def _is_author_exits(email) : con_d = db.connect_dev() query = 'SELECT id FROM author WHERE email=\'' + email + '\'' result = db.do_select(con_d, query) con_d.close() if len(result) == 0 : return False return True
def __get_hook_word_dict() : con = db.connect_dev() keyword_dict = {} query = "SELECT * FROM hooking_keyword" result = db.do_select(con, query) con.close() for (id, word) in result : keyword_dict[word.encode('utf-8')] = id return keyword_dict
def main() : result = _get_raw_data() con = db.connect_dev() for row in result : # article table article = _make_article_info(con, row) query = db.make_insert_query('article', article) db.do_insert(con, query) # hooking keyword table raw_content = row[2] hook_words = _extract_hook_word(raw_content.encode('utf-8')) for word in hook_words.keys() : words_in_article = {'article_URL': article['URL'], 'hooking_keyword_id': word, 'count': hook_words[word]} query = db.make_insert_query('article_hooking_keyword', words_in_article) db.do_insert(con, query) # author table expected_author_string = row[5] author_list = _make_author_list(con, expected_author_string) for author in author_list : if not _is_author_exits(author['email']) : query = db.make_insert_query('author', author) db.do_insert(con, query) # Get author_id query = "SELECT id FROM author WHERE email=\'" + author['email'] + "\'" author_id = db.do_select(con, query) author_id = author_id[0][0] print author_id, author['name'], author['email'], author['press_id'] # article_author table article_author = {'article_URL': article['URL'], 'author_id': author_id} query = db.make_insert_query('article_author', article_author) db.do_insert(con, query)
def __get_press_id_from(url) : con_d = db.connect_dev() query = 'SELECT id FROM press WHERE domain=\'' + url + '\'' result = db.do_select(con_d, query) con_d.close() return result[0][0]