extracts longlog metadata including updated date

common/longlog/2024-01-14.gmi

@@ -1,6 +1,6 @@
 # Harari’s Sapiens on Religion
 
-Published: 14 Jan 2024
+Published on: 14 Jan 2024
 
 I’ve been slowly re-reading Yuval Noah Harari’s 2014 classic, Sapiens, which apart from being ridiculously over-scoped and hilariously under-evidenced, is proving delightfully entertaining.
 

http/Makefile

@@ -6,12 +6,12 @@ out: microlog blog $(shell find src -type f) $(shell find share -type f) $(shell
 microlog: src/microlog
 
 src/microlog: $(shell find -L blogs/microlog -type f) bin/blog
-	./bin/blog -i microlog -l microlog -t
+	./bin/blog -i microlog -l microlog -t -p gmi2upphtml
 
 blog: src/blog
 
 src/blog: $(shell find -L blogs/blog -type f) bin/blog
-	./bin/blog -i blog -l blog
+	./bin/blog -i blog -l blog -p 'cut_longlog_meta | gmi2upphtml' -a extract_longlog_meta
 
 .PHONY: clean
 clean:

http/bin/blog

@@ -7,10 +7,14 @@ export LAYOUTS="$(realpath share/layouts)"
 usage()
 {
   >&2 cat << USAGE
-usage: blog -i LOG [-l LAYOUT] [-t]
+usage: blog -i LOG [-p POSTPROCESS_CMD] [-a EXTRACT_ARGS_CMD] [-l LAYOUT] [-t]
 
 options:
 
+    -a EXTRACT_ARGS_CMD
+        If provided, a shell command to extract arguments to the layout from the
+        source files. The shell will be provided with a variable, POST_TITLE.
+
     -i LOG
         The name of the source directory under \$BLOGS where log posts live.
 
@@ -18,24 +22,36 @@ options:
         If provided, the layout under \$LAYOUTS to use for the generated
         upphtml files. If not provided, uses the default layout.
 
+    -p POSTPROCESS_CMD
+        If provided, a shell command to post-process the source files into
+        upphtml. The shell will be provided with a variable, POST_TITLE.
+
     -t
         If enabled, use filenames as post titles
 USAGE
 }
 
+extract_args_cmd="head -n 0 | echo \"POST_TITLE=\\\"\${POST_TITLE}\\\"\""
 log_name=""
 layout="default"
+postprocess_cmd="echo"
 use_filenames_for_titles=0
 
-while getopts "i:l:t" opt
+while getopts "a:i:l:p:t" opt
 do
   case $opt in
+    a)
+      extract_args_cmd="$OPTARG"
+      ;;
     i)
       log_name="$OPTARG"
       ;;
     l)
       layout_name="$OPTARG"
       ;;
+    p)
+      postprocess_cmd="$OPTARG"
+      ;;
     t)
       use_filenames_for_titles=1
       ;;
@@ -78,11 +94,12 @@ do
   if [ $use_filenames_for_titles -eq 1 ]
   then
     title="$name"
-    cat "$entry" | gmi2upphtml "$title" > "$dest"
+    cat "$entry" |  eval "POST_TITLE=\"${title}\" ${postprocess_cmd}" > "$dest"
   else
     title="$(grep --regexp "^\# " --max-count 1 "$entry" | cut -c3-)"
-    cat "$entry" | grep --regexp "^\# " --invert-match | gmi2upphtml "$title" > "$dest"
+    export POST_TITLE="$title"
+    cat "$entry" | grep --regexp "^\# " --invert-match | eval "$postprocess_cmd" > "$dest"
   fi
-  echo "${title}" > "${dest}.args"
+  cat "$entry" | eval "$extract_args_cmd" > "${dest}.args"
   echo "Processed ${entry#$(dirname ${BLOGS})/} to ${dest#$(dirname ${SRC})/}"
 done

http/bin/cut_longlog_meta

@@ -0,0 +1,34 @@
+#!/bin/sh -e
+
+meta=""
+read_published_on=0
+while read line
+do
+  meta="${meta}
+${line}"
+  if [ "$(echo "$line" | cut -c1-14)" = "Published on: " ]
+  then
+    read_published_on=1
+    break
+  fi
+done
+
+if [ "$read_published_on" -eq 0 ]
+then
+  echo "$meta"
+  exit 0
+fi
+
+read line
+if [ "$(echo "$line" | cut -c1-14)" = "Updated: " ]
+then
+  meta="${meta}
+${line}"
+else
+  echo "$line"
+fi
+
+while read line
+do
+  echo "$line"
+done

http/bin/extract_longlog_meta

@@ -0,0 +1,34 @@
+#!/bin/sh -e
+
+POST_TITLE="${POST_TITLE:-}"
+DATE_PUBLISHED=""
+DATE_UPDATED=""
+
+while read line
+do
+  if [ "$(echo "$line" | cut -c1-2)" = "# " ]
+  then
+    POST_TITLE="$(echo "$line" | cut -c3-)"
+  elif [ "$(echo "$line" | cut -c1-14)" = "Published on: " ]
+  then
+    DATE_PUBLISHED="$(date --date="$(echo "$line" | cut -c15-)" +%Y-%m-%d)"
+  elif [ "$(echo "$line" | cut -c1-12)" = "Updated on: " ]
+  then
+    DATE_UPDATED="$(date --date="$(echo "$line" | cut -c13-)" +%Y-%m-%d)"
+  fi
+done
+
+if [ -n "$POST_TITLE" ]
+then
+  echo "POST_TITLE=\"${POST_TITLE}\""
+fi
+
+if [ -n "$DATE_PUBLISHED" ]
+then
+  echo "DATE_PUBLISHED=\"${DATE_PUBLISHED}\""
+fi
+
+if [ -n "$DATE_UPDATED" ]
+then
+  echo "DATE_UPDATED=\"${DATE_UPDATED}\""
+fi

http/bin/gmi2upphtml

@@ -1,6 +1,6 @@
 #!/bin/sh -e
 
-post_title="${1:-}"
+post_title="${POST_TITLE:-}"
 
 is_in_list=0
 is_in_link_block=0
@@ -118,8 +118,9 @@ do
   if [ "$(echo "$line" | cut -c1-2)" = "=>" ]
   then
     close_all_open_blocks_except link
-    href="$(escape_upphtml "$(echo "$line" | sed "s/^=>[[:space:]]*\([[:alnum:][:punct:]]\+\).*$/\1/")")"
+    href="$(escape_upphtml "$(echo "$line" | sed "s/^=>[[:space:]]\+//" | sed "s/[[:space:]]\+.*//")")"
-    title="$(escape_upphtml "$(echo "$line" | sed "s/^=>[[:space:]]*[[:alnum:][:punct:]]\+[[:space:]]\(.*\)$/\1/")")"
+    title="$(escape_upphtml "${line#=> *${href}}")"
+    title="${title:-${href}}"
     if [ $is_in_link_block -eq 0 ]
     then
       echo "<p>"

http/bin/mkws

@@ -51,14 +51,31 @@ do
   mkdir -p "${OUT}/${rel_dir}/"
 
   args="$( [ -f "${t}.args" ] && cat "${t}.args" || echo "" )"
+  if [ "$(head -n 1 "${layout}")" = "#!" ]
+  then
+    >"${layout}.tmp" echo "
+#!
+${args}
+$(tail -n +2 "${layout}")
+"
+  else
+    >"${layout}.tmp" echo "
+#!
+${args}
+#!
+$(cat "${layout}")
+"
+  fi
   if [ "$ext" = "html" ]
   then
-    pp "${layout}" "${t}" "$1" "$args" > \
+    pp "${layout}.tmp" "${t}" "${1}" > \
   		"${OUT}/${rel_dir}/${name}.${ext}"
 	else
-		pp "${t}" "$1" "$args" > \
+    pp "${t}" "${1}" > \
   		"${OUT}/${rel_dir}/${name}.${ext}"
   fi
+  rm "${layout}.tmp"
+
   echo "Written ${OUT}/${rel_dir}${name}.${ext}"
 done
 

http/share/layouts/blog.layout.upphtml

@@ -1,8 +1,7 @@
 #!
 SRC="$1"
 SITE_URL="$2"
-POST_TITLE="$3"
+DATE_PUBLISHED="${DATE_PUBLISHED:-$(basename "${SRC}" | cut -c1-10)}"
-DATE_PUBLISHED="$(basename "${SRC}" | cut -c1-10)"
 CANONICAL_URL="${SITE_URL}/blog/$(basename "${SRC}")"
 #!
 
@@ -25,20 +24,35 @@ pp "${SHARE}"/components/navbar.upphtml
 
 <aside>
 <span>This is a blog post by <a class='p-author h-card' href='/'>Joe Carstairs</a>.</span>
+#!
+if [ -z "${DATE_UPDATED}" ]; then
+#!
 <span>Published:
 #!
 DATE="${DATE_PUBLISHED}" CLASS="dt-published" pp "${SHARE}/components/formatted_date.upphtml"
 #!
 .</span>
+#!
+else
+#!
+<span>Updated:
+#!
+DATE="${DATE_UPDATED}" CLASS="dt-updated" pp "${SHARE}/components/formatted_date.upphtml"
+#!
+.</span>
+<span>Originally published:
+#!
+DATE="${DATE_PUBLISHED}" CLASS="dt-published" pp "${SHARE}/components/formatted_date.upphtml"
+#!
+.</span>
+#!
+fi
+#!
 <span hidden><a class='u-url uid' href='${CANONICAL_URL}'>Permalink</a></span>
 </aside>
 
 <header>
-<h1 class='h-name'>
+<h1 class='h-name'>${POST_TITLE}</h1>
-#!
-echo "${POST_TITLE}"
-#!
-</h1>
 </header>
 
 <section class='e-content'>

http/share/layouts/microlog.layout.upphtml

@@ -1,7 +1,6 @@
 #!
 SRC="$1"
 SITE_URL="$2"
-POST_TITLE="$3"
 DATE_PUBLISHED="$(basename "${SRC}" | cut -c1-10)"
 CANONICAL_URL="${SITE_URL}/microlog/$(basename "${SRC}")"
 #!