october

R/datasets.R

@@ -15,5 +15,6 @@
 #' @references <https://www.nytimes.com/interactive/2019/07/30/us/elections/debate-speaking-time.html>
 #' @references <https://www.nytimes.com/interactive/2019/07/31/us/elections/debate-speaking-time.html>
 #' @references <https://www.nytimes.com/interactive/2019/09/12/us/elections/debate-speaking-time.html>
+#' @references <https://www.nytimes.com/interactive/2019/10/15/us/elections/debate-speaking-time.html>
 #' @usage data("debates2019")
 NULL

data-raw/debates2019.R

@@ -21,6 +21,7 @@ if (!file.exists(here::here("data-raw/2019-06-27-us-elections-debate-speaking-ti
 if (!file.exists(here::here("data-raw/2019-07-30-us-elections-debate-speaking-time.html"))) download.file("https://www.nytimes.com/interactive/2019/07/30/us/elections/debate-speaking-time.html", here::here("data-raw/2019-07-30-us-elections-debate-speaking-time.html"))
 if (!file.exists(here::here("data-raw/2019-07-31-us-elections-debate-speaking-time.html"))) download.file("https://www.nytimes.com/interactive/2019/07/31/us/elections/debate-speaking-time.html", here::here("data-raw/2019-07-31-us-elections-debate-speaking-time.html"))
 if (!file.exists(here::here("data-raw/2019-09-12-us-elections-debate-speaking-time.html"))) download.file("https://www.nytimes.com/interactive/2019/09/12/us/elections/debate-speaking-time.html", here::here("data-raw/2019-09-12-us-elections-debate-speaking-time.html"))
+if (!file.exists(here::here("data-raw/2019-10-15-us-elections-debate-speaking-time.html"))) download.file("https://www.nytimes.com/interactive/2019/10/15/us/elections/debate-speaking-time.html", here::here("data-raw/2019-10-15-us-elections-debate-speaking-time.html"))
 
 read_html(here::here("data-raw/2019-06-26-us-elections-debate-speaking-time.html")) %>%
   html_nodes(xpath = ".//script[contains(., 'NYTG_DEMDEBATES')]") %>%
@@ -252,12 +253,59 @@ read_html(here::here("data-raw/2019-09-12-us-elections-debate-speaking-time.html
   ) %>%
   as_tibble() -> sep_day_1
 
+read_html(here::here("data-raw/2019-10-15-us-elections-debate-speaking-time.html")) %>%
+  html_nodes(xpath = ".//script[contains(., 'NYTG_DEMDEBATES')]") %>%
+  html_text() %>%
+  stri_split_lines() %>%
+  unlist() %>%
+  .[3] %>%
+  stri_replace_first_regex("^.*NYTG_DEMDEBATES = ", "") %>%
+  jsonlite::fromJSON() %>%
+  mutate(
+    elapsed = as.numeric(elapsed)/60,
+    debate_date = as.Date("2019-10-15"),
+    speaker = stri_trans_totitle(speaker),
+    timestamp = parse_time(timestamp),
+    debate_group = 4,
+    night = 1
+  ) %>%
+  mutate(
+    speaker = case_when(
+      speaker == "Orourke" ~ "O'Rourke",
+      TRUE ~ speaker
+    )
+  ) %>%
+  mutate(
+    topic = case_when(
+      topic == "" ~ "Other",
+      grepl("impeachment", topic) ~ "Impeachment",
+      grepl("economy", topic) ~ "Economy",
+      grepl("opioids", topic) ~ "Opioids",
+      grepl("candidate-age", topic) ~ "Age",
+      grepl("tech-companies", topic) ~ "Tech Companies",
+      grepl("middle-east policy", topic) ~ "Foreign Policy",
+      grepl("gun-control", topic) ~ "Gun Control",
+      grepl("income-inequality", topic) ~ "Income Inequality",
+      grepl("health-care", topic) ~ "Healthcare",
+      grepl("party-strategy", topic) ~ "Party Strategy",
+      grepl("womens-rights", topic) ~ "Women's Rights",
+      TRUE ~ topic
+    )
+  ) %>%
+  filter(
+    !is.na(timestamp),
+    speaker != "",
+    speaker != "Moderator"
+  ) %>%
+  as_tibble() -> oct_day_1
+
 bind_rows(
   jun_day_1,
   jun_day_2,
   jul_day_1,
   jul_day_2,
-  sep_day_1
+  sep_day_1,
+  oct_day_1
 ) -> debates2019
 
 usethis::use_data(debates2019, overwrite = TRUE)

man/debates2019.Rd

@@ -26,5 +26,7 @@ metadata and opening/closing statement records.
 \url{https://www.nytimes.com/interactive/2019/07/31/us/elections/debate-speaking-time.html}
 
 \url{https://www.nytimes.com/interactive/2019/09/12/us/elections/debate-speaking-time.html}
+
+\url{https://www.nytimes.com/interactive/2019/10/15/us/elections/debate-speaking-time.html}
 }
 \keyword{datasets}