feat: wikimedia foundation methods

NAMESPACE

@@ -13,5 +13,8 @@ export(linguee_translation_examples)
 export(linguee_word_translation)
 export(mymemory_translate)
 export(translate_file)
+export(wikimedia_detect_language)
+export(wikipedia_get_language_names)
+export(wmcloud_translate)
 importFrom(magrittr,"%>%")
 importFrom(rlang,":=")

R/wikimedia_detect_language.R

@@ -0,0 +1,45 @@
+#' Detect the language of a text
+#'
+#' This function sends a POST request to the Wikimedia Language ID API with the specified text,
+#' parses the JSON response, and returns the detected language.
+#'
+#' @param text The text whose language is to be detected.
+#'
+#' @return The detected language.
+#'
+#' @examples
+#' # Detect the language of a text
+#' wikimedia_detect_language("Hallo, wereld")
+#'
+#' @export
+wikimedia_detect_language <- function(text) {
+  # Define the URL of the API
+  url <- "https://api.wikimedia.org/service/lw/inference/v1/models/langid:predict"
+
+  # Create a list of parameters to send in the POST request
+  body <- list(
+    text = text
+  )
+
+  # Convert the list to a JSON string
+  json_body <- jsonlite::toJSON(body, auto_unbox = TRUE)
+
+  # Set the content type of the request to 'application/json'
+  headers <- c("Content-Type" = "application/json")
+
+  # Send the POST request and get the response
+  response <- httr::POST(url, body = json_body, httr::add_headers(headers))
+
+  # Check if the request was successful
+  if (httr::status_code(response) != 200) {
+    stop("Request failed with status ", httr::status_code(response))
+  }
+
+  # Parse the response
+  result <- httr::content(response, "parsed")
+
+  # Extract the detected language from the response
+  language <- result$wikicode
+
+  return(language)
+}

R/wikipedia_get_language_names.R

@@ -0,0 +1,42 @@
+#' Get language names
+#'
+#' This function sends a GET request to the Wikipedia API and returns the language names as a dataframe.
+#'
+#' @return A dataframe of language names.
+#'
+#' @examples
+#' # Get language names
+#' wikipedia_get_language_names()
+#'
+#' @export
+wikipedia_get_language_names <- function() {
+  # Define the URL of the API
+  url <- "https://en.wikipedia.org/w/api.php?action=query&liprop=autonym|name&meta=languageinfo&uselang=en&format=json&origin=*"
+
+  # Send the GET request and get the response
+  response <- httr::GET(url)
+
+  # Check if the request was successful
+  if (httr::status_code(response) != 200) {
+    stop("Request failed with status ", httr::status_code(response))
+  }
+
+  # Parse the response
+  result <- httr::content(response, "parsed")
+
+  # Extract the language names from the response
+  language_names <- result$query$languageinfo
+
+  # Get the keys of the language_names list
+  language_tags <- names(language_names)
+
+  # Combine the language_tags, names, and autonyms into a dataframe
+  df <- data.frame(
+    language_tag = language_tags,
+    name = sapply(language_names, function(x) x$name),
+    autonym = sapply(language_names, function(x) x$autonym),
+    row.names = NULL
+  )
+
+  return(df)
+}

R/wmcloud_translate.R

@@ -0,0 +1,69 @@
+#' Translate text using WMCloud
+#'
+#' This function sends a POST request to the WMCloud translation API with the specified parameters,
+#' parses the JSON response, and returns the translated text.
+#'
+#' @param text The text to translate.
+#' @param target_language The target language for the translation (default is "en").
+#' @param source_language The source language of the text (default is "en").
+#' @param format The format of the content ("html", "json", "markdown", "text", "svg", "webpage").
+#' @param model The model to use for the translation (only "nllb200-600M" is currently known to work).
+#'
+#' @return The translated text.
+#' @export
+wmcloud_translate <- function(text,
+                              target_language = "en",
+                              source_language = "en",
+                              format = "text",
+                              model = "nllb200-600M") {
+  # Define the URL of the API
+  url <- "https://translate.wmcloud.org/api/translate"
+
+  # List of valid formats
+  valid_formats <- c("html", "json", "markdown", "text", "svg", "webpage")
+
+  # List of valid models
+  valid_models <- c("nllb200-600M") # Add more models here
+
+  # Check if format and model are valid
+  if (!format %in% valid_formats) {
+    stop(paste("Invalid format. Must be one of:", paste(valid_formats, collapse = ", ")))
+  }
+  if (!model %in% valid_models) {
+    stop(paste("Invalid model. Must be one of:", paste(valid_models, collapse = ", ")))
+  }
+
+  # Create a list of parameters to send in the POST request
+  body <- list(
+    source_language = source_language,
+    target_language = target_language,
+    format = format,
+    model = model,
+    content = text
+  )
+
+  # Convert the list to a JSON string
+  json_body <- jsonlite::toJSON(body, auto_unbox = TRUE)
+
+  # Set the content type of the request to 'application/json'
+  headers <- c("Content-Type" = "application/json")
+
+  # Send the POST request and get the response
+  response <- httr::POST(url, body = json_body, httr::add_headers(headers))
+
+  # Check if the request was successful
+  if (httr::status_code(response) != 200) {
+    stop("Request failed with status ", httr::status_code(response))
+  }
+
+  # Parse the response
+  result <- httr::content(response, "parsed")
+
+  # Extract the translated text from the response
+  translation <- result$translation
+
+  return(translation)
+}
+
+
+

README.md

@@ -16,6 +16,8 @@ Currently, the package has functions to communicate with the following services:
 
 -   Linguee API
 
+-   Wikimedia Translation
+
 # Features
 To see which functions are available, please refer to the reference page of the `polyglotr` package. The [reference](https://Tomeriko96.github.io/polyglotr/reference/index.html) page provides a comprehensive list of functions available in the package, organized by category.
 

_pkgdown.yml

@@ -28,3 +28,10 @@ reference:
     desc: Methods using MyMemory Translation services
     contents:
       - mymemory_translate
+
+  - title: Wikimedia Foundation Methods
+    desc: Methods using Wikimedia Foundation Translation services
+    contents:
+      - wikimedia_detect_language
+      - wikipedia_get_language_names
+      - wmcloud_translate