From 4d6bfeecbabd23e591356be5adb4cd55c8087f9f Mon Sep 17 00:00:00 2001
From: schochastics <david@schochastics.net>
Date: Wed, 16 Oct 2024 10:04:39 +0200
Subject: [PATCH] added rp-online.de

---
 NAMESPACE                |  1 +
 R/deliver_rp_online_de.R | 25 +++++++++++++++++++++++++
 inst/status.csv          |  1 +
 3 files changed, 27 insertions(+)
 create mode 100644 R/deliver_rp_online_de.R

diff --git a/NAMESPACE b/NAMESPACE
index 7dab8af..b296dee 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -53,6 +53,7 @@ S3method(pb_deliver_paper,nypost_com)
 S3method(pb_deliver_paper,nytimes_com)
 S3method(pb_deliver_paper,parlamentnilisty_cz)
 S3method(pb_deliver_paper,prosieben_de)
+S3method(pb_deliver_paper,rp_online_de)
 S3method(pb_deliver_paper,rte_ie)
 S3method(pb_deliver_paper,rtl_de)
 S3method(pb_deliver_paper,rtl_nl)
diff --git a/R/deliver_rp_online_de.R b/R/deliver_rp_online_de.R
new file mode 100644
index 0000000..e8fb160
--- /dev/null
+++ b/R/deliver_rp_online_de.R
@@ -0,0 +1,25 @@
+#' @export
+pb_deliver_paper.rp_online_de <- function(x, verbose = NULL, pb, ...) {
+    pb_tick(x, verbose, pb)
+    # raw html is stored in column content_raw
+    html <- rvest::read_html(x$content_raw)
+
+    json_txt <- rvest::html_nodes(html, "script[type = \"application/ld+json\"] ")[1] %>% rvest::html_text()
+    json_df <- jsonlite::fromJSON(json_txt)
+
+    datetime <- lubridate::as_datetime(json_df$datePublished)
+    headline <- json_df$headline
+    author <- toString(json_df$author$name)
+    text <- html %>%
+        rvest::html_nodes("strong[data-cy=\"intro\"],div[data-cy=\"article_content\"] p") %>%
+        rvest::html_text2() %>%
+        paste(collapse = "\n")
+
+    s_n_list(
+        datetime,
+        author,
+        headline,
+        text,
+        json_df # dumping the whole json data of an article
+    )
+}
diff --git a/inst/status.csv b/inst/status.csv
index 1dc9d3e..6cf23f2 100644
--- a/inst/status.csv
+++ b/inst/status.csv
@@ -59,6 +59,7 @@
 "pagesix.com","![](https://img.shields.io/badge/status-requested-lightgrey)","","[#1](https://github.com/JBGruber/paperboy/issues/1)",NA
 "parlamentnilisty.cz","![](https://img.shields.io/badge/status-gold-%23ffd700.svg)","[@JBGruber](https://github.com/JBGruber/)","","http://www.parlamentnilisty.cz/export/rss.aspx"
 "prosieben.de","![](https://img.shields.io/badge/status-gold-%23ffd700.svg)","[@schochastics](https://github.com/schochastics)","[#23](https://github.com/JBGruber/paperboy/issues/23)",NA
+"rp-online.de","![](https://img.shields.io/badge/status-gold-%23ffd700.svg)","[@schochastics](https://github.com/schochastics)","[#23](https://github.com/JBGruber/paperboy/issues/23)", "https://rp-online.de/feed.rss"
 "rte.ie","![](https://img.shields.io/badge/status-gold-%23ffd700.svg)","[@JBGruber](https://github.com/JBGruber/)","","https://www.rte.ie/feeds/rss/?index=/news/"
 "rtl.de","![](https://img.shields.io/badge/status-gold-%23ffd700.svg)","[@schochastics](https://github.com/schochastics)","[#23](https://github.com/JBGruber/paperboy/issues/23)","https://www.rtl.de/rss/feed/news"
 "rtl.nl","![](https://img.shields.io/badge/status-gold-%23ffd700.svg)","[@JBGruber](https://github.com/JBGruber/)","","https://www.rtlnieuws.nl/rss.xml"