etc_cache: cache trying hard to be smart

etc · 2021-02-12 11:23:21

I’m trying

    RewriteCond %{DOCUMENT_ROOT}/textpattern/cache/$1 -f [OR]
    RewriteCond %{DOCUMENT_ROOT}/textpattern/cache/$1(\.html) -f [OR]
    RewriteCond %{DOCUMENT_ROOT}/textpattern/cache/$1(\.xml) -f [OR]
    RewriteCond %{DOCUMENT_ROOT}/textpattern/cache/$1index.html -f
    RewriteRule ^(.*) cache/$1%1 [PT,L]

but %1 capture group does not seem to work (see below). Will set an environment variable maybe…

But isn’t it reasonable to suggest that if a user wants to output an article as application/javascript, he should then url-title it name.js? This would simplify things a lot.

etc · 2021-02-12 11:23:51

philwareham wrote #328809:

Just to throw extra work into the mix, I’d also need an Nginx version of the .htaccess rules – sorry!

Not me!

Bloke · 2021-02-12 11:31:04

etc wrote #328810:

%1 capture group does not seem to work.

I’ve never used %, only $. Long shot: do percent replacements need to be in curly braces? %{1}

etc · 2021-02-12 11:43:31

Seemingly %n come from RewriteCond:

The problem is we need to capture them in the left hand term.

Bloke · 2021-02-12 11:47:14

Eeek, above my pay grade! I’m only a lowly regex hacker and they normally take me aaaages in regex101 to make them do what I want reliably.

Vienuolis · 2021-02-12 14:47:47

Textpattern powered websites will now serve visitors as hypertext flat-files, suspended in ‘maintenance mode’, too!

With this separation of dynamic SQL generation and static HTML delivery, in addition to the similar theming, and with the native HTML, CSS, TXP, PHP, also sections, categories, keywords, etc strict separation principle, and managed by smart and open-minded developers, Textpattern stands as exceptionally agile, neat, light, secure, and robust WWW publishing loom, unique on CMS market.

etc · 2021-02-12 14:58:11

Version 0.2 of etc_static is ready. What’s new:

pages with no-cache, no-store, private cache-control headers or any vary header will not be cached. In particular, that’s true in dev mode.
pages being visited by a logged-in user will not be cached, though he can be served from public cache. It means you need to log out to populate it.
cache will try to append the appropriate extension, so static files should be served with the correct mediatype.

The next big step is an admin interface, so etc_static will probably merge with etc_cache.

You’ll need to update .htaccess with the following:

# --- Append to .htaccess right after RewriteRule ^(.+) - [PT,L] --- 
    RewriteCond %{REQUEST_METHOD} =GET
    RewriteCond %{QUERY_STRING} ^$
    RewriteCond %{DOCUMENT_ROOT}/path/to/site/cache/$1 -f [OR]
    RewriteCond %{DOCUMENT_ROOT}/path/to/site/cache/$1index.html -f
    RewriteRule ^(.*) cache/$1 [PT,L]

    RewriteCond %{REQUEST_METHOD} =GET
    RewriteCond %{QUERY_STRING} ^$
    RewriteCond %{DOCUMENT_ROOT}/path/to/site/cache/$1.html -f
    RewriteRule ^(.*) cache/$1.html [PT,L]

Repeat the last block for every mediatype you intend to use without name extension, replacing html with the latter.

# Name: etc_static v0.2 
# Type: Admin/Public/AJAX plugin
# Static cache
# Author: Oleg Loukianov
# URL: https://github.com/bloatware/
# Recommended load order: 5

# .....................................................................
# This is a plugin for Textpattern CMS - http://textpattern.com/
# To install: textpattern > admin > plugins
# Paste the following text into the 'Install plugin' box:
# .....................................................................
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etc · 2021-02-12 15:07:48

If someone could translate these apache rules to Nginx, that would be ace.

philwareham · 2021-02-12 15:49:03

Pete is the resident Nginx expert – but I’ve pestered him enough this week. Will mention it to him in due course, I’m sure there are ways to duplicate your Apache rules within it.

Vienuolis · 2021-02-13 16:18:56

I guess that an URL Toolkit for the Hiawatha werbserver would much simpler, inserting two extra (bolder) lines only for the cache:

UrlToolkit {
ToolkitID = textpattern
RequestURI exists Return
Match ^/(files|images)/ Return
Match ^/(favicon.ico|robots.txt|sitemap.xml)$ Return
Match ^/themes/reguo/styles/reguo.css$ Return
Method GET Return
Match ^/(.*) Rewrite /cache/$1.html
Match [^?]*(\?.*)? Rewrite /index.php$1
}

Did I catch the right meaning of these Apache rules, Oleg?

etc · 2021-02-13 17:32:05

Dunno, I’m new to rewrites. The idea is:

if the request type is GET
and if the query string (i.e. ?something) is empty
and cache/request_uri or cache/request_uri/index.html or cache/request_uri.html exist

redirect request_uri to cache/request_uri(.html). For example,

/section/?pg=2   ->   leave it alone
/section/        ->   /cache/section/ (will serve index.html from there)
/section/title   ->   /cache/section/title.html

but only if these .html files exist.

Am I complicating? The first version did not append .html to cached files, but they were served without content-type header, though browsers considered them as html by default.

Pat64 · 2021-02-14 05:52:39

I’m sorry Oleg, but your second plugin version do not work anymore: the cache directory isn’t created, so there is no more conversion into static files…
Tested on a local MAMP PC server.

Textpattern CMS

Textpattern CMS support forum

#37 2021-02-12 11:23:21

Re: etc_cache: cache trying hard to be smart

#38 2021-02-12 11:23:51

Re: etc_cache: cache trying hard to be smart

philwareham wrote #328809:

#39 2021-02-12 11:31:04

Re: etc_cache: cache trying hard to be smart

etc wrote #328810:

#40 2021-02-12 11:43:31

Re: etc_cache: cache trying hard to be smart

#41 2021-02-12 11:47:14

Re: etc_cache: cache trying hard to be smart

#42 2021-02-12 14:47:47

Re: etc_cache: cache trying hard to be smart

#43 2021-02-12 14:58:11

Re: etc_cache: cache trying hard to be smart

#44 2021-02-12 15:07:48

Re: etc_cache: cache trying hard to be smart

#45 2021-02-12 15:49:03

Re: etc_cache: cache trying hard to be smart

#46 2021-02-13 16:18:56

Re: etc_cache: cache trying hard to be smart

#47 2021-02-13 17:32:05

Re: etc_cache: cache trying hard to be smart

#48 2021-02-14 05:52:39

Re: etc_cache: cache trying hard to be smart

Board footer