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)
}
contains_http_but_not_url <- function(text) {
  # Handle NA values explicitly
  ifelse(
    is.na(text),
    FALSE,
    str_detect(text, "http") & 
      !str_detect(text, "://") |
      str_detect(text, "login") |
      str_detect(text, "ssl") |
      str_detect(text, "tls") |
      (str_detect(text, "cert") & !str_detect(text, "certain"))
  )
}

#get count data for the repositories
#core 
core_csv <-"~/121325_work/121225_vd_data/core_2010-01-01_to_2024-12-31.csv"
core_df <- read.csv(core_csv, header = TRUE) 
core_df <- core_df |>
  mutate(commit_date = ymd_hms(commit_date)) |>
  mutate(code_location = case_when(
    grepl("login", diff_info, ignore.case = TRUE) ~ "login_in_location",
    grepl("auth", diff_info, ignore.case = TRUE) ~ "auth_in_location",
    grepl("security", diff_info, ignore.case = TRUE) ~ "security",
    TRUE ~ "Other"
  ))|>
  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"
  ))

c1_core_weekly <- core_df |>
  mutate(week_index = relative_week(commit_date, c1_event_date)) |>
  group_by(week_index, isAuthorWMF, code_location)|>
  summarise(count = n(), .groups = 'drop')|>
  filter(week_index >= -33 & week_index <= 13) |>
  mutate(source = 'c1')

c2_core_weekly <- core_df |>
  mutate(week_index = relative_week(commit_date, c2_event_date)) |>
  group_by(week_index, isAuthorWMF, code_location)|>
  summarise(count = n(), .groups = 'drop')|>
  filter(week_index >= -104 & week_index <= 13) |>
  mutate(source = 'c2')

c3_core_weekly <- core_df |>
  mutate(week_index = relative_week(commit_date, c3_event_date)) |>
  group_by(week_index, isAuthorWMF, code_location)|>
  summarise(count = n(), .groups = 'drop')|>
  filter(week_index >= -83 & week_index <= 13) |>
  mutate(source = 'c3')
#collate and save
core_counts <- rbind(c1_core_weekly, c2_core_weekly, c3_core_weekly)

core_counts <- core_counts |>
  filter(isAuthorWMF != "Gerrit") |>
  filter(code_location != "Other") 
core_commits_created <- ggplot(
  core_counts,
  aes(
    x=week_index,
    y=count, 
  )
) + 
  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 = core_counts |> filter(source == "c1"), 
             aes(xintercept = -29), 
             linetype = "dotted", color = "black", linewidth = 0.5) +
  geom_vline(data = core_counts |> filter(source == "c1"), 
             aes(xintercept = -9), 
             linetype = "dotted", color = "black", linewidth = 0.5) +
  geom_vline(data = core_counts |> filter(source == "c1"), 
             aes(xintercept = -4), 
             linetype = "3313", color = "black", linewidth = 0.5) +
  geom_vline(data = core_counts |> filter(source == "c2"), 
             aes(xintercept = -99), 
             linetype = "dotted", color = "black", linewidth = 0.5) +
  geom_vline(data = core_counts |> filter(source == "c2"), 
             aes(xintercept = -4),   
             linetype = "3313", color = "black", linewidth = 0.5) +
  geom_vline(data = core_counts |> filter(source == "c3"), 
             aes(xintercept = -97), 
             linetype = "dotted", color = "black", linewidth = 0.5) +
  geom_vline(data = core_counts |> filter(source == "c3"), 
             aes(xintercept = -3), 
             linetype = "3313", color = "black", linewidth = 0.5) +
  geom_vline(xintercept = 0, linetype = "dashed", color = "black", linewidth = 0.5) +
  geom_text(
    data = subset(core_counts, source == "c1" & week_index == 6),
    aes(x=week_index, y=120, label='Opt-out deployment'),
    size = 2.5) + 
  geom_text(
    data = subset(core_counts, source == "c1" & week_index == -33),
    aes(x=week_index, y=120, label='Opt-in Testing'),
    size = 2.5) + 
  geom_text(
    data = subset(core_counts, source == "c2" & 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")
core_commits_created