SQLite - Output count of all records per day including days with 0 records

Question

I have a sqlite3 database maintained on an AWS exchange that is regularly updated by a Python script. One of the things it tracks is when any team generates a new post for a given topic. The entries look something like this:

id	client	team	date	industry	city
895	acme industries	blueteam	2022-06-30	construction	springfield

I'm trying to create a table that shows me how many entries for construction occur each day. Right now, the entries with data populate, but they exclude dates with no entries. For example, if I search for just

SELECT date, count(id) as num_records
from mytable
WHERE industry = "construction"
group by date
order by date asc

I'll get results that looks like this:

date	num_records
2022-04-01	3
2022-04-04	1

How can I make sqlite output like this:

date	num_records
2022-04-02	3
2022-04-02	0
2022-04-03	0
2022-04-04	1

I'm trying to generate some graphs from this data and need to be able to include all dates for the target timeframe.

EDIT/UPDATE: The table does not already include every date; it only includes dates relevant to an entry. If no team posts work on a day, the date column will jump from day 1 (eg 2022-04-01) to day 3 (2022-04-03).

Answer 1

Given that your " mytable " table contains all dates you need as an assumption, you can first select all of your dates, then apply a LEFT JOIN to your own query, and map all resulting NULL values for the " num_records " field to "0" using the COALESCE function.

WITH cte AS (
    SELECT date, 
           COUNT(id) AS num_records
    FROM mytable
    WHERE industry = "construction"
    GROUP BY date
    ORDER BY date
)
SELECT dates.date, 
       COALESCE(cte.num_records, 0) AS num_records
FROM      (SELECT date FROM mytable) dates
LEFT JOIN cte
       ON dates.date = cte.date