Merge branch 'develop'

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: rosetteApi
 Type: Package
 Title: Rosette API
-Version: 1.5.0
+Version: 1.7.0
 Date: 2016-12-12
 Author: Sam Hausmann [aut, cre],
     Chris Park [ctb],

R/Api.R

@@ -56,6 +56,9 @@ api <- function(user_key, endpoint, parameters=FALSE, custom_headers=NULL, url_p
       "name-similarity"= error_check(
         post_endpoint(user_key, check_names(parameters, "name-similarity"), "name-similarity", url, custom_headers, url_parameters)
       ),
+      "name-deduplication"= error_check(
+        post_endpoint(user_key, check_deduplication(parameters, "name-deduplication"), "name-deduplication", url, custom_headers, url_parameters)
+      ),
       "relationships"= error_check(
         post_endpoint(user_key, check_content_parameters(parameters), "relationships", url, custom_headers, url_parameters)
       ),
@@ -74,6 +77,9 @@ api <- function(user_key, endpoint, parameters=FALSE, custom_headers=NULL, url_p
       "text-embedding"= error_check(
         post_endpoint(user_key, check_content_parameters(parameters), "text-embedding", url, custom_headers, url_parameters)
       ),
+      "transliteration"= error_check(
+        post_endpoint(user_key, check_content_parameters(parameters), "transliteration", url, custom_headers, url_parameters)
+      ),
       "syntax/dependencies"= error_check(
         post_endpoint(user_key, check_content_parameters(parameters), "syntax/dependencies", url, custom_headers, url_parameters)
       ),
@@ -87,7 +93,7 @@ api <- function(user_key, endpoint, parameters=FALSE, custom_headers=NULL, url_p
 #' Provides the binding Version
 #' @return current binding version
 get_binding_version <- function() {
-  bindingVersion <- "1.5.0"
+  bindingVersion <- "1.7.0"
   return(bindingVersion)
 }
 
@@ -138,6 +144,19 @@ check_names <- function(parameters, endpoint) {
   }
 }
 
+#' check if the required request parameters for name deduplication are correct
+#' @param parameters - parameters list to be passed to specified Rosette API endpoint
+#' @param endpoint - Rosette API endpoint to be utilized
+#' @return Returns list of verified parameters to be sent to Rosette API
+check_deduplication <- function(parameters, endpoint) {
+  params <- parameters
+  if (!("names" %in% names(params))) {
+    stop("Must supply a list of names to deduplicate")
+  } else {
+    return(parameters)
+  }
+}
+
 #' create a multipart
 #' @param parameters - parameters list to be passed to specified Rosette API endpoint
 #' @return Returns a multipart
@@ -225,6 +244,19 @@ serialize_name_parameters <- function(parameters) {
   return(jsonlite::toJSON(serialized_params, auto_unbox = TRUE))
 }
 
+#' serialize Rosette API parameters
+#' @param parameters - parameters list to be passed to name-deduplication
+#' @return Returns the serialized parameters for the Rosette API
+serialize_name_deduplication_parameters <- function(parameters) {
+  serialized_params <- list()
+  for (param in names(parameters)) {
+    if (param == "names" || param == "threshold") {
+    serialized_params[[param]] <- parameters[[param]]
+    }
+  }
+  return(jsonlite::toJSON(serialized_params, auto_unbox = TRUE))
+}
+
 #' Helper to check for file submission
 #' @param parameters - JSON parameters
 #' @return true if multipart
@@ -250,23 +282,25 @@ post_endpoint <- function(user_key, parameters, endpoint, url, custom_headers=NU
     content_type <- "application/json"
     if (endpoint == "name-translation" || endpoint == "name-similarity") {
       request_body <- serialize_name_parameters(parameters)
+    } else if (endpoint == "name-deduplication") {
+      request_body <- serialize_name_deduplication_parameters(parameters)
     } else {
       request_body <- serialize_parameters(parameters)
     }
   }
 
   if (is.null(url_parameters)) {
     response <- httr::POST(
-      paste(url, endpoint, sep=""), 
+      paste(url, endpoint, sep=""),
       encode = encoding,
-      httr::add_headers(get_headers(user_key, content_type = content_type, custom_headers = custom_headers)), 
+      httr::add_headers(get_headers(user_key, content_type = content_type, custom_headers = custom_headers)),
       body = request_body
     )
   } else {
     response <- httr::POST(
-      paste(url, endpoint, sep=""), 
+      paste(url, endpoint, sep=""),
       encode = encoding,
-      httr::add_headers(get_headers(user_key, content_type = content_type, custom_headers = custom_headers)), 
+      httr::add_headers(get_headers(user_key, content_type = content_type, custom_headers = custom_headers)),
       body = request_body,
       query = url_parameters
     )
@@ -284,12 +318,12 @@ post_endpoint <- function(user_key, parameters, endpoint, url, custom_headers=NU
 get_endpoint <- function(user_key, endpoint, url, custom_headers=NULL, url_parameters=NULL) {
   if (is.null(url_parameters)) {
     response <- httr::GET(
-      paste(url, endpoint, sep=""), 
+      paste(url, endpoint, sep=""),
       httr::add_headers(get_headers(user_key, custom_headers))
     )
   } else {
     response <- httr::GET(
-      paste(url, endpoint, sep=""), 
+      paste(url, endpoint, sep=""),
       httr::add_headers(get_headers(user_key, custom_headers)),
       query = url_parameters
     )

README.md

@@ -35,20 +35,21 @@ print(jsonlite::toJSON(result$content, pretty = TRUE)
 ```
 ## API Parameters
 | Parameter                     | Endpoint                                            | Required
-| -------------                 |-------------                                        |------------- 
-| content                    | categories, entities, language, morphology, relationships, sentences, sentiment, tokens, syntax_dependencies            | Either content or contentUri required |
-| contentUri                      | categories, entities, language, morphology, relationships, sentences, sentiment, tokens, syntax_dependencies       | Either content or contentUri required |
+| -------------                 |-------------                                        |-------------
+| content                    | categories, entities, language, morphology, relationships, sentences, sentiment, tokens, syntax/dependencies, transliteration (Yes)            | Either content or contentUri required |
+| contentUri                      | categories, entities, language, morphology, relationships, sentences, sentiment, tokens, syntax/dependencies       | Either content or contentUri required |
 | language                          | categories, entities, language, morphology, relationships, sentences, sentiment, tokens, name similarity                    | No |
 | documentFile                      | categories, entities, language, morphology, relationships, sentences, sentiment, tokens                  | No |
 | name1                 | name similarity               | Yes |
 | name2               | name similarity| Yes |
 | name    | name translation     | Yes |
-| targetLanguage           | name translation           | Yes |
+| targetLanguage           | name translation, transliteration (No)          | Yes |
 | entityType                 | name translation         | No |
 | sourceLanguageOfOrigin        | name translation | No |
 | sourceLanguageOfUse                         | name translation       | No |
-| sourceScript                     | name translation               | No |
-| targetScript                     | name translation                    | No |
+| sourceScript                     | name translation, transliteration               | No |
+| sourceLanguage | transliteration | Yes |
+| targetScript                     | name translation, transliteration                   | No |
 | targetScheme                        | name translation          | No |
 | options              | relationships        | No |
 | accuracyMode              | relationships        | Yes |

examples/categories.R

@@ -16,7 +16,7 @@ parameters[[ "contentUri" ]] <- categories_url_data
 if (is.na(opt$url)){
    result <- api(opt$key, "categories", parameters)
 } else {
-   result <- api(opt$key, "categories", parameters, NULL, opt$url)
+   result <- api(opt$key, "categories", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/entities.R

@@ -19,7 +19,7 @@ url_parameters <- list(output = "rosette")
 if (is.na(opt$url)){
    result <- api(opt$key, "entities", parameters, url_parameters = url_parameters)
 } else {
-   result <- api(opt$key, "entities", parameters, NULL, opt$url)
+   result <- api(opt$key, "entities", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/info.R

@@ -11,7 +11,7 @@ opt = parse_args(opt_parser)
 if (is.na(opt$url)){
    result <- api(opt$key, "info")
 } else {
-   result <- api(opt$key, "info", NULL, NULL, opt$url)
+   result <- api(opt$key, "info", FALSE, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/language.R

@@ -17,7 +17,7 @@ custom_headers <- c("X-RosetteAPI-App" = "r-application-test")
 if (is.na(opt$url)){
   result <- api(opt$key, "language", parameters = parameters, custom_headers = custom_headers)
 } else {
-  result <- api(opt$key, "language", parameters = parameters, custom_headers = headers, url = opt$url)
+  result <- api(opt$key, "language", parameters = parameters, custom_headers = custom_headers, NULL, url = opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/morphology_complete.R

@@ -17,7 +17,7 @@ parameters[[ "morphology" ]] <- "complete"
 if (is.na(opt$url)){
    result <- api(opt$key, "morphology", parameters)
 } else {
-   result <- api(opt$key, "morphology", parameters, NULL, opt$url)
+   result <- api(opt$key, "morphology", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/morphology_compound-components.R

@@ -17,7 +17,7 @@ parameters[[ "morphology" ]] <- "compound-components"
 if (is.na(opt$url)){
    result <- api(opt$key, "morphology", parameters)
 } else {
-   result <- api(opt$key, "morphology", parameters, NULL, opt$url)
+   result <- api(opt$key, "morphology", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/morphology_han-readings.R

@@ -17,7 +17,7 @@ parameters[[ "morphology" ]] <- "han-readings"
 if (is.na(opt$url)){
    result <- api(opt$key, "morphology", parameters)
 } else {
-   result <- api(opt$key, "morphology", parameters, NULL, opt$url)
+   result <- api(opt$key, "morphology", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/morphology_lemmas.R

@@ -17,7 +17,7 @@ parameters[[ "morphology" ]] <- "lemmas"
 if (is.na(opt$url)){
    result <- api(opt$key, "morphology", parameters)
 } else {
-   result <- api(opt$key, "morphology", parameters, NULL, opt$url)
+   result <- api(opt$key, "morphology", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/morphology_parts-of-speech.R

@@ -17,7 +17,7 @@ parameters[[ "morphology" ]] <- "parts-of-speech"
 if (is.na(opt$url)){
    result <- api(opt$key, "morphology", parameters)
 } else {
-   result <- api(opt$key, "morphology", parameters, NULL, opt$url)
+   result <- api(opt$key, "morphology", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/name_deduplication.R

@@ -0,0 +1,29 @@
+source("../R/Api.R")
+library(jsonlite)
+library("optparse")
+
+option_list = list( make_option(c("-k", "--key"), action="store", default=NA, type='character',
+              help="Rosette API key"), make_option(c("-u", "--url"), action="store", default=NA, type='character',
+              help="Rosette API url"))
+opt_parser = OptionParser(option_list=option_list)
+opt = parse_args(opt_parser)
+
+name_dedupe_data <- "John Smith,Johnathon Smith,Fred Jones"
+
+
+text <- unlist(strsplit(name_dedupe_data, ","))
+targetLanguage <- rep("eng", length(text))
+targetScript <- rep("Latn", length(text))
+
+names = data.frame(text, targetLanguage, targetScript)
+parameters <- list()
+parameters[[ "names" ]] <- names
+parameters[[ "threshold" ]] <- 0.75
+
+if (is.na(opt$url)){
+   result <- api(opt$key, "name-deduplication", parameters)
+} else {
+   result <- api(opt$key, "name-deduplication", parameters, NULL, NULL, opt$url)
+}
+print(jsonlite::toJSON(result$header, pretty = TRUE))
+print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/name_similarity.R

@@ -18,7 +18,7 @@ parameters[[ "name2" ]] <- matched_name_data2
 if (is.na(opt$url)){
    result <- api(opt$key, "name-similarity", parameters)
 } else {
-   result <- api(opt$key, "name-similarity", parameters, NULL, opt$url)
+   result <- api(opt$key, "name-similarity", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/name_translation.R

@@ -18,7 +18,7 @@ parameters[[ "targetScript" ]] <- "Latn"
 if (is.na(opt$url)){
    result <- api(opt$key, "name-translation", parameters)
 } else {
-   result <- api(opt$key, "name-translation", parameters, NULL, opt$url)
+   result <- api(opt$key, "name-translation", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/ping.R

@@ -11,7 +11,7 @@ opt = parse_args(opt_parser)
 if (is.na(opt$url)){
    result <- api(opt$key, "ping")
 } else {
-   result <- api(opt$key, "ping", NULL, NULL, opt$url)
+   result <- api(opt$key, "ping", FALSE, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/relationships.R

@@ -16,7 +16,7 @@ parameters[[ "content" ]] <- relationships_text_data
 if (is.na(opt$url)){
    result <- api(opt$key, "relationships", parameters)
 } else {
-   result <- api(opt$key, "relationships", parameters, NULL, opt$url)
+   result <- api(opt$key, "relationships", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/sentences.R

@@ -16,7 +16,7 @@ parameters[[ "content" ]] <- sentences_data
 if (is.na(opt$url)){
    result <- api(opt$key, "sentences", parameters)
 } else {
-   result <- api(opt$key, "sentences", parameters, NULL, opt$url)
+   result <- api(opt$key, "sentences", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/sentiment.R

@@ -17,7 +17,7 @@ parameters[[ "documentFile" ]] <- "true"
 if (is.na(opt$url)){
    result <- api(opt$key, "sentiment", parameters)
 } else {
-   result <- api(opt$key, "sentiment", parameters, NULL, opt$url)
+   result <- api(opt$key, "sentiment", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/syntax_dependencies.R

@@ -12,12 +12,11 @@ syntax_dependencies_data <- "Yoshinori Ohsumi, a Japanese cell biologist, was aw
 
 parameters <- list()
 parameters[[ "content" ]] <- syntax_dependencies_data
-parameters[[ "genre" ]] <- "social-media"
 
 if (is.na(opt$url)){
    result <- api(opt$key, "syntax/dependencies", parameters)
 } else {
-   result <- api(opt$key, "syntax/dependencies", parameters, NULL, opt$url)
+   result <- api(opt$key, "syntax/dependencies", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/text_embedding.R

@@ -16,7 +16,7 @@ parameters[[ "content" ]] <- embeddings_data
 if (is.na(opt$url)){
    result <- api(opt$key, "text-embedding", parameters)
 } else {
-   result <- api(opt$key, "text-embedding", parameters, NULL, opt$url)
+   result <- api(opt$key, "text-embedding", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/tokens.R

@@ -16,7 +16,7 @@ parameters[[ "content" ]] <- tokens_data
 if (is.na(opt$url)){
    result <- api(opt$key, "tokens", parameters)
 } else {
-   result <- api(opt$key, "tokens", parameters, NULL, opt$url)
+   result <- api(opt$key, "tokens", parameters, NULL, NULL, opt$url)
 }
 print(jsonlite::toJSON(result$header, pretty = TRUE))
 print(jsonlite::toJSON(result$content, pretty = TRUE))

examples/transliteration.R

@@ -0,0 +1,22 @@
+source("../R/Api.R")
+library(jsonlite)
+library("optparse")
+
+option_list = list( make_option(c("-k", "--key"), action="store", default=NA, type='character',
+              help="Rosette API key"), make_option(c("-u", "--url"), action="store", default=NA, type='character',
+              help="Rosette API url"))
+opt_parser = OptionParser(option_list=option_list)
+opt = parse_args(opt_parser)
+
+transliteration_data <- "معمر محمد أبو منيار القذاف"
+
+parameters <- list()
+parameters[[ "content" ]] <- transliteration_data
+
+if (is.na(opt$url)){
+   result <- api(opt$key, "transliteration", parameters)
+} else {
+   result <- api(opt$key, "transliteration", parameters, NULL, NULL, opt$url)
+}
+print(jsonlite::toJSON(result$header, pretty = TRUE))
+print(jsonlite::toJSON(result$content, pretty = TRUE))