From db63b272c4064cfc541f89f70aa8354ee6920905 Mon Sep 17 00:00:00 2001 From: Oleg Broytman Date: Fri, 22 Nov 2013 01:57:53 +0400 Subject: [PATCH] Refactor attrs2text() to be completely generic --- xml | 44 ++++++++++++++------------------------------ xml-ANNOUNCE | 2 +- 2 files changed, 15 insertions(+), 31 deletions(-) diff --git a/xml b/xml index 7c7ffd5..e78e47e 100755 --- a/xml +++ b/xml @@ -144,7 +144,7 @@ class XmlVfs(object): subpath = '%s %s' % (template % n, tag) subpath_encoded = subpath.encode(default_encoding, "replace") print "dr-xr-xr-x 1 user group 0 Jan 1 00:00 %s" % subpath_encoded - if self.hasattrs(element): + if self.getattrs(element): attr_text = self.attrs2text(element) print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/attributes" % ( len(attr_text), subpath_encoded) @@ -158,6 +158,14 @@ class XmlVfs(object): len(text), subpath_encoded) self._list(element, subpath) + def attrs2text(self, node): + attr_accumulator = [] + for name, value in self.getattrs(node): + name = self.getlocalname(name).encode(default_encoding, "replace") + value = value.encode(default_encoding, "replace") + attr_accumulator.append("%s=%s" % (name, value)) + return '\n'.join(attr_accumulator) + def has_ns(self, node): return False @@ -175,16 +183,10 @@ class MiniDOMXmlVfs(XmlVfs): def parse(self): self.document = xml.dom.minidom.parse(sys.argv[2]) - def hasattrs(self, node): - return bool(node.attributes) - - def attrs2text(self, node): + def getattrs(self, node): attrs = node.attributes attrs = [attrs.item(i) for i in range(attrs.length)] - return '\n'.join(["%s=%s" % - (a.name.encode(default_encoding, "replace"), - a.value.encode(default_encoding, "replace")) - for a in attrs]) + return [(a.name, a.value) for a in attrs] def collect_text(self, node): text_accumulator = [] @@ -218,8 +220,8 @@ class MiniDOMXmlVfs(XmlVfs): if use_elementtree or use_lxml: class CommonEtreeXmlVfs(XmlVfs): - def hasattrs(self, node): - return bool(node.attrib) + def getattrs(self, node): + return node.attrib.items() def collect_text(self, node): text_accumulator = [] @@ -278,16 +280,6 @@ if use_elementtree: self.document = ET.parse(sys.argv[2], PIParser()) - def attrs2text(self, node): - attr_accumulator = [] - for name, value in node.attrib.items(): - name = name.encode(default_encoding, "replace") - value = value.encode(default_encoding, "replace") - if name.startswith('{'): - name = name.split('}', 1)[1] # Remove XML namespace - attr_accumulator.append("%s=%s" % (name, value)) - return '\n'.join(attr_accumulator) - def getlocalname(self, name): if name.startswith('{'): name = name.split('}', 1)[1] # Remove XML namespace @@ -301,14 +293,6 @@ if use_lxml: def parse(self): self.document = etree.parse(sys.argv[2]) - def attrs2text(self, node): - attr_accumulator = [] - for name, value in node.attrib.items(): - name = etree.QName(name).localname.encode(default_encoding, "replace") - value = value.encode(default_encoding, "replace") - attr_accumulator.append("%s=%s" % (name, value)) - return '\n'.join(attr_accumulator) - def has_ns(self, node): return bool(node.nsmap) @@ -370,7 +354,7 @@ def mcxml_copyout(): xml_error('Unknown file') if path_comp == 'attributes': - if xmlvfs.hasattrs(node): + if xmlvfs.getattrs(node): text = xmlvfs.attrs2text(node) else: xml_error('There are no attributes') diff --git a/xml-ANNOUNCE b/xml-ANNOUNCE index 98ff711..8595910 100644 --- a/xml-ANNOUNCE +++ b/xml-ANNOUNCE @@ -6,7 +6,7 @@ WHAT IS IT WHAT'S NEW in version 0.6.0 (2013-11-22) - Refactored _list() to be completely generic. + Refactored _list() and attrs2text() to be completely generic. WHAT'S NEW in version 0.5.0 (2013-11-19) Added lxml.etree-based implementation. -- 2.39.5