library(tidyverse)
library(rvest)
library(janitor)
library(lubridate)
library(directlabels)
library(ggrepel)
library(ggthemes)

setwd("Development/r/2103.nhl")

leafs <- read_html("https://en.wikipedia.org/wiki/2020%E2%80%9321_Toronto_Maple_Leafs_season") %>%
  html_table(fill = TRUE) %>%
  .[[4]] %>%
  row_to_names(row_number = 2) %>%
  select(2,4) %>%
  filter(!Date %in% c("Game", "Date")) %>%
  filter(!grepl("schedule", Date)) %>%
  filter(Score != "–") %>%
  separate(Score, c("HomeScore", "VisitorScore"), sep = "–") %>%
  mutate(result = case_when(HomeScore > VisitorScore ~ 1,
                            TRUE ~ -1)) %>%
  mutate(Leafs = cumsum(result)) %>%
  select(Date, Leafs) %>%
  mutate(Date = mdy(paste(Date, "2021", sep = " ")))

jets <- read_html("https://en.wikipedia.org/wiki/2020%E2%80%9321_Winnipeg_Jets_season") %>%
  html_table(fill = TRUE) %>%
  .[[4]] %>%
  row_to_names(row_number = 3) %>%
  select(2:5) %>%
  filter(!Date %in% c("Game", "Date")) %>%
  filter(!grepl("Road|Legend", Date)) %>%
  filter(grepl("–", Score)) %>%
  separate(Score, c("VisitorScore", "HomeScore"), sep = "–") %>%
  mutate(result = case_when(HomeScore > VisitorScore & Home == "Winnipeg" ~ 1,
                            HomeScore > VisitorScore & Home != "Winnipeg" ~ -1,
                            HomeScore < VisitorScore & Visitor != "Winnipeg" ~ -1,
                            HomeScore < VisitorScore & Visitor == "Winnipeg" ~ 1)) %>%
  mutate(Jets = cumsum(result)) %>%
  select(Date, Jets) %>%
  mutate(Date = mdy(paste(Date, "2021", sep = " ")))


oilers <- read_html("https://en.wikipedia.org/wiki/2020%E2%80%9321_Edmonton_Oilers_season") %>%
  html_table(fill = TRUE) %>%
  .[[3]] %>%
  row_to_names(row_number = 3) %>%
  select(2:5) %>%
  filter(!Date %in% c("Game", "Date")) %>%
  filter(!grepl("Road|Legend", Date)) %>%
  filter(grepl("–", Score)) %>%
  separate(Score, c("VisitorScore", "HomeScore"), sep = "–") %>%
  mutate(result = case_when(HomeScore > VisitorScore & Home == "Edmonton" ~ 1,
                            HomeScore > VisitorScore & Home != "Edmonton" ~ -1,
                            HomeScore < VisitorScore & Visitor != "Edmonton" ~ -1,
                            HomeScore < VisitorScore & Visitor == "Edmonton" ~ 1)) %>%
  mutate(Oilers = cumsum(result)) %>%
  select(Date, Oilers) %>%
  mutate(Date = mdy(paste(Date, "2021", sep = " ")))

habs <- read_html("https://en.wikipedia.org/wiki/2020%E2%80%9321_Montreal_Canadiens_season") %>%
  html_table(fill = TRUE) %>%
  .[[3]] %>%
  row_to_names(row_number = 3) %>%
  select(2:5) %>%
  filter(!Date %in% c("Game", "Date")) %>%
  filter(!grepl("Road|Legend", Date)) %>%
  filter(grepl("–", Score)) %>%
  separate(Score, c("VisitorScore", "HomeScore"), sep = "–") %>%
  mutate(result = case_when(HomeScore > VisitorScore & Home == "Montreal" ~ 1,
                            HomeScore > VisitorScore & Home != "Montreal" ~ -1,
                            HomeScore < VisitorScore & Visitor != "Montreal" ~ -1,
                            HomeScore < VisitorScore & Visitor == "Montreal" ~ 1)) %>%
  mutate(Habs = cumsum(result)) %>%
  select(Date, Habs) %>%
  mutate(Date = mdy(paste(Date, "2021", sep = " ")))

flames <- read_html("https://en.wikipedia.org/wiki/2020%E2%80%9321_Calgary_Flames_season") %>%
  html_table(fill = TRUE) %>%
  .[[3]] %>%
  row_to_names(row_number = 3) %>%
  select(2:5) %>%
  filter(!Date %in% c("Game", "Date")) %>%
  filter(!grepl("Road|Legend", Date)) %>%
  filter(grepl("–", Score)) %>%
  separate(Score, c("VisitorScore", "HomeScore"), sep = "–") %>%
  mutate(result = case_when(HomeScore > VisitorScore & Home == "Calgary" ~ 1,
                            HomeScore > VisitorScore & Home != "Calgary" ~ -1,
                            HomeScore < VisitorScore & Visitor != "Calgary" ~ -1,
                            HomeScore < VisitorScore & Visitor == "Calgary" ~ 1)) %>%
  mutate(Flames = cumsum(result)) %>%
  select(Date, Flames) %>%
  mutate(Date = mdy(paste(Date, "2021", sep = " ")))

canucks <- read_html("https://en.wikipedia.org/wiki/2020%E2%80%9321_Vancouver_Canucks_season") %>%
  html_table(fill = TRUE) %>%
  .[[3]] %>%
  row_to_names(row_number = 3) %>%
  select(2:5) %>%
  filter(!Date %in% c("Game", "Date")) %>%
  filter(!grepl("Road|Legend", Date)) %>%
  filter(grepl("–", Score)) %>%
  separate(Score, c("VisitorScore", "HomeScore"), sep = "–") %>%
  mutate(result = case_when(HomeScore > VisitorScore & Home == "Vancouver" ~ 1,
                            HomeScore > VisitorScore & Home != "Vancouver" ~ -1,
                            HomeScore < VisitorScore & Visitor != "Vancouver" ~ -1,
                            HomeScore < VisitorScore & Visitor == "Vancouver" ~ 1)) %>%
  mutate(Canucks = cumsum(result)) %>%
  na.omit() %>%
  select(Date, Canucks) %>%
  mutate(Date = mdy(paste(Date, "2021", sep = " ")))

senators <- read_html("https://en.wikipedia.org/wiki/2020%E2%80%9321_Ottawa_Senators_season") %>%
  html_table(fill = TRUE) %>%
  .[[3]] %>%
  row_to_names(row_number = 3) %>%
  select(2:5) %>%
  filter(!Date %in% c("Game", "Date")) %>%
  filter(!grepl("Road|Legend", Date)) %>%
  filter(grepl("–", Score)) %>%
  separate(Score, c("VisitorScore", "HomeScore"), sep = "–") %>%
  mutate(result = case_when(HomeScore > VisitorScore & Home == "Ottawa" ~ 1,
                            HomeScore > VisitorScore & Home != "Ottawa" ~ -1,
                            HomeScore < VisitorScore & Visitor != "Ottawa" ~ -1,
                            HomeScore < VisitorScore & Visitor == "Ottawa" ~ 1)) %>%
  mutate(Senators = cumsum(result)) %>%
  select(Date, Senators) %>%
  mutate(Date = mdy(paste(Date, "2021", sep = " ")))


canucks %>%
  full_join(flames) %>%
  full_join(habs) %>%
  full_join(jets) %>%
  full_join(leafs) %>%
  full_join(oilers) %>%
  full_join(senators) %>%
  arrange(Date) %>%
  gather(Team, record, 2:8) %>%
  as_tibble() %>%
  na.omit() %>%
  group_by(Team) %>%
  mutate(label = if_else(Date == max(Date),
                         as.character(Team),
                         NA_character_)) %>%
  mutate(col = case_when(Team == "Leafs" ~ "Leafs",
                         Team == "Senators" ~ "Senators",
                         TRUE ~ "Other")) %>%
  ggplot(aes(Date, record, color = col, group = Team, linetype = Team)) +
  geom_point() +
  geom_path() +
  geom_hline(yintercept = 0) +
  geom_label_repel(aes(label = label), nudge_x = 1, na.rm = TRUE) +
  labs(x = "2021 NHL Season to Date",
       y = "",
       title = "A tale of seven cities: NHL North records to date",
       subtitle = "Cumulative games above / below .500 since January") +
  scale_color_manual(values = c("blue", "grey", "red")) +
  theme_bw() +
  theme(legend.position = "none")

ggsave("nhlplot.png", width = 10, height = 6)