Browse Source

feb 2019 2nd debate

master
boB Rudis 1 year ago
parent
commit
435ae7c0a1
No known key found for this signature in database GPG Key ID: 1D7529BE14E2BBA9
5 changed files with 739 additions and 39 deletions
  1. +3
    -0
      NEWS.md
  2. +40
    -36
      data-raw/2020-02-07-us-elections-debate-speaking-time.html
  3. +615
    -0
      data-raw/2020-02-19-us-elections-debate-speaking-time.html
  4. +81
    -3
      data-raw/debates2019.R
  5. BIN
      data/debates2019.rda

+ 3
- 0
NEWS.md View File

@ -1,3 +1,6 @@
0.5.1
* Data updated for latest debate (2020-02-19)
0.5.0
* Data updated for latest debate (2020-02-07)


+ 40
- 36
data-raw/2020-02-07-us-elections-debate-speaking-time.html
File diff suppressed because it is too large
View File


+ 615
- 0
data-raw/2020-02-19-us-elections-debate-speaking-time.html
File diff suppressed because it is too large
View File


+ 81
- 3
data-raw/debates2019.R View File

@ -25,7 +25,8 @@ if (!file.exists(here::here("data-raw/2019-10-15-us-elections-debate-speaking-ti
if (!file.exists(here::here("data-raw/2019-11-20-us-elections-debate-speaking-time.html"))) download.file("https://www.nytimes.com/interactive/2019/11/20/us/elections/debate-speaking-time.html", here::here("data-raw/2019-11-20-us-elections-debate-speaking-time.html"))
if (!file.exists(here::here("data-raw/2019-12-19-us-elections-debate-speaking-time.html"))) download.file("https://www.nytimes.com/interactive/2019/12/19/us/elections/debate-speaking-time.html", here::here("data-raw/2019-12-19-us-elections-debate-speaking-time.html"))
if (!file.exists(here::here("data-raw/2020-01-14-us-elections-debate-speaking-time.html"))) download.file("https://www.nytimes.com/interactive/2020/01/14/us/politics/debate-speaking-time.html", here::here("data-raw/2020-01-14-us-elections-debate-speaking-time.html"))
if (!file.exists(here::here("data-raw/2020-02-08-us-elections-debate-speaking-time.html"))) download.file("https://www.nytimes.com/interactive/2020/02/07/us/elections/debate-speaking-time.html", here::here("data-raw/2020-02-07-us-elections-debate-speaking-time.html"))
if (!file.exists(here::here("data-raw/2020-02-07-us-elections-debate-speaking-time.html"))) download.file("https://www.nytimes.com/interactive/2020/02/07/us/elections/debate-speaking-time.html", here::here("data-raw/2020-02-07-us-elections-debate-speaking-time.html"))
if (!file.exists(here::here("data-raw/2020-02-19-us-elections-debate-speaking-time.html"))) download.file("https://www.nytimes.com/interactive/2020/02/19/us/elections/debate-speaking-time.html", here::here("data-raw/2020-02-19-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')]") %>%
@ -337,6 +338,7 @@ read_html(here::here("data-raw/2019-11-20-us-elections-debate-speaking-time.html
grepl("immigration", topic) ~ "Immigration",
grepl("impeachment", topic) ~ "Impeachment",
grepl("income-inequality", topic) ~ "Income Inequality",
grepl("economic-inequality", topic) ~ "Income Inequality",
grepl("economy", topic) ~ "Economy",
grepl("middle-east policy", topic) ~ "Foreign Policy",
grepl("opioids", topic) ~ "Opioids",
@ -380,7 +382,9 @@ read_html(here::here("data-raw/2019-12-19-us-elections-debate-speaking-time.html
grepl("climate", topic) ~ "Climate",
grepl("closing", topic) ~ "Closing",
grepl("judges", topic) ~ "Judges",
grepl("transparency", topic) ~ "Transparency",
grepl("race", topic) ~ "Race",
grepl("economic-inequality", topic) ~ "Income Inequality",
grepl("education", topic) ~ "Education",
grepl("human-rights", topic) ~ "Human Rights",
grepl("criminal-justice", topic) ~ "Criminal Justice",
@ -489,7 +493,7 @@ read_html(here::here("data-raw/2020-02-07-us-elections-debate-speaking-time.html
debate_date = as.Date("2020-01-14"),
speaker = stri_trans_totitle(speaker),
timestamp = parse_time(timestamp),
debate_group = 7,
debate_group = 8,
night = 1
) %>%
filter(speaker != "") %>%
@ -541,6 +545,79 @@ read_html(here::here("data-raw/2020-02-07-us-elections-debate-speaking-time.html
) %>%
as_tibble() -> feb_07_day_1
read_html(here::here("data-raw/2020-02-19-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() %>%
as_tibble() %>%
mutate(
elapsed = as.numeric(elapsed)/60,
debate_date = as.Date("2020-01-14"),
speaker = stri_trans_totitle(speaker),
timestamp = parse_time(timestamp),
debate_group = 9,
night = 1
) %>%
filter(speaker != "") %>%
filter(!is.na(timestamp)) %>%
# distinct(topic) %>%
# print(n=nrow(.))
mutate(
topic = case_when(
topic == "" ~ "Other",
grepl("military", topic) ~ "Military",
grepl("climate", topic) ~ "Climate",
grepl("closing", topic) ~ "Closing",
grepl("judges", topic) ~ "Judges",
grepl("race", topic) ~ "Race",
grepl("Climate-change", topic) ~ "Climate",
grepl("poverty", topic) ~ "Poverty",
grepl("education", topic) ~ "Education",
grepl("bloomberg", topic) ~ "Bloomberg",
grepl("human-rights", topic) ~ "Human Rights",
grepl("criminal-justice", topic) ~ "Criminal Justice",
grepl("electability", topic) ~ "Electability",
grepl("election-reform", topic) ~ "Election Reform",
grepl("executive-power", topic) ~ "Executive Power",
grepl("candidate-age", topic) ~ "Age",
grepl("supreme-court", topic) ~ "Supreme Court",
grepl("foreign-policy", topic) ~ "Foreign Policy",
grepl("gun-control", topic) ~ "Gun Control",
grepl("child-care", topic) ~ "Child Care",
grepl("party-loyalty", topic) ~ "Party Loyalty",
grepl("health-care", topic) ~ "Healthcare",
grepl("immigration", topic) ~ "Immigration",
grepl("impeachment", topic) ~ "Impeachment",
grepl("economic-inequality", topic) ~ "Income Inequality",
grepl("income-inequality", topic) ~ "Income Inequality",
grepl("female-president", topic) ~ "Female President",
grepl("trade", topic) ~ "Trade",
grepl("economy", topic) ~ "Economy",
grepl("sexism", topic) ~ "Sexism",
grepl("middle-east policy", topic) ~ "Foreign Policy",
grepl("opioids", topic) ~ "Opioids",
grepl("party-strategy", topic) ~ "Party Strategy",
grepl("public-service", topic) ~ "Public Service",
grepl("tech-companies", topic) ~ "Tech Companies",
grepl("white-supremacist violence", topic) ~ "White-Supremacy",
grepl("womens-issues", topic) ~ "Women's Rights",
topic == "" ~ "Other",
TRUE ~ topic
)
) %>%
filter(
!is.na(timestamp),
speaker != "",
speaker != "Moderator"
) %>%
as_tibble() -> feb_19_day_1
bind_rows(
jun_day_1,
jun_day_2,
@ -551,7 +628,8 @@ bind_rows(
nov_day_1,
dec_day_1,
jan_20_day_1,
feb_07_day_1
feb_07_day_1,
feb_19_day_1
) %>%
mutate(topic = case_when(
grepl("elect.*form", topic, ignore.case = TRUE) ~ "Election Reform",


BIN
data/debates2019.rda View File


Loading…
Cancel
Save