Make the public html pages valid xhtml, and validate more RDFa cases.

[htsworkflow.git] / htsworkflow / util / ethelp.py

diff --git a/htsworkflow/util/ethelp.py b/htsworkflow/util/ethelp.py
index 19f6c9f8536d612c29406b2204b8f40c6ea53adb..e4fe897342acfe656d63c3155a51c715610a9dca 100644 (file)
--- a/htsworkflow/util/ethelp.py
+++ b/htsworkflow/util/ethelp.py
@@ -1,6 +1,15 @@
+"""ElementTree helper functions
 """
-ElementTree helper functions
-"""
+import logging
+import os
+LOGGER = logging.getLogger(__name__)
+
+import lxml.etree
+try:
+    XHTML_RDF_DTD = lxml.etree.DTD(external_id='-//W3C//DTD XHTML+RDFa 1.0//EN')
+except lxml.etree.DTDParseError as e:
+    LOGGER.warn("Unable to load XHTML DTD %s" % (str(e),))
+
 def indent(elem, level=0):
     """
     reformat an element tree to be 'pretty' (indented)
@@ -21,7 +30,7 @@ def indent(elem, level=0):
 
 def flatten(elem, include_tail=0):
     """
-    Extract the text from an element tree 
+    Extract the text from an element tree
     (AKA extract the text that not part of XML tags)
     """
     text = elem.text or ""
@@ -30,3 +39,36 @@ def flatten(elem, include_tail=0):
     if include_tail and elem.tail: text += elem.tail
     return text
 
+def validate_xhtml(html, base_url='http://localhost'):
+    """Helper for validating xhtml, mostly intended for test code
+
+    Defaults to assuming XHTML+RDFa
+    Returns None if there was a problem configuring validation
+    Logs messages from lxml.etree using python logging
+    Returns True if it passed validation
+    and False if it fails.
+    """
+    if XHTML_RDF_DTD is None:
+        return None
+
+    try:
+        root = lxml.etree.fromstring(html, base_url=base_url)
+    except lxml.etree.ParseError as e:
+        LOGGER.warn("Unable to parse document: %s" % (str(e),))
+        return False
+
+    if XHTML_RDF_DTD.validate(root):
+        # so unlikely.
+        return True
+
+    isgood = True
+    for msg in XHTML_RDF_DTD.error_log.filter_from_errors():
+        # I have no idea how to suppress this error
+        # but I need the xmlns attributes for of my RDFa 1.0 encoding
+        if 'ERROR:VALID:DTD_UNKNOWN_ATTRIBUTE' in str(msg):
+            continue
+        else:
+            LOGGER.error(msg)
+            isgood = False
+
+    return isgood