In [10]:

shhh <- function(expr) suppressPackageStartupMessages(suppressWarnings(suppressMessages(expr)))
shhh({
    library(tidyverse); library(glue); library(lubridate); library(scales)
})

Double Logged of Dialog Close Events¶

Task

Dialog close events (dialog-insert, dialog-done, dialog-abort, etc) were all getting logged twice for node events (a node is any component that is not editable withing the page. Said another way: any component that requires a dialog to edit (e.g. reference, image, template).

The fix was deployed on 8 October 2020.

Checks:

Verify that dialog events are only logging once in the database
There should be more open events then close events following the deployment date

In [113]:

# Collect vefu events

query <-
"SELECT
 date_format(dt, 'yyyy-MM-dd') AS date,
 event.editingSessionID AS sessionID,
 event.feature AS feature,
 If(event.action LIKE 'dialog-%', 'dialog_close_event', 'window_open_event') AS event_type,
 COUNT(*) as events
FROM
event.visualeditorfeatureuse
WHERE
-- fix deployed on 6 October 2020
    year = 2020 
    AND month = 10
-- review all dialog and window open actions
    AND (event.action IN ('dialog-done', 'dialog-insert', 'dialog-abort', 'dialog-remove') OR 
    event.action LIKE 'window-open-from-%')
-- remove bots
    AND useragent.is_bot = false 
GROUP BY
  date_format(dt, 'yyyy-MM-dd'),
  event.editingSessionID,
 event.feature,
  If(event.action LIKE 'dialog-%', 'dialog_close_event', 'window_open_event') 
"

In [114]:

collect_vefu_events <-  wmfdata::query_hive(query)

Don't forget to authenticate with Kerberos using kinit

In [115]:

collect_vefu_events$date <- as.Date(collect_vefu_events$date, format = "%Y-%m-%d")

In [139]:

duplicate_events_daily <- collect_vefu_events %>%
    spread(event_type, events) %>%
#review events with both dialog close and window-open-event
        filter(!is.na(dialog_close_event),
               !is.na(window_open_event),
               #review only node features
              feature %in% c('citoid', 'transclusion', 'media', 'reference', 'cite-book', 'cite-web', 'mwcite')) %>%
        mutate(is_duplicate = ifelse(dialog_close_event <= window_open_event, 'false', 'true')) %>% 
        arrange(date)
           

head(duplicate_events_daily,50)

#Need to investigate mWsave and cite events more - we don't start recording until the 13th or 14th with dialog-close-event and window-open events

A data.frame: 50 × 6
	date	sessionid	feature	dialog_close_event	window_open_event	is_duplicate
	<date>	<chr>	<chr>	<int>	<int>	<chr>
1	2020-10-01	0001a213d8b2c397591e	citoid	2	1	true
2	2020-10-01	001db27a38fc96a5a4ad	citoid	2	1	true
3	2020-10-01	003b70fd55795a20d6ab	transclusion	2	1	true
4	2020-10-01	00611273291617e0f1e2	transclusion	2	1	true
5	2020-10-01	00717b7ab5e1fcd56ac3	transclusion	2	1	true
6	2020-10-01	007263d77c643caec47b	transclusion	2	1	true
7	2020-10-01	0082a616346bef8d958c	citoid	2	1	true
8	2020-10-01	008344302959c32d8afc	citoid	2	1	true
9	2020-10-01	008879198041930f8cef	citoid	2	1	true
10	2020-10-01	00a97972c83cba8fa9a7	transclusion	2	1	true
11	2020-10-01	00af6c0be4a5b46f8590	transclusion	4	3	true
12	2020-10-01	00b226620a0ab6c88663	citoid	2	1	true
13	2020-10-01	00b26477187d073dc0ea	transclusion	2	2	false
14	2020-10-01	00c78b79be440da89f66	transclusion	2	1	true
15	2020-10-01	00c84f144789eaa4671c	transclusion	4	2	true
16	2020-10-01	00d8539350c95a15fcc9	transclusion	4	2	true
17	2020-10-01	00e75f40bf8976a9093f	transclusion	10	5	true
18	2020-10-01	0111834093cdea5b1a69	transclusion	2	1	true
19	2020-10-01	0111f9b0884b1ef77673	transclusion	2	1	true
20	2020-10-01	0114802093feacf36b2c	citoid	2	1	true
21	2020-10-01	0115f0e0efa30e31072b	citoid	2	1	true
22	2020-10-01	012abb5076c543fa6474	transclusion	6	3	true
23	2020-10-01	0135dfd71d231e4da210	transclusion	2	1	true
24	2020-10-01	0141f400de49e9608cf7	citoid	2	1	true
25	2020-10-01	01486f68ca71bb676d5f	citoid	2	1	true
26	2020-10-01	014b2d0dbefc4ca6ca40	transclusion	2	3	false
27	2020-10-01	0152f6e144390ccecb83	transclusion	2	1	true
28	2020-10-01	0153bba7d535780c0880	media	2	1	true
29	2020-10-01	015754417a186e0de223	transclusion	1	1	false
30	2020-10-01	01589699a9a3ccb12b61	transclusion	4	2	true
31	2020-10-01	015f10017d919561dc98	transclusion	2	1	true
32	2020-10-01	015f6380c19b3b40165c	transclusion	2	1	true
33	2020-10-01	0168f07f56a34cdec016	media	2	1	true
34	2020-10-01	0175d2d0cd509d3a1bc3	citoid	10	5	true
35	2020-10-01	0175d2d0cd509d3a1bc3	transclusion	6	3	true
36	2020-10-01	017f07d321c092735915	transclusion	2	2	false
37	2020-10-01	01883dc6f39479039a89	transclusion	2	1	true
38	2020-10-01	01c45ce282a8a4dbc329	citoid	2	1	true
39	2020-10-01	01ca2d932be004a17b36	citoid	2	2	false
40	2020-10-01	01d295e794156fc2eb0b	transclusion	2	4	false
41	2020-10-01	01dbe65f27c9aa4397af	citoid	4	2	true
42	2020-10-01	01e0b31553cd89772ebf	reference	2	1	true
43	2020-10-01	01e2962b8a6b2a9b58d8	transclusion	2	1	true
44	2020-10-01	020948a1d0b41146a0c7	citoid	2	1	true
45	2020-10-01	021231b0c863367074b3	citoid	6	3	true
46	2020-10-01	0221ece048c519cde207	citoid	4	2	true
47	2020-10-01	0221ece048c519cde207	transclusion	2	1	true
48	2020-10-01	0223733a8ae318d8ef96	media	2	1	true
49	2020-10-01	024981033d11a010c9f0	transclusion	2	1	true
50	2020-10-01	0257378ee716c70954a0	citoid	2	1	true

In [147]:

# Plot chart of duplicate events

p <- duplicate_events_daily %>%
    group_by(date, feature) %>%
    summarise(num_close_events = sum(dialog_close_event)) %>%
    ggplot(aes(x=date, y= num_close_events, color = feature)) +
    geom_line(size = 1.5) +
    scale_y_continuous() +
    labs (y = "Number of close events per day",
          x = "Date",
         title = "Daily dialog close events by feature")  +
     theme_bw() +
   theme(
        plot.title = element_text(hjust = 0.5),
        text = element_text(size=16),
        axis.text.x = element_text(angle=45, hjust=1),
       legend.position = 'bottom') 
 
p

options(repr.plot.width = 15, repr.plot.height = 10)

`summarise()` regrouping output by 'date' (override with `.groups` argument)

In [148]:

ggsave("figures/daily_dialog_close_events.png", p, width = 16, height = 8, units = "in", dpi = 300)

In [145]:

# plot duplicate events

p <- duplicate_events_daily %>%
    group_by(date, feature, is_duplicate) %>%
    summarise(num_close_events = sum(dialog_close_event)) %>%
    ggplot(aes(x=date, y= num_close_events, color = is_duplicate)) +
    facet_wrap(~feature) +
    geom_line(size = 1.5) +
    scale_y_continuous() +
    labs (y = "Number of duplicate events per day",
          x = "Date",
         title = "Daily duplicate close events by feature")  +
     theme_bw() +
   theme(
        plot.title = element_text(hjust = 0.5),
        text = element_text(size=16),
        axis.text.x = element_text(angle=45, hjust=1),
       legend.position = 'bottom') 
 
p

`summarise()` regrouping output by 'date', 'feature' (override with `.groups` argument)

In [146]:

ggsave("figures/daily_duplicate_events.png", p, width = 16, height = 8, units = "in", dpi = 300)

In [ ]: