Alt text

WITH 
-- Extract images from blog_entry.body (always HTML)
entry_images AS (
  SELECT 
    'https://simonwillison.net/' || to_char(created, 'YYYY/Mon/') || trim(leading '0' from to_char(created, 'DD')) || '/' || slug AS url,
    created,
    (REGEXP_MATCHES(body, '<img[^>]*?src="([^"]*)"[^>]*?alt="([^"]*)"', 'g'))[1] AS src,
    (REGEXP_MATCHES(body, '<img[^>]*?src="([^"]*)"[^>]*?alt="([^"]*)"', 'g'))[2] AS alt_text
  FROM blog_entry
  WHERE body ~ '<img[^>]*?src="[^"]*"[^>]*?alt="[^"]*"'
  
  UNION ALL
  
  SELECT 
    'https://simonwillison.net/' || to_char(created, 'YYYY/Mon/') || trim(leading '0' from to_char(created, 'DD')) || '/' || slug AS url,
    created,
    (REGEXP_MATCHES(body, '<img[^>]*?alt="([^"]*)"[^>]*?src="([^"]*)"', 'g'))[2] AS src,
    (REGEXP_MATCHES(body, '<img[^>]*?alt="([^"]*)"[^>]*?src="([^"]*)"', 'g'))[1] AS alt_text
  FROM blog_entry
  WHERE body ~ '<img[^>]*?alt="[^"]*"[^>]*?src="[^"]*"'
),

-- Extract images from blog_blogmark.commentary (HTML if use_markdown is false)
blogmark_html_images AS (
  SELECT 
    'https://simonwillison.net/' || to_char(created, 'YYYY/Mon/') || trim(leading '0' from to_char(created, 'DD')) || '/' || slug AS url,
    created,
    (REGEXP_MATCHES(commentary, '<img[^>]*?src="([^"]*)"[^>]*?alt="([^"]*)"', 'g'))[1] AS src,
    (REGEXP_MATCHES(commentary, '<img[^>]*?src="([^"]*)"[^>]*?alt="([^"]*)"', 'g'))[2] AS alt_text
  FROM blog_blogmark
  WHERE use_markdown = false AND commentary ~ '<img[^>]*?src="[^"]*"[^>]*?alt="[^"]*"'
  
  UNION ALL
  
  SELECT 
    'https://simonwillison.net/' || to_char(created, 'YYYY/Mon/') || trim(leading '0' from to_char(created, 'DD')) || '/' || slug AS url,
    created,
    (REGEXP_MATCHES(commentary, '<img[^>]*?alt="([^"]*)"[^>]*?src="([^"]*)"', 'g'))[2] AS src,
    (REGEXP_MATCHES(commentary, '<img[^>]*?alt="([^"]*)"[^>]*?src="([^"]*)"', 'g'))[1] AS alt_text
  FROM blog_blogmark
  WHERE use_markdown = false AND commentary ~ '<img[^>]*?alt="[^"]*"[^>]*?src="[^"]*"'
),

-- Extract markdown images from blog_blogmark.commentary (if use_markdown is true)
blogmark_md_images AS (
  SELECT 
    'https://simonwillison.net/' || to_char(created, 'YYYY/Mon/') || trim(leading '0' from to_char(created, 'DD')) || '/' || slug AS url,
    created,
    (REGEXP_MATCHES(commentary, '!\[([^\]]*)\]\(([^)]*)\)', 'g'))[2] AS src,
    (REGEXP_MATCHES(commentary, '!\[([^\]]*)\]\(([^)]*)\)', 'g'))[1] AS alt_text
  FROM blog_blogmark
  WHERE use_markdown = true AND commentary ~ '!\[[^\]]*\]\([^)]*\)'
),

-- Extract markdown images from blog_quotation.quotation
quotation_images AS (
  SELECT 
    'https://simonwillison.net/' || to_char(created, 'YYYY/Mon/') || trim(leading '0' from to_char(created, 'DD')) || '/' || slug AS url,
    created,
    (REGEXP_MATCHES(quotation, '!\[([^\]]*)\]\(([^)]*)\)', 'g'))[2] AS src,
    (REGEXP_MATCHES(quotation, '!\[([^\]]*)\]\(([^)]*)\)', 'g'))[1] AS alt_text
  FROM blog_quotation
  WHERE quotation ~ '!\[[^\]]*\]\([^)]*\)'
),

-- Extract markdown images from blog_note.body
note_images AS (
  SELECT 
    'https://simonwillison.net/' || to_char(created, 'YYYY/Mon/') || trim(leading '0' from to_char(created, 'DD')) || '/' || slug AS url,
    created,
    (REGEXP_MATCHES(body, '!\[([^\]]*)\]\(([^)]*)\)', 'g'))[2] AS src,
    (REGEXP_MATCHES(body, '!\[([^\]]*)\]\(([^)]*)\)', 'g'))[1] AS alt_text
  FROM blog_note
  WHERE body ~ '!\[[^\]]*\]\([^)]*\)'
),

-- Combine all results
all_images AS (
  SELECT url, src, alt_text, created FROM entry_images
  UNION ALL
  SELECT url, src, alt_text, created FROM blogmark_html_images
  UNION ALL
  SELECT url, src, alt_text, created FROM blogmark_md_images
  UNION ALL
  SELECT url, src, alt_text, created FROM quotation_images
  UNION ALL
  SELECT url, src, alt_text, created FROM note_images
)

-- Apply search filter and sort
SELECT created, alt_text, src, url
FROM all_images
WHERE 
  CASE 
    WHEN %(search)s = '' THEN true
    ELSE 
      alt_text ILIKE '%%' || %(search)s || '%%' OR 
      src ILIKE '%%' || %(search)s || '%%'
  END
ORDER BY created DESC