def parse_xml(cls, el, source_location):
"""
Creates a Rule object from a parsed XML <rule> element.
"""
assert el.tag == 'rule'
classes = el.get('class', '').split()
if not classes:
classes = ['default']
theme = None
actions = []
suppress_standard = asbool(el.get('suppress-standard'))
for child in el.iterchildren():
if child.tag == 'theme':
## FIXME: error if more than one theme
theme = Theme.parse_xml(child, source_location)
continue
if child.tag is etree.Comment:
continue
action = parse_action(child, source_location)
actions.append(action)
match = None
inst = cls(classes, actions, theme, match, suppress_standard, source_location)
for attr in RuleMatch.match_attrs:
if el.get(attr):
inst.match = RuleMatch.parse_xml(inst, el, source_location)
## FIXME: would last="1" be a good alternative to suppress-standard?
break
return inst
def parse_xml(cls, doc, source_location):
"""
Parses the given XML/etree document into an instance of this
class.
"""
assert doc.tag == 'ruleset'
matchers = []
clientsides = []
rules = []
default_theme = None
for el in doc.iterchildren():
if el.tag == 'match':
matcher = Match.parse_xml(el, source_location)
matchers.append(matcher)
elif el.tag == 'clientside':
matcher = ClientsideMatch.parse_xml(el, source_location)
clientsides.append(matcher)
elif el.tag == 'rule':
rule = Rule.parse_xml(el, source_location)
rules.append(rule)
elif el.tag == 'theme':
## FIXME: Add parse error
default_theme = Theme.parse_xml(el, source_location)
elif el.tag in ('proxy', 'server-settings', Comment):
# Handled elsewhere, so we just ignore this element
continue
else:
## FIXME: source location?
raise DeliveranceSyntaxError(
"Invalid tag %s (unknown tag name %r)" %
(tostring(el).split('>', 1)[0] + '>', el.tag),
element=el)
rules_by_class = {}
for rule in rules:
for class_name in rule.classes:
rules_by_class.setdefault(class_name, []).append(rule)
return cls(matchers,
clientsides,
rules_by_class,
default_theme=default_theme,
source_location=source_location)
def parse_xml(cls, doc, source_location):
"""
Parses the given XML/etree document into an instance of this
class.
"""
assert doc.tag == 'ruleset'
matchers = []
clientsides = []
rules = []
default_theme = None
for el in doc.iterchildren():
if el.tag == 'match':
matcher = Match.parse_xml(el, source_location)
matchers.append(matcher)
elif el.tag == 'clientside':
matcher = ClientsideMatch.parse_xml(el, source_location)
clientsides.append(matcher)
elif el.tag == 'rule':
rule = Rule.parse_xml(el, source_location)
rules.append(rule)
elif el.tag == 'theme':
## FIXME: Add parse error
default_theme = Theme.parse_xml(el, source_location)
elif el.tag in ('proxy', 'server-settings', Comment):
# Handled elsewhere, so we just ignore this element
continue
else:
## FIXME: source location?
raise DeliveranceSyntaxError(
"Invalid tag %s (unknown tag name %r)"
% (tostring(el).split('>', 1)[0]+'>', el.tag),
element=el)
rules_by_class = {}
for rule in rules:
for class_name in rule.classes:
rules_by_class.setdefault(class_name, []).append(rule)
return cls(matchers, clientsides, rules_by_class, default_theme=default_theme,
source_location=source_location)
def apply_rules(self, req, resp, resource_fetcher, log, default_theme=None):
"""
Apply the whatever the appropriate rules are to the request/response.
"""
extra_headers = parse_meta_headers(resp.body)
if extra_headers:
response_headers = HeaderDict(resp.headerlist + extra_headers)
else:
response_headers = resp.headers
try:
classes = run_matches(self.matchers, req, resp, response_headers, log)
except AbortTheme:
return resp
if 'X-Deliverance-Page-Class' in response_headers:
log.debug(self, "Found page class %s in headers", response_headers['X-Deliverance-Page-Class'].strip())
classes.extend(response_headers['X-Deliverance-Page-Class'].strip().split())
if 'deliverance.page_classes' in req.environ:
log.debug(self, "Found page class in WSGI environ: %s", ' '.join(req.environ["deliverance.page_classes"]))
classes.extend(req.environ['deliverance.page_classes'])
if not classes:
classes = ['default']
rules = []
theme = None
for class_name in classes:
## FIXME: handle case of unknown classes
## Or do that during compilation?
for rule in self.rules_by_class.get(class_name, []):
if rule not in rules:
rules.append(rule)
if rule.theme:
theme = rule.theme
if theme is None:
theme = self.default_theme
if theme is None and default_theme is not None:
theme = Theme(href=default_theme,
source_location=self.source_location)
if theme is None:
log.error(self, "No theme has been defined for the request")
return resp
try:
theme_href = theme.resolve_href(req, resp, log)
original_theme_resp = self.get_theme_response(
theme_href, resource_fetcher, log)
theme_doc = self.get_theme_doc(
original_theme_resp, theme_href,
should_escape_cdata=True,
should_fix_meta_charset_position=True)
resp = force_charset(resp)
body = resp.unicode_body
body = escape_cdata(body)
body = fix_meta_charset_position(body)
content_doc = self.parse_document(body, req.url)
run_standard = True
for rule in rules:
if rule.match is not None:
matches = rule.match(req, resp, response_headers, log)
if not matches:
log.debug(rule, "Skipping <rule>")
continue
rule.apply(content_doc, theme_doc, resource_fetcher, log)
if rule.suppress_standard:
run_standard = False
if run_standard:
## FIXME: should it be possible to put the standard rule in the ruleset?
standard_rule.apply(content_doc, theme_doc, resource_fetcher, log)
except AbortTheme:
return resp
remove_content_attribs(theme_doc)
## FIXME: handle caching?
if original_theme_resp.body.strip().startswith("<!DOCTYPE"):
tree = theme_doc.getroottree()
else:
tree = content_doc.getroottree()
if "XHTML" in tree.docinfo.doctype:
method = "xml"
else:
method = "html"
theme_str = tostring(theme_doc, include_meta_content_type=True)
theme_str = tree.docinfo.doctype + theme_str
theme_doc = document_fromstring(theme_str)
tree = theme_doc.getroottree()
resp.body = tostring(tree, method=method, include_meta_content_type=True)
resp.body = unescape_cdata(resp.body)
return resp
def apply_rules(self, req, resp, resource_fetcher, log, default_theme=None):
"""
Apply the whatever the appropriate rules are to the request/response.
"""
extra_headers = parse_meta_headers(resp.body)
if extra_headers:
response_headers = ResponseHeaders(resp.headerlist + extra_headers)
else:
response_headers = resp.headers
try:
classes = run_matches(self.matchers, req, resp, response_headers, log)
except AbortTheme:
return resp
if 'X-Deliverance-Page-Class' in response_headers:
log.debug(self, "Found page class %s in headers", response_headers['X-Deliverance-Page-Class'].strip())
classes.extend(response_headers['X-Deliverance-Page-Class'].strip().split())
if 'deliverance.page_classes' in req.environ:
log.debug(self, "Found page class in WSGI environ: %s", ' '.join(req.environ["deliverance.page_classes"]))
classes.extend(req.environ['deliverance.page_classes'])
if not classes:
classes = ['default']
rules = []
theme = None
for class_name in classes:
## FIXME: handle case of unknown classes
## Or do that during compilation?
for rule in self.rules_by_class.get(class_name, []):
if rule not in rules:
rules.append(rule)
if rule.theme:
theme = rule.theme
if theme is None:
theme = self.default_theme
if theme is None and default_theme is not None:
theme = Theme(href=default_theme,
source_location=self.source_location)
if theme is None:
log.error(self, "No theme has been defined for the request")
return resp
try:
theme_href = theme.resolve_href(req, resp, log)
original_theme_resp = self.get_theme_response(
theme_href, resource_fetcher, log)
theme_doc = self.get_theme_doc(
original_theme_resp, theme_href,
should_escape_cdata=True,
should_fix_meta_charset_position=True)
resp = force_charset(resp)
body = resp.unicode_body
body = escape_cdata(body)
body = fix_meta_charset_position(body)
content_doc = self.parse_document(body, req.url)
run_standard = True
for rule in rules:
if rule.match is not None:
matches = rule.match(req, resp, response_headers, log)
if not matches:
log.debug(rule, "Skipping <rule>")
continue
rule.apply(content_doc, theme_doc, resource_fetcher, log)
if rule.suppress_standard:
run_standard = False
if run_standard:
## FIXME: should it be possible to put the standard rule in the ruleset?
standard_rule.apply(content_doc, theme_doc, resource_fetcher, log)
except AbortTheme:
return resp
remove_content_attribs(theme_doc)
## FIXME: handle caching?
if original_theme_resp.body.strip().startswith("<!DOCTYPE"):
tree = theme_doc.getroottree()
else:
tree = content_doc.getroottree()
if "XHTML" in tree.docinfo.doctype:
method = "xml"
else:
method = "html"
theme_str = tostring(theme_doc, include_meta_content_type=True)
theme_str = tree.docinfo.doctype + theme_str
theme_doc = document_fromstring(theme_str)
tree = theme_doc.getroottree()
resp.body = tostring(tree, method=method, include_meta_content_type=True)
resp.body = unescape_cdata(resp.body)
return resp