Fixed #9 (Preprocessors do not have access to the user_id and sentence)

src/corporacreator/corpus.py

@@ -41,7 +41,7 @@ def _pre_process_corpus_data(self):
             "/", expand=True
         )[0] # TODO: Remove this line when the Gregor modifies the csv output to include user_id
         preprocessor = getattr(preprocessors, self.locale.replace("-","")) # Get locale specific preprocessor
-        self.corpus_data["sentence"] = self.corpus_data["sentence"].apply(func=preprocessor)
+        self.corpus_data["sentence"] = self.corpus_data[["user_id", "sentence"]].apply(func=lambda arg: preprocessor(*arg), axis=1)
 
     def _partition_corpus_data(self):
         self.other = self.corpus_data.loc[

src/corporacreator/preprocessors/br.py

@@ -1,7 +1,8 @@
-def br(sentence):
+def br(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/ca.py

@@ -1,7 +1,8 @@
-def ca(sentence):
+def ca(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/cv.py

@@ -1,7 +1,8 @@
-def cv(sentence):
+def cv(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/cy.py

@@ -1,7 +1,8 @@
-def cy(sentence):
+def cy(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/de.py

@@ -1,7 +1,8 @@
-def de(sentence):
+def de(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/en.py

@@ -1,7 +1,8 @@
-def en(sentence):
+def en(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/fr.py

@@ -1,7 +1,8 @@
-def fr(sentence):
+def fr(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/gaIE.py

@@ -1,7 +1,8 @@
-def gaIE(sentence):
+def gaIE(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/it.py

@@ -1,7 +1,8 @@
-def it(sentence):
+def it(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/kab.py

@@ -1,7 +1,8 @@
-def kab(sentence):
+def kab(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/ky.py

@@ -1,7 +1,8 @@
-def ky(sentence):
+def ky(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/sl.py

@@ -1,7 +1,8 @@
-def sl(sentence):
+def sl(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/tr.py

@@ -1,7 +1,8 @@
-def tr(sentence):
+def tr(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/tt.py

@@ -1,7 +1,8 @@
-def tt(sentence):
+def tt(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns:

src/corporacreator/preprocessors/zhTW.py

@@ -1,7 +1,8 @@
-def zhTW(sentence):
+def zhTW(user_id, sentence):
     """Cleans up the passed sentence, removing or reformatting invalid data.
 
     Args:
+      user_id (str): User ID of sentence's speaker
       sentence (str): Sentence to be cleaned up.
 
     Returns: