From efd32184ecbf610fd51df33f3da6d88dfed704fe Mon Sep 17 00:00:00 2001
From: Lukas Gehrke <lukasgehrke@Lukass-MacBook-Pro.local>
Date: Mon, 7 Jan 2019 06:51:43 +0100
Subject: [PATCH] Aendert if-Verzweigung in getBirthdate.R.
---
r/GetBirthdate.R | 56 +++++++++++++++++++++++++++---------------------
1 file changed, 31 insertions(+), 25 deletions(-)
diff --git a/r/GetBirthdate.R b/r/GetBirthdate.R
index 21f61a5..42dbb69 100644
--- a/r/GetBirthdate.R
+++ b/r/GetBirthdate.R
@@ -14,36 +14,42 @@ library(xml2)
#' will be checked for birthdate
#'
#' @param article Article in HTML-format
-#' @return String birthdate as string
+#' @return String birthdate as string|NULL
getBirthdate <- function(article) {
- # Check if there is an infobox
- if(!grepl("vcard", article)) {
- # Check first paragraph
- introduction <- getIntroduction(article)
- if(!introduction == "") {
- # Get birthdate inside of parentheses
- birthdate <- str_extract_all(introduction, "\\([^()]+\\)")[[1]]
- # Remove parentheses
- birthdate <- substring(birthdate, 2, nchar(birthdate)-1)
- } else {
- # Return Null if there is no birthdate
- return(NULL)
- }
- }
- # Try to get birthdate via infobox
- infoBox <- getInfoBox(article)
+ if(grepl("vcard", article)) {
- # Get the born field
- birthdate <- infoBox[infoBox$Desc %like% "Born",]$Content
+ # Check if there is an infobox
+ infoBox <- getInfoBox(article)
+
+ # Get the Born field
+ birthdate <- infoBox[infoBox$Desc %like% "Born",]$Content
+ # Remove everything except the birthdate:
+ # - Remove everything in round brackets
+ birthdate <- gsub("\\s*\\([^\\)]+\\)", "", birthdate)
+ # - Remove everything starting with newline
+ birthdate <- gsub("\\n.*$", "", birthdate)
+
+ return(birthdate)
- # Remove everything except the birthdate
- # Remove everything in round brackets
- birthdate <- gsub("\\s*\\([^\\)]+\\)", "", birthdate)
- # Remove everything starting with newline
- birthdate <- gsub("\\n.*$", "", birthdate)
- return(birthdate)
+ } else if(!getIntroduction(article) == "") {
+
+ # Check first paragraph
+ introduction <- getIntroduction(article)
+
+ # Get birthdate inside of parentheses
+ birthdate <- str_extract_all(introduction, "\\([^()]+\\)")[[1]]
+ # Remove parentheses
+ birthdate <- substring(birthdate, 2, nchar(birthdate)-1)
+
+ return(birthdate)
+
+ } else {
+
+ # Return Null if there is no birthdate
+ return(NULL)
+ }
}
### Converts info box to table
--
GitLab