From faf4f38dac36da5b2e35ff9e5df8d32ce00d7aae Mon Sep 17 00:00:00 2001
From: Linsberger Christian <linchr@gba.geolba.ac.at>
Date: Wed, 8 Sep 2021 09:04:52 +0200
Subject: [PATCH] structural changes

---
 01_find_new_entries.R     |  1 +
 02_not_in_db.R            |  2 +-
 03_find_URLs.R            | 31 +------------------------------
 04_find_and_replace_uri.R |  2 +-
 sparql.R => functions.R   | 29 +++++++++++++++++++++++++++++
 5 files changed, 33 insertions(+), 32 deletions(-)
 rename sparql.R => functions.R (56%)

diff --git a/01_find_new_entries.R b/01_find_new_entries.R
index ad9e139..6d9a014 100644
--- a/01_find_new_entries.R
+++ b/01_find_new_entries.R
@@ -4,6 +4,7 @@ library("tidyverse")
 library("xlsx")
 
 source("config/config.R")
+source("functions.R")
 
 #Übernimmt die Liste der zu durchsuchenden Excelfiles aus der Konfiguration
 file_list <- FIND_DISTINCTS_FILES
diff --git a/02_not_in_db.R b/02_not_in_db.R
index 57cac6f..a75cdd3 100644
--- a/02_not_in_db.R
+++ b/02_not_in_db.R
@@ -8,7 +8,7 @@ library("tidyverse")
 
 #Konfiguration einbinden
 source("config/config.R")
-source("sparql.R")
+source("functions.R")
 
 #Datenbank einlesen und doppelte entfernen
 inDB <- data.frame(get_current_geoera_lit_db()$citation)
diff --git a/03_find_URLs.R b/03_find_URLs.R
index 19d2df2..4352368 100644
--- a/03_find_URLs.R
+++ b/03_find_URLs.R
@@ -4,36 +4,7 @@ library("writexl")
 
 #Konfiguration einbinden
 source("config/config.R")
-
-## nicht meine Funktion - eingebunden von https://rdrr.io/cran/retractcheck/src/R/utils.R
-find_doi <- function (strings) {
-  regex <- '10\\.\\d{4,9}/[-._;()/:A-Z0-9]+'
-  doiLoc <- gregexpr(text = strings, pattern = regex, perl = TRUE, ignore.case = TRUE)
-  
-  i <- 1
-  res <- NULL
-  
-  # for each in the doiLoc list check whether match (!-1)
-  for ( i in 1:length(doiLoc) ) {
-    if ( doiLoc[[i]][1] != -1 ) {
-      for ( j in 1:length(doiLoc[[i]]) ) {
-        res <- c(res,
-                 substring(strings[i], doiLoc[[i]][j], doiLoc[[i]][j] + attr(doiLoc[[i]], 'match.length')[j] - 1))
-      }
-    }
-  }
-  
-  return(res)
-}
-
-
-## nicht meine Funktion - eingebunden von https://stackoverflow.com/questions/52911812/check-if-url-exists-in-r
-valid_url <- function(url_in,t=2){
-  con <- url(url_in)
-  check <- suppressWarnings(try(open.connection(con,open="rt",timeout=t),silent=T)[1])
-  suppressWarnings(try(close.connection(con),silent=T))
-  ifelse(is.null(check),TRUE,FALSE)
-}
+source("functions.R")
 
 
 url_pattern <- "http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\\(\\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+"
diff --git a/04_find_and_replace_uri.R b/04_find_and_replace_uri.R
index 5567e9f..cb9c778 100644
--- a/04_find_and_replace_uri.R
+++ b/04_find_and_replace_uri.R
@@ -8,7 +8,7 @@ library("tictoc")
 library("rlang")
 
 #Funktionen & Konfiguration einbinden
-source("sparql.R")
+source("functions.R")
 source("config/config.R")
 
 
diff --git a/sparql.R b/functions.R
similarity index 56%
rename from sparql.R
rename to functions.R
index c824852..cc2c62d 100644
--- a/sparql.R
+++ b/functions.R
@@ -59,3 +59,32 @@ get_current_geoera_lit_db <- function() {
   
 }
 
+## nicht meine Funktion - eingebunden von https://rdrr.io/cran/retractcheck/src/R/utils.R
+find_doi <- function (strings) {
+  regex <- '10\\.\\d{4,9}/[-._;()/:A-Z0-9]+'
+  doiLoc <- gregexpr(text = strings, pattern = regex, perl = TRUE, ignore.case = TRUE)
+  
+  i <- 1
+  res <- NULL
+  
+  # for each in the doiLoc list check whether match (!-1)
+  for ( i in 1:length(doiLoc) ) {
+    if ( doiLoc[[i]][1] != -1 ) {
+      for ( j in 1:length(doiLoc[[i]]) ) {
+        res <- c(res,
+                 substring(strings[i], doiLoc[[i]][j], doiLoc[[i]][j] + attr(doiLoc[[i]], 'match.length')[j] - 1))
+      }
+    }
+  }
+  
+  return(res)
+}
+
+
+## nicht meine Funktion - eingebunden von https://stackoverflow.com/questions/52911812/check-if-url-exists-in-r
+valid_url <- function(url_in,t=2){
+  con <- url(url_in)
+  check <- suppressWarnings(try(open.connection(con,open="rt",timeout=t),silent=T)[1])
+  suppressWarnings(try(close.connection(con),silent=T))
+  ifelse(is.null(check),TRUE,FALSE)
+}