standardebooks/web
1
0
Fork 0
mirror of https://github.com/standardebooks/web.git synced 2025-07-06 06:40:33 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Add <meta rel="canonical"> elements when deploying ebook

Browse source
This commit is contained in:
Alex Cabal 2025-01-05 13:07:59 -06:00
parent 692e418aee
commit cce0c328d1
1 changed files with 36 additions and 20 deletions
Download patch file Download diff file Expand all files Collapse all files

56
scripts/deploy-ebook-to-www
View file

@ -311,7 +311,7 @@ do
printf "Recomposing ebook ... "
fi
# Recompose the epub into a single file, but put it outside of the epub src for now so we don't stomp on it with the following sections.
# Recompose the epub into a single file, but put it outside of the epub source for now so we don't stomp on it with the following sections.
# We do this first because the tweaks below shouldn't apply to the single-page file.
se recompose-epub --xhtml --output "${workDir}"/single-page.xhtml --extra-css-file="${webRoot}/css/web.css" "${workDir}"
@ -322,7 +322,10 @@ do
sed --in-place --regexp-extended "s|<body([^>]*)>|<body\1><header><nav><ul><li><a href=\"/\">Standard Ebooks</a></li><li><a href=\"${bookUrl}\">Back to ebook</a></li></ul></nav></header>|" "${workDir}"/single-page.xhtml
# Adjust sponsored links in the colophon.
sed --in-place 's|<p><a href="http|<p><a href="http|g' "${workDir}"/single-page.xhtml
sed --in-place 's|<p><a href="http|<p><a rel="nofollow" href="http|g' "${workDir}"/single-page.xhtml
# Add a canonical `<meta>` element.
sed --in-place "s|</title>|</title>\n\t\t<link rel=\"canonical\" href=\"https://standardebooks.org${bookUrl}/text/single-page\" />|" "${workDir}"/single-page.xhtml
if [ "${verbose}" = "true" ]; then
printf "Done.\n"
@ -330,34 +333,47 @@ do
fi
# Make some compatibility adjustments for the individual XHTML files.
workTitle=$(grep --only-matching --extended-regexp "<dc:title id=\"title\">(.+?)</dc:title>" "${workDir}"/src/epub/content.opf | sed --regexp-extended "s/<[^>]+?>//g")
# Remove instances of the .xhtml filename extension in the source text.
find "${workDir}"/src/epub \( -type d -name .git -prune \) -o -type f -name "*.xhtml" -print0 | xargs -0 sed --in-place 's/\.xhtml//g'
for filename in $(find "${workDir}"/src/epub \( -type d -name .git -prune \) -o -type f -name "*.xhtml"); do
# Add our web stylesheet to XHTML files.
find "${workDir}"/src/epub \( -type d -name .git -prune \) -o -type f -name "*.xhtml" -print0 | xargs -0 sed --in-place --regexp-extended 's|</title>|</title>\n\t\t<link href="/css/web.css" media="screen" rel="stylesheet" type="text/css"/>|'
# Add our web stylesheet to XHTML files.
sed --in-place --regexp-extended 's|</title>|</title>\n\t\t<link href="/css/web.css" media="screen" rel="stylesheet" type="text/css"/>|' "${filename}"
# Add `@lang` attributes.
sed --in-place --regexp-extended 's/xml:lang="([^"]+?)"/xml:lang="\1" lang="\1"/g' "${filename}"
# Add the work title to `<title>` elements in the source text.
sed --in-place --regexp-extended "s|<title>|<title>${workTitle} - |g" "${filename}"
# Wrap book contents in a `<main>` element.
sed --in-place --regexp-extended "s|<body([^>]*)>|<body><main\1>|; s|</body>|</main></body>|" "${filename}"
# Add the header nav to each page.
sed --in-place --regexp-extended "s|<body([^>]*)>|<body\1><header><nav><ul><li><a href=\"/\">Standard Ebooks</a></li><li><a href=\"${bookUrl}\">Back to ebook</a></li><li><a href=\"${bookUrl}/text\">Table of contents</a></li></ul></nav></header>|" "${filename}"
# Add a canonical `<meta>` element.
fileUrl=$(echo "${filename}" | sed --regexp-extended "s|^.+/src/epub(/text/.+\.xhtml$)|https://standardebooks.org${bookUrl}\1|")
# Special case for `toc.xhtml`.
if [[ "${fileUrl}" =~ /toc\.xhtml$ ]]; then
fileUrl="https://standardebooks.org${bookUrl}/text"
fi
sed --in-place "s|</title>|</title>\n\t\t<link rel=\"canonical\" href=\"${fileUrl}\" />|" "${filename}"
# Remove instances of the `.xhtml` filename extension in the source text.
sed --in-place 's/\.xhtml//g' "${filename}"
done
# Remove `-epub-*` CSS properties from CSS files as they're invalid in a web context.
sed --in-place --regexp-extended "s|\s*\-epub\-[^;]+?;||g" "${workDir}"/src/epub/css/*.css
# Add lang attributes.
find "${workDir}"/src/epub \( -type d -name .git -prune \) -o -type f -name "*.xhtml" -print0 | xargs -0 sed --in-place --regexp-extended 's/xml:lang="([^"]+?)"/xml:lang="\1" lang="\1"/g'
# Add the work title to <title> tags in the source text.
workTitle=$(grep --only-matching --extended-regexp "<dc:title id=\"title\">(.+?)</dc:title>" "${workDir}"/src/epub/content.opf | sed --regexp-extended "s/<[^>]+?>//g")
find "${workDir}"/src/epub \( -type d -name .git -prune \) -o -type f -name "*.xhtml" -print0 | xargs -0 sed --in-place --regexp-extended "s|<title>|<title>${workTitle} - |g"
# Wrap book contents in a `<main>` element.
find "${workDir}"/src/epub \( -type d -name .git -prune \) -o -type f -name "*.xhtml" -print0 | xargs -0 sed --in-place --regexp-extended "s|<body([^>]*)>|<body><main\1>|; s|</body>|</main></body>|"
# Add the header nav to each page.
find "${workDir}"/src/epub \( -type d -name .git -prune \) -o -type f -name "*.xhtml" -print0 | xargs -0 sed --in-place --regexp-extended "s|<body([^>]*)>|<body\1><header><nav><ul><li><a href=\"/\">Standard Ebooks</a></li><li><a href=\"${bookUrl}\">Back to ebook</a></li><li><a href=\"${bookUrl}/text\">Table of contents</a></li></ul></nav></header>|"
# Add a chapter navigation footer to each page.
"${scriptsDir}"/inject-chapter-navigation-footer "${workDir}" "${bookUrl}"
# Adjust sponsored links in the colophon.
sed --in-place 's|<p><a href="http|<p><a href="http|g' "${workDir}"/src/epub/text/colophon.xhtml
sed --in-place 's|<p><a href="http|<p><a rel="nofollow" href="http|g' "${workDir}"/src/epub/text/colophon.xhtml
# Done adding compatibility!