From baef13eddecfe7881422619ca8edcd5438d33572 Mon Sep 17 00:00:00 2001
From: Linsberger Christian <linchr@gba.geolba.ac.at>
Date: Mon, 6 Sep 2021 16:02:30 +0200
Subject: [PATCH] 2 New Files, Renamed.

---
 00_Setup.R                                    |  8 +++++
 find_new_entries.R => 01_find_new_entries.R   |  2 +-
 02_not_in_db.R                                | 29 +++++++++++++++++++
 ...replace_uri.R => 03_find_and_replace_uri.R |  0
 4 files changed, 38 insertions(+), 1 deletion(-)
 create mode 100644 00_Setup.R
 rename find_new_entries.R => 01_find_new_entries.R (96%)
 create mode 100644 02_not_in_db.R
 rename find_and_replace_uri.R => 03_find_and_replace_uri.R (100%)

diff --git a/00_Setup.R b/00_Setup.R
new file mode 100644
index 0000000..efaa1cf
--- /dev/null
+++ b/00_Setup.R
@@ -0,0 +1,8 @@
+#### Generiert die Ordnerstruktur der Daten #### 
+
+source("config/config.R")
+
+ifelse(!dir.exists(file.path("data/")), dir.create(file.path("data/")), FALSE)
+ifelse(!dir.exists(file.path(PFAD_EXCEL)), dir.create(file.path(PFAD_EXCEL)), FALSE)
+ifelse(!dir.exists(file.path(PFAD_OUT)), dir.create(file.path(PFAD_OUT)), FALSE)
+ifelse(!dir.exists(file.path(PFAD_EXCEL)), dir.create(file.path(PFAD_DB_OUT)), FALSE)
\ No newline at end of file
diff --git a/find_new_entries.R b/01_find_new_entries.R
similarity index 96%
rename from find_new_entries.R
rename to 01_find_new_entries.R
index 5e63b2f..ad9e139 100644
--- a/find_new_entries.R
+++ b/01_find_new_entries.R
@@ -83,4 +83,4 @@ distincts <- data.frame(distincts[nchar(distincts$bibliographicCitation) >= THRE
 colnames(distincts) <- (c('bibliographicCitation'))
 
 #Schreibt die Distincts in eine Exceldatei
-write_xlsx(distincts, "data/Output/distincts_automated_gc3d.xlsx")
+write_xlsx(distincts, paste(PFAD_OUT, "distincts_automated_gc3d.xlsx", sep=""))
diff --git a/02_not_in_db.R b/02_not_in_db.R
new file mode 100644
index 0000000..5ac9392
--- /dev/null
+++ b/02_not_in_db.R
@@ -0,0 +1,29 @@
+library("readxl")
+library("writexl")
+library("dplyr")
+library("xlsx")
+library("stringi")
+
+source("config/config.R")
+source("sparql.R")
+
+vorhanden_df = data.frame(get_current_geoera_lit_db()$citation)
+vorhanden_df_2 = unique(vorhanden_df)
+
+colnames(vorhanden_df) = (c(COLUMN_NAME))
+
+neu_df = data.frame(read_excel(paste(PFAD_EXCEL, "distincts_kontrolliert.xlsx", sep="")))
+neu_df = data.frame(lapply(neu_df, stri_enc_toutf8))
+neu_df_2 = unique(neu_df)
+
+all_df = rbind(vorhanden_df, neu_df)
+all_df_2 = unique(all_df)
+
+#inner_join finds common elements between two data frames
+#anti_join finds elements the are exclusively in one of the data frames
+common = inner_join(neu_df, vorhanden_df)
+not_in_db = anti_join(neu_df, common)
+
+not_in_db_2 = anti_join(all_df, vorhanden_df)
+
+write_xlsx(not_in_db, paste(PFAD_OUT, "not_in_db_",format(Sys.time(), "%Y_%m_%d") ,".xlsx", sep=""))
diff --git a/find_and_replace_uri.R b/03_find_and_replace_uri.R
similarity index 100%
rename from find_and_replace_uri.R
rename to 03_find_and_replace_uri.R