Merge branch '43-make-create_annotation-parameteters-optional' into 'master'

Resolve "Make create_annotation parameteters optional"

Closes #43

See merge request !37

processing/wikiproc/R/nlp_annotate.R

@@ -25,8 +25,8 @@ init_nlp <- function(type, value) {
 #' Create annotations for the given text
 #'
 #' @param text Text to annotate
-#' @param article.id ArticleID used for cashing
-#' @param article.rev.id ArticleRevisionID used for cashing
+#' @param article.id ArticleID used for cashing. **Optional** as long as cache is not to be used.
+#' @param article.rev.id ArticleRevisionID used for cashing. **Optional** as long as cache is not to be used.
 #' @param use.cache Should cashed data be uses
 #' @param write.cache Should the generated annotations be cashed
 #' @param data.dir Directory the data should be read from and/or written to
@@ -34,22 +34,29 @@ init_nlp <- function(type, value) {
 #' @return Annotation object for use with cleanNLP methods
 #' @export
 create_annotations <- function(text, article.id, article.rev.id, use.cache = TRUE, write.cache = FALSE, data.dir = "data") {
-
-  # Generate filename, for some reason there paste0 will pad the article id with leading whitespaces
-  # To prevent this we stip 'em again
-
-  filename <- gsub(" ", "", paste(data.dir, "annotations", paste0(article.id, "-", article.rev.id, ".RDS"), sep = .Platform$file.sep), fixed = TRUE)
-
-  # Check if there is a cached version of the annotations for this article in this specific revision
-
-  if(use.cache & file.exists(filename)) {
-    res <- tryCatch({
-      data <- readRDS(filename)
-      data
-    }, error = function (e) {
-      cat("Cached data seems to be corrupted, redoing annotation.\n")
-    })
-    return(res)
+  
+  if (use.cache || write.cache) {
+    if (!missing(article.id) && !(article.rev.id)) {
+      
+      # Generate filename, for some reason there paste0 will pad the article id with leading whitespaces
+      # To prevent this we stip 'em again
+      
+      filename <- gsub(" ", "", paste(data.dir, "annotations", paste0(article.id, "-", article.rev.id, ".RDS"), sep = .Platform$file.sep), fixed = TRUE)
+      
+      # Check if there is a cached version of the annotations for this article in this specific revision
+      
+      if(use.cache & file.exists(filename)) {
+        res <- tryCatch({
+          data <- readRDS(filename)
+          data
+        }, error = function (e) {
+          cat("Cached data seems to be corrupted, redoing annotation.\n")
+        })
+        return(res)
+      }
+    } else if (use.cache) {
+      warning("use.cache was set to true without providing article id and revision id, cache will not be used.")
+    }
   }
   
   annotation <- cleanNLP::cnlp_annotate(text, as_strings = TRUE)
@@ -57,13 +64,18 @@ create_annotations <- function(text, article.id, article.rev.id, use.cache = TRU
   # Write cache if desired
 
   if(write.cache) {
-    if (!dir.exists("data")) {
-      dir.create("data")
+    if (!missing(article.id) && !missing(article.rev.id)) {
+      if (!dir.exists("data")) {
+        dir.create("data")
+      }
+      if (!dir.exists("data/annotations")) {
+        dir.create("data/annotations")
+      }
+      saveRDS(annotation, filename)
+    } else {
+      warning("write.cache was set to true without providing article id and revision id, cache will not be written")
     }
-    if (!dir.exists("data/annotations")) {
-      dir.create("data/annotations")
-    }
-    saveRDS(annotation, filename)
+    
   }
   
   # Return data
@@ -71,4 +83,4 @@ create_annotations <- function(text, article.id, article.rev.id, use.cache = TRU
   # But then again, it suggests snake case for variables...
   
   return(annotation)
-}
+}
\ No newline at end of file

processing/wikiproc/man/create_annotations.Rd

@@ -10,9 +10,9 @@ create_annotations(text, article.id, article.rev.id, use.cache = TRUE,
 \arguments{
 \item{text}{Text to annotate}
 
-\item{article.id}{ArticleID used for cashing}
+\item{article.id}{ArticleID used for cashing. **Optional** as long as cache is not to be used.}
 
-\item{article.rev.id}{ArticleRevisionID used for cashing}
+\item{article.rev.id}{ArticleRevisionID used for cashing. **Optional** as long as cache is not to be used.}
 
 \item{use.cache}{Should cashed data be uses}