A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.
Python artificial-intelligencedagdata-sciencedata-visualizationdataflowdeveloper-toolsmachine-learningnotebookspipelinepythonreactivesqlweb-app
import re
from pathlib import Path
from typing import Any
def on_page_markdown(markdown: str, page: Any = None, **kwargs: Any) -> str:
del kwargs
static_dir = Path("docs/")
if not static_dir.exists():
return markdown
# Find all src="/" patterns
pattern = r'src="(/[^"]+)"'
matches = re.finditer(pattern, markdown)
for match in matches:
src_path = match.group(1)
# Remove leading slash and check if file exists in _static
relative_path = src_path.lstrip("/")
full_path = static_dir / relative_path
if not full_path.exists():
print(f"\u26a0\ufe0f Warning: Static asset not found: {src_path}")
print(f" Expected at: {full_path}")
# Auto-generate meta description from page content if not set
if page and not page.meta.get("description"):
description = _extract_description(markdown)
if description:
page.meta["description"] = description
return markdown
def _extract_description(markdown: str) -> str:
"""Extract a meta description from the first meaningful paragraph."""
# Strip HTML blocks (style, script, div, p, table, video, etc.) before parsing
cleaned = re.sub(
r"<(style|script|video|table|figure)[^>]*>.*?</\1>",
"",
markdown,
flags=re.DOTALL | re.IGNORECASE,
)
lines = cleaned.split("\n")
paragraph_lines: list[str] = []
in_frontmatter = False
past_frontmatter = False
in_code_block = False
in_admonition = False
for line in lines:
stripped = line.strip()
# Skip YAML frontmatter
if stripped == "---":
if not past_frontmatter:
in_frontmatter = not in_frontmatter
if not in_frontmatter:
past_frontmatter = True
continue
if in_frontmatter:
continue
# Skip code blocks
if stripped.startswith("```"):
in_code_block = not in_code_block
continue
if in_code_block:
continue
# Skip headings
if stripped.startswith("#"):
# If we already collected paragraph text, stop
if paragraph_lines:
break
continue
# Skip HTML tags, admonitions, and directives
if stripped.startswith("<") or stripped.startswith("///"):
if stripped.startswith("///"):
in_admonition = not in_admonition
if paragraph_lines:
break
continue
if in_admonition:
continue
# Skip empty lines
if not stripped:
if paragraph_lines:
break
continue
# Skip markdown images and links-only lines
if re.match(r"^!?\[.*\]\(.*\)$", stripped):
continue
# Skip lines starting with ??? or !!! (admonition shorthand)
if stripped.startswith("???") or stripped.startswith("!!!"):
if paragraph_lines:
break
continue
# Skip lines that are just formatting (bold/italic markers, list items)
if re.match(r"^[*_\-|]", stripped) and not re.match(
r"^[*_]{1,2}\w", stripped
):
if paragraph_lines:
break
continue
paragraph_lines.append(stripped)
if not paragraph_lines:
return ""
text = " ".join(paragraph_lines)
# Remove markdown formatting
text = re.sub(r"\[([^\]]+)\]\([^)]+\)", r"\1", text) # [text](url) -> text
text = re.sub(r"`([^`]+)`", r"\1", text) # `code` -> code
text = re.sub(r"[*_]{1,2}([^*_]+)[*_]{1,2}", r"\1", text) # bold/italic
text = re.sub(r"\s+", " ", text).strip() # normalize whitespace
# Truncate to ~155 characters at a word boundary
max_length = 155
if len(text) > max_length:
text = text[:max_length].rsplit(" ", 1)[0] + "..."
return text