diff --git a/util/data_util.py b/util/data_util.py index 198da0e..94b2b36 100644 --- a/util/data_util.py +++ b/util/data_util.py @@ -102,7 +102,7 @@ def parse_department(string): result = [] if not string: return result - string = re.sub(r'\([^()]*\)|\[[^\[\]]*\]|\{[^\{\}]*\}|([^()]*)|[^⺀-鿿]', '', string)[:255] + string = re.sub(r'\([^()]*\)|\[[^\[\]]*]|\{[^{}]*}|([^()]*)|[^⺀-鿿]', '', string)[:255] if string == "科": return result result.append(string) diff --git a/util/util.py b/util/util.py index 4c3d645..41d7b60 100644 --- a/util/util.py +++ b/util/util.py @@ -131,7 +131,7 @@ def chinese_to_number(chinese): num = chinese_char_to_number(c) if num >= 0: if num == 0: - if number > 0 and unit[0] != None: + if number > 0 and unit[0] is not None: section += number * (unit[0] / 10) unit = [None, False] elif number > 0: @@ -139,21 +139,21 @@ def chinese_to_number(chinese): number = num else: unit = chinese_to_money_unit(c) - if unit[0] == None: + if unit[0] is None: raise ValueError(f"{chinese} has unknown unit '{c}' at: {i}") if unit[1]: section = (section + number) * unit[0] result += section section = 0 else: - unitNumber = number + unit_number = number if number == 0 and i == 0: - unitNumber = 1 + unit_number = 1 - section += unitNumber * unit[0] + section += unit_number * unit[0] number = 0 - if number > 0 and unit[0] != None: + if number > 0 and unit[0] is not None: number *= unit[0] / 10 return result + section + number diff --git a/visual_model_test/visual_model_test.py b/visual_model_test/visual_model_test.py index 5dad92f..c791218 100644 --- a/visual_model_test/visual_model_test.py +++ b/visual_model_test/visual_model_test.py @@ -50,7 +50,7 @@ def visual_model_test(model_type, test_img, task_path, schema): imgs = image_util.split(test_img) layout = [] temp_files_paths = [] - doc_parser = DocParser(layout_analysis=False) + # doc_parser = DocParser(layout_analysis=False) for img in imgs: with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as temp_file: # angle = image_util.parse_rotation_angles(img["img"])[0]