troll-spelling-corrector: improve the detection and correction algorithms in the most horrendous way

modules/troll-spelling-corrector/index.py

@@ -1,5 +1,6 @@
 import re
 
+''' Grammar reference
 all_possible_match_strings = [
     "силка",  "силки",
     "силки",  "силок",
@@ -11,23 +12,113 @@ all_possible_match_strings = [
 
     "лінк",   "лінка",  "лінки",
     "лінку",  "лінки",  "лінок",
-    "лінці",  "лінці",  "лінкам",
+    "лінку",  "лінці",  "лінкам",
     "лінк",   "лінку",  "лінки",
     "лінком", "лінкою", "лінками",
     "лінку",  "лінці",  "лінках",
     "лінке",  "лінко",  "лінки"
 ]
+'''
 
-unique_match_strings = set(all_possible_match_strings)
+EXTRACT_PADDING = 11
+EXTRACT_INCLUDE_WHOLE_WORDS = True
 
-regex_matchers = [re.compile(fr"\b{i}\b") for i in unique_match_strings]
+corrections = [       # IDs
+    "посилання",      # 0
+    "посиланню",      # 1
+    "посиланням",     # 2
+    "на посиланні",   # 3
+    "посилань",       # 4
+    "посиланнями",    # 5
+    "посиланнях"      # 6
+]
+
+replacements = [
+    ["силка", "силки", "силку", "силко",
+     "лінк", "лінка", "лінки", "лінку", "лінке", "лінко"],
+    ["силці",
+     "лінку", "лінці"],
+    ["силкам", "силкою",
+     "лінкам", "лінком", "лінкою"],
+    ["на силці",
+     "на лінку", "на лінці"],
+    ["силок",
+     "лінок"],
+    ["силками",
+     "лінками"],
+    ["силках",
+     "лінках"]
+]
+
+#unique_match_strings = set(all_possible_match_strings)
+
+#ua_alphabet = "абвгґдеєжзиіїйклмнопрстуфхцчшщьюя"
+#regex_matchers = [re.compile(fr"((?<=[^{ua_alphabet}])|(?<=\b)|(?<=^)|(?<= )){i}((?=[^{ua_alphabet}])|(?=\b)|(?=$)|(?= ))", re.DEBUG)
+#        for i in unique_match_strings]
+
+ua_alphabet = "абвгґдеєжзиіїйклмнопрстуфхцчшщьюя"
+
+for i, group in enumerate(replacements):
+    for j, match_word in enumerate(group):
+        #replacements[i][j] = re.compile(fr"((?<=[^{ua_alphabet}])|(?<=\b)|(?<=^)|(?<= )){i}((?=[^{ua_alphabet}])|(?=\b)|(?=$)|(?= ))")
+        replacements[i][j] = [match_word, re.compile(fr"((?<=[^{ua_alphabet}])|(?<=\b)|(?<=^)|(?<= )){match_word}((?=[^{ua_alphabet}])|(?=\b)|(?=$)|(?= ))")]
+        #print(replacements[i][j])
+
+#_ = [print(fr"(?<=[^абвгґдеєжзиіїйклмнопрстуфхцчшщьюя]){i}(?=[^абвгґдеєжзиіїйклмнопрстуфхцчшщьюя])") for i in unique_match_strings]
 
 def process(message, path):
     lowercase_message = message.text.lower()
 
-    for m in regex_matchers:
+    for correct_word_id, group in enumerate(replacements):
-        result = m.match(lowercase_message)
+        for match_word, matcher in group:
-        if result:
+            result = matcher.search(lowercase_message)
-            return "*посилання 🌚", None
+
+            if result:
+                l = len(message.text)
+
+                mistake_start = result.start()
+                mistake_end = result.end()
+
+                print(mistake_start, mistake_end)
+
+                original_text_before = message.text[max(mistake_start-EXTRACT_PADDING,0):mistake_start]
+                original_text_after = message.text[mistake_end:min(mistake_end+EXTRACT_PADDING,l)]
+                original_text_mistake = message.text[mistake_start:mistake_end]
+                
+                if EXTRACT_INCLUDE_WHOLE_WORDS:
+                    while 0 <= mistake_start - EXTRACT_PADDING - 1 < l and \
+                            message.text[mistake_start-EXTRACT_PADDING-1].isalnum():
+                        mistake_start -= 1
+                        original_text_before = message.text[max(mistake_start-EXTRACT_PADDING,0):result.start()]
+
+                    while 0 <= mistake_end + EXTRACT_PADDING < l and \
+                            message.text[mistake_end+EXTRACT_PADDING].isalnum():
+                        mistake_end += 1
+                        original_text_after = message.text[result.end():min(mistake_end+EXTRACT_PADDING,l)]
+
+
+                if len(message.text[:mistake_start]) > EXTRACT_PADDING:
+                    original_text_before_continue = "..."
+                else:
+                    original_text_before_continue = ""
+
+                if len(message.text[mistake_end:]) > EXTRACT_PADDING:
+                    original_text_after_continue = "..."
+                else:
+                    original_text_after_continue = ""
+
+                original_extract = original_text_before_continue + original_text_before \
+                        + original_text_mistake + original_text_after + original_text_after_continue
+
+                correct_word = corrections[correct_word_id]
+                if original_text_mistake == match_word.capitalize():
+                    correct_word = corrections[correct_word_id].capitalize()
+                elif original_text_mistake == match_word.upper():
+                    correct_word = corrections[correct_word_id].upper()
+
+                fixed_extract = original_text_before_continue + original_text_before \
+                        + correct_word + original_text_after + original_text_after_continue
+
+                return f'"{original_extract}" -> "{fixed_extract}" 🌚', None
     else:
         return "", None