode is *not* NA
- (i.e., they can both be NA, or one can be NA).
+(i.e., they can both be NA, or one can be NA).
}
+\keyword{internal}
diff --git a/man/guessGender.Rd b/man/guessGender.Rd
@@ -21,6 +21,6 @@ This function uses the genderize.io API to supply estimates of the gender one
or more names.
}
\examples{
-guessGender(c("Eamon", "Sean"), countryCode = "US")
+guessGender(c("Natalie", "Liam", "Eamon"), countryCode = "US")
}
diff --git a/man/lookupNameVector.Rd b/man/lookupNameVector.Rd
@@ -1,25 +0,0 @@
-% Generated by roxygen2 (4.1.1): do not edit by hand
-% Please edit documentation in R/guessGender.R
-\name{lookupNameVector}
-\alias{lookupNameVector}
-\title{Look up a vector of names.}
-\usage{
-lookupNameVector(nameVector, countryCode = NA, languageCode = NA,
- apiKey = NA)
-}
-\arguments{
-\item{nameVector}{A vector containing one or more names to look up.}
-
-\item{countryCode}{An optional ISO 3166-1 alpha-2 country code.}
-
-\item{languageCode}{An optional ISO 639-1 language code. Only one of
-countryCode or languageCode can be specified.}
-
-\item{apiKey}{An optional API key for genderize.io.}
-}
-\description{
-This function actually implements the genderize.io API. Can only query 10
-names at a time.
-}
-\keyword{internal}
-
diff --git a/man/lookupNameVectorGenderize.Rd b/man/lookupNameVectorGenderize.Rd
@@ -0,0 +1,25 @@
+% Generated by roxygen2 (4.1.1): do not edit by hand
+% Please edit documentation in R/guessGender.R
+\name{lookupNameVectorGenderize}
+\alias{lookupNameVectorGenderize}
+\title{Look up a vector of names on genderize.io.}
+\usage{
+lookupNameVectorGenderize(nameVector, countryCode = NA, languageCode = NA,
+ apiKey = NA)
+}
+\arguments{
+\item{nameVector}{A vector containing one or more names to look up.}
+
+\item{countryCode}{An optional ISO 3166-1 alpha-2 country code.}
+
+\item{languageCode}{An optional ISO 639-1 language code. Only one of
+countryCode or languageCode can be specified.}
+
+\item{apiKey}{An optional API key for genderize.io.}
+}
+\description{
+This function actually implements the genderize.io API. Can only query 10
+names at a time.
+}
+\keyword{internal}
+
