v2: Generate sitemap.xml

7 files changed, 167 insertions, 5 deletions

diff --git a/TODOs.md b/TODOs.md
index 22e3a8c..7db3e7a 100644
--- a/TODOs.md
+++ b/TODOs.md
@@ -534,9 +534,9 @@ FIXMEs:
   - envsubst, with errors on undefined variables
   - `date -d` isn't POSIX: simple replacement
 - translation not coming from english (like pt -> fr)
+- add linkchecker
 - how to use tmp fd over tmpfile?
 - po4a for markdown...
 - render slides
-- sitemap
 - shrink CSS
 - build makemake output in parallel
diff --git a/v2/.gitignore b/v2/.gitignore
index 84d3bd9..ac25e0d 100644
--- a/v2/.gitignore
+++ b/v2/.gitignore
@@ -15,6 +15,7 @@
 /src/content/**/*.html
 /src/content/**/*.extrahtml
 /src/content/**/*.i18n
+/src/content/**/*.mapentry
 /src/content/**/*.md.next
 /src/content/**/*.md.prev
 /src/content/**/*.md.tmp
diff --git a/v2/dynamic.mk b/v2/dynamic.mk
index ef7d24c..c0a3132 100644
--- a/v2/dynamic.mk
+++ b/v2/dynamic.mk
@@ -7,7 +7,7 @@ include generated.mk
 
 
 .SUFFIXES:
-.SUFFIXES: .page .md .content .conf .html .categoryentry .indexentry .xmlentry .uncolored
+.SUFFIXES: .page .md .content .conf .html .categoryentry .indexentry .xmlentry .mapentry .uncolored
 
 
 .page.md:
@@ -33,6 +33,9 @@ include generated.mk
 .md.xmlentry:
 	xmlentry $< > $@
 
+.md.mapentry:
+	mapentry $< > $@
+
 .uncolored:
 	printf '%s\n%s\n%s\n' \
 		"`head -n2 $<`" \
@@ -71,7 +74,9 @@ i18n-index.sentinel: po/i18n.mappings
 clean:
 	rm -rf \
 		$(all-generated) $(pages-mds) $(svgs) public/ *.sentinel       \
-		generated.mk po/po4a.cfg      src/content/.gitignore           \
+		generated.mk po/po4a.cfg po/*.mo                               \
+		src/lib/generated.conf        src/lib/generated.*.conf         \
+		src/content/.gitignore        src/content/sitemap.xml          \
 		src/content/*/*.category      src/content/*/*/*.category       \
 		src/content/*/*.categorysort  src/content/*/*/*.categorysort   \
 		src/content/*/*.html          src/content/*/*/*.html           \
@@ -83,7 +88,6 @@ clean:
 		src/content/*/feed.*.xml      src/content/*/*/feed.*.xml       \
 		src/content/*/*.md.next       src/content/*/*/*.md.next        \
 		src/content/*/*.md.prev       src/content/*/*/*.md.prev        \
-		src/lib/generated.conf src/lib/generated.*.conf po/*.mo        \
 
 
 public: symlink-mappings.sentinel copy-content.sentinel copy-containers.sentinel
@@ -94,6 +98,7 @@ content = \
 	$(all-generated.xml)                 \
 	$(static-content)                    \
 	$(svgs)                              \
+	src/content/sitemap.xml              \
 
 containers = \
 	$(all-generated.snippets)            \
@@ -101,6 +106,9 @@ containers = \
 	$(all-generated.extrafeeds)          \
 
 
+src/content/sitemap.xml: $(all-generated.mapentry)
+	sitemap > $@
+
 symlink-mappings.sentinel: src/redirect.mappings
 	awk '{print $$2}' src/redirect.mappings | \
 		xargs dirname | \
diff --git a/v2/src/bin/makemake b/v2/src/bin/makemake
index 22e92c9..c6e3407 100755
--- a/v2/src/bin/makemake
+++ b/v2/src/bin/makemake
@@ -71,6 +71,7 @@ EXTENSIONS='
 .html
 .snippets
 .htmlbody
+.mapentry
 '
 
 COLL_EXTENSIONS='
@@ -119,7 +120,7 @@ for lang in $(langs); do
 		exts | sed "s|^\(.*\)\$|$c.$lang\1 = \$($c.$lang.md:.md=\1)|"
 		exts | sed "s|^\(.*\)\$|\$($c.$lang\1)|" | varlist "$c.$lang"
 
-		mds | sed 's/^\(.*\)\.md$/\1.conf \1.content: \1.md/'
+		mds | sed 's/^\(.*\)\.md$/\1.mapentry \1.conf \1.content: \1.md/'
 		mds | sed 's/^\(.*\)\.md$/\1.categoryentry \1.indexentry \1.html: \1.conf \1.content/'
 		mds | sed 's/^\(.*\)\.md$/\1.snippets \1.htmlbody: \1.html/'
 		if [ "$c" = 'pages' ]; then
diff --git a/v2/src/bin/mapentry b/v2/src/bin/mapentry
new file mode 100755
index 0000000..34b6e75
--- /dev/null
+++ b/v2/src/bin/mapentry
@@ -0,0 +1,75 @@
+#!/bin/sh
+set -eu
+
+usage() {
+	cat <<-'EOF'
+		Usage:
+		  mapentry FILENAME
+		  mapentry -h
+	EOF
+}
+
+help() {
+	cat <<-'EOF'
+
+
+		Options:
+		  -h, --help    show this message
+
+		  FILENAME      the name of the input .md file
+
+
+		Process FILENAME, and generate a "mapentry": an "<url>...</url>"
+		entry to be included into the final sitemap.xml
+
+
+		Examples:
+
+		  Generate the sitemap entry entry for a TIL:
+
+		    $ xmlentry src/til/a-til.md > src/til/a-til.mapentry
+	EOF
+}
+
+
+for flag in "$@"; do
+	case "$flag" in
+		--)
+			break
+			;;
+		--help)
+			usage
+			help
+			exit
+			;;
+		*)
+			;;
+	esac
+done
+
+while getopts 'h' flag; do
+	case "$flag" in
+		h)
+			usage
+			help
+			exit
+			;;
+		*)
+			usage >&2
+			exit 2
+			;;
+	esac
+done
+shift $((OPTIND - 1))
+
+. src/lib.sh
+
+FILENAME="${1:-}"
+eval "$(assert_arg "$FILENAME" 'FILENAME')"
+
+
+url_absolute="$(url-for "$FILENAME" | absolute)"
+lastmod="$(git log -1 --format=%ad --date=short "$FILENAME")"
+export url_absolute lastmod
+
+envsubst < src/lib/mapentry.xml
diff --git a/v2/src/bin/sitemap b/v2/src/bin/sitemap
new file mode 100755
index 0000000..09f7eb1
--- /dev/null
+++ b/v2/src/bin/sitemap
@@ -0,0 +1,72 @@
+#!/bin/sh
+set -eu
+
+usage() {
+	cat <<-'EOF'
+		Usage:
+		  sitemap
+		  sitemap -h
+	EOF
+}
+
+help() {
+	cat <<-'EOF'
+
+
+		Options:
+		  -h, --help    show this message
+
+
+		Process all generate "*.mapentry" files, combine them all
+		together into the final "sitemap.xml" file.
+
+
+		Examples:
+
+		  Generate the final sitemap.xml:
+
+		    $ sitemap > src/sitemap.xml
+	EOF
+}
+
+
+for f in "$@"; do
+	case "$f" in
+		--)
+			break
+			;;
+		--help)
+			usage
+			help
+			exit
+			;;
+		*)
+			;;
+	esac
+done
+
+while getopts 'h' flag; do
+	case "$flag" in
+		h)
+			usage
+			help
+			exit
+			;;
+		*)
+			usage >&2
+			exit 2
+			;;
+	esac
+done
+shift $((OPTIND - 1))
+
+
+
+printf '<?xml version="1.0" encoding="UTF-8"?>\n'
+printf '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n'
+
+find "$CONTENT_PREFIX"/ -type f -name '*.mapentry' |
+	LANG=POSIX.UTF-8 sort |
+	xargs cat
+
+printf '</urlset>\n'
diff --git a/v2/src/lib/mapentry.xml b/v2/src/lib/mapentry.xml
new file mode 100644
index 0000000..5c715c0
--- /dev/null
+++ b/v2/src/lib/mapentry.xml
@@ -0,0 +1,5 @@
+  <url>
+    <loc>$url_absolute</loc>
+    <lastmod>$lastmod</lastmod>
+    <changefreq>monthly</changefreq>
+  </url>