From e416974906c819ebb64d70aaa4c332dc698e58bc Mon Sep 17 00:00:00 2001
From: nwickel <n.wickelmaier@iwm-tuebingen.de>
Date: Thu, 14 Sep 2023 12:29:04 +0200
Subject: [PATCH] Fixed some stuff in artwork extraction

---
 code/functions.R         | 5 ++++-
 code/overview_artworks.R | 4 ++--
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/code/functions.R b/code/functions.R
index e2972f8..857adec 100644
--- a/code/functions.R
+++ b/code/functions.R
@@ -377,7 +377,6 @@ add_topic <- function(data, topics) {
 
 extract_artworks <- function(artworks, files = paste0(artworks, ".xml"),
                              path = path) {
-
   out <- NULL
   i <- 1
 
@@ -397,6 +396,10 @@ extract_artworks <- function(artworks, files = paste0(artworks, ".xml"),
       names(xmllist) <- varnames
       xmllist[which(sapply(xmllist, is.null))] <- NA
     }
+    # remove ugly quotes
+    xmllist <- lapply(xmllist, function(x) gsub("„|“", "", x))
+    # remove HTML tags
+    xmllist <- lapply(xmllist, function(x) gsub("<br/>", " ", x))
     xmldat <- as.data.frame(xmllist)
     xmldat$artwork <- artwork
     # trim white space from strings
diff --git a/code/overview_artworks.R b/code/overview_artworks.R
index 7a5c380..f203547 100644
--- a/code/overview_artworks.R
+++ b/code/overview_artworks.R
@@ -2,7 +2,7 @@
 
 source("functions.R")
 
-fname <- "overview_artworks.tex"
+fname <- "../haum/overview_artworks.tex"
 if (file.exists(fname)) file.remove(fname)
 
 path <- "C:/Users/nwickelmaier/Nextcloud/Documents/MDS/2023ss/60100_master_thesis/data/ContentEyevisit/eyevisit_cards_light"
@@ -40,7 +40,7 @@ for (artwork in dat$artwork) {
   writeLines("\\end{center}", fout)
   writeLines("", fout)
   writeLines("\\noindent", fout)
-  #writeLines(paste0("Info: ", dat[dat$artwork == artwork, "misc"]), fout)
+  writeLines(paste0("Info: ", dat[dat$artwork == artwork, "misc"]), fout)
   writeLines("\\newpage", fout)
 }