prain: web UI with a model monitor

A clear chat page (markdown, code blocks with copy, stop, new chat, kept in the browser,
drafts shown as such) and a monitor sidebar, a sheet on phones: tokens/s, a live map of the
30 x 128 experts (in VRAM, used now, copied in), VRAM hit rate, copies, context, prompt cache,
GPU VRAM/load/temperature/power/fan. GET /status merges vLLM /metrics, amdgpu sysfs and the
stats file, which now lists the experts in VRAM. /warm answers 202 when a warm-up starts.
README: the int4 kernels, their A/B, 2048-token chunks as a trade, prompt logprobs limit.

tests/test_web_page.py

@@ -0,0 +1,49 @@
+import json
+import shutil
+import subprocess
+import unittest
+from pathlib import Path
+
+PAGE = (Path(__file__).parent.parent / "prain" / "web.html").read_text()
+
+
+@unittest.skipUnless(shutil.which("node"), "needs node")
+class Markdown(unittest.TestCase):
+    """The page's md() (model output -> innerHTML) run in node."""
+
+    def md(self, text):
+        script = PAGE[PAGE.index("<script>") + 8:PAGE.index("/* ---------- thread")]
+        script = script.split("/* ---------- markdown", 1)[1].split("*/", 1)[1]
+        out = subprocess.run(["node", "-e", script + f"\nprocess.stdout.write(md({json.dumps(text)}))"],
+                             capture_output=True, text=True, check=True)
+        return out.stdout
+
+    def test_html_in_model_output_is_text(self):
+        html = self.md('<img src=x onerror="alert(1)"> **<b>hi</b>** `<i>` [x](javascript:alert(1))')
+        self.assertNotIn("<img", html)
+        self.assertNotIn("<b>", html)
+        self.assertNotIn('href="javascript', html)
+        self.assertIn("&lt;img src=x onerror=&quot;alert(1)&quot;&gt;", html)
+        self.assertIn("<strong>&lt;b&gt;hi&lt;/b&gt;</strong>", html)
+
+    def test_blocks(self):
+        html = self.md("# Title\n\nSome *very* **bold** text with `code` and "
+                       "[a link](https://x.org/?a=1&b=2).\n\n```python\nif a < b:\n    pass\n```\n"
+                       "3. three\n4. four\n\n- one\n  more\n- two\n\n| a | b |\n|---|---|\n| 1 | 2 |\n"
+                       "\n> quoted\n\n---\nend")
+        for part in ("<h3>Title</h3>", "<em>very</em>", "<strong>bold</strong>", "<code>code</code>",
+                     '<a href="https://x.org/?a=1&amp;b=2" target="_blank" rel="noopener noreferrer">',
+                     '<span>python</span>', "<pre><code>if a &lt; b:\n    pass</code></pre>",
+                     '<ol start="3"><li>three</li><li>four</li></ol>',
+                     "<ul><li>one more</li><li>two</li></ul>",
+                     "<th>a</th><th>b</th></tr></thead><tbody><tr><td>1</td><td>2</td>",
+                     "<blockquote><p>quoted</p></blockquote>", "<hr>", "<p>end</p>"):
+            self.assertIn(part, html)
+
+    def test_an_open_fence_while_streaming_is_code(self):
+        self.assertIn("<pre><code>x = 1\n# not a heading</code></pre>",
+                      self.md("Here:\n```\nx = 1\n# not a heading"))
+
+
+if __name__ == "__main__":
+    unittest.main()