performance: pre-render html fully instead of at runtime

We still parsed the raw html at runtime and fed that back into the
jinja template. We now do that as the ssg build phase, which gives
much better performance.

packages/solara-enterprise/solara_enterprise/ssg.py

@@ -3,6 +3,7 @@
 import threading
 import time
 import typing
+import urllib
 from pathlib import Path
 from typing import List, Optional
 
@@ -124,10 +125,18 @@ def ssg_crawl_route(base_url: str, route: solara.Route, build_path: Path, thread
                 page.locator("#kernel-busy-indicator").wait_for(state="hidden")
                 # page.wait_
                 time.sleep(0.5)
-                html = page.content()
+                raw_html = page.content()
             except Exception:
                 logger.exception("Failure retrieving content for url: %s", url)
                 raise
+            request_path = urllib.parse.urlparse(url).path
+
+            import solara.server.server
+
+            # the html from playwright is not what we want, pass it through the jinja template again
+            html = solara.server.server.read_root(request_path, ssg_data=_ssg_data(raw_html))
+            if html is None:
+                raise Exception(f"Failed to render {url}")
             path.write_text(html, encoding="utf-8")
             rprint(f"Wrote to {path}")
             page.goto("about:blank")
@@ -140,12 +149,8 @@ def ssg_crawl_route(base_url: str, route: solara.Route, build_path: Path, thread
     return results
 
 
-def ssg_data(path: str) -> Optional[SSGData]:
+def ssg_content(path: str) -> Optional[str]:
     license.check("SSG")
-    html = ""
-    # pre_rendered_css = ""
-    styles = []
-    title = "Solara ☀️"
     # still not sure why we sometimes end with a double slash
     if path.endswith("//"):
         path = path[:-2]
@@ -164,38 +169,46 @@ def ssg_data(path: str) -> Optional[SSGData]:
             html_path = html_path.with_suffix(".html")
         if html_path.exists() and html_path.is_file():
             logger.info("Using pre-rendered html at %r", html_path)
-
-            from bs4 import BeautifulSoup, Tag
-
-            soup = BeautifulSoup(html_path.read_text("utf8"), "html.parser")
-            node = soup.find(id="app")
-            # TODO: add classes...
-            if node and isinstance(node, Tag):
-                # only render children
-                html = "".join(str(x) for x in node.contents)
-            title_tag = soup.find("title")
-            if title_tag:
-                title = title_tag.text
-
-            # include all meta tags
-            rendered_metas = soup.find_all("meta")
-            metas = []
-            for meta in rendered_metas:
-                # but only the ones added by solara
-                if meta.attrs.get("data-solara-head-key"):
-                    metas.append(str(meta))
-
-            # include all styles
-            rendered_styles = soup.find_all("style")
-            for style in rendered_styles:
-                style_html = str(style)
-                # in case we want to skip the mathjax css
-                # if "MJXZERO" in style_html:
-                #     continue
-                # pre_rendered_css += style_html
-                styles.append(style_html)
-                logger.debug("Include style (size is %r mb):\n\t%r", len(style_html) / 1024**2, style_html[:200])
-            return SSGData(title=title, html=html, styles=styles, metas=metas)
+            return html_path.read_text("utf8")
         else:
             logger.error("Count not find html at %r", html_path)
     return None
+
+
+def _ssg_data(html: str) -> Optional[SSGData]:
+    license.check("SSG")
+    from bs4 import BeautifulSoup, Tag
+
+    # pre_rendered_css = ""
+    styles = []
+    title = "Solara ☀️"
+
+    soup = BeautifulSoup(html, "html.parser")
+    node = soup.find(id="app")
+    # TODO: add classes...
+    if node and isinstance(node, Tag):
+        # only render children
+        html = "".join(str(x) for x in node.contents)
+    title_tag = soup.find("title")
+    if title_tag:
+        title = title_tag.text
+
+    # include all meta tags
+    rendered_metas = soup.find_all("meta")
+    metas = []
+    for meta in rendered_metas:
+        # but only the ones added by solara
+        if meta.attrs.get("data-solara-head-key"):
+            metas.append(str(meta))
+
+    # include all styles
+    rendered_styles = soup.find_all("style")
+    for style in rendered_styles:
+        style_html = str(style)
+        # in case we want to skip the mathjax css
+        # if "MJXZERO" in style_html:
+        #     continue
+        # pre_rendered_css += style_html
+        styles.append(style_html)
+        logger.debug("Include style (size is %r mb):\n\t%r", len(style_html) / 1024**2, style_html[:200])
+    return SSGData(title=title, html=html, styles=styles, metas=metas)

solara/__main__.py

@@ -11,12 +11,11 @@
 
 import rich
 import rich_click as click
+import solara
 import uvicorn
 from rich import print as rprint
-from uvicorn.main import LEVEL_CHOICES, LOOP_CHOICES
-
-import solara
 from solara.server import settings
+from uvicorn.main import LEVEL_CHOICES, LOOP_CHOICES
 
 from .server import telemetry
 
@@ -450,6 +449,7 @@ def ssg(app: str, port: int, host: str, headed: bool):
     """Static site generation"""
     settings.ssg.headed = headed
     settings.ssg.enabled = True
+    settings.main.mode = "production"  # always override this
     os.environ["SOLARA_APP"] = app
     from solara.server.starlette import ServerStarlette
 

solara/server/server.py

@@ -12,7 +12,6 @@
 import ipywidgets
 import jinja2
 import requests
-
 import solara
 import solara.routing
 
@@ -247,7 +246,15 @@ def busy_idle(parent):
         return False
 
 
-def read_root(path: str, root_path: str = "", render_kwargs={}, use_nbextensions=True) -> Optional[str]:
+def read_root(path: str, root_path: str = "", render_kwargs={}, use_nbextensions=True, ssg_data=None) -> Optional[str]:
+    if settings.ssg.enabled and ssg_data is None:
+        # simply return the pre-rendered html
+        from solara_enterprise import ssg
+
+        content = ssg.ssg_content(path)
+        if content is not None:
+            return content
+
     default_app = app.apps["__default__"]
     routes = default_app.routes
     router = solara.routing.Router(path, routes)
@@ -267,15 +274,11 @@ def read_root(path: str, root_path: str = "", render_kwargs={}, use_nbextensions
     pre_rendered_css = ""
     pre_rendered_metas = ""
     title = "Solara ☀️"
-    if settings.ssg.enabled:
-        from solara_enterprise import ssg
-
-        ssg_data = ssg.ssg_data(path)
-        if ssg_data is not None:
-            pre_rendered_html = ssg_data["html"]
-            pre_rendered_css = "\n".join(ssg_data["styles"])
-            pre_rendered_metas = "\n    ".join(ssg_data["metas"])
-            title = ssg_data["title"]
+    if ssg_data is not None:
+        pre_rendered_html = ssg_data["html"]
+        pre_rendered_css = "\n".join(ssg_data["styles"])
+        pre_rendered_metas = "\n    ".join(ssg_data["metas"])
+        title = ssg_data["title"]
 
     if settings.assets.proxy:
         # solara acts as a proxy