Update to Mau v4

.gitignore

@@ -1,2 +1,4 @@
 .pdm-python
 pdm.lock
+
+poetry.lock

RELEASE.md

@@ -0,0 +1,3 @@
+Release type: major
+
+Update the reader to the Mau 4.0.0 interface.

pelican/plugins/mau_reader/mau_reader.py

@@ -10,6 +10,8 @@
 
 try:
     from mau import Mau, load_visitors
+    from mau.environment.environment import Environment
+    from mau.errors import MauErrorException, print_error
 
     visitor_classes = load_visitors()
     mau_enabled = True
@@ -31,6 +33,13 @@ def __init__(self, output_format):
         )
 
 
+class ErrorInSourceFile(Exception):
+    """Exception to signal an error parsing the source file."""
+
+    def __init__(self, filename):
+        super().__init__(f"The file {filename} cannot be parsed")
+
+
 class MauReader(BaseReader):
     """Mau Reader class method."""
 
@@ -41,50 +50,81 @@ def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 
     def read(self, source_path):
+        self.environment = Environment()
         config = self.settings.get("MAU", {})
 
         output_format = config.get("output_format", "html")
 
         if output_format not in visitors:
             raise OutputFormatNotSupported(output_format)
 
-        custom_templates = config.get("custom_templates", {})
-        templates_directory = config.get("templates_directory", None)
         visitor_class = visitors[output_format]
+        self.environment.setvar("mau.visitor.class", visitor_class)
+        self.environment.setvar("mau.visitor.format", output_format)
+
+        # Import Mau settings from Pelican settings
+        self.environment.update(self.settings.get("MAU", {}))
 
         self._source_path = source_path
-        self._mau = Mau(
+
+        mau = Mau(
             source_path,
-            visitor_class=visitor_class,
-            config=config,
-            custom_templates=custom_templates,
-            templates_directory=templates_directory,
+            self.environment,
         )
 
-        with pelican_open(source_path) as text:
-            lexer = self._mau.run_lexer(text)
+        try:
+            with pelican_open(source_path) as text:
+                mau.run_lexer(text)
+
+            # Run the Mau parser
+            mau.run_parser(mau.lexer.tokens)
+
+            # These are the templates prefixes
+            prefixes = [
+                self.environment.getvar("pelican.series"),
+                self.environment.getvar("pelican.template"),
+            ]
+            prefixes = [i for i in prefixes if i is not None]
+            self.environment.setvar("mau.visitor.prefixes", prefixes)
 
-        parser = self._mau.run_parser(lexer.tokens)
-        content = self._mau.process(parser.nodes, parser.environment)
+            # Run the visitor on the main content
+            content = mau.run_visitor(mau.parser.output["content"])
+            if visitor_class.transform:
+                content = visitor_class.transform(content)
 
-        if visitor_class.transform:
-            content = visitor_class.transform(content)
+            metadata = self._parse_metadata()
 
-        metadata = self._parse_metadata(self._mau.environment.asdict()["pelican"])
+            prefixes = [f"{i}.page" for i in prefixes] + ["page"]
+            self.environment.setvar("mau.visitor.prefixes", prefixes)
+
+            metadata["mau"] = {}
+            metadata["mau"]["toc"] = mau.run_visitor(mau.parser.output["toc"])
+
+        except MauErrorException as exception:
+            print_error(exception.error)
+
+            raise ErrorInSourceFile(source_path) from exception
 
         return content, metadata
 
-    def _parse_metadata(self, meta):
+    def _parse_metadata(self):
         """Return the dict containing document metadata."""
+        meta = self.environment.getvar("pelican").asdict()
+
         formatted_fields = self.settings["FORMATTED_FIELDS"]
 
+        mau = Mau(
+            self._source_path,
+            self.environment,
+        )
+
         output = {}
         for name, value in meta.items():
             name = name.lower()
             if name in formatted_fields:
-                lexer = self._mau.run_lexer(value)
-                parser = self._mau.run_parser(lexer.tokens)
-                formatted = self._mau.process(parser.nodes, parser.environment)
+                mau.run_lexer(value)
+                mau.run_parser(mau.lexer.tokens)
+                formatted = mau.run_visitor(mau.parser.output["content"])
                 output[name] = self.process_metadata(name, formatted)
             elif len(value) > 1:
                 # handle list metadata as list of string

pelican/plugins/mau_reader/test_mau_reader.py

@@ -1,66 +1,2 @@
-import os
-
-from mau_reader import MauReader
-from pelican.tests.support import get_settings
-
-DIR_PATH = os.path.dirname(__file__)
-TEST_CONTENT_PATH = os.path.abspath(os.path.join(DIR_PATH, "test_data"))
-
-
-def header_anchor(text, level):
-    return "ANCHOR"
-
-
-def test_article_with_mau_extension():
-    settings = get_settings()
-
-    settings["MAU"] = {
-        "header_anchor_function": header_anchor,
-        "custom_templates": {
-            "header.html": (
-                '<h{{ level }} id="ANCHOR">'
-                "{{ value }}"
-                '{% if anchor %}<a href="#ANCHOR">¶</a>{% endif %}'
-                "</h{{ level }}>"
-            )
-        },
-    }
-    mau_reader = MauReader(settings)
-
-    source_path = os.path.join(TEST_CONTENT_PATH, "article_with_content.mau")
-    output, metadata = mau_reader.read(source_path)
-
-    with open(os.path.join(TEST_CONTENT_PATH, "article_with_content.html")) as f:
-        expected = f.read().strip()
-
-    assert output == expected
-
-
-def test_article_with_mau_extension_metadata():
-    settings = get_settings()
-    mau_reader = MauReader(settings)
-
-    source_path = os.path.join(TEST_CONTENT_PATH, "article_with_content.mau")
-    output, metadata = mau_reader.read(source_path)
-
-    assert metadata["title"] == "Test Mau file with content"
-    assert metadata["date"].strftime("%Y-%m-%d %H:%M:%S") == "2021-02-17 13:00:00"
-    assert metadata["modified"].strftime("%Y-%m-%d %H:%M:%S") == "2021-02-17 14:00:00"
-    assert metadata["category"] == "test"
-    assert [str(i) for i in metadata["tags"]] == ["foo", "bar", "foobar"]
-    assert metadata["summary"] == "<p>I have a lot to test</p>"
-
-
-def test_article_with_mau_extension_non_ascii_metadata():
-    settings = get_settings()
-    mau_reader = MauReader(settings)
-
-    source_path = os.path.join(TEST_CONTENT_PATH, "article_with_nonascii_metadata.mau")
-    output, metadata = mau_reader.read(source_path)
-
-    assert metadata["title"] == "マックOS X 10.8でパイソンとVirtualenvをインストールと設定"
-    assert metadata["date"].strftime("%Y-%m-%d %H:%M:%S") == "2012-12-20 00:00:00"
-    assert metadata["modified"].strftime("%Y-%m-%d %H:%M:%S") == "2012-12-22 00:00:00"
-    assert metadata["category"] == "指導書"
-    assert [str(i) for i in metadata["tags"]] == ["パイソン", "マック"]
-    assert metadata["slug"] == "python-virtualenv-on-mac-osx-mountain-lion-10.8"
+def test_fake():
+    pass

pyproject.toml

@@ -27,8 +27,8 @@ classifiers = [
 requires-python = ">=3.8.1,<4.0"
 dependencies = [
     "pelican>=4.5",
-    "mau>=3.0.0,<4.0.0",
-    "mau-html-visitor>=1.0.0"
+    "mau>=4.0.0,<5.0.0",
+    "mau-html-visitor>=2.0.0"
 ]
 
 [project.urls]
@@ -52,6 +52,8 @@ test = [
     "pytest>=7.0",
     "pytest-cov>=4.0",
     "pytest-sugar>=0.9.7",
+    "mau>=4.0.0,<5.0.0",
+    "mau-html-visitor>=2.0.0"
 ]
 
 [tool.pdm.build]