improve sql query for tag denylist

Unfortunately, excludeTags defined in the _config.yml
cannot be used by the SQL Queries defined on a per-page
basis, since the globs are dynamic in nature, as applied
by Jekyll, so these are not picked up by our
current jekyll-sqlite implementation for generating
the query bind params.

For now, we hardcode the sql deny list, but will file
an issue on the jekyll-sqlite side to see if I ca
improve it there.

_config.yml

@@ -34,6 +34,11 @@ defaults:
  layout: events
  permalink: /:path/:basename/
  subscribe: true
+ # While this is available under the page
+ # this is dynamic, and not available for use
+ # in sql queries, or copied to external layouts
+ # TODO: Fix jekyll-sqlite and multi-page layouts to support this
+ excludeTags: '["BUSINESS", "LOW-QUALITY", "NOTINBLR", "DANDIYA", "WOOWOO"]'
  layouts:
  - layout: icalendar.ics
  permalink: /:path/:basename/cal.ics
@@ -43,65 +48,37 @@ defaults:
  - data: events
  file: events.db
  query: |
- SELECT *
- FROM events
- WHERE EXISTS (
- SELECT 1
- FROM json_each(:tags) AS tags
- WHERE tags.value IN (
- SELECT value
- FROM json_each(event_json->'$.keywords')
- )
- AND tags.value NOT IN ("WOOWOO", "LOW-QUALITY", "BUSINESS", "NOTINBLR", "DANDIYA")
+ WITH 
+ allowlist_tags AS (
+ SELECT value FROM json_each(:tags)
+ ),
+ denylist_tags AS (
+ SELECT value FROM json_each('["BUSINESS", "LOW-QUALITY", "NOTINBLR", "DANDIYA", "WOOWOO"]')
  )
- ORDER BY event_json -> '$.startDate'
- ;
- - data: today
- file: events.db
- query: |
- SELECT event_json, url, date(json_extract(event_json,'$.startDate')) as startDate, date(json_extract(event_json,'$.endDate')) as endDate
+ SELECT *
  FROM events
- WHERE EXISTS (
+ WHERE 
+ -- Check if no denylist tags are present
+ NOT EXISTS (
  SELECT 1
- FROM json_each(:tags) AS tags
- WHERE tags.value IN (
-  SELECT value
-  FROM json_each(event_json->'$.keywords')
+ FROM denylist_tags
+ WHERE denylist_tags.value IN (
+ SELECT value
+ FROM json_each(event_json->'$.keywords')
  )
- ) AND (
- (startDate >= date('now', 'localtime') AND endDate <= date('now', 'localtime'))
- OR 
- (startDate <= date('now', 'localtime') AND endDate >= date('now', '+1 day', 'localtime'))
- )
- ORDER by startDate;
- 
- - data: tomorrow
- file: events.db
- query: |
- SELECT event_json, url, date(json_extract(event_json,'$.startDate')) as startDate, date(json_extract(event_json,'$.endDate')) as endDate
- FROM events
- WHERE EXISTS (
+ )
+ AND
+ -- Check if at least one allowlist tag is present
+ EXISTS (
  SELECT 1
- FROM json_each(:tags) AS tags
- WHERE tags.value IN (
-  SELECT value
-  FROM json_each(event_json->'$.keywords')
+ FROM allowlist_tags
+ WHERE allowlist_tags.value IN (
+ SELECT value
+ FROM json_each(event_json->'$.keywords')
  )
- -- Find all events either starting and ending tomorrow
- -- or starting today or earlier, and end ending after tomorrow
- ) AND (
- (startDate >= date('now', '+1 day', 'localtime') AND endDate <= date('now', '+1 day', 'localtime'))
- OR 
- (startDate <= date('now', '+1 day', 'localtime') AND endDate >= date('now', '+2 day', 'localtime'))
- )
- ORDER by startDate
+ )
+ ORDER BY event_json -> '$.startDate';
 sqlite:
- - data: tags.events
- file: events.db
- query: |
- SELECT event_json, event_json -> '$.keywords' as keywords from events
- WHERE EXISTS (SELECT 1 FROM json_each(keywords) WHERE value = :id)
-
  - data: geo_events
  file: events.db
  query: |
@@ -110,67 +87,3 @@ sqlite:
  event_json->'$.location.geo.longitude' as lng,
  event_json-> '$.location.name' as name
  from events WHERE event_json -> '$.location.geo' IS NOT NULL
-
- - data: events
- file: events.db
- query: |
- SELECT substr(url, 9, instr(substr(url, 9), '/') - 1) AS domain,
- url,
- json_extract(event_json, '$.name') AS name,
- json_extract(event_json, '$.description') AS description,
- json_extract(event_json, '$.keywords') AS keywords,
- json_extract(event_json, '$.startDate') AS startDate,
- json_extract(event_json, '$.eventStatus') AS eventStatus,
- json_extract(event_json, '$.location.name') AS location,
- json_extract(event_json, '$.organizer.name') AS organizer
-
- FROM events WHERE
- (
- (
- keywords NOT LIKE '%karaoke%' AND name not like '%karaoke%' AND -- MUSIC
- keywords NOT LIKE '%dj night%' AND name not like '%dj night%' AND -- MUSIC
- keywords NOT LIKE '%bollywood night%' AND name not like '%bollywood night%' AND -- MUSIC
- keywords NOT LIKE '%ladies night%' AND name not like '%ladies night%' AND -- MUSIC
- keywords NOT LIKE '%ft. %' AND name not like '%ft. %' AND -- MUSIC
- keywords NOT LIKE '%camping%' AND name not like '%camping%' AND -- TRIP
- keywords NOT LIKE '%trek%' AND name not like '%trek%' AND -- TRIP
- keywords NOT LIKE '%techno%' AND name not like '%techno%' AND -- MUSIC
- description NOT LIKE '%karaoke%' AND -- MUSIC
- keywords NOT LIKE '%tipsy%' AND name not like '%tipsy%' AND -- MUSIC
- keywords NOT LIKE '%summer camp%' AND name not like '%summer camp%' AND -- KIDS
- description NOT LIKE '%get sloshed%' AND -- DRINKING
- description NOT LIKE '%magic mocktails%' AND -- DRINKING
- name NOT LIKE '%jollywood%' AND -- ADS
- name NOT LIKE '%thrifty x%' AND -- DATING
- name NOT LIKE '% vs %' AND -- SPORTS
- name NOT LIKE '% admissions %' AND -- ADS
- keywords NOT LIKE '%parties%' AND -- MUSIC
- name NOT LIKE '%deck gigs%' AND -- MUSIC
- name NOT LIKE '%5 day program%' AND -- LONG
- name NOT GLOB '*IPL*' AND -- SPORTS
- name NOT GLOB '*GP*' AND -- SPORTS
- name NOT GLOB '*F1*' AND -- SPORTS
- name NOT GLOB '*T20*' AND -- SPORTS
- name NOT GLOB '*RCB*' AND -- SPORTS
- name NOT GLOB '*MI*' AND -- SPORTS
- name NOT LIKE '%pet paw%' AND -- PETS
- keywords NOT LIKE '%summer vacation%' AND -- KIDS
- (organizer NOT LIKE 'Odyssey vibes' OR organizer IS NULL) AND -- TRIP
- (organizer NOT LIKE '%Medicine%' OR organizer IS NULL) AND -- CONFERENCE
- (location NOT LIKE 'Small World' OR location IS NULL) AND -- BADRATING
- (organizer NOT LIKE '%sheena - banjara%' OR organizer IS NULL) AND -- TRIP
- (location NOT LIKE '%HeyBrewty Wellness Studio%' OR location is NULL) AND -- WOOWOO
- (location NOT LIKE '%blue butterfly%' OR location is NULL) AND -- MUSIC
- (description NOT LIKE '%techno%') AND -- MUSIC
- (name NOT LIKE '%DnBIndia presents%') AND -- MUSIC
-
- -- Not sure about these events: https://insider.in/search?q=burgerman
- -- (organizer NOT LIKE '%burgerman%' OR organizer IS NULL) AND
-
- -- See https://insider.in/search?q=boche%20club
- (organizer NOT LIKE '%brewcraft hospitality%' OR organizer IS NULL) AND -- MUSIC
- description not LIKE '%happy hour%' -- MUSIC
- )
- OR (keywords IS NULL OR description IS NULL OR url LIKE '%sumukha%')
- )
- ORDER BY startDate ASC

_includes/head.html

@@ -10,3 +10,6 @@
 <link rel="icon" type="image/png" sizes="32x32" href="/img/favicon-32x32.png" />
 <link rel="icon" type="image/png" sizes="16x16" href="/img/favicon-16x16.png" />
 <link rel="manifest" href="/assets/site.webmanifest" />
+{% if page.noindex %}
+<meta name="robots" content="noindex, nofollow" />
+{% endif %}

_layouts/events.html

@@ -51,6 +51,7 @@
 </script>
 
 <noscript>
+<h3>{{eventCount}} Events</h3>
 <ul>
  {% for e in page.events %} {% assign E =e.event_json | parse_json %}
  <li>

_plugins/icalendar.rb

@@ -95,7 +95,13 @@ module Jekyll
  module ToIcal
  def to_ical(event_json)
  ics = Icalendar::Event.new
- event = JSON.parse event_json
+ begin
+ event = JSON.parse event_json
+ rescue StandardError => e
+ print("Could not parse")
+ print(event_json)
+ return ""
+ end
  ics.color = COLOR_MAP[event['@type']]
  ics.description = event['description']
  # set geolocation if available in schema.org/Place

_plugins/multi_layout.rb

@@ -25,6 +25,11 @@ def create_layout_views(page)
  dir = File.dirname(page.relative_path)
  Page.new(page.site, page.site.source, dir, page.name).tap do |new_page|
  new_page.data = page.data.clone
+ # Default configuration is not copied, but dynamically
+ # applied. We need to copy this somehow as well.
+ # new_page.data.default_proc = proc do |_, key|
+ # site.frontmatter_defaults.find(relative_path, type, key)
+ # end
  new_page.data["layout"] = config["layout"]
  new_page.data["permalink"] = config["permalink"]
  end

cal/unwanted.md

@@ -1,23 +1,27 @@
 ---
 title: Unwanted Events
 tags: '["LOW-QUALITY", "NOTINBLR", "WOOWOO", "DANDIYA", "BUSINESS"]'
-subscribe: false
 sqlite:
 - data: events
  file: events.db
  query: |
+ WITH 
+ allowlist_tags AS (
+ SELECT value FROM json_each(:tags)
+ )
  SELECT *
  FROM events
  WHERE EXISTS (
  SELECT 1
- FROM json_each(:tags) AS tags
- WHERE tags.value IN (
- SELECT value
- FROM json_each(event_json->'$.keywords')
- )
- )
- ORDER BY event_json -> '$.startDate'
- ;
+ FROM allowlist_tags
+ WHERE allowlist_tags.value IN (
+ SELECT value
+ FROM json_each(event_json->'$.keywords')
+ ))
+ ORDER BY event_json -> '$.startDate';
+subscribe: false
+# This is such a trash event page, that we don't want links here to impact our rankings
+noindex: true
 --- 
 This is a meta calendar that lists events that have been excluded
 from other calendars. This is primarily meant for debugging.