Beispiel #1
0
def api_upload():
    results = {}
    status_code = 200
    headers = {'Content-Type': 'application/json'}
    if request.method == 'POST':
        parsed_request = request.get_json()
        if 'files' not in parsed_request:
            status_code = 400
            results['error'] = 'Unrecognized JSON object in request'
            return (json.dumps(results), status_code, headers)
        processed = []
        for file_contents in parsed_request['files']:
            tokenized_text = word_tokenize(file_contents)
            keywords = get_keyword(file_contents)
            highlighted_words = get_highlighted_words(tokenized_text, keywords)
            abstract_id = insert_document(tokenized_text, keywords, abstracts)
            result = {}
            result['tokenized_text'] = tokenized_text
            result['keywords'] = keywords
            result['highlighted'] = highlighted_words
            result['abstract_id'] = abstract_id
            processed.append(result)
        results['results'] = processed
    else:  # unknown requests type
        results['error'] = 'Method Not Allowed'
        status_code = 405
    return (json.dumps(results), status_code, headers)
Beispiel #2
0
def keyword_output():
    if request.method == 'POST':
        text = request.form['text']
        tokenized_text = word_tokenize(text)
        keywords = get_keyword(text)
        highlighted = get_highlighted_words(tokenized_text, keywords)
        abstract_id = insert_document(tokenized_text, keywords, abstracts)
        related_abstracts = get_related_abstracts(abstract_id, keywords)
        create_file(text, keywords, abstract_id)
        return render_template('output.html', hashtags=keywords, tokenized_text=tokenized_text,
                               highlighted_text=highlighted, abstract_id=abstract_id,
                               related_abstracts=related_abstracts)
    else:
        return 'Something has gone terribly wrong.'
Beispiel #3
0
def get_data(opened_file):
    text = opened_file.read()
    decoded_text = text.decode("ISO-8859-1")
    tokenized_text = word_tokenize(decoded_text)
    keywords = get_keyword(decoded_text)
    highlighted_words = get_highlighted_words(tokenized_text, keywords)
    abstract_id = insert_document(tokenized_text, keywords, abstracts)
    # data that will be stored in output file
    filedata = {'abstract_id': abstract_id, 'filename': opened_file.filename, 'hashtags': keywords,
                'original_abstract': decoded_text, }
    # create a new file for each file uploaded with the data from the results
    resultsfile = open('results' + '-' + str(abstract_id) + '.txt', 'w+')
    jsonifieddata = json.dumps(filedata)
    resultsfile.write(jsonifieddata)
    resultsfile.close()
    # data needed for webpage items to be populated
    data = {'filename': opened_file.filename, 'hashtags': keywords, 'text': decoded_text,
            'tokenized_text': tokenized_text,
            'highlighted_text': highlighted_words, 'abstract_id': abstract_id}
    return data
Beispiel #4
0
def get_keywords():
    if request.method == 'POST':
        text = request.form['text']
        return json.dumps(get_keyword(text))
    else:
        return 'Something has gone terribly wrong.'