126 lines
3.5 KiB
Python
126 lines
3.5 KiB
Python
import lxml.etree as ET
|
|
from fastapi import FastAPI, Response
|
|
from fastapi.middleware.cors import CORSMiddleware
|
|
|
|
from formex_viewer.formex4 import FormexArticleConverter
|
|
from formex_viewer.main import (
|
|
CellarClient,
|
|
CellarIdentifier,
|
|
ContentType,
|
|
Language,
|
|
SystemName,
|
|
)
|
|
|
|
origins = [
|
|
"http://localhost:5173",
|
|
]
|
|
app = FastAPI()
|
|
app.add_middleware(
|
|
CORSMiddleware,
|
|
allow_origins=origins,
|
|
allow_credentials=True,
|
|
allow_methods=["*"],
|
|
allow_headers=["*"],
|
|
)
|
|
|
|
type CacheKey = tuple[str, Language]
|
|
|
|
CACHE: dict[CacheKey, str] = {}
|
|
|
|
|
|
def _get_fmx4_data(celex_id: str, language: Language) -> str:
|
|
"""
|
|
Fetch the FMX4 data from the server.
|
|
"""
|
|
|
|
if (celex_id, language) in CACHE:
|
|
return CACHE[(celex_id, language)]
|
|
|
|
client = CellarClient(language)
|
|
cellar_id = CellarIdentifier(
|
|
system_name=SystemName.CELEX,
|
|
system_id=celex_id,
|
|
)
|
|
fmx4_data = client.publication_text(cellar_id, ContentType.ZIP_FMX4)
|
|
|
|
CACHE[(celex_id, language)] = fmx4_data
|
|
|
|
return fmx4_data
|
|
|
|
|
|
@app.get("/{celex_id}/articles")
|
|
def article_ids(celex_id: str, language: Language = Language.ENG):
|
|
"""
|
|
Fetch the article IDs from the server.
|
|
"""
|
|
fmx4_data = _get_fmx4_data(celex_id, language)
|
|
xml = ET.fromstring(fmx4_data.encode("utf-8"))
|
|
|
|
article_xpath = "//ARTICLE/@IDENTIFIER"
|
|
article_ids = xml.xpath(article_xpath)
|
|
article_ids = [int(article_id.lstrip("0")) for article_id in article_ids]
|
|
article_ids.sort()
|
|
return article_ids
|
|
|
|
|
|
@app.get("/{celex_id}/toc/{language}")
|
|
def toc(celex_id: str, language: Language = Language.ENG):
|
|
"""
|
|
Fetch the table of contents from the server.
|
|
"""
|
|
fmx4_data = _get_fmx4_data(celex_id, language)
|
|
xml = ET.fromstring(fmx4_data.encode("utf-8"))
|
|
|
|
toc = []
|
|
|
|
for division in xml.xpath("//DIVISION"):
|
|
print(division)
|
|
title = ti_el[0] if (ti_el := division.xpath("TITLE/TI//text()")) else ""
|
|
subtitle = sti_el[0] if (sti_el := division.xpath("TITLE/STI//text()")) else ""
|
|
articles = []
|
|
for article in division.xpath("ARTICLE") or []:
|
|
art_id = article.get("IDENTIFIER")
|
|
if not art_id:
|
|
continue
|
|
art_title = ti_el[0] if (ti_el := article.xpath("TI.ART//text()")) else ""
|
|
art_subtitle = (
|
|
sti_el[0] if (sti_el := article.xpath("STI.ART//text()")) else ""
|
|
)
|
|
articles.append(
|
|
{
|
|
"id": int(art_id.lstrip("0")),
|
|
"type": "article",
|
|
"title": art_title,
|
|
"subtitle": art_subtitle,
|
|
}
|
|
)
|
|
toc.append(
|
|
{
|
|
"title": title,
|
|
"type": "division",
|
|
"subtitle": subtitle,
|
|
"articles": articles,
|
|
}
|
|
)
|
|
|
|
return toc
|
|
|
|
|
|
@app.get("/{celex_id}/articles/{article_id}/{language}")
|
|
def article(celex_id: str, article_id: int, language: Language = Language.ENG):
|
|
"""
|
|
Fetch an article from the server.
|
|
"""
|
|
fmx4_data = _get_fmx4_data(celex_id, language)
|
|
xml = ET.fromstring(fmx4_data.encode("utf-8"))
|
|
|
|
article_xpath = "//ARTICLE"
|
|
articles = xml.xpath(article_xpath)
|
|
for article in articles:
|
|
num = article.get("IDENTIFIER").lstrip("0")
|
|
if num == str(article_id):
|
|
return Response(
|
|
FormexArticleConverter().convert_article(article),
|
|
media_type="text/html",
|
|
)
|