library(tidyverse)
library(dplyr)
library(lubridate)
library(stringr)
c1_event_date <- as.Date("2013-07-01")
c2_event_date <- as.Date("2013-08-28")
c3_event_date <- as.Date("2015-07-02")
relative_week <- function(date, ref_date) {
  as.integer(as.numeric(difftime(date, ref_date, units = "days")) %/% 7)
}
#get count data for the repositories
ve_csv <-"~/121325_work/121225_vd_data/extension_VisualEditor_2000-01-01_to_2016-12-31.csv"
ve_df <- read.csv(ve_csv, header = TRUE) 
ve_df <- ve_df |>
  mutate(commit_date = ymd_hms(commit_date)) |>
  mutate(isAuthorWMF = case_when(
    grepl("krinkle@fastmail\\.com", author_email, ignore.case = TRUE) ~ "TRUE",
    grepl("@wikimedia\\.org", author_email, ignore.case = TRUE) ~ "TRUE",
    grepl("@wikimedia\\.de", author_email, ignore.case = TRUE) ~ "TRUE",
    grepl("@gerrit\\.wikimedia\\.org", author_email, ignore.case = TRUE) ~ "Gerrit",
    TRUE ~ "FALSE"
  ))|>
  group_by(author_email) |> 
  mutate(
    isFirstCommit = commit_date == min(commit_date)
  ) |>
  ungroup()

ve_counts <- ve_df |>
  mutate(week_index = relative_week(commit_date, c1_event_date)) |>
  group_by(week_index, isAuthorWMF, isFirstCommit)|>
  summarise(count = n(), .groups = 'drop')|>
  filter(week_index >= -33 & week_index <= 13) |>
  mutate(source = 'c1')

ve_counts <- ve_counts |>
  filter(isAuthorWMF != "Gerrit")|>
  filter(source == 'c1') 

ve_new_commits_created <- ggplot(
  ve_counts,
  aes(
    x=week_index,
    y=count,
    fill=isAuthorWMF,
    color=isFirstCommit
  )
) + 
  facet_grid(source ~ ., 
             scales = "free_y",
             labeller = labeller(source = c("c1" = "VisualEditor", 
                                            "c2" = "HTTPS-login", 
                                            "c3" = "HTTP-deprecation"))) +  
  geom_col(position = position_dodge(width = 0.9), width = 0.8) +
  geom_vline(data = ve_counts |> filter(source == "c1"), 
             aes(xintercept = -29), 
             linetype = "dotted", color = "black", linewidth = 0.5) +
  geom_vline(data = ve_counts|> filter(source == "c1"), 
             aes(xintercept = -9), 
             linetype = "dotted", color = "black", linewidth = 0.5) +
  geom_vline(data = ve_counts |> filter(source == "c1"), 
             aes(xintercept = -4), 
             linetype = "3313", color = "black", linewidth = 0.5) +
  geom_vline(xintercept = 0, linetype = "dashed", color = "black", linewidth = 0.5) +
  geom_text(
    data = subset(ve_counts, source == "c1" & week_index == 6),
    aes(x=week_index, y=120, label='Opt-out deployment'),
    size = 2.5) + 
  geom_text(
    data = subset(ve_counts, source == "c1" & week_index == -33),
    aes(x=week_index, y=120, label='Opt-in Testing'),
    size = 2.5) + 
  geom_text(
    data = subset(ve_counts, source == "c1" & week_index == -12),
    aes(x=week_index, y=20, label='Deployment Announcement'),
    size = 2.5) + 
  theme_minimal() + 
  scale_fill_viridis_d() + 
  labs(
    x = "Weeks from Feature Deployment", 
    y = "Count of mediawiki/core commits Created", 
    fill = "Commit Author Affiliated with WMF?"
  ) + 
  theme(legend.position = "top")
ve_new_commits_created