def extract_related_reports(self):
""" 연관 보고서 리스트 추출
Returns
-------
list of RelatedReport
연관 보고서리스트 반환
"""
if self.html is None:
self._get_report()
results = []
soup = self.html
family = soup.find('select', id='family')
related_reports = family.find_all('option')
for report in related_reports:
value = report.attrs.get('value')
if compare_str(value, 'null'):
continue
rpt_nm = re.sub(r'\s+', ' ', report.text).strip()
rcp_no = value.split('=')[1]
if compare_str(self.rcp_no, rcp_no):
if self.info.get('rpt_nm') is None:
self.info['rpt_nm'] = rpt_nm
continue
info = {'rcp_no': rcp_no, 'rpt_nm': rpt_nm, 'parent': self}
results.append(RelatedReport(**info))
self._related_reports = sorted(results,
key=lambda x: x.rcp_no,
reverse=True)
return self._related_reports
def extract_attached_reports(self):
""" 첨부된 보고서 리스트 추출 및 반환
Returns
-------
list of AttachedReport
첨부된 보고서 리스트
"""
if self.html is None:
self._get_report()
soup = self.html
attached = soup.find('p', class_='f_none')
attached_list = attached.find_all('option')
attached_reports = []
for docs in attached_list:
rpt_nm = re.sub(r'\s+', ' ', docs.text).strip()
docs_url = docs.attrs.get('value')
if compare_str(docs_url, 'null'):
pass
else:
info = dict()
parsed = parse_qs(docs_url)
info['rcp_no'] = parsed.get('rcpNo')[0]
info['dcm_no'] = parsed.get('dcmNo')[0]
info['rpt_nm'] = rpt_nm
info['parent'] = self
attached_reports.append(AttachedReport(**info))
self._attached_reports = sorted(attached_reports,
key=lambda x: x.rcp_no,
reverse=True)
return self._attached_reports
def run_test(self):
fs = self.crp.get_financial_statement(start_dt=self.start_dt,
separate=self.separate,
report_tp=self.report_tp)
for test in self.test_set:
tp = test['fs_tp']
date = test['date']
column = test['column']
item = test['item']
expected = test['expected']
df = fs[tp]
date_column = find_all_columns(df=df, query=date)[0]
label_column = find_all_columns(df=df, query=column)[0]
actual = None
for idx in range(len(df)):
text = df[label_column].iloc[idx].replace(' ', '')
if compare_str(text, item):
actual = df[date_column].iloc[idx]
if actual != expected:
pytest.fail("Test failed: crp_cd='{}', fs_tp='{}', ".format(
self.crp.crp_cd, tp) +
"start_dt='{}', report_tp='{}', ".format(
self.start_dt, fs.info['report_tp']) +
"date='{}', column='{}',".format(date, column) +
"item='{}', actual='{}', expected='{}'".format(
item, actual, expected))
def get_value_from_dataset(classification, dataset, concept_id):
""" dataset에서 값을 추출하는 함수 """
def str_to_float(val):
try:
return float(val)
except ValueError:
return val
if isinstance(classification, dict):
classification = [classification]
results = list()
added_title = list()
for cls in classification:
value = float('nan')
for data in dataset[cls['cls_id']]:
if compare_str(data.concept.id, concept_id):
value = str_to_float(data.value)
break
title = get_title(cls, 'en')
if title in added_title:
index = added_title.index(title)
if not math.isnan(value):
results[index] = value
else:
results.append(value)
added_title.append(title)
return results
def get_table_by_code(self, code: str) -> Union[Table, None]:
""" Table 코드와 일치하는 Table 반환
Parameters
----------
code: str
Table 코드번호
Returns
-------
Table or None
코드 번호에 맞는 Table 또는 None
"""
for table in self.tables:
if compare_str(table.code, code):
return table
return None
def search_financial_statement(crp_cd: str, start_dt: str, end_dt: str = None,
fs_tp: Tuple[str] = ('fs', 'is', 'ci', 'cf'), separate: bool = False,
report_tp: str = 'annual', lang: str = 'ko',
separator: bool = True) -> FinancialStatement:
"""
재무제표 검색
Parameters
----------
crp_cd: str
종목코드
start_dt: str
검색 시작일자(YYYYMMDD)
end_dt: str, optional
검색 종료일자(YYYYMMDD)
fs_tp: tuple of str, optional
'fs' 재무상태표, 'is' 손익계산서, 'ci' 포괄손익계산서, 'cf' 현금흐름표
separate: bool, optional
개별재무제표 여부
report_tp: str, optional
'annual' 1년, 'half' 반기, 'quarter' 분기
lang: str, optional
'ko' 한글, 'en' 영문
separator: bool, optional
1000단위 구분자 표시 여부
Returns
-------
FinancialStatement
제무제표 검색 결과
"""
if is_notebook():
from tqdm import tqdm_notebook as tqdm
else:
from tqdm import tqdm
# 재무제표 검색 결과
statements = None
# 사업보고서 검색(최종보고서)
reports = search_report(crp_cd=crp_cd, start_dt=start_dt, end_dt=end_dt,
bsn_tp='A001', page_count=100, fin_rpt=True)
if len(reports) == 0:
# todo 감사보고서를 이용하여 재무제표 검색
raise RuntimeError('Could not find an annual report')
next_index = None
for idx, _ in enumerate(reports):
# 가장 최근 보고서의 경우 XBRL 파일을 이용하여 재무제표 검색
latest_report = reports[idx]
latest_xbrl = latest_report.xbrl
# XBRL 파일이 존재할 때
if latest_xbrl:
if separate is False and not latest_xbrl.exist_consolidated():
raise NotFoundConsolidated('Could not find consolidated financial statements')
# XBRL 정보를 이용하여 재무제표 정보 초기화
analyzed_results = analyze_xbrl(latest_report, fs_tp=fs_tp, separate=separate, lang=lang,
show_abstract=False, show_class=True,
show_depth=10, show_concept=True, separator=separator)
statements = copy.deepcopy(analyzed_results)
break
else:
statements = analyze_html(latest_report, fs_tp=fs_tp, separate=separate, lang=lang)
# Report 에 재무제표 정보 없이 수정 사항만 기록된 경우 다음 리포트 검색
if statements is not None:
next_index = idx + 1
break
if separate is False and all([statements[tp] is None for tp in statements]):
raise NotFoundConsolidated('Could not find consolidated financial statements')
label_df = None
for report in tqdm(reports[next_index:], desc='Annual reports', unit='report'):
statements, label_df = merge_fs(statements, label_df, report, fs_tp=fs_tp, separate=separate, lang=lang)
if compare_str(report_tp, 'half') or compare_str(report_tp, 'quarter'):
half = search_report(crp_cd=crp_cd, start_dt=start_dt, end_dt=end_dt,
bsn_tp=['A002'], page_count=100, fin_rpt=True)
for report in tqdm(half, desc='Semiannual reports', unit='report'):
statements, label_df = merge_fs(statements, label_df, report, fs_tp=fs_tp, separate=separate, lang=lang)
if compare_str(report_tp, 'quarter'):
quarter = search_report(crp_cd=crp_cd, start_dt=start_dt, end_dt=end_dt,
bsn_tp=['A003'], page_count=100, fin_rpt=True)
for report in tqdm(quarter, desc='Quarterly report', unit='report'):
statements, label_df = merge_fs(statements, label_df, report, fs_tp=fs_tp, separate=separate, lang=lang)
statements = drop_empty_columns(statements)
label_df = drop_empty_columns(label_df)
statements = sorting_columns(statements)
label_df = sorting_columns(label_df)
info = {
'crp_cd': crp_cd,
'start_dt': start_dt,
'end_dt': end_dt,
'separate': separate,
'report_tp': report_tp,
'lang': lang,
'separator': separator
}
return FinancialStatement(statements, label_df, info)
def convert_thead_into_columns(fs_tp: str, fs_table: dict, separate: bool = False,
lang: str = 'ko'):
""" thead에서 DataFrame의 columns을 추출하는 Method"""
def column_ko_to_en(ko):
ko_to_en = {
'과목': 'label_ko',
'주석': 'comment'
}
en = ko_to_en.get(ko)
return en if en else ko
thead = fs_table['table'].thead
if thead is None:
tt = fs_table['table'].tbody.tr.extract()
thead = BeautifulSoup('<thead></thead>', 'html.parser')
thead.thead.append(tt)
for td in thead.tr.find_all('td'):
td.name = 'th'
th_colspan_list = [int(th.attrs.get('colspan', 1)) for th in thead.tr.find_all('th')]
date_info = extract_date_from_header(fs_table['header'])
# Regular Expression for title
regex = str_to_regex('과목 OR 주석')
fs_string = {
'fs': 'Statement of financial position',
'is': 'Income statement',
'ci': 'Statement of comprehensive income',
'cf': 'Statement of cash flows'
}
str_unit = extract_unit_from_header(fs_table['header'])
str_unit = str_to_regex('원 OR USD').search(str_unit)
if str_unit:
str_unit = str_unit.group(0)
str_unit = 'KRW' if compare_str('원', str_unit) else 'USD'
for key in fs_string:
fs_string[key] = fs_string[key] + '(Unit: {})'.format(str_unit)
label = {
'ko': {
True: '별도재무제표',
False: '연결재무제표'
},
'en': {
True: 'Separate',
False: 'Consolidated'
}
}
# 최대 Col
col_length = sum(th_colspan_list)
# 최대 Row
row_length = len(thead.find_all('tr'))
row_length = row_length + 1 if row_length == 1 else row_length
# row-sapn, col-span을 처리하기 위한 Matrix
columns_matrix = [[None for y in range(col_length)] for x in range(row_length)]
for idx, tr in enumerate(thead.find_all('tr')):
start_idx = 0
for ele_idx, element in enumerate(columns_matrix[idx]):
if element is None:
start_idx = ele_idx
break
for jdx, th in enumerate(tr.find_all('th')):
row_span = int(th.attrs.get('rowspan', 1))
col_span = int(th.attrs.get('colspan', 1))
text = re.sub(r'\s+', '', th.text)
date_list = [datetime(1900, 1, 1)]
if idx == 0:
if jdx == 0:
text = '과목'
elif regex.search(text) is None:
if len(date_info) > 0:
date_list = date_info.pop(0)
else:
import warnings
date = '-'.join([date.strftime('%Y%m%d') for date in date_list])
warnings_text = "Date data length does not match table header."\
+ "So last date was set using last data({}). ".format(date)
warnings.warn(warnings_text, RuntimeWarning)
text = '-'.join([date.strftime('%Y%m%d') for date in date_list])
if regex.search(text):
row_span = 2
for mdx in range(row_span):
for ndx in range(col_span):
new_text = text
if mdx == 0 and regex.search(text):
new_text = fs_string[fs_tp]
columns_matrix[idx + mdx][start_idx + ndx] = new_text
start_idx = start_idx + ndx + 1
regex_3month = re.compile(r'3개월')
regex_total = str_to_regex(r'누적 OR 금액')
columns = []
for jdx in range(len(columns_matrix[0])):
column = []
sec_item = []
for idx in range(len(columns_matrix)):
item = columns_matrix[idx][jdx]
if idx == 0:
column.append(item)
continue
elif idx == 1 and (item is None or regex.search(item) is None):
sec_item.append(label[lang][separate])
else:
pass
if item is None:
pass
elif compare_str(column[0], item):
continue
elif regex_3month.search(item):
# extract date info
date_info = [datetime.strptime(date_str, '%Y%m%d') for date_str in column[0].split('-')]
# calculating start_dt
delta = relativedelta(months=3)
start_dt = date_info[1] - delta
start_dt = start_dt.replace(day=1)
end_dt = date_info[1]
column[0] = '-'.join([date.strftime('%Y%m%d') for date in [start_dt, end_dt]])
elif regex_total.search(item):
pass
else:
sec_item.append(column_ko_to_en(item))
if sec_item[0] in ['label_ko', 'comment']:
column.append(sec_item[0])
else:
column.append(tuple(sec_item))
columns.append(column)
return columns
