seirdy.one/scripts/xhtmlize.sh

#!/bin/sh
# takes an arg for the output dir.
# Runs xhtmlize-single-file.sh on every single html file in the output dir.
# exits if xhtmlize-single-file fails.

# no pipefail here since there are no pipes.
set -e -u

output_dir="$1"
script_dir="$(dirname "$0")"
temp_resume="$(mktemp)"

# I have an alias for a redirect. I also define the redirect in Nginx, but this is there for the envs.net/~seirdy mirror. Hugo aliases don't have trailing slashes; this will trip up xmllint.
sed -i -e 's|<html lang="en-us"|<html xmlns="http://www.w3.org/1999/xhtml" lang="en-us" xml:lang="en-us"|' -E -e 's#<((link|meta) .*)">#<\1" />#' "$output_dir/resume/index.html"
mv "$output_dir/resume/index.html" "$temp_resume"

{
	printf '\t' && sed -e '7q;d' "$output_dir/index.html"
} >"$output_dir/tmp.css"
cleanup() {
	rm -f "$output_dir/tmp.css"
	mv "$temp_resume"  "$output_dir/resume/index.html"
}
trap cleanup EXIT

export XMLLINT_INDENT='	'
export OUTPUT_DIR="$output_dir"
find "$output_dir" -type f -name '*.html' -exec sh "$script_dir/xhtmlize-single-file.sh" {} \;
find "$output_dir" -type f -name '*.xml' -exec xmllint --noblanks --encode UTF-8 --noent {} --output {} \;
Create both index.html and index.xhtml files Combines well with content-negotiation based on the "Accept" header. 2022-05-30 16:09:40 -07:00			`#!/bin/sh`
			`# takes an arg for the output dir.`
xhtmlize: switch back to using xmllint Switch back to using xmllint, but now extract + replace the CSS to address the escaping. 2022-06-01 22:11:40 -07:00			`# Runs xhtmlize-single-file.sh on every single html file in the output dir.`
xhtmlize: better formatting, exit on err - make xhtmlize exit early if any file fails to be xhtml-ized - add whitespace before self-closing slash. 2022-06-13 08:27:09 -07:00			`# exits if xhtmlize-single-file fails.`
xhtmlize: switch back to using xmllint Switch back to using xmllint, but now extract + replace the CSS to address the escaping. 2022-06-01 22:11:40 -07:00
Opportunistically use pipefail 2023-11-25 13:40:52 -08:00			`# no pipefail here since there are no pipes.`
xhtmlize: switch back to using xmllint Switch back to using xmllint, but now extract + replace the CSS to address the escaping. 2022-06-01 22:11:40 -07:00			`set -e -u`

			`output_dir="$1"`
			`script_dir="$(dirname "$0")"`
Move resume to "about" Define a redirect alias too. This also entails special xhtmlize behavior as redirects in Hugo don't use trailing slashes. 2023-11-26 14:28:41 -08:00			`temp_resume="$(mktemp)"`

			`# I have an alias for a redirect. I also define the redirect in Nginx, but this is there for the envs.net/~seirdy mirror. Hugo aliases don't have trailing slashes; this will trip up xmllint.`
Make redirect alias valid XHTML5 2023-11-26 14:57:42 -08:00			`sed -i -e 's\|<html lang="en-us"\|<html xmlns="http://www.w3.org/1999/xhtml" lang="en-us" xml:lang="en-us"\|' -E -e 's#<((link\|meta) .*)">#<\1" />#' "$output_dir/resume/index.html"`
Move resume to "about" Define a redirect alias too. This also entails special xhtmlize behavior as redirects in Hugo don't use trailing slashes. 2023-11-26 14:28:41 -08:00			`mv "$output_dir/resume/index.html" "$temp_resume"`
xhtmlize: switch back to using xmllint Switch back to using xmllint, but now extract + replace the CSS to address the escaping. 2022-06-01 22:11:40 -07:00
Quit using HTML-Tidy See https://github.com/htacg/tidy-html5/issues/1094. The most recent commit without that regression can't handle `dl` elements with `div` children. 2023-11-15 09:21:26 -08:00			`{`
			`printf '\t' && sed -e '7q;d' "$output_dir/index.html"`
			`} >"$output_dir/tmp.css"`
xhtmlize: switch back to using xmllint Switch back to using xmllint, but now extract + replace the CSS to address the escaping. 2022-06-01 22:11:40 -07:00			`cleanup() {`
xhtmlize: avoid race between parallel jobs Prevent parallel jobs from accessing the same tmp.css 2022-08-01 20:54:25 -07:00			`rm -f "$output_dir/tmp.css"`
Move resume to "about" Define a redirect alias too. This also entails special xhtmlize behavior as redirects in Hugo don't use trailing slashes. 2023-11-26 14:28:41 -08:00			`mv "$temp_resume" "$output_dir/resume/index.html"`
xhtmlize: switch back to using xmllint Switch back to using xmllint, but now extract + replace the CSS to address the escaping. 2022-06-01 22:11:40 -07:00			`}`
			`trap cleanup EXIT`

			`export XMLLINT_INDENT=' '`
xhtmlize: avoid race between parallel jobs Prevent parallel jobs from accessing the same tmp.css 2022-08-01 20:54:25 -07:00			`export OUTPUT_DIR="$output_dir"`
Add HTML-Tidy's generator tag manually It auto-inserts it to the start of <head> but I want it at the end. 2023-11-15 02:26:52 -08:00			`find "$output_dir" -type f -name '*.html' -exec sh "$script_dir/xhtmlize-single-file.sh" {} \;`
Minify feeds with xmllint This works particularly well on my Atom feeds with embedded XHTML. 2023-11-25 19:05:53 -08:00			`find "$output_dir" -type f -name '*.xml' -exec xmllint --noblanks --encode UTF-8 --noent {} --output {} \;`