Preparing v0.4.5

Author: KaruroChori

RELEASE.md

@@ -1,14 +1,31 @@
 ## v0.4.5
 
 This release has not visible changes in its functionality (aside for some subtle bug fixes).  
-The main objectives were:
+Which is good, it means the language itself is getting stable.  
+The main objectives for this release were:
 
 - Reducing complexity and verbosity for most of the code
 - Removing obsolete or unused features from earlier revisions
-- Better documentation of the public interface.
+- Better documentation of its public interface.
+- Improved tests (so that I can validate the implementation so far)
+
+So, a boring one! But it means that a "final" v1 is getting close.  
+Basically all that is left are better tests to see if all commands are working as expected, better documentation and some code optimization.
+
+## Fixes
+
+- Fixed behaviour of filters in cycles. Now they always used meta-expressions only (no prefix `:` in that case).
+- Fixed behaviour of `for-range` with negative increments to match specs.
+- Fixed behaviour of `~!txt` to return the concatenation of all _cdata_ and _pcdata_ children, not just the first one.
 
 ## Breaking changes
 
 The interface to build the preprocessor is now a configuration object.  
 This is done to avoid verbosity, and to allow for the definition of more fields if so needed.  
 In most cases, using the new interface is just as simple as placing braces between the original args.
+
+## Internal changes
+
+- Code cleanup.
+- Speed up `O(nm)` children detection in cycles avoiding multiple passes.
+- Rework the mechanisms used in `order-by`

TODO.md

@@ -3,16 +3,17 @@
 ## `v0.4.5`
 
 - [x] Code cleanup.
-- [x] Speed up `O(nm)` children detection in cycles.
-- [ ] `string_view` version of `strnatcmp` to optimize the dot comparisons
+- [x] Speed up `O(nm)` children detection in cycles avoiding multiple passes.
 - [x] Rework the mechanisms used in `order-by`
-- [ ] and fill missing comparisons in the RPN vm
+- [x] More tests covering the language features
 
 ## `v0.4.7`
 
 - [ ] Properly show ctx information while logging
 - [ ] In the UI for the demo, add a loader as the first boot is quite slow
 - [ ] Fill in the docs at least of the public interface
+- [ ] `string_view` version of `strnatcmp` to optimize the dot comparisons
+- [ ] and fill missing comparisons in the RPN vm
 
 ## `v0.4.x`
 

docs/releases/v0.4.5.md

@@ -0,0 +1,29 @@
+This release has not visible changes in its functionality (aside for some subtle bug fixes).  
+Which is good, it means the language itself is getting stable.  
+The main objectives for this release were:
+
+- Reducing complexity and verbosity for most of the code
+- Removing obsolete or unused features from earlier revisions
+- Better documentation of its public interface.
+- Improved tests (so that I can validate the implementation so far)
+
+So, a boring one! But it means that a "final" v1 is getting close.  
+Basically all that is left are better tests to see if all commands are working as expected, better documentation and some code optimization.
+
+## Fixes
+
+- Fixed behaviour of filters in cycles. Now they always used meta-expressions only (no prefix `:` in that case).
+- Fixed behaviour of `for-range` with negative increments to match specs.
+- Fixed behaviour of `~!txt` to return the concatenation of all _cdata_ and _pcdata_ children, not just the first one.
+
+## Breaking changes
+
+The interface to build the preprocessor is now a configuration object.  
+This is done to avoid verbosity, and to allow for the definition of more fields if so needed.  
+In most cases, using the new interface is just as simple as placing braces between the original args.
+
+## Internal changes
+
+- Code cleanup.
+- Speed up `O(nm)` children detection in cycles avoiding multiple passes.
+- Rework the mechanisms used in `order-by`

meson.build

@@ -1,7 +1,7 @@
 project(
   'vs-templ',
   ['cpp'],
-  version: '0.4.4',
+  version: '0.4.5',
   meson_version: '>= 1.1',
   default_options: ['cpp_std=c++20'],
 )

src/vs-templ.cpp

@@ -249,7 +249,13 @@ std::optional<symbol> preprocessor::resolve_expr(const std::string_view& _str, c
                 ref=as<const pugi::xml_node>(tmp.value());
             }
         }
-        if(strcmp(str+idx+1,"!txt")==0) return ref.text().as_string();
+        if(strcmp(str+idx+1,"!txt")==0){
+            std::string txt;
+            for(auto child: ref.children()){
+                if(child.type()==pugi::node_pcdata || child.type()==pugi::node_cdata) txt+=child.text().as_string();
+            }
+            return txt;
+        }
         else if(strcmp(str+idx+1,"!tag")==0) return ref.name();
         else return ref.attribute(str+idx+1).as_string();
     }

test/cases/complex-paths.xml

@@ -1 +1,28 @@
-<?xml version="1.0" encoding="UTF-8"?>
+<?xml version="1.0" encoding="UTF-8"?>
+<test xmlns:s="vs.templ">
+    <data attr-1="0">
+        <root attr-1="1">
+            <child attr-1="2">text<![CDATA[a]]>a</child>
+        </root>
+    </data>
+
+    <template>
+        <doc>
+            <a><s:value src="/~attr-1"/></a>
+            <a><s:value src="/root/~attr-1"/></a>
+            <a><s:value src="/root/child/~attr-1"/></a>
+            <a><s:value src="/root/child/~!txt"/></a>
+            <a><s:value src="/root/child/~!tag"/></a>
+        </doc>
+    </template>
+
+    <expects>
+        <doc>
+            <a>0</a>
+            <a>1</a>
+            <a>2</a>
+            <a>textaa</a>
+            <a>child</a>
+        </doc>
+    </expects>
+</test>

test/meson.build

@@ -5,7 +5,7 @@ vs_templ_test = executable(
     install: false,
 )
 
-cases = ['id', 'for-range', 'for.tag', 'for-props.tag', 'calc', 'when.prop']
+cases = ['id', 'for-range', 'for.tag', 'for-props.tag', 'calc', 'when.prop', 'complex-paths']
 
 foreach case : cases