experimental argument types_src for get_dbpedia_uris() #27

NAMESPACE

@@ -5,12 +5,12 @@ export(as_annotation)
 export(as_subcorpus)
 export(dbpedia_get_wikidata_uris)
 export(dbpedia_spotlight_status)
-export(map_types_to_class)
 export(namespaced_xpath)
 export(sparql_query)
 export(wikidata_query)
 export(xml_enrich)
 exportMethods(add_wikidata_uris)
+exportMethods(entity_types_map)
 exportMethods(get_annotation_table)
 exportMethods(get_dbpedia_uris)
 import(methods)

R/dbpedia.R

@@ -354,7 +354,19 @@ setGeneric("get_dbpedia_uris", function(x, ...) standardGeneric("get_dbpedia_uri
 #'   types = "Company",
 #'   api = "http://api.dbpedia-spotlight.org/en/annotate"
 #' )
-setMethod("get_dbpedia_uris", "character", function(x, language = getOption("dbpedia.lang"), max_len = 5600L, confidence = 0.35, api = getOption("dbpedia.endpoint"), types = character(), support = 20, verbose = TRUE){
+setMethod(
+  "get_dbpedia_uris", "character",
+  function(
+    x,
+    language = getOption("dbpedia.lang"),
+    max_len = 5600L,
+    confidence = 0.35,
+    api = getOption("dbpedia.endpoint"),
+    types = character(),
+    support = 20,
+    types_src = c("DBpedia", "Wikidata"),
+    verbose = TRUE
+  ){
 
   if (nchar(x) > max_len){
     if (verbose) cli_alert_warning(
@@ -425,6 +437,28 @@ setMethod("get_dbpedia_uris", "character", function(x, language = getOption("dbp
       )
     }
   )]
+
+  if (length(types_src) > 0L){
+    src_all <- unique(unlist(lapply(resources_min[["types"]], names)))
+    src_unused <- setdiff(src_all, types_src)
+    if (length(src_unused) > 0L)
+      cli_alert_info(
+        "dropping available types from: {paste(src_unused, collapse = ' / ')}"
+      )
+    for (src in types_src){
+      types_vec <- unlist(lapply(
+        lapply(resources_min[["types"]], `[[`, src),
+        function(x){
+          if (is.null(x))
+            NA_character_
+          else
+            sprintf("|%s|", paste(x, collapse = "|"))
+        }
+      ))
+
+      resources_min[, (paste(src, "type", sep = "_")) := types_vec]
+    }
+  }
 
   resources_min
 })
@@ -475,6 +509,11 @@ setMethod("get_dbpedia_uris", "AnnotatedPlainTextDocument", function(x, language
 #'   vector is empty (default), no restrictions are applied.
 #' @param support The number of indegrees at Wikidata. Useful for limiting the 
 #'   the number of results by excluding insignificant entities.
+#' @param types_src A `character` vector specifying knowledge bases as sources
+#'   for entity types. If provided, columns following the pattern '(src)_type'
+#'   (e.g. "DBpedia_types") with entity types (`NA` if not available) will be
+#'   added to the table. Values are wrapped and separated by vertical bars.
+#'   `types_src` defaults to "DBpedia" and "Wikidata".
 #' @param verbose A `logical` value - whether to display messages.
 #' @param progress A `logical` value - whether to show progress.
 #' @param s_attribute A length-one `character` vector indicating a s-attribute.
@@ -765,7 +804,8 @@ setMethod(
             verbose = if (progress) FALSE else verbose
           )[, "doc" := docname]
         }
-      )
+      ),
+      fill = TRUE
     )
 
     if (progress) cli_progress_done(.envir = env)

R/entity_types.R

@@ -12,19 +12,24 @@
 #' @param other a `character vector` with the name of the class of all types not
 #'   matched by the `mapping_vector`.
 #' @param verbose A `logical` value - whether to display messages.
+#' @param ... Further arguments.
 #' @importFrom data.table is.data.table
 #' @importFrom cli format_error cli_alert_info
 #' @details If there is more than one match between the retrieved types and the
 #'   `mapping vector`, unique classes are sorted alphabetically and collapsed.
 #' @return Function adds classes to input data.table by reference.
-#' @exportMethod generic
+#' @exportMethod entity_types_map
 #' @rdname entity_types_map
-setGeneric("entity_types_map", function(x, ...)
-  standardGeneric("entity_types_map"))
+setGeneric(
+  "entity_types_map",
+  function(x, ...) standardGeneric("entity_types_map")
+)
 
 
 #' @rdname entity_types_map
 #' @examples
+#' library(quanteda)
+#' 
 #' inaugural_paragraphs <- data_corpus_inaugural %>%
 #'   corpus_subset(Year == 2009) %>% # limit to Barack Obama 2009
 #'   corpus_reshape(to = "paragraphs")
@@ -116,4 +121,4 @@ setMethod(
 
   x[, class := entity_types_map(x = x[["types"]])]
   x
-}
+})