Hierarchical TOC + routing

src/formex_viewer/server.py

@@ -28,7 +28,7 @@ type CacheKey = tuple[str, Language]
 CACHE: dict[CacheKey, str] = {}
 
 
-def _get_fmx4_data(celex_id: str, language: Language) -> str:
+def _get_fmx4_data(celex_id: str, language: Language) -> ET.Element:
     """
     Fetch the FMX4 data from the server.
     """
@@ -43,9 +43,10 @@ def _get_fmx4_data(celex_id: str, language: Language) -> str:
     )
     fmx4_data = client.publication_text(cellar_id, ContentType.ZIP_FMX4)
 
-    CACHE[(celex_id, language)] = fmx4_data
+    xml = ET.fromstring(fmx4_data.encode("utf-8"))
+    CACHE[(celex_id, language)] = xml
 
-    return fmx4_data
+    return xml
 
 
 @app.get("/{celex_id}/articles")
@@ -53,8 +54,7 @@ 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"))
+    xml = _get_fmx4_data(celex_id, language)
 
     article_xpath = "//ARTICLE/@IDENTIFIER"
     article_ids = xml.xpath(article_xpath)
@@ -65,18 +65,16 @@ def article_ids(celex_id: str, language: Language = Language.ENG):
 
 @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"):
+    def _handle_division(division: ET.Element, level: int):
         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 ""
+
+        subdivisions = []
+        for subdivision in division.xpath("DIVISION") or []:
+            subdivisions.append(_handle_division(subdivision, level + 1))
+
         articles = []
         for article in division.xpath("ARTICLE") or []:
             art_id = article.get("IDENTIFIER")
@@ -94,14 +92,23 @@ def toc(celex_id: str, language: Language = Language.ENG):
                     "subtitle": art_subtitle,
                 }
             )
-        toc.append(
-            {
-                "title": title,
-                "type": "division",
-                "subtitle": subtitle,
-                "articles": articles,
-            }
-        )
+
+        return {
+            "type": "division",
+            "title": title,
+            "subtitle": subtitle,
+            "level": level,
+            "content": subdivisions + articles,
+        }
+
+    """
+    Fetch the table of contents from the server.
+    """
+    xml = _get_fmx4_data(celex_id, language)
+    toc = []
+
+    for division in xml.xpath("//ENACTING.TERMS/DIVISION"):
+        toc.append(_handle_division(division, 0))
 
     return toc
 
@@ -111,8 +118,7 @@ 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"))
+    xml = _get_fmx4_data(celex_id, language)
 
     article_xpath = "//ARTICLE"
     articles = xml.xpath(article_xpath)