unit tests: Rewrite documentation table parsing code

The old logic was completely broken, and didn't even assert that the
specified section was found at all. The CPU families test was broken
because of this. Luckily, the table didn't go out of sync with the
code.

It now also doesn't assume that each section has only one table. This
fixes the test now that we document the buildtype/optimization/debug
mapping in a second table inside the `Universal options` section.

1 files changed, 44 insertions, 32 deletions

diff --git a/run_unittests.py b/run_unittests.py
index b9dcd41..0249183 100755
--- a/run_unittests.py
+++ b/run_unittests.py
@@ -1274,39 +1274,54 @@ class DataTests(unittest.TestCase):
                 self.assertIn(opt, md)
         self.assertNotIn('b_unknown', md)
 
+    @staticmethod
+    def _get_section_content(name, sections, md):
+        for section in sections:
+            if section and section.group(1) == name:
+                try:
+                    next_section = next(sections)
+                    end = next_section.start()
+                except StopIteration:
+                    end = len(md)
+                # Extract the content for this section
+                return md[section.end():end]
+        raise RuntimeError('Could not find "{}" heading'.format(name))
+
     def test_builtin_options_documented(self):
         '''
         Test that universal options and base options are documented in
         Builtin-Options.md.
         '''
+        from itertools import tee
         md = None
         with open('docs/markdown/Builtin-options.md', encoding='utf-8') as f:
             md = f.read()
         self.assertIsNotNone(md)
 
         found_entries = set()
-        sections = list(re.finditer(r"^## (.+)$", md, re.MULTILINE)) + [None]
-
-        for s1, s2 in zip(sections[:], sections[1:]):
-            if s1.group(1) == "Universal options":
-                # Extract the content for this section
-                end = s2.start() if s2 is not None else len(md)
-                content = md[s1.end():end]
-                subsections = list(re.finditer(r"^### (.+)$", content, re.MULTILINE)) + [None]
-
-                for sub1, sub2 in zip(subsections[:], subsections[1:]):
-                    if sub1.group(1) == "Directories" or sub1.group(1) == "Core options":
-                        # Extract the content for this subsection
-                        sub_end = sub2.start() if sub2 is not None else len(content)
-                        subcontent = content[sub1.end():sub_end]
-                        # Find the list entries
-                        arches = [m.group(1) for m in re.finditer(r"^\| (\w+) .* \|", subcontent, re.MULTILINE)]
-                        # Drop the header
-                        arches = set(arches[1:])
-
-                        self.assertEqual(len(found_entries & arches), 0)
-                        found_entries |= arches
-            break
+        sections = re.finditer(r"^## (.+)$", md, re.MULTILINE)
+        # Extract the content for this section
+        content = self._get_section_content("Universal options", sections, md)
+        subsections = tee(re.finditer(r"^### (.+)$", content, re.MULTILINE))
+        subcontent1 = self._get_section_content("Directories", subsections[0], content)
+        subcontent2 = self._get_section_content("Core options", subsections[1], content)
+        for subcontent in (subcontent1, subcontent2):
+            # Find the option names
+            options = set()
+            # Match either a table row or a table heading separator: | ------ |
+            rows = re.finditer(r"^\|(?: (\w+) .* | *-+ *)\|", subcontent, re.MULTILINE)
+            # Skip the header of the first table
+            next(rows)
+            # Skip the heading separator of the first table
+            next(rows)
+            for m in rows:
+                value = m.group(1)
+                # End when the `buildtype` table starts
+                if value is None:
+                    break
+                options.add(value)
+            self.assertEqual(len(found_entries & options), 0)
+            found_entries |= options
 
         self.assertEqual(found_entries, set([
             *mesonbuild.coredata.builtin_options.keys(),
@@ -1318,16 +1333,13 @@ class DataTests(unittest.TestCase):
             md = f.read()
         self.assertIsNotNone(md)
 
-        sections = list(re.finditer(r"^## (.+)$", md, re.MULTILINE))
-        for s1, s2 in zip(sections[::2], sections[1::2]):
-            if s1.group(1) == "CPU families":
-                # Extract the content for this section
-                content = md[s1.end():s2.start()]
-                # Find the list entries
-                arches = [m.group(1) for m in re.finditer(r"^\| (\w+) +\|", content, re.MULTILINE)]
-                # Drop the header
-                arches = set(arches[1:])
-                self.assertEqual(arches, set(mesonbuild.environment.known_cpu_families))
+        sections = re.finditer(r"^## (.+)$", md, re.MULTILINE)
+        content = self._get_section_content("CPU families", sections, md)
+        # Find the list entries
+        arches = [m.group(1) for m in re.finditer(r"^\| (\w+) +\|", content, re.MULTILINE)]
+        # Drop the header
+        arches = set(arches[1:])
+        self.assertEqual(arches, set(mesonbuild.environment.known_cpu_families))
 
     def test_markdown_files_in_sitemap(self):
         '''