library(tidyverse) library(dplyr) library(lubridate) 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) } known_affil_emails <- c("krinkle@fastmail.com", "roan.kattouw@gmail.com", "trevorparscal@gmail.com", "krinklemail@gmail.com", "moriel@gmail.com") #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(isAuthorWMF = case_when( (author_email %in% known_affil_emails) ~ "TRUE", grepl("l10n-bot@translatewiki\\.net", author_email, ignore.case = TRUE) ~ "localization", 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)|> 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)|> 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)|> summarise(count = n(), .groups = 'drop')|> filter(week_index >= -83 & week_index <= 13) |> mutate(source = 'c3') #collate and save core_weekly <- rbind(c1_core_weekly, c2_core_weekly, c3_core_weekly) write.csv(core_weekly, "~/121325_work/aggregate_dfs/121325_core_weekly.csv", row.names = FALSE) #operations wmfconfig_csv <-"~/121325_work/121225_vd_data/mediawiki-config_2010-01-01_to_2024-12-31.csv" wmfconfig_df <- read.csv(wmfconfig_csv, header = TRUE) wmfconfig_df <- wmfconfig_df |> mutate(commit_date = ymd_hms(commit_date)) |> mutate(isAuthorWMF = case_when( (author_email %in% known_affil_emails) ~ "TRUE", grepl("l10n-bot@translatewiki\\.net", author_email, ignore.case = TRUE) ~ "localization", 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_config_weekly <- wmfconfig_df |> mutate(week_index = relative_week(commit_date, c1_event_date)) |> group_by(week_index, isAuthorWMF)|> summarise(count = n(), .groups = 'drop')|> filter(week_index >= -33 & week_index <= 13) |> mutate(source = 'c1') c2_config_weekly <- wmfconfig_df |> mutate(week_index = relative_week(commit_date, c2_event_date)) |> group_by(week_index, isAuthorWMF)|> summarise(count = n(), .groups = 'drop')|> filter(week_index >= -104 & week_index <= 13) |> mutate(source = 'c2') c3_config_weekly <- wmfconfig_df |> mutate(week_index = relative_week(commit_date, c3_event_date)) |> group_by(week_index, isAuthorWMF)|> summarise(count = n(), .groups = 'drop')|> filter(week_index >= -83 & week_index <= 13) |> mutate(source = 'c3') #collate and save config_weekly <- rbind(c1_config_weekly, c2_config_weekly, c3_config_weekly) write.csv(config_weekly, "~/121325_work/aggregate_dfs/121325_config_weekly.csv", row.names = FALSE) #puppet puppet_csv <-"~/121325_work/121225_vd_data/puppet_2000-01-01_to_2016-12-31.csv" puppet_df <- read.csv(puppet_csv, header = TRUE) puppet_df <- puppet_df |> mutate(commit_date = ymd_hms(commit_date)) |> mutate(isAuthorWMF = case_when( (author_email %in% known_affil_emails) ~ "TRUE", grepl("l10n-bot@translatewiki\\.net", author_email, ignore.case = TRUE) ~ "localization", 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_puppet_weekly <- puppet_df |> mutate(week_index = relative_week(commit_date, c1_event_date)) |> group_by(week_index, isAuthorWMF)|> summarise(count = n(), .groups = 'drop')|> filter(week_index >= -33 & week_index <= 13) |> mutate(source = 'c1') c2_puppet_weekly <- puppet_df |> mutate(week_index = relative_week(commit_date, c2_event_date)) |> group_by(week_index, isAuthorWMF)|> summarise(count = n(), .groups = 'drop')|> filter(week_index >= -104 & week_index <= 13) |> mutate(source = 'c2') c3_puppet_weekly <- puppet_df |> mutate(week_index = relative_week(commit_date, c3_event_date)) |> group_by(week_index, isAuthorWMF)|> summarise(count = n(), .groups = 'drop')|> filter(week_index >= -83 & week_index <= 13) |> mutate(source = 'c3') puppet_weekly <- rbind(c1_puppet_weekly, c2_puppet_weekly, c3_puppet_weekly) write.csv(puppet_weekly, "~/121325_work/aggregate_dfs/121325_puppet_weekly.csv", row.names = FALSE) #extension_ve 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( (author_email %in% known_affil_emails) ~ "TRUE", grepl("l10n-bot@translatewiki\\.net", author_email, ignore.case = TRUE) ~ "localization", 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_ve_weekly <- ve_df |> mutate(week_index = relative_week(commit_date, c1_event_date)) |> group_by(week_index, isAuthorWMF)|> summarise(count = n(), .groups = 'drop')|> filter(week_index >= -33 & week_index <= 13) |> mutate(source = 'c1') write.csv(c1_ve_weekly, "~/121325_work/aggregate_dfs/121325_ve_weekly.csv", row.names = FALSE)