Code for blogpost https://www.hvitfeldt.me/2018/01/01/recreate-sunshine-report/

sunshine-report.R

library(rvest)
library(tidyverse)
library(lubridate)
library(glue)
#library(ehlib) # devtools::install_github("EmilHvitfeldt/ehlib")

str_between <- function(string, start, end) {
  stringr::str_extract(string,
                       stringr::str_c(start, '(.*?)', end, collapse = '')) %>%
    stringr::str_replace(start, "") %>%
    stringr::str_replace(end, "")
}

str_before <- function(string, pattern) {
  stringr::str_extract(string, stringr::str_c(".+?(?=", pattern, ")"))
}

all_dates_in <- function(year) {
  if(ymd(glue::glue("{year}0101")) > as.Date(Sys.time())) {
    stop("Please select a past or current year.")
  }
  
  start <- ymd(glue::glue("{year}0101"))
  
  if(as.Date(Sys.time()) > ymd(glue::glue("{year}1231"))) {
    end <- ymd(glue::glue("{year}1231"))
  } else {
    end <- as.Date(Sys.time())
  }
  
  seq(start, end, by = "day")
}

airport_code <- "KCQT"

full_data <- map(dates, ~{
  weather_data_html(.x, airport_code)
  Sys.sleep(5)
  cat(month(dates), "/", mday(dates), "\n", sep = "")
  })

map_lgl(full_data, ~ is.null(.x$error))

ampm_minutes <- function(x) {
  as.numeric(str_between(x, ":", " ")) +
  as.numeric(str_replace(str_before(x, ":"), "12", "0")) * 60 +
  60 * 12 * str_detect(x, "PM")
}

data_wrangling <- function(html_url, date) {
  
  # Sun rise time
    sun_rise <- html_url %>%
    html_nodes('div[id="astronomy-mod"] table') %>%
    html_text() %>%
    .[1] %>%
    str_between("Time\n\t\t", "\n\t\t")
  # Sun set time
  sun_set <- html_url %>%
    html_nodes('div[id="astronomy-mod"] table') %>%
    html_text() %>%
    .[1] %>%
    str_between("\n\t\t", "\n\t\tCivil")

  # Table
  table <- html_url %>%
    html_nodes('table[id="obsTable"]') %>%
    html_table() %>% 
    .[[1]]
  
  # Time column standardization 
  is_daylight <- any("Time (PDT)" == names(table),
                     "Time (MDT)" == names(table),
                     "Time (CDT)" == names(table),
                     "Time (EDT)" == names(table))
  
  time_names <- str_c("Time", c(" (PDT)", " (MDT)", " (CDT)", " (EDT)",
                                " (PST)", " (MST)", " (CST)", " (EST)"))
  
  names(table) <- if_else(names(table) %in% time_names,
                          "Time",
                          names(table))
  
  table %>%
    mutate(sun_set = sun_set,
           sun_rise = sun_rise,
           date = date,
           yday = yday(date), 
           day_minutes = ampm_minutes(Time) - is_daylight * 60,
           set_minutes = ampm_minutes(sun_set) - is_daylight * 60,
           rise_minutes = ampm_minutes(sun_rise) - is_daylight * 60,
           sun_up = day_minutes > (rise_minutes + 90) & 
                    day_minutes < (set_minutes - 30))
}

full_data2 <- map2_df(full_data[-303], dates[-303], ~ .x$result %>%
                      data_wrangling(.y))

x_axis <- dates %>% month() %>% table() %>% cumsum()
names(x_axis) <- month.abb[1:12]

y_axis <- 1:24 * 60
names(y_axis) <- str_c(c(12, rep(1:12, 2, length.out = 23)), 
                       rep(c("AM", "PM"), each = 12))

full_data2 %>%
  mutate(con = Conditions == "Clear",
         day_minutes2 = lag(day_minutes)) %>%
  filter(sun_up) %>%
  ggplot(aes(fill = con)) +
  geom_rect(aes(xmin = yday, xmax = yday + 1,
                ymin = day_minutes, ymax = day_minutes2)) +
  geom_line(aes(yday, set_minutes)) +
  geom_line(aes(yday, rise_minutes)) +
  scale_fill_manual(values = c("grey40", NA)) +
  theme_minimal() +
  guides(fill = "none") +
  theme(
  panel.grid.major.y = element_blank(),
  panel.grid.minor.y = element_blank(),
  panel.grid.minor.x = element_blank(), 
  axis.text.x.bottom = element_text(hjust = 1.7)
  ) +
  scale_x_continuous(breaks = x_axis, position = "right") +
  scale_y_continuous(breaks = y_axis, limits = c(0, 24 * 60)) +
  labs(x = NULL, y = NULL, title = "Sunshine report of Los Angeles 2017")