Merge pull request #23 from uogbuji/develop

3.2.1 release prep

example/legacyparse.py

@@ -1,4 +1,6 @@
-import asyncio
+'''
+Not sure what this file was all about. WIll delete soon
+'''
 
 from amara3.uxml.xml import parse, buffer_handler
 evs = []
@@ -7,6 +9,6 @@
 print(evs)
 
 
-@asyncio.coroutine
-def handle_start_element(ev)
+def handle_start_element(ev):
+    pass
 

exec/microx

@@ -93,7 +93,6 @@ import sys
 import os
 import time
 import argparse
-from asyncio import coroutine
 from itertools import islice, chain
 
 import logging
@@ -187,7 +186,6 @@ def run(command_name, command_detail, sources=None, foreach=None, partition=None
                         if search_for in node.xml_value:
                             print(xpath_to(elem, show_attrs), file=out)
 
-    @coroutine
     def sink():
         while True:
             root = yield

pylib/uxml/html5.py

@@ -61,6 +61,9 @@ def hasContent(self):
         """
         return bool(self.xml_children)
 
+    def xml_encode(self):
+        return self.toxml()
+
 
 def qname_to_local(qname):
     return qname.split(':')[-1]
@@ -130,23 +133,33 @@ def cloneNode(self):
         return element(self.xml_name, attrs=attrs)
 
 
-#class comment(tree.comment):
-class comment(object):
+# class comment(tree.comment):
+class comment(node):
     type = 6
     value = tree.text.xml_value
     def __init__(self, data):
         self.data = data
+        # XXX Next 2 wererequired by html5lib tree API,
+        # but maybe only for __str__, which is now overridden
+        self.attributes = {}
+        self.name = '#comment'
         return
 
     def toxml(self):
-        return "<!--%s-->" % self.data
+        return f'<!--{self.data}-->'
+
+    def __str__(self):
+        return f'<#comment {self.data}>'
 
 #Note: element = 6
 
-class document(object):
+class document(node):
     type = 9
     def __init__(self):
         self.root_nodes = []
+        # Next 2 required by html5lib tree API
+        self.attributes = {}
+        self.name = '#document'
         return
 
     def appendChild(self, node):
@@ -155,13 +168,13 @@ def appendChild(self, node):
     def toxml(self):
         return "<!--%s-->" % self.data
 
-
-class doctype(object):
+class doctype(node):
     type = 10
     def __init__(self, name, publicId, systemId):
         self.name = name
         self.publicId = publicId
         self.systemId = systemId
+        self.attributes = {} # Required by html5lib tree API
         return
 
     def toxml(self):

pylib/uxml/tree.py

@@ -9,10 +9,9 @@
 
 import sys
 import weakref
-import asyncio
 from xml.sax.saxutils import escape, quoteattr
 
-from amara3.uxml.parser import parse, parser, parsefrags, event
+from amara3.uxml.parser import parser, event  # parse, parsefrags
 
 # NO_PARENT = object()
 
@@ -69,15 +68,20 @@ def xml_encode(self, indent=None, depth=0):
                 if indent:
                     strbits.append('\n')
                     strbits.append(indent*depth)
-            else:
+            elif isinstance(child, str):
                 strbits.append(escape(child))
+            else:
+                strbits.append(child.xml_encode())
         strbits.extend(['</', self.xml_name, '>'])
         return ''.join(strbits)
 
+    # What's the difference from strval?
     @property
     def xml_value(self):
         '''
+        Accumulated text in all descendant elements (similar to XPath text value)
         '''
+        # Recursive action
         return ''.join(map(lambda x: x.xml_value, self.xml_children))
 
     #Really just an alias that forbids specifying position
@@ -161,7 +165,6 @@ def __init__(self):
         self._root = None
         self._parent = None
 
-    @asyncio.coroutine
     def _handler(self):
         while True:
             ev = yield

pylib/uxml/treeiter.py

@@ -5,10 +5,9 @@
 #
 # -----------------------------------------------------------------------------
 
-import asyncio
-import collections
+from collections.abc import Iterable
 
-from .parser import parser, parsefrags, event
+from .parser import parser, event  # parsefrags
 from .tree import element, text, name_test
 
 
@@ -56,10 +55,10 @@ def __init__(self, patterns, sinks, prime_sinks=True):
         self._stateses = [None] * self._pattern_count
         self._evstacks = [[]] * self._pattern_count
         self._building_depths = [0] * self._pattern_count
-        #if asyncio.iscoroutine(sink):
+        # if asyncio.iscoroutine(sink):
         if prime_sinks:
             for sink in self._sinks:
-                if isinstance(sink, collections.Iterable):
+                if isinstance(sink, Iterable):
                     next(sink)  # Prime coroutine
         self._currents = [None] * self._pattern_count
         self._prep_patterns()
@@ -121,7 +120,6 @@ def _match_state(self, ix):
                 return False
         return False
 
-    @asyncio.coroutine
     def _handler(self):
         while True:
             ev = yield

pylib/uxml/version.py

@@ -1,2 +1,2 @@
 # http://legacy.python.org/dev/peps/pep-0440/
-version_info = ('3', '2', '0')
+version_info = ('3', '2', '1')

pylib/uxml/xml.py

@@ -5,19 +5,18 @@
 #
 # -----------------------------------------------------------------------------
 
-import asyncio
 import xml.parsers.expat
-from xml.sax.saxutils import escape  # also quoteattr?
+# from xml.sax.saxutils import escape  # also quoteattr?
 
 from . import tree
-from .parser import parser, parsefrags, event
+from .parser import event  # parser, parsefrags
 
 
 class expat_callbacks(object):
     def __init__(self, handler, prime_handler=True):
         self._handler = handler
         self._elem_stack = []
-        #if asyncio.iscoroutine(handler):
+        # if asyncio.iscoroutine(handler):
         if prime_handler:
             next(handler)  # Prime coroutine
         return

pylib/uxml/xmliter.py

@@ -5,11 +5,10 @@
 #
 # -----------------------------------------------------------------------------
 
-import asyncio
 import xml.parsers.expat
 
 from . import treeiter
-from .xml import expat_callbacks, ns_expat_callbacks
+from .xml import expat_callbacks  # ns_expat_callbacks
 
 
 def buffer_handler(accumulator):

test/uxml/test_chunk_parse_write.py

@@ -30,4 +30,3 @@ def record_sink(writer_obj):
 if __name__ == '__main__':
     raise SystemExit("Run with py.test")
 
-#@coroutine

test/uxml/test_html5.py

@@ -5,7 +5,6 @@
 import sys
 import io
 import logging
-from asyncio import coroutine
 
 import pytest #Consider also installing pytest_capturelog
 from amara3.uxml import tree
@@ -34,5 +33,15 @@ def test_basic_nav(doc):
         assert elem.xml_parent is root, (elem, root)
 
 
+DOC2 = '<html><head><title>HELLO</title></head><body><p>WORLD<!--Comment--></body></html>'
+DOC2_NORMALIZED = '<html><head><title>HELLO</title></head><body><p>WORLD<!--Comment--></p></body></html>'
+
+def test_xml_encode_with_comment():
+    root = html5.parse(io.StringIO(DOC2))
+    logging.debug('root: {}'.format(repr(root)))
+    # Round trip
+    assert root.xml_encode() == DOC2_NORMALIZED
+
+
 if __name__ == '__main__':
     raise SystemExit("Run with py.test")

test/uxml/test_parser.py

@@ -1,5 +1,4 @@
 import pytest
-from asyncio import coroutine
 
 from amara3.uxml.parser import parse, parser, parsefrags, event
 
@@ -103,7 +102,6 @@
 
 #def test_basic():
 
-@coroutine
 def handler(accumulator):
     while True:
         event = yield

test/uxml/test_tree.py

@@ -4,7 +4,6 @@
 
 import sys
 import logging
-from asyncio import coroutine
 
 import pytest #Consider also installing pytest_capturelog
 from amara3.uxml import tree

test/uxml/test_treegc.py

@@ -4,7 +4,6 @@
 
 import sys
 import gc
-from asyncio import coroutine
 
 import pytest
 from amara3.uxml import tree
@@ -24,7 +23,6 @@
 
 @pytest.mark.parametrize('doc,pat,expected', TREESEQ_CASES)
 def test_ts_gc(doc, pat, expected):
-    @coroutine
     def sink(accumulator):
         old_e = None
         while True:

test/uxml/test_treeutil.py

@@ -7,7 +7,6 @@
 import pytest
 from amara3.uxml import tree
 from amara3.uxml.treeutil import *
-#from amara3.util import coroutine
 
 DOC1 = '<a><b>1</b><b>2</b><b>3</b></a>'
 DOC2 = '<a><b>1</b><c>2</c><d>3</d></a>'