From a0a764502272b3a5303ddf439149a5bec6acb421 Mon Sep 17 00:00:00 2001
From: Uriel <u@berlinblue.org>
Date: Fri, 6 Feb 2009 20:42:15 +0100
Subject: [PATCH 1/3] Make fproc_cache take an optional second arg that can be
 a file or dir name, if provided that file/path will be used as input for the
 first argument removing the need for a temporary file.

---
 bin/fproc_cache.rc | 27 +++++++++++++++++++++------
 1 file changed, 21 insertions(+), 6 deletions(-)

diff --git a/bin/fproc_cache.rc b/bin/fproc_cache.rc
index 11619e6..55096b9 100755
--- a/bin/fproc_cache.rc
+++ b/bin/fproc_cache.rc
@@ -1,19 +1,34 @@
 #!/usr/bin/env rc
 #. 9.rc # Not really needed when calling from werc, only would be needed if you use fproc_cache.rc standalone
-# TODO: Allow to provide two arguments, second arg is a file name, use filename+size+mtime for hashing.
-#    Useful when dealing with many big files (eg., thumb gallery generation).
+
+a=()
+tmpf=()
 
 proc=$1
-tmpfile=/tmp/fmttmp.$pid
-score=`{{tee $tmpfile || exit 1} | sha1sum}
+shift
 
+if(~ $#* 0) {
+    tmpf=/tmp/fmttmp.$pid
+    f=$tmpf
+    score=`{{tee $tmpf || exit 1} | sha1sum}
+}
+if not {
+    f=$1
+    if(~ $f */) {
+        score=`{du -an $f | sha1sum || exit 1} # XXX using -n(bytes) instead of -t(lastmod) because sitemap proc touches files in tree.
+        a=$f
+        f=/dev/null
+    }
+    if not
+        score=`{sha1sum $f || exit 1}
+}
 cachedir=/tmp/fproc_cache/$score
 mkdir -p $cachedir >[2]/dev/null
 
 if(test -s $cachedir/$proc)
     cat $cachedir/$proc
 if not
-    if($proc < $tmpfile | tee $cachedir/$pid)
+    if($proc $a < $f | tee $cachedir/$pid)
         mv $cachedir/$pid $cachedir/$proc
 
-rm $tmpfile $cachedir/$pid >[2]/dev/null &
+rm $tmpf $cachedir/$pid >[2]/dev/null &

From 9ff545ec83cb90adeab414f886697a56dd332853 Mon Sep 17 00:00:00 2001
From: Uriel <u@berlinblue.org>
Date: Fri, 6 Feb 2009 20:43:11 +0100
Subject: [PATCH 2/3] Rewrite most sitemap generation code, much cleaner,
 modularize file title/desc extraction, take advantage of new fproc_cache to
 cache sitemap.

---
 lib/sitemap.tpl | 80 +++++++++++++++++++++++--------------------------
 1 file changed, 38 insertions(+), 42 deletions(-)

diff --git a/lib/sitemap.tpl b/lib/sitemap.tpl
index 3ade1e9..3818233 100644
--- a/lib/sitemap.tpl
+++ b/lib/sitemap.tpl
@@ -5,65 +5,61 @@ tmpfile=/tmp/werc_sitemap_$pid.txt
 echo '' > $tmpfile
 saveddf=$dirfilter
 
-fn getMdDesc {
-    sed 's/^(.......................................................................................................[^ ]*).*$/\1/g; 1q' < $1 
+fn get_md_title {
+    sed 's/^(................................................................[^ ]*).*$/\1/g; 1q' < $1 
+}
+
+fn get_html_title {
+    # H1 is not reliable because htmlroff doesn't use it :(
+    #desc=`{cat $1 | sed 32q | grep '<[Hh]1>' | sed 's/<[Hh]1>(.*)(<\/[Hh]1>|$)/\1/;s/<[^>]*>//g;1q'}
+    # Pick the first line of body  instead
+    desc=`{sed -n '/<[Bb][Oo][Dd][Yy]/,/./s/(<[^>]*>|$)//gp' < $1}
+    if(~ $#desc 0)
+        desc=`{sed 's/<[^>]*>//g; 1q' < $1}
+}
+
+fn get_file_title {
+        
+    if(~ $1 */) {
+        if(test -f $1/index.md)
+            get_md_title $1/index.md
+        if not if(test -f $1/index.html)
+            get_html_title $1/index.html
+    }
+    if not if(~ $1 *.md)
+        get_md_title $1
+    if not if(~ $1 *.html)
+        get_html_title $1
+    if not
+        echo ''
 }
 
-# XXX Instead of recursion should use du(1) or similar.
 fn listDir {
     d=$1
-    if(~ $#d 0)
-        d=''
     dirfilter=$saveddf
-    blogDirs=()
     if(test -f $d/_werc/config)
         . ./$d/_werc/config
 
     echo '<ul class="sitemap-list">'
 
-    # Don't hide blog dirs for now
-    #if(! ~ $#blogDirs 0 || ~ $1 */blog */Blog)
-    #    echo '' 
-    #if not 
-    if(! ~ $#redirectPermanent 1)
-    {
-
-    for(i in `{ls -dF $d^*/ $d^*.md $d^*.html $d^*.txt >[2]/dev/null | sed $dirfilter$dirclean}) {
-        desc=''
-        if(test -f $i.md)
-            desc=`{getMdDesc $i.md}
-        if not if(~ $i */ && test -f $i/index.md)
-            desc=`{getMdDesc $i/index.md}
-        if not if(test -f $i.html) {
-            # H1 is not reliable because htmlroff doesn't use it :(
-            #desc = `{cat $i.html |sed 32q | grep '<[Hh]1>' |sed 's/<[Hh]1>(.*)(<\/[Hh]1>|$)/\1/;s/<[^>]*>//g;1q'}
-            # Pick the first line of body  instead
-            desc=`{sed -n '/<[Bb][Oo][Dd][Yy]/,/./s/(<[^>]*>|$)//gp' < $i.html}
-            if(~ $#desc 0)
-                desc=`{sed 's/<[^>]*>//g; 1q' < $i.html}
-            #desc=`{/bin/sed -e '0,/<[Bb][Oo][Dd][Yy]/d;s/<[^>]*>//g;/^$/d' < $i.html >[2]/dev/null | sed 1q}
-        }
-
+    for(i in `{ls -dF $d^*/ $d^*.md $d^*.html $d^*.txt >[2]/dev/null | sed $dirfilter}) {
+        desc=`{get_file_title $i}
+        u=`{echo $i|sed 's!'$sitedir'!!; '$dirclean's!/index$!/!; '}
         if(! ~ $#desc 0 && ! ~ $desc '')
             desc=' - '$"desc
-        tit=`{echo /$i|sed 's/_/ /g; s,.*/([^/]+)/?$,\1,'}
-        echo '<li><a href="/'$i'">'^$"tit^'</a>' $"desc '</li>' 
-        echo -n $base_url^$i >> $tmpfile
-        if(test -d $i) {
-            echo / >> $tmpfile	
+        n=`{echo /$u|sed 's/_/ /g; s,.*/([^/]+)/?$,\1,'}
+        echo '<li><a href="'$base_url$u'">'^$"n^'</a>' $"desc '</li>' 
+        echo $base_url^$u >> $tmpfile
+        if(test -d $i)
             @{ listDir $i }
-        }
-        if not
-            echo >> $tmpfile
-    }
     }
     echo '</ul>'
 }
 
-cd $sitedir 
-listDir ''
+fproc_cache.rc listDir $sitedir/
+#listDir $sitedir/
 
-cp $tmpfile ./sitemap.txt
-rm $tmpfile
+if(test -s $tmpfile)
+    mv $tmpfile $sitedir/sitemap.txt &
 
 %}

From 2e1f0dd32f2ee4794044567ea9c3a8ef091694ef Mon Sep 17 00:00:00 2001
From: Uriel <u@berlinblue.org>
Date: Fri, 6 Feb 2009 22:09:32 +0100
Subject: [PATCH 3/3] Convert fltr_cache into a function. Enable output
 buffering which had been commented out for unknwon reasons.

---
 bin/fltr_cache.rc  | 35 +++++++++++++++++++++++++++++++++++
 bin/fproc_cache.rc | 34 ----------------------------------
 bin/werc.rc        |  3 ++-
 etc/initrc         | 12 ++++++------
 lib/sitemap.tpl    |  2 +-
 5 files changed, 44 insertions(+), 42 deletions(-)
 create mode 100755 bin/fltr_cache.rc
 delete mode 100755 bin/fproc_cache.rc

diff --git a/bin/fltr_cache.rc b/bin/fltr_cache.rc
new file mode 100755
index 0000000..f09c963
--- /dev/null
+++ b/bin/fltr_cache.rc
@@ -0,0 +1,35 @@
+#!/usr/bin/env rc
+
+fn fltr_cache {
+    a=()
+    tmpf=()
+
+    proc=$1
+    shift
+
+    if(~ $#* 0) {
+        tmpf=/tmp/fmttmp.$pid
+        f=$tmpf
+        score=`{{tee $tmpf || exit 1} | sha1sum}
+    }
+    if not {
+        f=$1
+        if(~ $f */) {
+            score=`{du -an $f | sha1sum || exit 1} # XXX using -n(bytes) instead of -t(lastmod) because sitemap proc touches files in tree.
+            a=$f
+            f=/dev/null
+        }
+        if not
+            score=`{sha1sum $f || exit 1}
+    }
+    cachedir=/tmp/fltr_cache/$score
+    mkdir -p $cachedir >[2]/dev/null
+
+    if(test -s $cachedir/$proc)
+        cat $cachedir/$proc
+    if not
+        if($proc $a < $f | tee $cachedir/$pid)
+            mv $cachedir/$pid $cachedir/$proc
+
+    rm $tmpf $cachedir/$pid >[2]/dev/null &
+}
diff --git a/bin/fproc_cache.rc b/bin/fproc_cache.rc
deleted file mode 100755
index 55096b9..0000000
--- a/bin/fproc_cache.rc
+++ /dev/null
@@ -1,34 +0,0 @@
-#!/usr/bin/env rc
-#. 9.rc # Not really needed when calling from werc, only would be needed if you use fproc_cache.rc standalone
-
-a=()
-tmpf=()
-
-proc=$1
-shift
-
-if(~ $#* 0) {
-    tmpf=/tmp/fmttmp.$pid
-    f=$tmpf
-    score=`{{tee $tmpf || exit 1} | sha1sum}
-}
-if not {
-    f=$1
-    if(~ $f */) {
-        score=`{du -an $f | sha1sum || exit 1} # XXX using -n(bytes) instead of -t(lastmod) because sitemap proc touches files in tree.
-        a=$f
-        f=/dev/null
-    }
-    if not
-        score=`{sha1sum $f || exit 1}
-}
-cachedir=/tmp/fproc_cache/$score
-mkdir -p $cachedir >[2]/dev/null
-
-if(test -s $cachedir/$proc)
-    cat $cachedir/$proc
-if not
-    if($proc $a < $f | tee $cachedir/$pid)
-        mv $cachedir/$pid $cachedir/$proc
-
-rm $tmpf $cachedir/$pid >[2]/dev/null &
diff --git a/bin/werc.rc b/bin/werc.rc
index a274c7c..709d7b5 100755
--- a/bin/werc.rc
+++ b/bin/werc.rc
@@ -3,6 +3,7 @@
 . ./werclib.rc
 . ./wercconf.rc
 . ./corehandlers.rc
+. ./fltr_cache.rc
 cd ..
 
 forbidden_uri_chars='[^a-zA-Z0-9_+\-\/\.]'
@@ -105,7 +106,7 @@ fn werc_exec_request {
     if(! ~ $#debug 0)
         dprint $"SERVER_NAME^$"REQUEST_URI - $"HTTP_USER_AGENT - $"REQUEST_METHOD - $"handler_body_main - $"master_template
 
-    template $headers $master_template #| awk_buffer
+    template $headers $master_template | awk_buffer
     echo $res_tail
 }
 
diff --git a/etc/initrc b/etc/initrc
index c02e0f2..d5c9f57 100644
--- a/etc/initrc
+++ b/etc/initrc
@@ -9,17 +9,17 @@
 
 # Location of your Plan 9 from User Space installation (usually /usr/local/plan9)
 plan9port=$PLAN9
+#plan9port=/usr/local/plan9
 
 # Path, make sure the plan9port /bin directory is included before /bin
 # Keep '.' in path! It is needed.
 path=($plan9port/bin/ . ./bin/ ./bin/contrib/ /bin/ /usr/bin/)
 
-# Set this to your favorite markdown formatter, eg., markdown.pl (fproc_cache
-# is a wrapper around a script it takes as an argument, in the default
-# configuration markdown.pl, that caches output)
-# Note that some werc components assume a markdown-like formatter, but all
-# major functionality should should be formatter agnostic.
-formatter=(fproc_cache.rc markdown.pl)
+# Set this to your favorite markdown formatter, eg., markdown.pl (fltr_cache
+# takes as an argument a filter, in the default configuration markdown.pl, that
+# caches output) Note that some werc components assume a markdown-like
+# formatter, but all major functionality should should be formatter agnostic.
+formatter=(fltr_cache markdown.pl)
 
 # Enable debugging, to disable set to ()
 debug=true
diff --git a/lib/sitemap.tpl b/lib/sitemap.tpl
index 3818233..7967856 100644
--- a/lib/sitemap.tpl
+++ b/lib/sitemap.tpl
@@ -56,7 +56,7 @@ fn listDir {
     echo '</ul>'
 }
 
-fproc_cache.rc listDir $sitedir/
+fltr_cache listDir $sitedir/
 #listDir $sitedir/
 
 if(test -s $tmpfile)