char/imapfilter-config

git clone https://git.t4t.associates/char/imapfilter-config

Charlotte Somrewrite6a9e1052026-10-24

local cases = {
    { name = 'shipping', from = 'Amazon <shipment-tracking@amazon.co.uk>', subject = 'Dispatched: your order', folder = 'notifications' },
    { name = 'bare mixed-case sender', from = 'SHIPMENT-TRACKING@AMAZON.CO.UK', subject = 'Out for delivery', folder = 'notifications' },
    { name = 'receipt beats mailing list', from = 'Apple <no_reply@email.apple.com>', subject = 'Your receipt from Apple.', ['list-unsubscribe'] = '<https://example.invalid>', folder = 'notifications' },
    { name = 'Stripe receipt', from = 'Example <receipts+acct_example@stripe.com>', subject = 'Your Example receipt [#1234]', folder = 'notifications' },
    { name = 'hosting payment', from = 'Host <email@client.host4fun.com>', subject = 'Payment Confirmation : Invoice #123', folder = 'notifications' },
    { name = 'support conversation', from = 'Host <email@client.host4fun.com>', subject = '[Ticket ID: 123] Service Unavailable?' },
    { name = 'bank statement', from = 'Monzo <hello@email.monzo.com>', subject = 'Your September statement', folder = 'notifications' },
    { name = 'bank account change', from = 'Monzo <hello@email.monzo.com>', subject = 'Your account details changed' },
    { name = 'Bandcamp release', from = 'Bandcamp <noreply@bandcamp.com>', subject = 'New release from an artist', folder = 'automated mail' },
    { name = 'Bandcamp purchase', from = 'Bandcamp <noreply@bandcamp.com>', subject = 'Thank you!', folder = 'notifications' },
    { name = 'Ableton beta', from = 'Live 12 <beta@beta.ableton.com>', subject = 'We have just released Live 12.5b1', folder = 'automated mail' },
    { name = 'policy update', from = 'service@example.invalid', subject = 'Changes to our User Agreement', folder = 'automated mail' },
    { name = 'mailing list', from = 'newsletter@example.invalid', subject = 'Weekly roundup', ['list-id'] = 'weekly.example.invalid', folder = 'automated mail' },
    { name = 'bulk precedence', from = 'newsletter@example.invalid', subject = 'Monthly roundup', precedence = 'bulk', folder = 'automated mail' },
    { name = 'non-bulk precedence', from = 'person@example.invalid', subject = 'Hello', precedence = 'normal' },
    { name = 'login notification', from = 'service@example.invalid', subject = 'You signed in from a new device', folder = 'notifications' },
    { name = 'password reset', from = 'Steam <noreply@steampowered.com>', subject = 'Reset your password', ['list-unsubscribe'] = '<https://example.invalid>' },
    { name = 'verification code', from = 'Wise <noreply@wise.com>', subject = 'Your verification code' },
    { name = 'magic link', from = 'service@example.invalid', subject = 'Your secure link is here', ['list-id'] = 'service.example.invalid' },
    { name = 'failed payment', from = 'Wise <noreply@wise.com>', subject = 'Payment failed' },
    { name = 'Stripe payment failure', from = 'Example <invoice+statements+acct_example@stripe.com>', subject = 'Payment failed for your invoice' },
    { name = 'domain expiration', from = 'Cloudflare <noreply@notify.cloudflare.com>', subject = 'Domain Expiration Notification', ['list-unsubscribe'] = '<https://example.invalid>' },
    { name = 'security warning', from = 'Steam <noreply@steampowered.com>', subject = 'Suspicious activity: new device' },
    { name = 'breach newsletter', from = 'service@example.invalid', subject = 'Notice of data breach', ['list-id'] = 'service.example.invalid' },
    { name = 'flagged receipt', from = 'Apple <no_reply@email.apple.com>', subject = 'Your receipt from Apple.', flagged = true },
    { name = 'confirmed phishing', from = 'Mail Delivery System <vada@servicedive.com>', subject = 'Verify your email or be blocked', ['list-unsubscribe'] = '<https://example.invalid>', folder = 'Junk' },
    { name = 'flagged spam', from = 'vada@servicedive.com', subject = 'Verify your email', flagged = true },
    { name = 'sender suffix lookalike', from = 'orders@asos.com.attacker.invalid', subject = 'Your order' },
    { name = 'sender prefix lookalike', from = 'not-orders@asos.com', subject = 'Your order' },
    { name = 'unreviewed invoice', from = 'accounts@example.invalid', subject = 'Your invoice is available' },
    { name = 'personal conversation', from = 'friend@example.invalid', subject = 'Want to catch up?' },
    { name = 'rotating spam mailbox', from = 'New Person <random.123@devthusiastapp.com>', subject = 'Hello', folder = 'Junk' },
    { name = 'spam subdomain', from = 'Campaign <random@outreach.eu.devthusiaststudio.com>', subject = 'Hello', folder = 'Junk' },
    { name = 'bare mixed-case spam domain', from = 'RANDOM@DEVTHUSIASTAPP.COM', subject = 'Hello', folder = 'Junk' },
    { name = 'other dedicated spam domain', from = 'new-person@sendnovada.com', subject = 'Hello', folder = 'Junk' },
    { name = 'flagged spam domain', from = 'random@devthusiastapp.com', subject = 'Hello', flagged = true },
    { name = 'domain suffix lookalike', from = 'person@devthusiastapp.com.example.invalid', subject = 'Hello' },
    { name = 'domain prefix lookalike', from = 'person@notdevthusiastapp.com', subject = 'Hello' },
    { name = 'domain in display name', from = '"random@devthusiastapp.com" <friend@example.invalid>', subject = 'Hello' },
    { name = 'domain in local part', from = 'devthusiastapp.com@example.invalid', subject = 'Hello' },
    { name = 'exact sender in display name', from = '"vada@servicedive.com" <friend@example.invalid>', subject = 'Hello' },
    { name = 'unrelated Gmail sender', from = 'friend@gmail.com', subject = 'Hello' },
    { name = 'unrelated Tiscali sender', from = 'friend@tiscali.it', subject = 'Hello' },
    { name = 'unreviewed mailbox on possibly compromised domain', from = 'person@club-de-race.com', subject = 'Hello' },
    { name = 'Devthusiast rotating domain', from = 'random@new-campaign.invalid', subject = 'Re: Gifted Invite Enclosed (Github Community Partnership)', body = 'Your lifetime membership to Devthusiast awaits.', folder = 'Junk' },
    { name = 'campaign subject without brand', from = 'person@example.invalid', subject = 'Gifted Invite Enclosed (Github Community Partnership)', body = 'A different message.' },
    { name = 'brand mention without campaign subject', from = 'friend@example.invalid', subject = 'Spam filters', body = 'Have you heard of devthusiast?' },
    { name = 'generic partnership', from = 'person@example.invalid', subject = 'GitHub partnership', body = 'Would you like to collaborate?' },
    { name = 'flagged campaign', from = 'random@new-campaign.invalid', subject = 'Gifted Invite Enclosed (Github Community Partnership)', body = 'Welcome to devthusiast.', flagged = true },
    { name = 'decoded MIME subject', from = 'Windows <Windowsinsiderprogram@e-mails.microsoft.com>', subject = 'Action required: verify your account', raw_subject = '=?UTF-8?B?QWN0aW9uIHJlcXVpcmVkOiB2ZXJpZnkgeW91ciBhY2NvdW50?=' },
}

for _, dry_run in ipairs({ false, true }) do
    local moved, preview = {}, {}
    local inbox = setmetatable({}, { __index = Mailbox })

    function inbox:send_query(query, messages)
        local result = Set {}
        local field, value = query:match('^HEADER ([^ ]+) "(.*)"$')
        if not field then field, value = query:match('^(%u+) "(.*)"$') end
        for _, message in ipairs(messages or self:_send_query()) do
            local case = cases[message[2]]
            local matches = query == 'FLAGGED' and case.flagged
            if field then
                local header = case[field:lower()]
                matches = header and header:lower():find(value:lower(), 1, true)
            end
            if matches then result[#result + 1] = message end
        end
        return result
    end

    function inbox:_send_query()
        local result = Set {}
        for id in ipairs(cases) do result[#result + 1] = { self, id } end
        return result
    end

    function inbox:_fetch_fields(fields, messages)
        local result = {}
        for _, id in ipairs(messages) do
            local field = fields[1]:lower()
            local value = cases[id][field]
            if field == 'subject' then value = cases[id].raw_subject or value end
            if value then result[id] = fields[1] .. ': ' .. value .. '\r\n' end
        end
        return result
    end

    function inbox:move_messages(destination, messages)
        assert(not dry_run, 'Dry run attempted to move mail')
        for _, message in ipairs(messages) do
            local id = message[2]
            assert(not moved[id], 'Message filed twice: ' .. cases[id].name)
            moved[id] = destination.folder
        end
        return true
    end

    local account = setmetatable({ INBOX = inbox }, {
        __index = function(_, folder) return { folder = folder } end,
    })
    local env = setmetatable({
        IMAP = function() return account end,
        dofile = function() return { { username = 'test', password_env = 'TEST_PASSWORD' } } end,
        os = { getenv = function(name)
            if name == 'IMAPFILTER_DRY_RUN' then return dry_run and '1' or nil end
            if name == 'HOME' then return '/nonexistent' end
            if name == 'TEST_PASSWORD' then return 'not-a-real-password' end
        end },
        print = function(line)
            local folder, id = line:match('^MATCH\ttest\t([^\t]+)\t(%d+)$')
            if id then
                id = tonumber(id)
                assert(not preview[id], 'Message previewed twice: ' .. cases[id].name)
                preview[id] = folder
            end
        end,
    }, { __index = _G })
    assert(loadfile('config.lua', 't', env))()
    for id, case in ipairs(cases) do
        local actual = (dry_run and preview or moved)[id]
        assert(actual == case.folder, case.name .. ': expected ' .. (case.folder or 'INBOX') .. ', got ' .. (actual or 'INBOX'))
    end
end
print(#cases .. ' classification cases passed in live and dry-run modes (no network)')