From 0047a42f37a03d81021cb963470c4f6bb86482e4 Mon Sep 17 00:00:00 2001
From: olikelly00 <oliverkelly1995@hotmail.co.uk>
Date: Wed, 5 Jun 2024 12:46:54 +0100
Subject: [PATCH] negator.py updated - Team Cheemu

---
 src/harmony/matching/negator.py | 100 +++++++++++++++++++++++++++++++-
 1 file changed, 98 insertions(+), 2 deletions(-)

diff --git a/src/harmony/matching/negator.py b/src/harmony/matching/negator.py
index 3ded536..fe789a9 100644
--- a/src/harmony/matching/negator.py
+++ b/src/harmony/matching/negator.py
@@ -40,6 +40,13 @@ def get_change_en(doc) -> dict:
         if tok.text.lower() in {"always", "rather", "really", "very", "totally", "utterly", "absolutely", "completely",
                                 "frequently", "often", "sometimes", "generally", "usually"}:
             return {tok.i: ("replace", "never")}
+        # Team Cheemu: added these if statements to handle negative contractions (eg. can't, won't, shan't)
+        if tok.text.lower() == "ca" and doc[tok.i + 1].text.lower() == "n't":
+            return {tok.i: ("replace", "can"), tok.i + 1: ("replace", "")}
+        if tok.text.lower() == "wo" and doc[tok.i + 1].text.lower() == "n't":
+            return {tok.i: ("replace", "will"), tok.i + 1: ("replace", "")}
+        if tok.text.lower() == "sha" and doc[tok.i + 1].text.lower() == "n't":
+            return {tok.i: ("replace", "shall"), tok.i + 1: ("replace", "")}
         if tok.text.lower() in {"never", "not", "n't"}:
             return {tok.i: ("replace", "")}
         if tok.text.lower() in {"cannot"}:
@@ -73,18 +80,107 @@ def get_change_pt(doc) -> dict:
     return {0: ("insert_before", "não")}
 
 
+def get_change_es(doc) -> dict:
+    """
+    # Team Cheemu: Identify how to change a Spanish sentence from positive to negative or vice versa.
+    :param doc:
+    :return:
+    """
+    for tok in doc:
+        if tok.text.lower() in {"siempre", "bastante", "realmente", "muy", "mucho", "totalmente", "totalmente", "absolutamente",
+                                "completamente",
+                                "frecuentemente", "frequentemente", "veces"}:
+            return {tok.i: ("replace", "nunca")}
+        if tok.text.lower() in {"nunca", "jamás", "ni", "no"}:
+            return {tok.i: ("replace", "")}
+    result = {}
+    if len(result) > 0:
+        return result
+    return {0: ("insert_before", "no")}
+
+
+
+def get_change_it(doc) -> dict:
+    """
+    # Team Cheemu: Identify how to change an Italian sentence from positive to negative or vice versa.
+    :param doc:
+    :return:
+    """
+    for tok in doc:
+        if tok.text.lower() in {"sempre", "abbastanza", "realmente", "davvero", "veramente", "molto", "molta", "molti", "molte", "totalmente", "assolutamente",
+                                "completamente",
+                                "frequentemente", "qualche volta", "a volte", "ogni tanto"}:
+            return {tok.i: ("replace", "mai")}
+        if tok.text.lower() in {"mai", "né", "non", "nessuno", "nulla", "niente"}:
+            return {tok.i: ("replace", "")}
+    result = {}
+    for tok in doc:
+        if tok.text.lower() in {"è", "sono", "ero", "erano", "avevano", "avevo", "ho avuto", "sono stato", "sono stata", "sono stati", "siamo stati", "sono state"}:
+            result[tok.i] = "insert_before", "non"
+    if len(result) > 0:
+        return result
+    return {0: ("insert_before", "non")}
+
+def get_change_de(doc) -> dict:
+    """
+    # Team Cheemu: Identify how to change a German sentence from positive to negative or vice versa.
+    :param doc:
+    :return:
+    """
+    for tok in doc:
+        if tok.text.lower() in {"immer", "ziemlich", "wirklich", "sehr", "viel", "total", "absolut",
+                                "vollständig",
+                                "häufig", "manchmal"}:
+            return {tok.i: ("replace", "nie")}
+        if tok.text.lower() in {"nie", "niemals", "weder", "nicht"}:
+            return {tok.i: ("replace", "")}
+    result = {}
+    if len(result) > 0:
+        return result
+    return {0: ("insert_before", "nicht")}
+# if we had time: add functionality to handle german word order using Spacy
+
+
+def get_change_fr(doc) -> dict:
+    """
+    # Team Cheemu: Identify how to change a French sentence from positive to negative or vice versa.
+    :param doc:
+    :return:
+    """
+    for tok in doc:
+        if tok.text.lower() in {"toujours", "assez", "vraiment", "très", "beaucoup de", "totalement", "absolumment",
+                                "complètement", "plus", "trop de", "plein de",
+                                "souvent", "de temps en temps"}:
+            return {tok.i: ("replace", "nie")}
+        if tok.text.lower() in {"personne", "jamais", "ni", "rien", "pas", "non", "ne", "n'", "nulle", "aucun", "aucune", "guère"}:
+            return {tok.i: ("replace", "")}
+    result = {}
+    if len(result) > 0:
+        return result
+    return {0: ("insert_before", "ne pas")}
+
 def negate(text: str, language: str) -> str:
     """
-    Converts negative sentences to pos and vice versa.
+    Converts negative sentences to positive and vice versa.
     Not meant to generate 100% accurate natural language, it's to go into transformer model and is not shown to a human.
     :param text:
-    :param language: "en" or "pt"
+    :param language:
+    "en" for English, "pt" for Portuguese, "es" for Spanish, "it" for Italian, "de" for German, "fr" for French.
     :return: the sentence negated
     """
     doc = nlp(text)
 
     if language == "pt":
         changes = get_change_pt(doc)
+    # Team Cheemu: added handling of four additional languages
+    elif language == "es":
+        changes = get_change_es(doc)
+    elif language == "it":
+        changes = get_change_it(doc)
+    elif language == "fr":
+        changes = get_change_fr(doc)
+    elif language == "de":
+        changes = get_change_de(doc)
     else:
         changes = get_change_en(doc)