From fba552577ec4667397f3818ac9a36185504fd98d Mon Sep 17 00:00:00 2001
From: Linsberger Christian <linchr@gba.geolba.ac.at>
Date: Tue, 7 Sep 2021 08:12:19 +0200
Subject: [PATCH] added a script that finds URLs an puts them in a new column

---
 find_URLs.R | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)
 create mode 100644 find_URLs.R

diff --git a/find_URLs.R b/find_URLs.R
new file mode 100644
index 0000000..1f19cdd
--- /dev/null
+++ b/find_URLs.R
@@ -0,0 +1,28 @@
+library("stringr")
+library("readxl")
+library("writexl")
+
+#Konfiguration einbinden
+source("config/config.R")
+
+url_pattern <- "http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\\(\\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+"
+
+#Excel mit den Zitaten die hinzugefügt werden sollen einlesen
+df <- data.frame(read_excel(paste(PFAD_EXCEL, "distincts_kontrolliert.xlsx", sep="")))
+df <- data.frame(lapply(df, stri_enc_toutf8))
+
+#Neue Spalte für die DownloadLinks erstellen
+df['downloadLink'] <- NA
+
+#Schleife um nach Links zu suchen
+for (element in 1:length(df$bibliographicCitation)){
+  
+  url <- str_extract(df$bibliographicCitation[element], url_pattern)
+  
+  if (!is.na(url)) {
+    df$downloadLink[element] <- url
+  }
+  
+}
+
+write_xlsx(df, paste(PFAD_OUT, "distincts_automatisch_mit_URL",format(Sys.time(), "%Y_%m_%d") ,".xlsx", sep=""))
\ No newline at end of file