entity_types_map (#40) and types_src (#41) more robust

DESCRIPTION

@@ -1,8 +1,8 @@
 Package: dbpedia
 Type: Package
 Title: R Wrapper for DBpedia Spotlight
-Version: 0.1.2
-Date: 2024-02-26
+Version: 0.1.2.9001
+Date: 2024-03-06
 Authors@R: c(
   person("Andreas", "Blaette", role = c("aut", "cre"), email = "andreas.blaette@uni-due.de", comment = c(ORCID = "0000-0001-8970-8010")),
   person("Christoph", "Leonhardt", role = "aut")

NEWS.md

@@ -1,3 +1,9 @@
+## dbpedia v0.1.2.9001
+* `entity_types_map()` now creates assignments again (#40) and returns them as character vectors
+* `entity_types_map()` also passes all arguments when used with data.table objects
+* `types_src` works in `get_dbpedia_uris()` for documents with a single type (#41)
+* messages for `types_src` follow verbosity set by the argument `verbose`
+
 ## dbpedia v0.1.2
 * `get_dbpedia_uris()` has new argument `types` to filter results.
 * `dbpedia_spotlight_status()` without warnings if docker not available / not running #32.

R/dbpedia.R

@@ -424,24 +424,33 @@ setMethod(
     new = c("dbpedia_uri", "text", "start", "types")
   )
   setcolorder(resources_min, c("start", "text", "dbpedia_uri", "types"))
-  
+
   resources_min[, "start" := as.integer(resources_min[["start"]]) + 1L]
+
+  # See issue 41.
+  types_list <- strsplit(x = resources_min[["types"]], split = ",")
+
   resources_min[, "types" := lapply(
-    strsplit(x = resources_min[["types"]], split = ","),
+    types_list,
     function(x){
       if (length(x) == 0L) return(list())
       spl <- strsplit(x, split = ":")
-      split(
+      types <- split(
         x = unlist(lapply(spl, `[`, 2L)),
         f = unlist(lapply(spl, `[`, 1L))
       )
+      if (length(types) == 1L & length(types_list) == 1L) {
+        list(types)
+      } else {
+        types
+      }
     }
   )]
-  
+
   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)
+    if (length(src_unused) > 0L & isTRUE(verbose))
       cli_alert_info(
         "dropping available types from: {paste(src_unused, collapse = ' / ')}"
       )
@@ -481,17 +490,17 @@ setMethod("get_dbpedia_uris", "AnnotatedPlainTextDocument", function(x, language
 
 
 #' Get DBpedia links.
-#' 
-#' #' @details
-#' `expand_to_token` is a rather experimental feature that resolves mismatches
-#' between entity spans and token spans by expanding the former to the last
-#' character position of the corresponding token. See issue #26 in the `dbpedia`
-#' GitHub repository.
-#' The configuration of the `httr::GET()` calls used can be controlled using
-#' `httr::config()`. A relevant scenario is SSL verification issues that can be
-#' addressed using `httr::set_config(httr::config(ssl_verifypeer = 0L))` (at own
-#' risk!). The error "HTTP/2 stream 1 was not closed cleanly before end of the
-#' underlying stream" can be addressed using
+#'
+#' @details - `expand_to_token` is a rather experimental feature that resolves
+#' mismatches between entity spans and token spans by expanding the former to
+#' the last character position of the corresponding token. See issue #26 in the
+#' `dbpedia` GitHub repository.
+#' - The configuration of the `httr::GET()` calls
+#' used can be controlled using `httr::config()`. A relevant scenario is SSL
+#' verification issues that can be addressed using
+#' `httr::set_config(httr::config(ssl_verifypeer = 0L))` (at own risk!). The
+#' error "HTTP/2 stream 1 was not closed cleanly before end of the underlying
+#' stream" can be addressed using
 #' `httr::set_config(httr::config(http_verson = 1.1))`
 #'
 #' @param x A `subcorpus` (`xml`, ...) object. Will be coerced to
@@ -505,9 +514,9 @@ setMethod("get_dbpedia_uris", "AnnotatedPlainTextDocument", function(x, language
 #'   as threshold before DBpedia Spotlight includes a link into the report.
 #' @param api An URL of the DBpedia Spotlight API.
 #' @param types A `character` vector to restrict result returned to certain
-#'   entity types, such as 'Company' or 'Organization'. If the `character` 
+#'   entity types, such as 'Company' or 'Organization'. If the `character`
 #'   vector is empty (default), no restrictions are applied.
-#' @param support The number of indegrees at Wikidata. Useful for limiting the 
+#' @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'
@@ -530,9 +539,9 @@ setMethod("get_dbpedia_uris", "AnnotatedPlainTextDocument", function(x, language
 #' @return A `data.table` with the following columns:
 #' - *dbpedia_uri*: The DBpedia URI.
 #' - *text*: Text that has been annotated
-#' - *types*: Recognized entity types, for each row a named list, if available 
-#'   entries such as 'DBpedia', 'Schema', 'Wikidata', 'DUL'
-#' Depending on the input object, further columns may be available.
+#' - *types*: Recognized entity types, for each row a named list, if available
+#'   entries such as 'DBpedia', 'Schema', 'Wikidata', 'DUL'.
+#'   Depending on the input object, further columns may be available.
 #' @exportMethod get_dbpedia_uris
 #' @importFrom cli cli_alert_warning cli_progress_step cli_alert_danger
 #'   cli_progress_done cli_alert_info

R/entity_types.R

@@ -1,12 +1,12 @@
-#' Map types returned by DBpedia Spotlight to a limited set of classes
+#' Map types returned by DBpedia Spotlight to a limited set of categories
 #'
 #' This function takes the output of `get_dbpedia_uris()` and compares values in
 #' the `types` column with a named character vector. The main purpose of this
-#' function is to reduce the number of types to a limited set of classes.
+#' function is to reduce the number of types to a limited set of categories.
 #'
 #' @param x A `data.table` with DBpedia URIs.
-#' @param mapping_vector A `named character vector` with desired class names (as
-#'   names) and types from the DBpedia ontology as values. For example:
+#' @param mapping_vector A `named character vector` with desired category names
+#'   (as names) and types from the DBpedia ontology as values. For example:
 #'   c("PERSON" = "DBpedia:Person"). Can contain more than one pair of class and
 #'   type.
 #' @param other a `character vector` with the name of the class of all types not
@@ -70,32 +70,28 @@ setMethod(
     if (!is.character(other) | length(other) > 1)
       stop(format_error("{.var other} not character vector of length {.val 1}."))
 
-    lapply(
+    sapply(
       x,
       function(el){
-        # types is a list of lists. Transform to single character vector.
-        type_list <- unlist(el, recursive = FALSE)
-
-        # An unintended consequence here is that you may get DBpedia1, DBpedia2, ...
-
-        types_with_class_raw <- lapply(
-          seq_along(type_list),
+        types_with_category_raw <- lapply(
+          seq_along(el),
           function(i) {
-            list_name <- names(type_list)[[i]]
-            list_elements <- type_list[[i]]
+            list_name <- names(el)[[i]]
+            list_elements <- el[[i]]
             paste0(list_name, ":", list_elements)
           })
-        types_with_class <- intersect(unlist(types_with_class_raw), mapping_vector)
-
-        if (length(types_with_class) > 0L) {
-          match_idx <- which(mapping_vector %in% types_with_class)
-
-          class_name <- paste(
+
+        types_with_category <- intersect(unlist(types_with_category_raw), mapping_vector)
+
+        if (length(types_with_category) > 0L) {
+          match_idx <- which(mapping_vector %in% types_with_category)
+
+          category <- paste(
             sort(unique(names(mapping_vector)[match_idx])),
             collapse = "|"
           )
         } else {
-          class_name <- other
+          category <- other
         }
       }
     )
@@ -116,9 +112,9 @@ setMethod(
 
   if (verbose)
     cli_alert_info(
-      "mapping values in column {.var types} to new column {.var class}"
+      "mapping values in column {.var types} to new column {.var category}"
     )
 
-  x[, class := entity_types_map(x = x[["types"]])]
+  x[, category := entity_types_map(x = x[["types"]], mapping_vector = mapping_vector, other = other, verbose = verbose)]
   x
 })

R/utils.R

@@ -105,11 +105,11 @@ as_chunks <- function(x, size){
 #' Transform table with DBpedia URIs to subcorpus.
 #'
 #' @param x A `data.table` with DBpedia URIs.
-#' @param highlight_by A `character vector` of the column in which entity names
-#'   are annotated. Defaults to NULL.
+#' @param highlight_by A `character vector` of the column in which the types of
+#'   entities are annotated. Defaults to NULL.
 #' @details If a `character vector` is supplied to `highlight_by`, selected
 #'   entity types  (PERSON, LOCATION, ORGANIZATION, MISC) are assigned specific
-#'   color codes. Other entities in the column are assigned a single color.
+#'   color codes. Other types in the column are assigned a single color.
 #' @importFrom fs path
 #' @export
 as_subcorpus <- function(x, highlight_by = NULL){