import json

import logging

import os

import re

import shutil

import subprocess

from html.parser import HTMLParser

from http.server import HTTPServer, SimpleHTTPRequestHandler

from multiprocessing import Pool

from pathlib import Path

from typing import Any

import mkdocs.utils

import typer

import yaml

from jinja2 import Template

from ruff.__main__ import find_ruff_bin

from slugify import slugify as py_slugify

logging.basicConfig(level=logging.INFO)

SUPPORTED_LANGS = {

"de",

"en",

"es",

"fr",

"ja",

"ko",

"pt",

"ru",

"tr",

"uk",

"zh",

"zh-hant",

}

app = typer.Typer()

mkdocs_name = "mkdocs.yml"

missing_translation_snippet = """

{!../../docs/missing-translation.md!}

"""

non_translated_sections = (

f"reference{os.sep}",

"release-notes.md",

"fastapi-people.md",

"external-links.md",

"newsletter.md",

"management-tasks.md",

"management.md",

"contributing.md",

)

docs_path = Path("docs")

en_docs_path = Path("docs/en")

en_config_path: Path = en_docs_path / mkdocs_name

site_path = Path("site").absolute()

build_site_path = Path("site_build").absolute()

header_pattern = re.compile(r"^(#{1,6}) (.+?)(?:\s*\{\s*(#.*)\s*\})?\s*$")

header_with_permalink_pattern = re.compile(r"^(#{1,6}) (.+?)(\s*\{\s*#.*\s*\})\s*$")

code_block3_pattern = re.compile(r"^\s*```")

code_block4_pattern = re.compile(r"^\s*````")

# Pattern to match markdown links: [text](url) → text

md_link_pattern = re.compile(r"\[([^\]]+)\]\([^)]+\)")

def strip_markdown_links(text: str) -> str:

"""Replace markdown links with just their visible text."""

return md_link_pattern.sub(r"\1", text)

class VisibleTextExtractor(HTMLParser):

"""Extract visible text from a string with HTML tags."""

def __init__(self):

super().__init__()

self.text_parts = []

def handle_data(self, data):

self.text_parts.append(data)

def extract_visible_text(self, html: str) -> str:

self.reset()

self.text_parts = []

self.feed(html)

return "".join(self.text_parts).strip()

def slugify(text: str) -> str:

return py_slugify(

text,

replacements=[

("`", ""), # `dict`s -> dicts

("'s", "s"), # it's -> its

("'t", "t"), # don't -> dont

("**", ""), # **FastAPI**s -> FastAPIs

],

)

def get_en_config() -> dict[str, Any]:

return mkdocs.utils.yaml_load(en_config_path.read_text(encoding="utf-8"))

def get_lang_paths() -> list[Path]:

return sorted(docs_path.iterdir())

def lang_callback(lang: str | None) -> str | None:

if lang is None:

return None

lang = lang.lower()

return lang

def complete_existing_lang(incomplete: str):

lang_path: Path

for lang_path in get_lang_paths():

if lang_path.is_dir() and lang_path.name.startswith(incomplete):

yield lang_path.name

@app.callback()

def callback() -> None:

# For MacOS with Cairo

os.environ["DYLD_FALLBACK_LIBRARY_PATH"] = "/opt/homebrew/lib"

@app.command()

def new_lang(lang: str = typer.Argument(..., callback=lang_callback)):

"""

Generate a new docs translation directory for the language LANG.

"""

new_path: Path = Path("docs") / lang

if new_path.exists():

typer.echo(f"The language was already created: {lang}")

raise typer.Abort()

new_path.mkdir()

new_config_path: Path = Path(new_path) / mkdocs_name

new_config_path.write_text("INHERIT: ../en/mkdocs.yml\n", encoding="utf-8")

new_llm_prompt_path: Path = new_path / "llm-prompt.md"

new_llm_prompt_path.write_text("", encoding="utf-8")

print(f"Successfully initialized: {new_path}")

update_languages()

@app.command()

def build_lang(

lang: str = typer.Argument(

..., callback=lang_callback, autocompletion=complete_existing_lang

),

) -> None:

"""

Build the docs for a language.

"""

lang_path: Path = Path("docs") / lang

if not lang_path.is_dir():

typer.echo(f"The language translation doesn't seem to exist yet: {lang}")

raise typer.Abort()

typer.echo(f"Building docs for: {lang}")

build_site_dist_path = build_site_path / lang

if lang == "en":

dist_path = site_path

# Don't remove en dist_path as it might already contain other languages.

# When running build_all(), that function already removes site_path.

# All this is only relevant locally, on GitHub Actions all this is done through

# artifacts and multiple workflows, so it doesn't matter if directories are

# removed or not.

else:

dist_path = site_path / lang

shutil.rmtree(dist_path, ignore_errors=True)

current_dir = os.getcwd()

os.chdir(lang_path)

shutil.rmtree(build_site_dist_path, ignore_errors=True)

subprocess.run(["mkdocs", "build", "--site-dir", build_site_dist_path], check=True)

shutil.copytree(build_site_dist_path, dist_path, dirs_exist_ok=True)

os.chdir(current_dir)

typer.secho(f"Successfully built docs for: {lang}", color=typer.colors.GREEN)

index_sponsors_template = """

### Keystone Sponsor

{% for sponsor in sponsors.keystone -%}

<a href="{{ sponsor.url }}" target="_blank" title="{{ sponsor.title }}"><img src="{{ sponsor.img }}"></a>

{% endfor %}

### Gold and Silver Sponsors

{% for sponsor in sponsors.gold -%}

<a href="{{ sponsor.url }}" target="_blank" title="{{ sponsor.title }}"><img src="{{ sponsor.img }}"></a>

{% endfor -%}

{%- for sponsor in sponsors.silver -%}

<a href="{{ sponsor.url }}" target="_blank" title="{{ sponsor.title }}"><img src="{{ sponsor.img }}"></a>

{% endfor %}

"""

def remove_header_permalinks(content: str):

lines: list[str] = []

for line in content.split("\n"):

match = header_with_permalink_pattern.match(line)

if match:

hashes, title, *_ = match.groups()

line = f"{hashes} {title}"

lines.append(line)

return "\n".join(lines)

def generate_readme_content() -> str:

en_index = en_docs_path / "docs" / "index.md"

content = en_index.read_text("utf-8")

content = remove_header_permalinks(content) # remove permalinks from headers

match_pre = re.search(r"</style>\n\n", content)

match_start = re.search(r"<!-- sponsors -->", content)

match_end = re.search(r"<!-- /sponsors -->", content)

sponsors_data_path = en_docs_path / "data" / "sponsors.yml"

sponsors = mkdocs.utils.yaml_load(sponsors_data_path.read_text(encoding="utf-8"))

if not (match_start and match_end):

raise RuntimeError("Couldn't auto-generate sponsors section")

if not match_pre:

raise RuntimeError("Couldn't find pre section (<style>) in index.md")

frontmatter_end = match_pre.end()

pre_end = match_start.end()

post_start = match_end.start()

template = Template(index_sponsors_template)

message = template.render(sponsors=sponsors)

pre_content = content[frontmatter_end:pre_end]

post_content = content[post_start:]

new_content = pre_content + message + post_content

# Remove content between <!-- only-mkdocs --> and <!-- /only-mkdocs -->

new_content = re.sub(

r"<!-- only-mkdocs -->.*?<!-- /only-mkdocs -->",

"",

new_content,

flags=re.DOTALL,

)

return new_content

@app.command()

def generate_readme() -> None:

"""

Generate README.md content from main index.md

"""

readme_path = Path("README.md")

old_content = readme_path.read_text("utf-8")

new_content = generate_readme_content()

if new_content != old_content:

print("README.md outdated from the latest index.md")

print("Updating README.md")

readme_path.write_text(new_content, encoding="utf-8")

raise typer.Exit(1)

print("README.md is up to date ✅")

@app.command()

def build_all() -> None:

"""

Build mkdocs site for en, and then build each language inside, end result is located

at directory ./site/ with each language inside.

"""

update_languages()

shutil.rmtree(site_path, ignore_errors=True)

langs = [

lang.name

for lang in get_lang_paths()

if (lang.is_dir() and lang.name in SUPPORTED_LANGS)

]

cpu_count = os.cpu_count() or 1

process_pool_size = cpu_count * 4

typer.echo(f"Using process pool size: {process_pool_size}")

with Pool(process_pool_size) as p:

p.map(build_lang, langs)

@app.command()

def update_languages() -> None:

"""

Update the mkdocs.yml file Languages section including all the available languages.

"""

old_config = get_en_config()

updated_config = get_updated_config_content()

if old_config != updated_config:

print("docs/en/mkdocs.yml outdated")

print("Updating docs/en/mkdocs.yml")

en_config_path.write_text(

yaml.dump(updated_config, sort_keys=False, width=200, allow_unicode=True),

encoding="utf-8",

)

raise typer.Exit(1)

print("docs/en/mkdocs.yml is up to date ✅")

@app.command()

def serve() -> None:

"""

A quick server to preview a built site with translations.

For development, prefer the command live (or just mkdocs serve).

This is here only to preview a site with translations already built.

Make sure you run the build-all command first.

"""

typer.echo("Warning: this is a very simple server.")

typer.echo("For development, use the command live instead.")

typer.echo("This is here only to preview a site with translations already built.")

typer.echo("Make sure you run the build-all command first.")

os.chdir("site")

server_address = ("", 8008)

server = HTTPServer(server_address, SimpleHTTPRequestHandler)

typer.echo("Serving at: http://127.0.0.1:8008")

server.serve_forever()

@app.command()

def live(

lang: str = typer.Argument(

None, callback=lang_callback, autocompletion=complete_existing_lang

),

dirty: bool = False,

) -> None:

"""

Serve with livereload a docs site for a specific language.

This only shows the actual translated files, not the placeholders created with

build-all.

Takes an optional LANG argument with the name of the language to serve, by default

en.

"""

# Enable line numbers during local development to make it easier to highlight

if lang is None:

lang = "en"

lang_path: Path = docs_path / lang

# Enable line numbers during local development to make it easier to highlight

args = ["mkdocs", "serve", "--dev-addr", "127.0.0.1:8008"]

if dirty:

args.append("--dirty")

subprocess.run(

args, env={**os.environ, "LINENUMS": "true"}, cwd=lang_path, check=True

)

def get_updated_config_content() -> dict[str, Any]:

config = get_en_config()

languages = [{"en": "/"}]

new_alternate: list[dict[str, str]] = []

# Language names sourced from https://quickref.me/iso-639-1

# Contributors may wish to update or change these, e.g. to fix capitalization.

language_names_path = Path(__file__).parent / "../docs/language_names.yml"

local_language_names: dict[str, str] = mkdocs.utils.yaml_load(

language_names_path.read_text(encoding="utf-8")

)

for lang_path in get_lang_paths():

if lang_path.name in {"en", "em"} or not lang_path.is_dir():

continue

if lang_path.name not in SUPPORTED_LANGS:

# Skip languages that are not yet ready

continue

code = lang_path.name

languages.append({code: f"/{code}/"})

for lang_dict in languages:

code = list(lang_dict.keys())[0]

url = lang_dict[code]

if code not in local_language_names:

print(

f"Missing language name for: {code}, "

"update it in docs/language_names.yml"

)

raise typer.Abort()

use_name = f"{code} - {local_language_names[code]}"

new_alternate.append({"link": url, "name": use_name})

config["extra"]["alternate"] = new_alternate

return config

@app.command()

def ensure_non_translated() -> None:

"""

Ensure there are no files in the non translatable pages.

"""

print("Ensuring no non translated pages")

lang_paths = get_lang_paths()

error_paths = []

for lang in lang_paths:

if lang.name == "en":

continue

for non_translatable in non_translated_sections:

non_translatable_path = lang / "docs" / non_translatable

if non_translatable_path.exists():

error_paths.append(non_translatable_path)

if error_paths:

print("Non-translated pages found, removing them:")

for error_path in error_paths:

print(error_path)

if error_path.is_file():

error_path.unlink()

else:

shutil.rmtree(error_path)

raise typer.Exit(1)

print("No non-translated pages found ✅")

@app.command()

def langs_json():

langs = []

for lang_path in get_lang_paths():

if lang_path.is_dir() and lang_path.name in SUPPORTED_LANGS:

langs.append(lang_path.name)

print(json.dumps(langs))

@app.command()

def generate_docs_src_versions_for_file(file_path: Path) -> None:

target_versions = ["py39", "py310"]

full_path_str = str(file_path)

for target_version in target_versions:

if f"_{target_version}" in full_path_str:

logging.info(

f"Skipping {file_path}, already a version file for {target_version}"

)

return

base_content = file_path.read_text(encoding="utf-8")

previous_content = {base_content}

for target_version in target_versions:

version_result = subprocess.run(

[

find_ruff_bin(),

"check",

"--target-version",

target_version,

"--fix",

"--unsafe-fixes",

"-",

],

input=base_content.encode("utf-8"),

capture_output=True,

)

content_target = version_result.stdout.decode("utf-8")

format_result = subprocess.run(

[find_ruff_bin(), "format", "-"],

input=content_target.encode("utf-8"),

capture_output=True,

)

content_format = format_result.stdout.decode("utf-8")

if content_format in previous_content:

continue

previous_content.add(content_format)

# Determine where the version label should go: in the parent directory

# name or in the file name, matching the source structure.

label_in_parent = False

for v in target_versions:

if f"_{v}" in file_path.parent.name:

label_in_parent = True

break

if label_in_parent:

parent_name = file_path.parent.name

for v in target_versions:

parent_name = parent_name.replace(f"_{v}", "")

new_parent = file_path.parent.parent / f"{parent_name}_{target_version}"

new_parent.mkdir(parents=True, exist_ok=True)

version_file = new_parent / file_path.name

else:

base_name = file_path.stem

for v in target_versions:

if base_name.endswith(f"_{v}"):

base_name = base_name[: -len(f"_{v}")]

break

version_file = file_path.with_name(f"{base_name}_{target_version}.py")

logging.info(f"Writing to {version_file}")

version_file.write_text(content_format, encoding="utf-8")

@app.command()

def generate_docs_src_versions() -> None:

"""

Generate Python version-specific files for all .py files in docs_src.

"""

docs_src_path = Path("docs_src")

for py_file in sorted(docs_src_path.rglob("*.py")):

generate_docs_src_versions_for_file(py_file)

@app.command()

def copy_py39_to_py310() -> None:

"""

For each docs_src file/directory with a _py39 label that has no _py310

counterpart, copy it with the _py310 label.

"""

docs_src_path = Path("docs_src")

# Handle directory-level labels (e.g. app_b_an_py39/)

for dir_path in sorted(docs_src_path.rglob("*_py39")):

if not dir_path.is_dir():

continue

py310_dir = dir_path.parent / dir_path.name.replace("_py39", "_py310")

if py310_dir.exists():

continue

logging.info(f"Copying directory {dir_path} -> {py310_dir}")

shutil.copytree(dir_path, py310_dir)

# Handle file-level labels (e.g. tutorial001_py39.py)

for file_path in sorted(docs_src_path.rglob("*_py39.py")):

if not file_path.is_file():

continue

# Skip files inside _py39 directories (already handled above)

if "_py39" in file_path.parent.name:

continue

py310_file = file_path.with_name(

file_path.name.replace("_py39.py", "_py310.py")

)

if py310_file.exists():

continue

logging.info(f"Copying file {file_path} -> {py310_file}")

shutil.copy2(file_path, py310_file)

@app.command()

def update_docs_includes_py39_to_py310() -> None:

"""

Update .md files in docs/en/ to replace _py39 includes with _py310 versions.

For each include line referencing a _py39 file or directory in docs_src, replace

the _py39 label with _py310.

"""

include_pattern = re.compile(r"\{[^}]*docs_src/[^}]*_py39[^}]*\.py[^}]*\}")

count = 0

for md_file in sorted(en_docs_path.rglob("*.md")):

content = md_file.read_text(encoding="utf-8")

if "_py39" not in content:

continue

new_content = include_pattern.sub(

lambda m: m.group(0).replace("_py39", "_py310"), content

)

if new_content != content:

md_file.write_text(new_content, encoding="utf-8")

count += 1

logging.info(f"Updated includes in {md_file}")

print(f"Updated {count} file(s) ✅")

@app.command()

def remove_unused_docs_src() -> None:

"""

Delete .py files in docs_src that are not included in any .md file under docs/.

"""

docs_src_path = Path("docs_src")

# Collect all docs .md content referencing docs_src

all_docs_content = ""

for md_file in docs_path.rglob("*.md"):

all_docs_content += md_file.read_text(encoding="utf-8")

# Build a set of directory-based package roots (e.g. docs_src/bigger_applications/app_py39)

# where at least one file is referenced in docs. All files in these directories

# should be kept since they may be internally imported by the referenced files.

used_package_dirs: set[Path] = set()

for py_file in docs_src_path.rglob("*.py"):

if py_file.name == "__init__.py":

continue

rel_path = str(py_file)

if rel_path in all_docs_content:

# Walk up from the file's parent to find the package root

# (a subdirectory under docs_src/<topic>/)

parts = py_file.relative_to(docs_src_path).parts

if len(parts) > 2:

# File is inside a sub-package like docs_src/topic/app_xxx/...

package_root = docs_src_path / parts[0] / parts[1]

used_package_dirs.add(package_root)

removed = 0

for py_file in sorted(docs_src_path.rglob("*.py")):

if py_file.name == "__init__.py":

continue

# Build the relative path as it appears in includes (e.g. docs_src/first_steps/tutorial001.py)

rel_path = str(py_file)

if rel_path in all_docs_content:

continue

# If this file is inside a directory-based package where any sibling is

# referenced, keep it (it's likely imported internally).

parts = py_file.relative_to(docs_src_path).parts

if len(parts) > 2:

package_root = docs_src_path / parts[0] / parts[1]

if package_root in used_package_dirs:

continue

# Check if the _an counterpart (or non-_an counterpart) is referenced.

# If either variant is included, keep both.

# Handle both file-level _an (tutorial001_an.py) and directory-level _an

# (app_an/main.py)

counterpart_found = False

full_path_str = str(py_file)

if "_an" in py_file.stem:

# This is an _an file, check if the non-_an version is referenced

counterpart = full_path_str.replace(

f"/{py_file.stem}", f"/{py_file.stem.replace('_an', '', 1)}"

)

if counterpart in all_docs_content:

counterpart_found = True

else:

# This is a non-_an file, check if there's an _an version referenced

# Insert _an before any version suffix or at the end of the stem

stem = py_file.stem

for suffix in ("_py39", "_py310"):

if suffix in stem:

an_stem = stem.replace(suffix, f"_an{suffix}", 1)

break

else:

an_stem = f"{stem}_an"

counterpart = full_path_str.replace(f"/{stem}.", f"/{an_stem}.")

if counterpart in all_docs_content:

counterpart_found = True

# Also check directory-level _an counterparts

if not counterpart_found:

parent_name = py_file.parent.name

if "_an" in parent_name:

counterpart_parent = parent_name.replace("_an", "", 1)

counterpart_dir = str(py_file).replace(

f"/{parent_name}/", f"/{counterpart_parent}/"

)

if counterpart_dir in all_docs_content:

counterpart_found = True

else:

# Try inserting _an into parent directory name

for suffix in ("_py39", "_py310"):

if suffix in parent_name:

an_parent = parent_name.replace(suffix, f"_an{suffix}", 1)

break

else:

an_parent = f"{parent_name}_an"

counterpart_dir = str(py_file).replace(

f"/{parent_name}/", f"/{an_parent}/"

)

if counterpart_dir in all_docs_content:

counterpart_found = True

if counterpart_found:

continue

logging.info(f"Removing unused file: {py_file}")

py_file.unlink()

removed += 1

# Clean up directories that are empty or only contain __init__.py / __pycache__

for dir_path in sorted(docs_src_path.rglob("*"), reverse=True):

if not dir_path.is_dir():

continue

remaining = [

f

for f in dir_path.iterdir()

if f.name != "__pycache__" and f.name != "__init__.py"

]

if not remaining:

logging.info(f"Removing empty/init-only directory: {dir_path}")

shutil.rmtree(dir_path)

print(f"Removed {removed} unused file(s) ✅")

@app.command()

def add_permalinks_page(path: Path, update_existing: bool = False):

"""

Add or update header permalinks in specific page of En docs.

"""

if not path.is_relative_to(en_docs_path / "docs"):

raise RuntimeError(f"Path must be inside {en_docs_path}")

rel_path = path.relative_to(en_docs_path / "docs")

# Skip excluded sections

if str(rel_path).startswith(non_translated_sections):

return

visible_text_extractor = VisibleTextExtractor()

updated_lines = []

in_code_block3 = False

in_code_block4 = False

permalinks = set()

with path.open("r", encoding="utf-8") as f:

lines = f.readlines()

for line in lines:

# Handle codeblocks start and end

if not (in_code_block3 or in_code_block4):

if code_block4_pattern.match(line):

in_code_block4 = True

elif code_block3_pattern.match(line):

in_code_block3 = True

else:

if in_code_block4 and code_block4_pattern.match(line):

in_code_block4 = False

elif in_code_block3 and code_block3_pattern.match(line):

in_code_block3 = False

# Process Headers only outside codeblocks

if not (in_code_block3 or in_code_block4):

match = header_pattern.match(line)

if match:

hashes, title, _permalink = match.groups()

if (not _permalink) or update_existing:

slug = slugify(

visible_text_extractor.extract_visible_text(

strip_markdown_links(title)

)

)

if slug in permalinks:

# If the slug is already used, append a number to make it unique

count = 1

original_slug = slug

while slug in permalinks:

slug = f"{original_slug}_{count}"

count += 1

permalinks.add(slug)

line = f"{hashes} {title} {{ #{slug} }}\n"

updated_lines.append(line)

with path.open("w", encoding="utf-8") as f:

f.writelines(updated_lines)

@app.command()

def add_permalinks_pages(pages: list[Path], update_existing: bool = False) -> None:

"""

Add or update header permalinks in specific pages of En docs.

"""

for md_file in pages:

add_permalinks_page(md_file, update_existing=update_existing)

@app.command()

def add_permalinks(update_existing: bool = False) -> None:

"""

Add or update header permalinks in all pages of En docs.

"""

for md_file in en_docs_path.rglob("*.md"):

add_permalinks_page(md_file, update_existing=update_existing)

if __name__ == "__main__":

app()