final

website-checker/app.py

@@ -35,6 +35,20 @@ IGNORE_SELECTORS = [
 
 CHUNK_SIZE = 9000  # stay under 10k limit
 
+# LanguageTool rule IDs to silently ignore
+IGNORED_RULES = {
+    "LEERZEICHEN_VOR_SATZZEICHEN",
+    "WHITESPACE_RULE",
+    "WHITESPACE_BEFORE_PUNCTUATION",
+    "LEERZEICHEN_VOR_DOPPELPUNKT",
+    "LEERZEICHEN_VOR_SEMIKOLON",
+    "LEERZEICHEN_VOR_AUSRUFEZEICHEN",
+    "LEERZEICHEN_VOR_FRAGEZEICHEN",
+}
+
+# Regex: space(s) directly before punctuation — catch any remaining cases
+SPACE_BEFORE_PUNCT_RE = re.compile(r"\s+[,.:;!?\"\u201C\u201D\u201E\u201F]")
+
 
 def normalize_url(url: str) -> str:
     """Remove fragment and trailing slash for dedup."""
@@ -289,6 +303,15 @@ def check_text_with_languagetool(
             if resp.status_code == 200:
                 result = resp.json()
                 for match in result.get("matches", []):
+                    rule_id = match.get("rule", {}).get("id", "")
+                    if rule_id in IGNORED_RULES:
+                        continue
+                    # Extra filter: skip any match that is just whitespace before punctuation
+                    m_off = match.get("offset", 0)
+                    m_len = match.get("length", 0)
+                    matched_text = chunk_text[m_off:m_off + m_len]
+                    if SPACE_BEFORE_PUNCT_RE.fullmatch(matched_text):
+                        continue
                     match["offset"] += offset
                     all_matches.append(match)
             elif resp.status_code in (401, 403):