apply the "less diacritics" rules for lat -> cyr

myv_translit.py

@@ -138,12 +138,28 @@ _cyr2lat_first_e = [
     {'find_what': '\\bě', 'replacer': 'e', 're': True},
 ]
 
+_lat2cyr_first_e = [
+    {'find_what': '\\bE', 'replacer': 'Ě', 're': True},
+    {'find_what': '\\be', 'replacer': 'ě', 're': True},
+]
+
 _cyr2lat_soft_l_after_vowels = [
     # joint acutes | disjoint acutes
     {'find_what': '([iěeIĚE])(Ĺ|Ĺ)', 'replacer': '\\1L', 're': True},
     {'find_what': '([iěeIĚE])(ĺ|ĺ)', 'replacer': '\\1l', 're': True},
 ]
 
+_lat2cyr_soft_l_after_vowels = [
+    # add the soft sign, but only if the next letter is not softening
+    {'find_what': '([иэеИЭЕ])(Л)\\b', 'replacer': '\\1Ль', 're': True},
+    {'find_what': '([иэеИЭЕ])(л)\\b', 'replacer': '\\1ль', 're': True},
+    {'find_what': '([иэеИЭЕ])(Л)([^ьъиеюяю])', 'replacer': '\\1ЛЬ\\3', 're': True},
+    {'find_what': '([иэеИЭЕ])(л)([^ьъиеюяю])', 'replacer': '\\1ль\\3', 're': True},
+    # special cases when L is still hard
+    # todo: fix all the exclusions from the list in https://t.me/ravo_club/9776
+    {'find_what': '([иэеИЭЕ][Лл])([Ьь])(ГАД|ГАВТ|гад|гавт)', 'replacer': '\\1\\3', 're': True},
+]
+
 _lat2cyr = [
     {'find_what': 'Ŕ', 'replacer': 'Ŕ', 're': False},
     {'find_what': 'Ĺ', 'replacer': 'Ĺ', 're': False},
@@ -299,8 +315,11 @@ def cyr2lat(text, joint_acute=True, first_e_with_hacek=True, soft_l_after_vowels
 
 
 def lat2cyr(text, joint_acute=True, first_e_with_hacek=True, soft_l_after_vowels=True):
-    # todo: support all the optional settings
+    if not first_e_with_hacek:
+        text = transliterate_with_rules(text, _lat2cyr_first_e)
     text = transliterate_with_rules(text, _lat2cyr)
+    if not soft_l_after_vowels:
+        text = transliterate_with_rules(text, _lat2cyr_soft_l_after_vowels)
     text = transliterate_with_rules(text, _lat2cyr_special_cases)
     return text
 

test_translit.py

@@ -53,6 +53,23 @@ def test_consistency():
         line_cyr2 = lat2cyr(line_lat)
         assert line_cyr == line_cyr2
 
+    for line_cyr in lines:
+        line_lat = cyr2lat(line_cyr, joint_acute=False)
+        line_cyr2 = lat2cyr(line_lat, joint_acute=False)
+        assert line_cyr == line_cyr2
+
+    for line_cyr in lines:
+        line_lat = cyr2lat(line_cyr, soft_l_after_vowels=False)
+        line_cyr2 = lat2cyr(line_lat, soft_l_after_vowels=False)
+        assert line_cyr == line_cyr2
+
+    for line_cyr in lines:
+        if ' ежос' in line_cyr:  # normally, this does not happen in the Erzya language
+            continue
+        line_lat = cyr2lat(line_cyr, first_e_with_hacek=False)
+        line_cyr2 = lat2cyr(line_lat, first_e_with_hacek=False)
+        assert line_cyr == line_cyr2
+
 
 def test_zontik():
     with open('examples/zontik_cyr.txt', 'r') as f: