Improve feed update dating code, add fallback to feed uuid file lastmod.

[redo-blog] / processor / feed.xml.do

diff --git a/processor/feed.xml.do b/processor/feed.xml.do
index 2cdb7b20572f8ce86ede6331e765d788ab4e5e4b..513d538e08f416717b1a8b1c83bd94f52896f270 100644 (file)
--- a/processor/feed.xml.do
+++ b/processor/feed.xml.do
@@ -2,21 +2,25 @@
 
 # Pull in global dependencies.
 . ./helpers.sh
-redo-ifchange url
-redo-ifchange author
-redo-ifchange uuid
-redo-ifchange title 
+metadata_dir=metadata
+author_file="$metadata_dir"/author
+uuid_file="$metadata_dir"/uuid
+title_file="$metadata_dir"/title
+url_file="$metadata_dir"/url
+redo-ifchange "$url_file"
+redo-ifchange "$author_file"
+redo-ifchange "$uuid_file"
+redo-ifchange "$title_file"
 
 # Build some variables. XML-escape even file contents that should not contain
 # dangerous characters, just to avoid any XML trouble.
-base_url=`cat url | head -1`
-url_protocol=`echo $base_url | cut -d ':' -f 1`
-url_basepath=`echo $base_url | cut -d '/' -f 3-`
-url_basepath_escaped=`escape_url "$url_basepath"`
-basepath="$url_protocol""://""$url_basepath_escaped"
-title=`read_and_escape_file title | head -1`
-author=`read_and_escape_file author | head -1`
-uuid=`read_and_escape_file uuid | head -1`
+srcdir=`pwd`
+basepath=$(get_basepath "${metadata_dir}/")
+title=`read_and_escape_file "$title_file" | head -1`
+author=`read_and_escape_file "$author_file" | head -1`
+uuid=`read_and_escape_file "$uuid_file" | head -1`
+tmp_snippets_dir=.tmp_feed_snippets
+feed_gen_date=$(stat -c%Y "${uuid_file}")
 
 # Write majority of feed head.
 cat << EOF
@@ -29,42 +33,43 @@ printf "<title type=\"html\">%s</title>\n" "$title"
 printf "<author><name>%s</name></author>\n" "$author"
 printf "<id>urn:uuid:%s</id>\n" "$uuid"
 
-# Iterate through most recent entries (go by lastmod date of source files) to
-# build feed head "updated" element, and individual entries.
-first_run=0
-files=`ls -1t *.rst *.md | head -10 | tr '\n' $'\0'`
-oldIFS="$IFS"
-IFS=$'\0'
-for file in $files; do
-  lastmod=`stat -c%y "$file"`
-  lastmod_rfc3339=`date -u "+%Y-%m-%dT%TZ" -d "$lastmod"`
-  if [ "$first_run" -lt "1" ]; then
-    IFS="$oldIFS"
-    printf "<updated>%s</updated>\n\n" "$lastmod_rfc3339" 
-    first_run=1
+# Generate feed entry snippets.
+mkdir -p "$tmp_snippets_dir"
+for file in ./*.rst ./*.md; do
+  if [ -e "$file" ]; then
+    uuid_file="${metadata_dir}/${file%.*}.uuid"
+    redo-ifchange "$uuid_file"
+    published=$(stat -c%Y "${uuid_file}")
+    snippet_file=./${metadata_dir}/"${file%.*}.feed_snippet"
+    redo-ifchange "$snippet_file"
+    ln -s "$srcdir/$snippet_file" "./${tmp_snippets_dir}/${published}"
   fi
+done
 
-  # Build some variables and dependencies.
-  intermediate_file="${file%.*}.intermediate"
-  htmlfile=`escape_url "${file%.*}.html"`
-  redo-ifchange "$intermediate_file"
-  redo-ifchange "$uuidfile"
-  title=`read_and_escape_file "$intermediate_file" | head -1`
-  uuidfile="${file%.*}.uuid"
-  uuid=`read_and_escape_file "$uuidfile" | head -1`
-  body=`read_and_escape_file "$intermediate_file" | sed 1d`
-  published=`stat -c%y "$uuidfile"`
-  published_rfc3339=`date -u "+%Y-%m-%dT%TZ" -d "$published"`
+# Derive feed modification date from snippets. Fallback to uuid file mod date.
+n_snippet_files=`ls -1 ./${metadata_dir}/*.feed_snippet 2>/dev/null | wc -l`
+if [ $n_snippet_files != 0 ]
+then
+  mod_dates=$(grep -hE "^<updated>" ./${metadata_dir}/*.feed_snippet | sed -E 's/<.?updated>//g')
+fi
+last_mod_unix=$feed_gen_date
+for date in $mod_dates; do
+  date_unix=$(date -u "+%s" -d "${date}")
+  if [ "$date_unix" -gt "$last_mod_unix" ]; then
+    last_mod_unix=$date_unix
+  fi
+done
+last_mod_rfc3339=`date -u "+%Y-%m-%dT%TZ" -d "@${last_mod_unix}"`
+printf "<updated>%s</updated>\n\n" "$last_mod_rfc3339"
 
-  # Write entry.
-  printf "<entry>\n"
-  printf "<title type=\"html\">%s</title>\n" "$title"
-  printf "<id>urn:uuid:%s</id>\n" "$uuid" 
-  printf "<updated>%s</updated>\n" "$lastmod_rfc3339" 
-  printf "<published>%s</published>\n" "$published_rfc3339" 
-  printf "<link href=\"%s%s\" />\n" "$basepath" "$htmlfile"
-  printf "<content type=\"html\">\n%s\n</content>\n" "$body"
-  printf "</entry>\n\n"
+# Write feed entries.
+for file in ./${tmp_snippets_dir}/*; do
+  if [ -e "$file" ]; then
+    cat "${file}"
+    printf "\n"
+  fi
 done
+rm -rf "$tmp_snippets_dir"
 
+# Write feed foot.
 printf "</feed>"