Refactor _list() to be completely generic
authorOleg Broytman <phd@phdru.name>
Thu, 21 Nov 2013 21:48:09 +0000 (01:48 +0400)
committerOleg Broytman <phd@phdru.name>
Thu, 21 Nov 2013 21:48:09 +0000 (01:48 +0400)
xml
xml-ANNOUNCE

diff --git a/xml b/xml
index 5ed82db..7c7ffd5 100755 (executable)
--- a/xml
+++ b/xml
@@ -37,13 +37,13 @@ The VFS was inspired by a FUSE xmlfs: https://github.com/halhen/xmlfs
 
 """
 
-__version__ = "0.5.1"
+__version__ = "0.6.0"
 __author__ = "Oleg Broytman <phd@phdru.name>"
 __copyright__ = "Copyright (C) 2013 PhiloSoft Design"
 __license__ = "GPL"
 
-default_implementation = None  # Can be None for default choice,
-                               # 'lxml', 'elementtree' or 'minidom'
+force_implementation = None  # Can be None for default choice,
+                             # 'lxml', 'elementtree' or 'minidom'
 
 use_minidom = True
 use_elementtree = False
@@ -115,12 +115,49 @@ locale.setlocale(locale.LC_ALL, '')
 
 
 class XmlVfs(object):
+    """Abstract base class"""
+
+    supports_namespaces = False
+
     def __init__(self):
         self.parse()
 
     def list(self):
         self._list(self.getroot())
 
+    def _list(self, node, path=''):
+        n = len(self.getchildren(node))
+        if n:
+            width = int(math.log10(n)) + 1
+            template = "%%0%dd" % width
+        else:
+            template = "%d"
+        n = 0
+        for element in self.getchildren(node):
+            if not self.istag(element):
+                continue
+            n += 1
+            tag = self.getlocalname(self.gettag(element))
+            if path:
+                subpath = '%s/%s %s' % (path, template % n, tag)
+            else:
+                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):
+                attr_text = self.attrs2text(element)
+                print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/attributes" % (
+                    len(attr_text), subpath_encoded)
+            if self.supports_namespaces and self.has_ns(element):
+                ns_text = self.ns2text(element)
+                print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/namespaces" % (
+                    len(ns_text), subpath_encoded)
+            text = self.collect_text(element)
+            if text:
+                print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/text" % (
+                    len(text), subpath_encoded)
+            self._list(element, subpath)
+
     def has_ns(self, node):
         return False
 
@@ -163,46 +200,21 @@ class MiniDOMXmlVfs(XmlVfs):
             if text: text_accumulator.append(text)
         return '\n'.join(text_accumulator).encode(default_encoding, "replace")
 
-    def _list(self, node, path=''):
-        childNodes = node.childNodes
-        n = 0
-        for element in childNodes:
-            if element.localName:
-                n += 1
-        if n:
-            width = int(math.log10(n)) + 1
-            template = "%%0%dd" % width
-        else:
-            template = "%d"
-        n = 0
-        for element in childNodes:
-            if element.localName:
-                n += 1
-                if path:
-                    subpath = '%s/%s %s' % (path, template % n, element.localName)
-                else:
-                    subpath = '%s %s' % (template % n, element.localName)
-                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):
-                    attr_text = self.attrs2text(element)
-                    print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/attributes" % (
-                        len(attr_text), subpath_encoded)
-                text = self.collect_text(element)
-                if text:
-                    print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/text" % (
-                        len(text), subpath_encoded)
-                self._list(element, subpath)
-
     def getroot(self):
         return self.document
 
     def getchildren(self, node):
         return node.childNodes
 
+    def gettag(self, node):
+        return node.localName
+
     def istag(self, node):
         return bool(node.localName)
 
+    def getlocalname(self, name):
+        return name
+
 
 if use_elementtree or use_lxml:
     class CommonEtreeXmlVfs(XmlVfs):
@@ -229,6 +241,9 @@ if use_elementtree or use_lxml:
         def getchildren(self, node):
             return list(node)
 
+        def gettag(self, node):
+            return node.tag
+
         def istag(self, node):
             return isinstance(node.tag, basestring)
 
@@ -269,44 +284,20 @@ if use_elementtree:
                 name = name.encode(default_encoding, "replace")
                 value = value.encode(default_encoding, "replace")
                 if name.startswith('{'):
-                    name = name.split('}', 1)[1] # Remove XML namespace
+                    name = name.split('}', 1)[1]  # Remove XML namespace
                 attr_accumulator.append("%s=%s" % (name, value))
             return '\n'.join(attr_accumulator)
 
-        def _list(self, node, path=''):
-            n = len(node)
-            if n:
-                width = int(math.log10(n))+1
-                template = "%%0%dd" % width
-            else:
-                template = "%d"
-            n = 0
-            for element in node:
-                if not isinstance(element.tag, basestring):
-                    continue
-                n += 1
-                tag = element.tag
-                if tag.startswith('{'):
-                    tag = tag.split('}', 1)[1] # Remove XML namespace
-                if path:
-                    subpath = '%s/%s %s' % (path, template % n, tag)
-                else:
-                    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):
-                    attr_text = self.attrs2text(element)
-                    print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/attributes" % (
-                        len(attr_text), subpath_encoded)
-                text = self.collect_text(element)
-                if text:
-                    print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/text" % (
-                        len(text), subpath_encoded)
-                self._list(element, subpath)
+        def getlocalname(self, name):
+            if name.startswith('{'):
+                name = name.split('}', 1)[1]  # Remove XML namespace
+            return name
 
 
 if use_lxml:
     class LxmlEtreeXmlVfs(CommonEtreeXmlVfs):
+        supports_namespaces = True
+
         def parse(self):
             self.document = etree.parse(sys.argv[2])
 
@@ -329,60 +320,29 @@ if use_lxml:
                 ns_accumulator.append("%s=%s" % (name, value))
             return '\n'.join(ns_accumulator)
 
-        def list(self):
-            self._list(self.getroot())
-
-        def _list(self, node, path=''):
-            n = len(node)
-            if n:
-                width = int(math.log10(n))+1
-                template = "%%0%dd" % width
-            else:
-                template = "%d"
-            n = 0
-            for element in node:
-                if not isinstance(element.tag, basestring):
-                    continue
-                n += 1
-                tag = etree.QName(element.tag).localname
-                if path:
-                    subpath = '%s/%s %s' % (path, template % n, tag)
-                else:
-                    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):
-                    attr_text = self.attrs2text(element)
-                    print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/attributes" % (
-                        len(attr_text), subpath_encoded)
-                if element.nsmap:
-                    ns_text = self.ns2text(element)
-                    print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/namespaces" % (
-                        len(ns_text), subpath_encoded)
-                text = self.collect_text(element)
-                if text:
-                    print "-r--r--r-- 1 user group %d Jan 1 00:00 %s/text" % (
-                        len(text), subpath_encoded)
-                self._list(element, subpath)
-
         def getroot(self):
             return [self.document.getroot()]
 
+        def getlocalname(self, name):
+            return etree.QName(name).localname
+
 
 def build_xmlvfs():
-    if default_implementation is None:
+    if force_implementation is None:
         if use_lxml:
             return LxmlEtreeXmlVfs()
         elif use_elementtree:
             return ElementTreeXmlVfs()
         else:
             return MiniDOMXmlVfs()
-    elif default_implementation == 'minidom':
+    elif force_implementation == 'minidom':
         return MiniDOMXmlVfs()
-    elif default_implementation == 'elementtree':
+    elif force_implementation == 'elementtree':
         return ElementTreeXmlVfs()
-    elif default_implementation == 'lxml':
+    elif force_implementation == 'lxml':
         return LxmlEtreeXmlVfs()
+    else:
+        raise ValueError('Unknown implementation "%s", expected "minidom", "elementtree" or "lxml"' % force_implementation)
 
 
 def mcxml_list():
index 263f1d9..98ff711 100644 (file)
@@ -5,8 +5,8 @@ WHAT IS IT
    View an XML file in Midnight Commander as a filesystem.
 
 
-WHAT'S NEW in version 0.5.1 (2013-11-20)
-   Fixed a bug in handling comments.
+WHAT'S NEW in version 0.6.0 (2013-11-22)
+   Refactored _list() to be completely generic.
 
 WHAT'S NEW in version 0.5.0 (2013-11-19)
    Added lxml.etree-based implementation.