Lemmatization from BSC/projecte_aina lookups

documentAssembler = DocumentAssembler() \
    .setInputCol("text") \
    .setOutputCol("document")

ex_list = ["aprox\.","pàg\.","p\.ex\.","gen\.","feb\.","abr\.","jul\.","set\.","oct\.","nov\.","dec\.","dr\.","dra\.","sr\.","sra\.","srta\.","núm\.","st\.","sta\.","pl\.","etc\.", "ex\."]
ex_list_all = []
ex_list_all.extend(ex_list)
ex_list_all.extend([x[0].upper() + x[1:] for x in ex_list])
ex_list_all.extend([x.upper() for x in ex_list])

tokenizer = Tokenizer() \
     .setInputCols(['sentence']).setOutputCol('token')\
     .setInfixPatterns(["(d|D)(els)","(d|D)(el)","(a|A)(ls)","(a|A)(l)","(p|P)(els)","(p|P)(el)",\
                            "([A-zÀ-ú_@]+)(-[A-zÀ-ú_@]+)",\
                             "(d'|D')([·A-zÀ-ú@_]+)(\.|\"|;|:|!|\?|\-|\(|\)|”|“|'|,)+","(l'|L')([·A-zÀ-ú_]+)(\.|\"|;|:|!|\?|\-|\(|\)|”|“|'|,)+", \
                             "(l'|l'|s'|s'|d'|d'|m'|m'|n'|n'|D'|D'|L'|L'|S'|S'|N'|N'|M'|M')([A-zÀ-ú_]+)",\
                             """([A-zÀ-ú·]+)(\.|,|\)|\?|!|;|\:|\"|”)(\.|,|\)|\?|!|;|\:|\"|”)+""",\
                             "([A-zÀ-ú·]+)('l|'ns|'t|'m|'n|-les|-la|-lo|-li|-los|-me|-nos|-te|-vos|-se|-hi|-ne|-ho)(\.|,|;|:|\?|,)+",\
                             "([A-zÀ-ú·]+)('l|'ns|'t|'m|'n|-les|-la|-lo|-li|-los|-me|-nos|-te|-vos|-se|-hi|-ne|-ho)",\
                             "(\.|\"|;|:|!|\?|\-|\(|\)|”|“|')+([0-9A-zÀ-ú_]+)",\
                             "([0-9A-zÀ-ú·]+)(\.|\"|;|:|!|\?|\(|\)|”|“|'|,|%)",\
                             "(\.|\"|;|:|!|\?|\-|\(|\)|”|“|,)+([0-9]+)(\.|\"|;|:|!|\?|\-|\(|\)|”|“|,)+",\
                             "(d'|D'|l'|L')([·A-zÀ-ú@_]+)('l|'ns|'t|'m|'n|-les|-la|-lo|-li|-los|-me|-nos|-te|-vos|-se|-hi|-ne|-ho)(\.|\"|;|:|!|\?|\-|\(|\)|”|“|,)", \
                             "([\.|\"|;|:|!|\?|\-|\(|\)|”|“|,]+)([\.|\"|;|:|!|\?|\-|\(|\)|”|“|,]+)"]) \
         .setExceptions(ex_list_all).fit(data)

lemmatizer = LemmatizerModel.pretrained("lemmatizer_bsc","ca") \
    .setInputCols(["token"]) \
    .setOutputCol("lemma")

pipeline = Pipeline(stages=[documentAssembler, tokenizer, lemmatizer]) 

example = spark.createDataFrame([["Bons dies, al mati"]], ["text"]) 

results = pipeline.fit(example).transform(example)