Add minhead filter, and resync opml-top100.ini

examples/opml-top100.ini

@@ -31,13 +31,18 @@ activity_threshold = 90
 # filters to be run
 filters = excerpt.py
 
+bill_of_materials:
+  .htaccess
+  favicon.ico
+  robots.txt
+
 # filter parameters
 [excerpt.py]
 omit = img p br
 width = 500
 
 # add memes to output
-[index.html.tmpl]
+[index.html.xslt]
 filters = mememe.plugin
 
 [mememe.plugin]

filters/minhead.py

@@ -0,0 +1,36 @@
+#
+# Ensure that all headings are below a permissible maximum (like h3).
+# If not, all heading levels will be changed to conform.
+# Note: this may create "illegal" heading levels, like h7 and beyond.
+#
+
+import sys
+from xml.dom import minidom, XHTML_NAMESPACE
+
+# determine permissible minimimum heading
+if '--min' in sys.argv:
+  minhead = int(sys.argv[sys.argv.index('--min')+1])
+else:
+  minhead=3
+
+# parse input stream
+doc = minidom.parse(sys.stdin)
+
+# search for headings below the permissable minimum
+first=minhead
+for i in range(1,minhead):
+  if doc.getElementsByTagName('h%d' % i):
+    first=i
+    break
+
+# if found, bump all headings so that the top is the permissible minimum
+if first < minhead:
+  for i in range(6,0,-1):
+    for oldhead in doc.getElementsByTagName('h%d' % i):
+      newhead = doc.createElementNS(XHTML_NAMESPACE, 'h%d' % (i+minhead-first))
+      for child in oldhead.childNodes:
+        newhead.appendChild(child)
+      oldhead.parentNode.replaceChild(newhead, oldhead)
+
+# return (possibly modified) document
+print doc.toxml('utf-8')

tests/data/filter/minhead.ini

@@ -0,0 +1,3 @@
+[Planet]
+filters = minhead.py?min=3
+filter_directories = filters

tests/data/filter/minhead.xml

@@ -0,0 +1,3 @@
+<div xmlns="http://www.w3.org/1999/xhtml">
+<h1>title</h1><h3>mid</h3><h5>bottom</h5>
+</div>