Merge pull request #46 from keith-hall/st4_fixes

ST4 fixes

Author: rosshadden

.python-version

@@ -0,0 +1 @@
+3.8

README.md

@@ -87,6 +87,37 @@ The recommended way to install the Sublime Text XPath plugin is via [Package Con
 
 ## Troubleshooting
 
+## Context menu items disabled
+
+This can happen if the `lxml` dependency didn't load properly. You'll see errors in the ST console.
+
+### Mac
+On a Mac with Apple silicon, the version of lxml installed by Package Control 4 doesn't seem to work.
+In ST console, we can see that ST build 4180 is using Python 3.8.12:
+```python
+import sys; sys.version_info
+```
+So you can build lxml manually using this version of Python. You will need to download the source code release asset, as lxml's git repository doesn't contain some `.c` files which are required to build. lxml v5.1.1 for sure works:
+- Navigate to https://github.com/lxml/lxml/releases/tag/lxml-5.1.1
+- Download `lxml-5.1.1.tar.gz`
+- extract it
+
+```sh
+brew install pyenv
+pyenv init # follow instructions
+
+pyenv install 3.8.12
+pyenv shell 3.8.12
+
+cd ~/Downloads/lxml-5.1.1
+python setup.py build
+```
+
+- then copy `~/Downloads/lxml-5.1.1/build/lib.macosx-15.1-arm64-3.8/lxml` into `~/Library/Application Support/Sublime Text/Lib/python38/lxml`, overwriting anything already there.
+- Restart ST.
+
+(If you were to try downloading `lxml-5.1.1-cp38-cp38-macosx_10_9_universal2.whl` for example, and extracting that into ST's lib folder mentioned above, when you restart ST, you would be told that an `.so` file in `Lib/python38/lxml/` folder isn't trusted, and there would be no option to "allow". You could go in Mac Settings -> Privacy and Security and it should show up there with an option to allow it. But you'd still see that the `lxml` dependency fails to load in ST, and the only solution seems to be building from source on the Mac.)
+
 ### CDATA Nodes
 
 When working with XML documents, you are probably used to the Document Object Model (DOM), where CDATA nodes are separate to text nodes.  XPath sees `text()` nodes as all adjacent CDATA and text node siblings together.

example_xml_ns.xml

@@ -1,31 +1,31 @@
 <?xml version="1.0"?>
 <test>
-	<hello xmlns="hello_ns">
-		<!-- from an element name standpoint, you could expect this element to be reachable by the path "/test/hello".
-		However, XPath 1.0 does not have the concept of a default namespace. If the XML document being queried defines a default namespace, the XPath expression should map the namespace to a prefix for easier access.
-		This is what this plugin does for you. Because this is the first prefix-less namespace declared in document order, it will become "default", assuming "default" is set as the default namespace prefix to use.
-		But, because there is at least one other prefix-less namespace in the document - which resolves to a different uri - it will become "default1" instead. -->
-		
-		<!-- The scope of a default namespace declaration extends from the beginning of the start-tag in which it appears to the end of the corresponding end-tag, excluding the scope of any inner default namespace declarations. -->
-		<world xmlns="world_ns"><!-- second prefix-less namespace in doc order, becomes "default2" -->
-			<example /><!-- path here is "/test/default1:hello/default2:world/default2:example" -->
-		</world>
-	</hello>
-	<more xmlns="more_ns" xmlns:an="another_ns"><!-- third prefix-less namespace in doc order, becomes "default3".  First "an" prefix declared in doc-order.  Because there is more than one unique namespace uri declared with this prefix, it will become "an1" for query purposes. -->
-		<an:another></an:another><!-- path here is "/test/default3:more[1]/an1:another" -->
-	</more>
-	<more xmlns="more_ns" xmlns:an="yet_another_ns" xmlns:unique="single"><!-- this prefixless namespace has the same uri as one already used, "default3", so this is also "default3". Because the an prefix is being declared again to a different uri than the previous one, it becomes "an2". As the "unique" prefix is unique, it remains as "unique", with no numeric suffix. -->
-		<an:yet_another></an:yet_another><!-- path here is "/test/default3:more[2]/an2:yet_another" -->
-		<unique:example></unique:example><!-- path here is "/test/default3:more[2]/unique:example" -->
-	</more>
-	<foo xmlns="world_ns" /><!-- same uri as default2 used, path is therefore "/test/default2:foo" -->
-	<numeric xmlns:an="yans"><!-- this one becomes "an4", because "an" has been used twice already and "an3" is explicitly defined later in the document -->
-		<an:test /><!-- /test/numeric/an4:test -->
-	</numeric>
-	<numeric xmlns:an3="numbered_ns"><!-- specific namespace prefix with a numeric suffix declared -->
-		<an3:okay></an3:okay>
-	</numeric>
-	<text attr1="hello" attr2='world'>sample text<more some_value
-		=
-		"foobar" another_value = "super" xmlns:abc="abc" abc:another_value="value" /> lorem ipsum etc.</text>abc<![CDATA[def]]>ghi<hij><![CDATA[klm]]></hij><![CDATA[nop]]>
+    <hello xmlns="hello_ns">
+        <!-- from an element name standpoint, you could expect this element to be reachable by the path "/test/hello".
+        However, XPath 1.0 does not have the concept of a default namespace. If the XML document being queried defines a default namespace, the XPath expression should map the namespace to a prefix for easier access.
+        This is what this plugin does for you. Because this is the first prefix-less namespace declared in document order, it will become "default", assuming "default" is set as the default namespace prefix to use.
+        But, because there is at least one other prefix-less namespace in the document - which resolves to a different uri - it will become "default1" instead. -->
+
+        <!-- The scope of a default namespace declaration extends from the beginning of the start-tag in which it appears to the end of the corresponding end-tag, excluding the scope of any inner default namespace declarations. -->
+        <world xmlns="world_ns"><!-- second prefix-less namespace in doc order, becomes "default2" -->
+            <example /><!-- path here is "/test/default1:hello/default2:world/default2:example" -->
+        </world>
+    </hello>
+    <more xmlns="more_ns" xmlns:an="another_ns"><!-- third prefix-less namespace in doc order, becomes "default3".  First "an" prefix declared in doc-order.  Because there is more than one unique namespace uri declared with this prefix, it will become "an1" for query purposes. -->
+        <an:another></an:another><!-- path here is "/test/default3:more[1]/an1:another" -->
+    </more>
+    <more xmlns="more_ns" xmlns:an="yet_another_ns" xmlns:unique="single"><!-- this prefixless namespace has the same uri as one already used, "default3", so this is also "default3". Because the an prefix is being declared again to a different uri than the previous one, it becomes "an2". As the "unique" prefix is unique, it remains as "unique", with no numeric suffix. -->
+        <an:yet_another></an:yet_another><!-- path here is "/test/default3:more[2]/an2:yet_another" -->
+        <unique:example></unique:example><!-- path here is "/test/default3:more[2]/unique:example" -->
+    </more>
+    <foo xmlns="world_ns" /><!-- same uri as default2 used, path is therefore "/test/default2:foo" -->
+    <numeric xmlns:an="yans"><!-- this one becomes "an4", because "an" has been used twice already and "an3" is explicitly defined later in the document -->
+        <an:test /><!-- /test/numeric/an4:test -->
+    </numeric>
+    <numeric xmlns:an3="numbered_ns"><!-- specific namespace prefix with a numeric suffix declared -->
+        <an3:okay></an3:okay>
+    </numeric>
+    <text attr1="hello" attr2='world'>sample text<more some_value
+        =
+        "foobar" another_value = "super" xmlns:abc="abc" abc:another_value="value" /> lorem ipsum etc.</text>abc<![CDATA[def]]>ghi<hij><![CDATA[klm]]></hij><![CDATA[nop]]>
 </test>

lxml_parser.py

@@ -158,6 +158,11 @@ def element_start(self, tag, attrib=None, nsmap=None, location=None):
 
     def create_element(self, tag, attrib=None, nsmap=None):
         LocationAwareElement.TAG = tag
+        if nsmap: # a change made in lxml 3.8.0 / 3.5.0b1 requires us to pass None instead of an empty prefix string
+            if '' in nsmap:
+                nsmap[None] = nsmap['']
+                del nsmap['']
+        
         return LocationAwareElement(attrib=attrib, nsmap=nsmap)
 
     def element_end(self, tag, location=None):
@@ -299,11 +304,10 @@ def unique_namespace_prefixes(namespaces, replaceNoneWith = 'default', start = 1
 
 def get_results_for_xpath_query(query, tree, context = None, namespaces = None, **variables):
     """Given a query string and a document trees and optionally some context elements, compile the xpath query and execute it."""
-    nsmap = {}
-    if namespaces is not None:
+    nsmap = dict()
+    if namespaces:
         for prefix in namespaces.keys():
-            if namespaces[prefix][0] != '':
-                nsmap[prefix] = namespaces[prefix][0]
+            nsmap[prefix] = namespaces[prefix][0]
 
     xpath = etree.XPath(query, namespaces = nsmap)
 

sublime_lxml.py

@@ -3,8 +3,8 @@
 from .sublime_helper import get_scopes
 import re
 
-RE_TAG_NAME_END_POS = re.compile('[>\s/]')
-RE_TAG_ATTRIBUTES = re.compile('\s+((\w+(?::\w+)?)\s*=\s*(?:"([^"]*)"|\'([^\']*)\'))')
+RE_TAG_NAME_END_POS = re.compile(r'[>\s/]')
+RE_TAG_ATTRIBUTES = re.compile(r'\s+((\w+(?::\w+)?)\s*=\s*(?:"([^"]*)"|\'([^\']*)\'))')
 
 # TODO: consider subclassing etree.ElementBase and adding as methods to that
 def getNodeTagRegion(view, node, position_type):