From ce6f0ca6abb0b13245a91d585b62b456bf0e7e09 Mon Sep 17 00:00:00 2001
From: Adrien Barbaresi <barbaresi@bbaw.de>
Date: Tue, 5 Nov 2024 17:25:23 +0100
Subject: [PATCH] add rules for Estonian

---
 simplemma/strategies/defaultrules/__init__.py |  2 ++
 simplemma/strategies/defaultrules/et.py       | 29 +++++++++++++++++++
 tests/strategies/defaultrules/test_rules.py   | 10 ++++++-
 3 files changed, 40 insertions(+), 1 deletion(-)
 create mode 100644 simplemma/strategies/defaultrules/et.py

diff --git a/simplemma/strategies/defaultrules/__init__.py b/simplemma/strategies/defaultrules/__init__.py
index cd2344f..2107dfd 100644
--- a/simplemma/strategies/defaultrules/__init__.py
+++ b/simplemma/strategies/defaultrules/__init__.py
@@ -4,6 +4,7 @@
 
 from .de import apply_de
 from .en import apply_en
+from .et import apply_et
 from .fi import apply_fi
 from .nl import apply_nl
 from .pl import apply_pl
@@ -12,6 +13,7 @@
 DEFAULT_RULES: Dict[str, Callable[[str], Optional[str]]] = {
     "de": apply_de,
     "en": apply_en,
+    "et": apply_et,
     "fi": apply_fi,
     "nl": apply_nl,
     "pl": apply_pl,
diff --git a/simplemma/strategies/defaultrules/et.py b/simplemma/strategies/defaultrules/et.py
new file mode 100644
index 0000000..d798163
--- /dev/null
+++ b/simplemma/strategies/defaultrules/et.py
@@ -0,0 +1,29 @@
+import re
+from typing import Optional
+
+from .generic import apply_rules
+
+DEFAULT_RULES = {
+    # adjectives
+    # https://en.wiktionary.org/wiki/-line
+    re.compile(
+        r"(?:lise|list|lisse|lisesse|lises|lisest|lisele|lisel|liselt|liseks|liseni|lisena|liseta|lisega|lised|liste|lisi|listesse|lisisse|listes|lisis|listest|lisist|listele|lisile|listel|lisil|listelt|lisilt|listeks|lisiks|listeni|listena|listeta|listega)$"
+    ): "line",
+    # nouns
+    # https://en.wiktionary.org/wiki/-kond
+    re.compile(
+        r"(?:konna|konda|konnasse|konnas|konnast|konnale|konnal|konnalt|konnaks|konnani|konnana|konnata|konnaga|konnad|kondade|kondi|kondasid|kondadesse|konnisse|kondades|konnis|kondadest|konnist|kondadele|konnile|kondadel|konnil|kondadelt|konnilt|kondadeks|konniks|kondadeni|kondadena|kondadeta|kondadega)$"
+    ): "kond",
+    # https://en.wiktionary.org/wiki/-nik
+    re.compile(
+        r"(?:niku|nikku|nikusse|nikus|nikust|nikule|nikul|nikult|nikuks|nikuni|nikuna|nikuta|nikuga|nikud|nike|nikudde|nikke|nikusid|nikesse|nikkudesse|nikes|nikkudes|nikest|nikkudest|nikele|nikkudele|nikel|nikkudel|nikelt|nikkudelt|nikeks|nikkudeks|nikeni|nikkudeni|nikena|nikkudena|niketa|nikkudeta|nikega|nikkudega)$"
+    ): "nik",
+}
+
+
+def apply_et(token: str) -> Optional[str]:
+    "Apply pre-defined rules for Estonian."
+    if len(token) < 10 or token[0].isupper():
+        return None
+
+    return apply_rules(token, DEFAULT_RULES)
diff --git a/tests/strategies/defaultrules/test_rules.py b/tests/strategies/defaultrules/test_rules.py
index 37ea474..bc6c61f 100644
--- a/tests/strategies/defaultrules/test_rules.py
+++ b/tests/strategies/defaultrules/test_rules.py
@@ -4,11 +4,19 @@
 def test_DEFAULT_RULES() -> None:
     """Test rules on all available languages."""
     rules_strategy = RulesStrategy()
+
     assert rules_strategy.get_lemma("Pfifferlinge", "de") == "Pfifferling"
-    assert rules_strategy.get_lemma("Pfifferlinge", "en") is None
     assert rules_strategy.get_lemma("atonements", "de") is None
+
     assert rules_strategy.get_lemma("atonements", "en") == "atonement"
+    assert rules_strategy.get_lemma("Pfifferlinge", "en") is None
+
     assert rules_strategy.get_lemma("brieven", "nl") == "brief"
+
     assert rules_strategy.get_lemma("liikenaisessa", "fi") == "liikenainen"
+
     assert rules_strategy.get_lemma("pracowaliście", "pl") == "pracować"
+
     assert rules_strategy.get_lemma("безгра́мотностью", "ru") == "безгра́мотность"
+
+    assert rules_strategy.get_lemma("keelkondade", "et") == "keelkond"