From ae12ba408599533d9c94513cc967ffac76576e5e Mon Sep 17 00:00:00 2001
From: xxyzz <gitpull@protonmail.com>
Date: Tue, 25 Feb 2025 10:19:32 +0800
Subject: [PATCH] [de] add "Hilfsverb haben", "Hilfsverb sein" tags in Flexion
 pages

---
 src/wiktextract/extractor/de/flexion.py | 13 ++++++++-----
 src/wiktextract/extractor/de/tags.py    |  3 +++
 2 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/src/wiktextract/extractor/de/flexion.py b/src/wiktextract/extractor/de/flexion.py
index 130380b8..bf21cfb2 100644
--- a/src/wiktextract/extractor/de/flexion.py
+++ b/src/wiktextract/extractor/de/flexion.py
@@ -14,13 +14,13 @@
 from .models import Form, WordEntry
 from .tags import GRAMMATICAL_TAGS, translate_raw_tags
 
-LEVEL2_TAGS = frozenset(["untrennbar"])
-
 
 def parse_flexion_page(
     wxr: WiktextractContext, word_entry: WordEntry, page_title: str
 ) -> None:
     # https://de.wiktionary.org/wiki/Hilfe:Flexionsseiten
+    LEVEL2_TAGS = ["Hilfsverb haben", "Hilfsverb sein"]
+
     flexion_page = wxr.wtp.get_page_body(
         page_title, wxr.wtp.NAMESPACE_DATA["Flexion"]["id"]
     )
@@ -37,10 +37,13 @@ def parse_flexion_page(
                 section_str = clean_node(wxr, None, node.largs)
                 for word in section_str.split(" "):
                     word = word.strip(", ")
-                    if (
-                        word in LEVEL2_TAGS or word in GRAMMATICAL_TAGS
-                    ) and not page_title.endswith(f":{word}"):
+                    if word in GRAMMATICAL_TAGS and not page_title.endswith(
+                        f":{word}"
+                    ):
                         shared_raw_tags.append(word)
+                for raw_tag in LEVEL2_TAGS:
+                    if raw_tag in section_str:
+                        shared_raw_tags.append(raw_tag)
             case NodeKind.TEMPLATE:
                 if node.template_name.startswith("Deklinationsseite"):
                     process_deklinationsseite_template(
diff --git a/src/wiktextract/extractor/de/tags.py b/src/wiktextract/extractor/de/tags.py
index e43157ce..944bc627 100644
--- a/src/wiktextract/extractor/de/tags.py
+++ b/src/wiktextract/extractor/de/tags.py
@@ -338,6 +338,9 @@
     # Vorlage:Deutsch Verb schwach trennbar reflexiv
     "Nebensatzkonjugation": "subordinate-clause",
     "Hauptsatzkonjugation": "main-clause",
+    "regelmäßig": "regular",
+    "untrennbar": "inseparable",
+    "trennbar": "separable",
 }
 
 GRAMMATICAL_TAGS = {