From 3f0130c75b20f93c48a2b79187886a36fbc77b6f Mon Sep 17 00:00:00 2001
From: Marco Lents <lentsmarco@gmail.com>
Date: Sat, 15 Nov 2025 15:30:07 +0100
Subject: [PATCH] fix disclosuure handling

---
 crawler.py | 33 ++++++++++++++++++++++++++++++---
 1 file changed, 30 insertions(+), 3 deletions(-)

diff --git a/crawler.py b/crawler.py
index c47432e..46aa103 100644
--- a/crawler.py
+++ b/crawler.py
@@ -88,8 +88,7 @@ Biographie: {self.cv}
 # Mandat 
 {self.mandate[0]},  {self.mandate[1]}
 
-# Veröffentlichungspflichtige Angaben
-{self.disclosures.replace(DISCLOSURE_DISLAIMER, "")}
+# Veröffentlichungspflichtige Angaben {funcs_to_str(self.disclosures)}
             """
         return txt
 
@@ -125,6 +124,7 @@ def main():
     )
     parser.add_argument("-o", "--out")
     parser.add_argument("--debug", action="store_true")
+    parser.add_argument("--no-git", action="store_true")
     args = parser.parse_args()
     if not args.out:
         raise ValueError("must supply out directory")
@@ -142,6 +142,9 @@ def main():
     save_raw(bios, args.out)
     save_individuals(bios, args.out)
 
+    if args.no_git:
+        return
+
     repo.git.add("*")
     repo.index.commit(datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
     origin = repo.remote(name="origin")
@@ -209,7 +212,7 @@ def get_bio(url, name, sleep_for):
         soup.find(class_="m-biography__subHeading --mandate").text,
         soup.find(string=re.compile(r"^Wahlkreis \d*:")),
     )
-    disclosures = soup.find(class_="m-biography__infoDisclaimer").text.strip()
+    disclosures = get_disclosures(soup.find(class_="m-biography__infos"))
 
     bio = Biography(
         name,
@@ -228,6 +231,30 @@ def get_bio(url, name, sleep_for):
     return bio
 
 
+def get_disclosures(elem):
+    if not elem:
+        return None
+    divs = elem.find_all("div", class_=lambda cls: cls != "m-biography__infoDisclaimer")
+    out = []
+    for div in divs:
+        current_heading = ""
+        current_body = []
+        for child in div.children:
+            if child.name == "h3":
+                if current_body != []:
+                    out.append((current_heading, current_body))
+                current_heading = child.text.strip()
+                current_body = []
+                continue
+            if not child.name:
+                continue
+            if child.text.strip() == "":
+                continue
+            current_body.append(child.text.strip())
+        out.append((current_heading, current_body))
+    return out
+
+
 def get_functions(elem):
     out = []
     current_heading = None