From e31bfbb985265ff4820d7d64bbd7d2f85d47d02c Mon Sep 17 00:00:00 2001
From: Candace Savonen <cansav09@gmail.com>
Date: Wed, 30 Oct 2024 14:30:03 -0400
Subject: [PATCH] Pair programming URL accommodation adds

---
 scripts/url-check.R | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/scripts/url-check.R b/scripts/url-check.R
index 113c1a8..b90e284 100644
--- a/scripts/url-check.R
+++ b/scripts/url-check.R
@@ -134,7 +134,10 @@ get_urls <- function(file) {
   urls <- unlist(url_list)
 
   # Remove trailing characters
-  urls <- gsub("\\'\\:$|\\'|\\:$|\\.$", "", urls)
+  urls <- gsub("\\'\\:$|\\'|\\:$|\\.$|\\)$|\\,$", "", urls)
+  
+  # Remove URLs that are in the ignore 
+  if( ignore_urls[1] != "") urls <- grep(paste0(ignore_urls, collapse = "|"), urls, invert = TRUE, value = TRUE)
 
   if (length(urls) > 0 ){
     # Remove trailing characters
@@ -150,8 +153,7 @@ get_urls <- function(file) {
 all_urls <- lapply(files, get_urls)
 
 # Write the file
-all_urls_df <- dplyr::bind_rows(all_urls) %>%
-  dplyr::filter(!(urls %in% ignore_urls))
+all_urls_df <- dplyr::bind_rows(all_urls)
 
 if (nrow(all_urls_df) > 0) {
   all_urls_df <- all_urls_df %>%