Problem
There is no reason in the world to give Google the IP addresses of each of our visitors just to serve cute fonts.
Solution
For privacy reasons we setup a Google Font reverse proxy cache with Apache's mod_cache.
This is a drop-in replacement of Google Font. For example you can update right now your URLs from fonts.googleapis.com to google-font-proxy.reyboz.it.
See it in action:
- https://fonts.googleapis.com/css?family=Open+Sans%3Aregular%2C700|Khand%3A500&subset=latin-ext%2Clatin-ext%2Clatin-ext%2Clatin-ext%2Clatin-ext%2Clatin-ext%2Clatin-ext%2Clatin-ext
- https://google-font-proxy.reyboz.it/css?family=Open+Sans%3Aregular%2C700|Khand%3A500&subset=latin-ext%2Clatin-ext%2Clatin-ext%2Clatin-ext%2Clatin-ext%2Clatin-ext%2Clatin-ext%2Clatin-ext
The nice part is that after the first request, fonts are then literally stored locally. We can do this because most of Google Fonts are under a Free license.
NOTE: Yes, different user agents should trigger different fonts. If this does not happen, feel free to file a Task.
WARNING: If you noticed a problem in the licenses, file a Task. There shouldn't be any because we use Free fonts.
VirtualHost
apt install apache2 a2enmod cache a2enmod substitute
/etc/apache2/sites-available/reyboz.google-font-proxy.conf
# # See https://gitpull.it/T776 # <VirtualHost *:80> ServerName google-font-proxy.reyboz.it Include /etc/apache2/my-includes/google-font-proxy.conf </VirtualHost> <VirtualHost *:443> ServerName google-font-proxy.reyboz.it SSLEngine on SSLCertificateFile /etc/letsencrypt/live/google-font-proxy.reyboz.it/cert.pem SSLCertificateKeyFile /etc/letsencrypt/live/google-font-proxy.reyboz.it/privkey.pem SSLCertificateChainFile /etc/letsencrypt/live/google-font-proxy.reyboz.it/chain.pem Include /etc/apache2/my-includes/google-font-proxy.conf Include /etc/apache2/my-includes/ssl-hardening.conf </VirtualHost>
/etc/apache2/my-includes/google-font-proxy.conf
#
# This is a shared Apache HTTPd configuration for a Google Font proxy
# License: public domain (CC 0)
# Author: Valerio Bozzolan (2021)
#
# https://gitpull.it/T776
#
# where you want to put a nice homepage
DocumentRoot /home/www-data/reyboz.it/google-font-proxy/www
#
# Your canonical domain
#
# useful to store same cache entries from different hostnames
# so, if you share this to multiple ServerName/ServerAlias, you save some kilobytes! asd
CacheKeyBaseURL "http://google-font-proxy.reyboz.it/"
# allow to proxy via https://
SSLProxyEngine On
# fix mod_cache for proxies
CacheQuickHandler off
# disable unuseful and slow features
<Location />
AllowOverride none
</Location>
# forward static font files and cache locally
<Location /s/>
# hi Google! Give me the font file
ProxyPass https://fonts.gstatic.com/s/
ProxyPassReverse https://fonts.gstatic.com/s/
# cache future requests
# CacheEnable disk
</Location>
<LocationMatch "^/(?<SUFFIX>css|css2|icon)$">
# append the value "User-Agent" to the Vary HTTP header
# but only if the dont-vary environment is not set.
# don't know why we have to check the dont-vary env
Header append Vary User-Agent env=!dont-vary
# hi Google! Give me the CSS file
ProxyPass https://fonts.googleapis.com
ProxyPassReverse https://fonts.googleapis.com
#
# Replace some URLS to fonts.gstatic.com
#
# Note that Sobstitute will not work without the next line
# https://stackoverflow.com/questions/32603182/apache-httpd-substitute-wont-work
#
AddOutputFilterByType INFLATE;SUBSTITUTE;DEFLATE text/css
Substitute "s|https://fonts.gstatic.com/s/|/s/|n"
# cache future requests
CacheEnable disk
</LocationMatch>
#
# Forward JavaScript requests to Google and cache locally
#
# Example:
# https://ajax.googleapis.com/ajax/libs/jquery/1.10.2/jquery.min.js
#
<Location /ajax>
# cache future requests
CacheEnable disk
CacheHeader on
# hi Google! Give me the JavaScript file
ProxyPass https://ajax.googleapis.com/ajax
ProxyPassReverse https://ajax.googleapis.com/ajax
</Location>
# cache even with Cache-Control: private
CacheStorePrivate On
# please cache also these cases or it will not work
CacheIgnoreNoLastMod On
CacheIgnoreCacheControl On
# add X-Cache with HIT or REVALIDATE or MISS
# useful for debug purposes
CacheHeader on
# remove "Link: <https://fonts.gstatic.com>; rel=preconnect; crossorigin"
Header unset Link
#
# Cache directory root
#
# no need to change the default cache pathname in Debian
#
# Check if the system is cleaning it with:
## systemctl status apache-htcacheclean
#
#CacheRoot /var/cache/apache2/mod_cache_disk
# eventually enable serious debug
#LogLevel debug
#CustomLog "/var/log/apache2/cached-requests.log" common env=cache-hit
#CustomLog "/var/log/apache2/uncached-requests.log" common env=cache-miss
#CustomLog "/var/log/apache2/revalidated-requests.log" common env=cache-revalidate
#CustomLog "/var/log/apache2/invalidated-requests.log" common env=cache-invalidate
#LogFormat "%{cache-status}e " cachelog
#CustomLog /var/log/apache2/cachelog.log cachelogProposed extra to be tested, made by @andreaganduglia in order to do not send the original user agent but just some fixed ones (also removing accepted languages, eventual cookies, and eventual referer):
ProxyAddHeaders Off
RequestHeader unset Cookie
RequestHeader unset Accept-Language
RequestHeader unset Referer
# Generic User-Agent
RequestHeader set User-Agent "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/100.0.0.0 Safari/537.36"
SetEnvIf User-Agent ^(.*)(Chrome/)[0-9]{1,}(.*)$ ua_chrome
RequestHeader set User-Agent "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/104.0.0.0 Safari/537.36" env=ua_chrome
SetEnvIf User-Agent ^(.*)(Firefox/)[0-9]{1,}(.*)$ ua_firefox
RequestHeader set User-Agent "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:103.0) Gecko/20100101 Firefox/103.0" env=ua_firefox
SetEnvIf User-Agent ^(.*)(Version/)[0-9]{1,}\s+Safari/(.*)$ ua_safari
RequestHeader set User-Agent "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/15.6 Safari/605.1.15" env=ua_safari
SetEnvIf User-Agent ^(.*)(MSIE)\s+[0-9]{1,}(.*)$ ua_ie
RequestHeader set User-Agent "Mozilla/5.0 (Windows NT 10.0; Trident/7.0; rv:11.0) like Gecko" env=ua_ieTODO: Check if the "referer" is normally sent to Google Fonts in both Firefox and Chrome with default settings.
Let's Encrypt
$ certbot certonly --webroot --webroot-path=/var/www/html -d google-font-proxy.reyboz.it Saving debug log to /var/log/letsencrypt/letsencrypt.log Plugins selected: Authenticator webroot, Installer None Obtaining a new certificate Performing the following challenges: http-01 challenge for google-font-proxy.reyboz.it Using the webroot path /var/www/html for all unmatched domains. Waiting for verification... Cleaning up challenges IMPORTANT NOTES: - Congratulations! Your certificate and chain have been saved at: /etc/letsencrypt/live/google-font-proxy.reyboz.it/fullchain.pem Your key file has been saved at: /etc/letsencrypt/live/google-font-proxy.reyboz.it/privkey.pem Your cert will expire on 2021-06-29. To obtain a new or tweaked version of this certificate in the future, simply run certbot again. To non-interactively renew *all* of your certificates, run "certbot renew" - If you like Certbot, please consider supporting our work by: Donating to ISRG / Let's Encrypt: https://letsencrypt.org/donate Donating to EFF: https://eff.org/donate-le
Cache directory
Already created by Debian here:
/var/cache/apache2/mod_cache_disk
Systemd service
This service is needed for a general cleanup of the cache directory to keep it at a maximum size.
It was already provided by Debian:
systemctl enable --now apache-htcacheclean