Team Ai
Apppublic

CalebKoster/Translation_Note_Alignment

sourceHugging Faceupdated 3y agoView on Hugging Face
0likes
tfidf.cpython-39.pyc23 linesDownload Raw Back to __pycache__
1a

2��e�@slddlmZddlZddlZddlmZddlm	Z	gd�Z3dd�e4D�Z5dd	d6�Zddd�Zdd
d�Z
dS)�)�TfidfVectorizerN)�islice)�uroman)A�i�i�7iii�ipi�i/iZi"iB%i(i�+i�.i0i�1iJ2ix6iW@i�Di=EiHJi�Oi6Pi/Ui�ViYWi�Wi4XiIXiyXi�XiIYi~Yi�YiwZi�Zi�^i�aifiqii_mioi�pi�qiZri�ri]si�sitiDti�tiui6uiOui~vi�viSwi�wi�wixixi.xi�ycCsg|]}|d�qS)r�)�.0�xrr�:c:\Users\caleb\Bible Translation Project\guidance\tfidf.py�8<listcomp>M�r9FcCs�t|ddd��f}t|�D]L\}}||kr|rJt|���Wd�S|��Wd�SqWd�n1sz0YdS)N�r�utf-8��encoding)�open�	enumerater�strip)�	file_path�line_number�romanize�file�i�linerrr	�extract_interested_versePs:rcCs�g}g}t|ddd���}t|dd�D]F\}}|tvr\|r\d�|�}|rNt|�}|�|�g}|�|���q$|r�d�|�}|r�t|�}|�|�Wd�n1s�0Y|S)Nrr
rr)�start� )rr�verses�joinr�appendr)rr�	documentsZcurrent_documentrrrZjoined_doc_stringrrr	�segment_corpus\s$101112(r cs�t||d�}tdd�}|�|�}|��}t�fdd�tt�D��d}t�||�	���13�}tt||���t
|�d|d�}	|	r�tdd�}14|15�|	g�|16��}�fdd�|D�}tt|��d	d17�dd��}
|
Sd
SdS)N)r)��)Zngram_rangec3s|]\}}|�kr|VqdS)Nr)rr�v)�interested_linerr	�	<genexpr>yrz*analyze_verse_in_corpus.<locals>.<genexpr>rcsi|]}|��|d��qS)r)�get)r�ngram)�scores_dictrr	�18<dictcomp>�rz+analyze_verse_in_corpus.<locals>.<dictcomp>cSs|dS)Nrr)�itemrrr	�<lambda>�rz)analyze_verse_in_corpus.<locals>.<lambda>T)�key�reversezVerse not found.)r rZ
fit_transformZget_feature_names_out�nextrr�np�array�todense�flatten�dict�zipr�fit�sorted�items)rr$rrZtfidf_vectorizerZtfidf_matrixZ
feature_namesZdocument_index�scoresZinterested_verseZtfidf_vectorizer_verseZverse_ngrams�verse_scoresZsorted_verse_scoresr)r$r(r	�analyze_verse_in_corpusrs 192021r:)F)F)F)Zsklearn.feature_extraction.textr�pandas�pd�numpyr/�	itertoolsrrrrrr r:rrrr	�<module>sE2223