Extract idl without html5lib (#3760)

Easier than setting up pip/venv for e.g. html5lib's `six` requirement.

Author: kdashg

specs/extract-idl.py

@@ -3,7 +3,6 @@
 from datetime import date
 from string import Template
 import sys
-import html5lib
 
 LICENSE = """
 // Copyright (c) $YEAR The Khronos Group Inc.
@@ -30,35 +29,69 @@
 """
 
 htmlfilename = sys.argv[1]
-htmlfile = open(htmlfilename)
-try:
-    doc = html5lib.parse(htmlfile, treebuilder="dom")
-finally:
-    htmlfile.close()
 
-def elementHasClass(el, classArg):
-    """
-    Return true if and only if classArg is one of the classes of el
-    """
-    classes = [ c for c in el.getAttribute("class").split(" ") if c != "" ]
-    return classArg in classes
+USE_HTML5LIB = False
+if not USE_HTML5LIB:
+    import re
+    RE_IDL = re.compile(r'<pre class="idl">\n?(.*?)</pre>', re.DOTALL)
 
-def elementTextContent(el):
-    """
-    Implementation of DOM Core's .textContent
-    """
-    textContent = ""
-    for child in el.childNodes:
-        if child.nodeType == 3: # Node.TEXT_NODE
-            textContent += child.data
-        elif child.nodeType == 1: # Node.ELEMENT_NODE
-            textContent += elementTextContent(child)
-        else:
-            # Other nodes are ignored
-            pass
-    return textContent
+    IDL_EXAMPLE = '''
+    <pre class="idl">
+[Exposed=(Window,Worker)]
+interface <dfn data-dfn-type="interface" id="WebGLContextEvent">WebGLContextEvent</dfn> : <a href="http://www.w3.org/TR/domcore/#event">Event</a> {
+    constructor(DOMString type, optional WebGLContextEventInit eventInit = {});
+    readonly attribute DOMString statusMessage;
+};
 
-preList = doc.getElementsByTagName("pre")
-idlList = [elementTextContent(p) for p in preList if elementHasClass(p, "idl") ]
+// EventInit is defined in the DOM4 specification.
+dictionary WebGLContextEventInit : <a href="http://www.w3.org/TR/domcore/#eventinit">EventInit</a> {
+    DOMString statusMessage = "";
+};</pre>
+'''
+    assert RE_IDL.search(IDL_EXAMPLE)
+
+
+    from pathlib import Path
+    SRC_PATH = Path(htmlfilename)
+    matches = RE_IDL.finditer(SRC_PATH.read_text())
+    idlList = [m[1] for m in matches]
+
+    RE_TAG = re.compile(r'<(.*?)>')
+    idlList = [re.sub(RE_TAG, '', s) for s in idlList]
+
+    import html
+    idlList = [html.unescape(s) for s in idlList]
+else:
+    import html5lib
+    htmlfile = open(htmlfilename)
+    try:
+        doc = html5lib.parse(htmlfile, treebuilder="dom")
+    finally:
+        htmlfile.close()
+
+    def elementHasClass(el, classArg):
+        """
+        Return true if and only if classArg is one of the classes of el
+        """
+        classes = [ c for c in el.getAttribute("class").split(" ") if c != "" ]
+        return classArg in classes
+
+    def elementTextContent(el):
+        """
+        Implementation of DOM Core's .textContent
+        """
+        textContent = ""
+        for child in el.childNodes:
+            if child.nodeType == 3: # Node.TEXT_NODE
+                textContent += child.data
+            elif child.nodeType == 1: # Node.ELEMENT_NODE
+                textContent += elementTextContent(child)
+            else:
+                # Other nodes are ignored
+                pass
+        return textContent
+
+    preList = doc.getElementsByTagName("pre")
+    idlList = [elementTextContent(p) for p in preList if elementHasClass(p, "idl") ]
 licenseTemplate = Template(LICENSE)
 print(licenseTemplate.substitute(YEAR=date.today().year) + "\n\n".join(idlList))