Add simple analytics

[McPringle]

Log page requests

Step 1: Extract the data from the request and response headers

• URL
• date and time
• IP address
• user-agent
• accept-language
• referrer
• status code

Step 2: Process the data asynchronously

• date (as is) and time (remove seconds)
• page requested (based on URL)
• browser and version (based on user-agent)
• OS and version (based on user-agent)
• country (based on IP address)
• language (based on accept-language)
• referrer (as is)

Step 3: Respect privacy

• save the result from data processing in step 2 only
• don't save the raw header data
• don't save the IP address
• don't save the user-agent string
• use separate tables (e.g. page_requests, browsers, languages, etc), so we will not be able to make connections between them
• store date and time (down to minutes, no seconds) for page_requests only
• for all other data (browsers, languages, etc), store just the date (no time) -> no connection by timestamp possible

Brainstorming: Is it okay to save the complete referrer? It should not contain any sensitive data. Just the page the user was linked from. IMHO it should be okay.

Step 4: Documentation on the website

• why are we tracking
• what are we tracking
• how are we tracking