Show the updated columns from my journal tables

Question

In my Postgresql db, I keep a journal of all updates against some of my tables.

The journal tables are of the form:

| Column          | Type   |
| --------------- | ------ |
| journal_id      | bigint | 
| id              | text   | 
| col_1           | text   | 
| col_i           | any    |

• journal_id is an auto increment sequence.
• every time the resource is updated, it creates a new record in the journal table for the same id and a new journal_id

I want to create a function show_column_differences to show the diff between the updates: which columns got updated and the before / after values.

So if the table contains:

| journal_id | id  | col_1 | col_i |
| 1          | foo | v1_1  | vi_1  |
| 2          | foo | v1_2  | vi_1  |
| 3          | foo | v1_3  | vi_3  |

It would return

> SELECT * FROM show_column_differences('my_table', 'foo');
| journal_id | previous_journal_id | col_name | value | previous_value |
| 1          | null                | col_1    | v1_1  | null           |
| 1          | null                | col_i    | vi_1  | null           |
| 2          | 1                   | col_1    | v1_2  | v1_1           |
| 3          | 2                   | col_1    | v1_3  | v1_2           |
| 3          | 2                   | col_i    | vi_3  | vi_1           |

I have created a function leveraging the LAG with OVER (PARTITION BY id ORDER BY journal_id) but I keep running into issues to get the dynamic list of columns from the record r:

CREATE OR REPLACE FUNCTION show_column_differences(
    p_table_name TEXT,
    p_id TEXT
)
RETURNS TABLE (
    journal_id TEXT,
    column_name TEXT,
    old_value TEXT,
    new_value TEXT
) AS $$
DECLARE
    r RECORD;
    col_name TEXT;
    old_col_name TEXT;
    col_value TEXT;
    old_col_value TEXT;
    column_list TEXT[];
    select_old_columns TEXT := '';
BEGIN
    -- Compute the list of columns once
    SELECT array_agg(c.column_name::TEXT)
    INTO column_list
    FROM information_schema.columns c
    WHERE c.table_name = p_table_name

    -- Build the list of old_journal columns with alias
    FOREACH col_name IN ARRAY column_list LOOP
        select_old_columns := select_old_columns || format(', old_journal.%I AS old_%I', col_name, col_name);
    END LOOP;

    -- CTE to compute the lag and join with the old version of each row
    FOR r IN EXECUTE format('
        WITH journals AS (
            SELECT *, lag(journal_id) OVER (PARTITION BY id ORDER BY journal_id ASC) AS old_journal_id
            FROM %I
            WHERE id = %L
            ORDER BY journal_id ASC
        )
        SELECT
            journals.* %s
        FROM journals
        JOIN journals AS old_journal
        ON journals.old_journal_id = old_journal.journal_id
    ', p_table_name, p_id, select_old_columns)
    LOOP
        -- Loop over each column and compare old_value and new_value using the precomputed column list
        FOREACH col_name IN ARRAY column_list LOOP
            old_col_name := 'old_' || col_name;
            -- How to access the "col_name" and "old_col_name field of the record?

            IF old_value IS DISTINCT FROM new_value THEN
                -- Set OUT parameters
                -- ...
                RETURN NEXT;
            END IF;
        END LOOP;
    END LOOP;
END;
$$ LANGUAGE plpgsql;

• How can I access r.old_col_name where old_col_name is a variable?
• I am open to completely different way to achieve the diff

Thanks for the help