Rework how highlighting works.

See static-src/README.rst for details about how it works.

7 files changed, 130 insertions, 7 deletions

diff --git a/muppet/__main__.py b/muppet/__main__.py
index 943f9ae..b064cf1 100644
--- a/muppet/__main__.py
+++ b/muppet/__main__.py
@@ -47,7 +47,8 @@ def __main() -> None:
         # print(module)
         setup_module('output', module, path_base=args.path_base)
 
-    os.system("cp -r static output")
+    os.system('make -C static-src --silent install-full PREFIX=$PWD/output')
+    os.system("cp -r static/* output/static/")
 
 
 if __name__ == '__main__':
diff --git a/muppet/format.py b/muppet/format.py
index 97c38d1..870dfd3 100644
--- a/muppet/format.py
+++ b/muppet/format.py
@@ -1117,7 +1117,7 @@ def format_class(d_type: dict[str, Any]) -> str:
     # print(name, file=sys.stderr)
     out += print_docstring(name, d_type['docstring'])
 
-    out += '<pre><code class="puppet">'
+    out += '<pre class="highlight-muppet"><code class="puppet">'
     out += render(renderer, data)
     out += '</code></pre>'
     return out
@@ -1136,7 +1136,7 @@ def format_type_alias(d_type: dict[str, Any]) -> str:
     # print(name, file=sys.stderr)
     out += print_docstring(name, d_type['docstring'])
     out += '\n'
-    out += '<pre><code class="puppet">'
+    out += '<pre class="highlight-muppet"><code class="puppet">'
     t = parse_puppet(d_type['alias_of'])
     data = parse(t, 0, ['root'])
     out += render(renderer, data)
@@ -1152,7 +1152,7 @@ def format_defined_type(d_type: dict[str, Any]) -> str:
     # print(name, file=sys.stderr)
     out += print_docstring(name, d_type['docstring'])
 
-    out += '<pre><code class="puppet">'
+    out += '<pre class="highlight-muppet"><code class="puppet">'
     t = parse_puppet(d_type['source'])
     out += render(renderer, parse(t, 0, ['root']))
     out += '</code></pre>\n'
@@ -1201,9 +1201,10 @@ def format_puppet_function(function: dict[str, Any]) -> str:
         signature['signature']
         signature['docstring']
     if t in ['ruby3x', 'ruby4x']:
+        # TODO manual highlight here
         out += f'<pre><code class="ruby">{function["source"]}</code></pre>\n'
     elif t == 'puppet':
-        out += '<pre><code class="puppet">'
+        out += '<pre class="highlight-muppet"><code class="puppet">'
         try:
             t = parse_puppet(function['source'])
             out += str(parse(t, 0, ['root']))
diff --git a/muppet/output.py b/muppet/output.py
index 0e83c58..7e2266b 100644
--- a/muppet/output.py
+++ b/muppet/output.py
@@ -30,6 +30,7 @@ from collections.abc import (
 from .util import group_by
 from .puppet.strings import isprivate
 from .breadcrumbs import breadcrumbs
+from .syntax_highlight import highlight
 
 
 # TODO replace 'output' with base, or put this somewhere else
@@ -291,8 +292,15 @@ def setup_module(base: str, module: ModuleEntry, *, path_base: str) -> None:
 
         # TODO option to add .txt extension (for web serverse which
         # treat .pp as application/binary)
-        with open(os.path.join(dir, 'source.pp.txt'), 'w') as f:
-            f.write(puppet_class['source'])
+        with open(os.path.join(dir, 'source.pp.txt'), 'wb') as f:
+            with open(module.file(puppet_class['file']), 'rb') as g:
+                f.write(g.read())
+
+        with open(os.path.join(dir, 'source.pp.html'), 'w') as f:
+            template = jinja.get_template('code_page.html')
+            with open(module.file(puppet_class['file']), 'r') as g:
+                f.write(template.render(content=highlight(g.read(), 'puppet'),
+                                        path_base=path_base))
 
         with open(os.path.join(dir, 'source.json'), 'w') as f:
             json.dump(puppet_class, f, indent=2)
diff --git a/muppet/syntax_highlight/__init__.py b/muppet/syntax_highlight/__init__.py
new file mode 100644
index 0000000..bcadf81
--- /dev/null
+++ b/muppet/syntax_highlight/__init__.py
@@ -0,0 +1,39 @@
+"""
+Syntax highlight with the best available backend.
+
+Syntax highlight the given code by the best, available, syntax
+highlighter, returning the result as HTML. The default highlighter
+simply wraps the output in
+``<pre><code class={language}>{code}</code></pre>``
+(with escaping). This means that it can still be handled by JavaScript if so desired.
+"""
+
+from . import pygments
+from . import andre_simon
+from . import plain
+
+from typing import cast
+
+
+for module in [pygments, andre_simon, plain]:
+    if module.available:
+        backend = module
+        break
+else:
+    # This should never happen, since ``plain`` should always be
+    # available.
+    raise ValueError("No applicable highlight module")
+
+
+def highlight(source: str, language: str) -> str:
+    """
+    Highlight the given source as language, retuning HTML.
+
+    :param source:
+        Source code to highlight.
+    :param language:
+        Language of the source code.
+    :returns:
+        An HTML string.
+    """
+    return cast(str, backend.highlight(source, language))
diff --git a/muppet/syntax_highlight/andre_simon.py b/muppet/syntax_highlight/andre_simon.py
new file mode 100644
index 0000000..7bbcd43
--- /dev/null
+++ b/muppet/syntax_highlight/andre_simon.py
@@ -0,0 +1,36 @@
+"""
+Syntax highlighting through the highlight(1) command.
+
+The name "Andre Simon" is choosen based of the author, and that
+"highlight" by itself it really non-descriptive.
+"""
+
+import subprocess
+
+try:
+    subprocess.run(['highlight', '--version'], stdout=subprocess.DEVNULL)
+    available = True
+except FileNotFoundError:
+    available = False
+
+
+def highlight(code: str, language: str) -> str:
+    """Highlight code through the ``highlight`` command."""
+    # TODO line- vs pygments line_
+    cmd = subprocess.run(['highlight',
+                          '--out-format', 'html',
+                          '--fragment',
+                          '--line-numbers',
+                          '--anchors',
+                          '--anchors-prefix=line',
+                          '--class-name=NONE',
+                          '--syntax', language,
+                          '--enclose-pre'],
+                         capture_output=True,
+                         text=True,
+                         input=code,
+                         check=True)
+    return f"""
+    <!-- Generated through highlight(1), as language {language} -->
+    <div class="highlight-andre-simon">{cmd.stdout}</div>
+    """
diff --git a/muppet/syntax_highlight/plain.py b/muppet/syntax_highlight/plain.py
new file mode 100644
index 0000000..0d1234f
--- /dev/null
+++ b/muppet/syntax_highlight/plain.py
@@ -0,0 +1,14 @@
+"""Non-highlighter useful as a backup."""
+
+import html
+
+available = True
+
+
+def highlight(code: str, language: str) -> str:
+    """Return the code "highlighted" by wrapping it in <pre> tags."""
+    out = f'<pre><code class="{language}">{html.escape(code)}</code></pre>'
+    return f"""
+    <!-- "Genererated" as plain output -->
+    <div class"highlight-plain">{out}</div>
+    """
diff --git a/muppet/syntax_highlight/pygments.py b/muppet/syntax_highlight/pygments.py
new file mode 100644
index 0000000..05e8562
--- /dev/null
+++ b/muppet/syntax_highlight/pygments.py
@@ -0,0 +1,24 @@
+"""Syntax highlighting through pygments."""
+
+try:
+    from pygments.formatters import HtmlFormatter
+    from pygments.lexers import get_lexer_by_name
+    import pygments
+    available = True
+except ModuleNotFoundError:
+    available = False
+
+
+def highlight(code: str, language: str) -> str:
+    """Highlight code through pygments."""
+    out = pygments.highlight(code, get_lexer_by_name(language),
+                             HtmlFormatter(cssclass='highlight-pygments',
+                                           lineanchors='line',
+                                           linenos='table',
+                                           # linenos='inline'
+                                           anchorlinenos=True,
+                                           ))
+    return f"""
+    <!-- Generated through pygments, as {language} -->
+    {out}
+    """