feat: extraire urls de legisquebec

alexi/extract.py

@@ -435,6 +435,7 @@ def analyse(self, iob: Iterable[T_obj], pdf_path: Path):
 
     def output_json(self):
         """Sauvegarder les metadonnées"""
+        self.metadata["urls"] = sorted(self.resolver.urls)
         with open(self.outdir / "index.json", "wt") as outfh:
             LOGGER.info("Génération de %s", self.outdir / "index.json")
             json.dump(self.metadata, outfh, indent=2, ensure_ascii=False)

alexi/link.py

@@ -55,6 +55,7 @@ def __init__(self, metadata: Optional[dict] = None):
         self.metadata = {"docs": {}} if metadata is None else metadata
         self.numeros = {}
         self.titles = {}
+        self.urls = set()
         for docpath, info in self.metadata["docs"].items():
             self.numeros[info["numero"]] = docpath
             self.titles[normalize_title(info["titre"])] = docpath
@@ -202,6 +203,7 @@ def resolve_external(self, text: str) -> Optional[str]:
             url = "https://www.legisquebec.gouv.qc.ca/fr/document/lc/Q-2"
         else:
             return None
+        self.urls.add(url)
         for m in SEC_RE.finditer(text):
             sectype = m.group("sec")
             num = m.group("num")