diw/diwScoring2.py

import tkinter as tk
from tkinter import filedialog, messagebox
from datetime import datetime
import difflib
import json
from pathlib import Path
import os
from lxml import etree as ET
import re
from difflib import SequenceMatcher
import pandas as pd
import base64
# from xpathSearch import XMLPathHandler

class XMLScorer:
    # 채점 기준 경로 초기화
    def __init__(self, scoring_criteria_path):
        # 채점 기준 로드
        self.scoring_criteria = self._load_scoring_criteria(scoring_criteria_path)
        self.total_score = 0
        self.partial_score = 0
        self.typo_score = 0
        
    def set_typo_score(self, score):
        self.typo_score = score
        
    def get_typo_score(self):
        return self.typo_score
    
    # 채점 기준파일 로드(JSON 파일)
    def _load_scoring_criteria(self, file_path):
        with open(file_path, 'r', encoding='utf-8') as f:
            return json.load(f)

    # XML 파일에서 element의 값을 찾아 반환  
    def query_xml(self, root, *args):
        first_xpath = args[0]
        second_xpath = args[1]
        points = args[2]
        category = args[3]
        
        if ("특수문자" in category) and (second_xpath is not None):
            try:
                result = root.xpath(first_xpath)
                # 결과값이 리스트형인데 내부에 정보가 없는경우
                # 결과값이 없음
                if type(result) is list and len(result) == 0:
                    return None
                elif result < points:
                    result = root.xpath(second_xpath)
                    return result
                else:
                    return result

            except ET.XPathEvalError as e:
                return None
            
        elif second_xpath is not None:
            try:
                result1 = root.xpath(first_xpath)
                result2 = root.xpath(second_xpath)
                if (type(result1) is list and len(result1) == 0) and (type(result2) is list and len(result2) == 0):
                    return None
                return result1 if result1 else result2
            
            except ET.XPathEvalError as e:
                return None
            
        else:
            try:
                result = root.xpath(first_xpath)
                if type(result) is list and len(result) == 0:
                    return None
                return result
            except ET.XPathEvalError as e:
                return None
            
    def chart_query_xml(self, tree, xpath, namespaces):
        result = tree.xpath(xpath, namespaces=namespaces)
        if type(result) is list and len(result) == 0:
            return None
        
        return result
    
    # 유사한 텍스트 찾기
    def find_similar_text(self, root, target_text, threshold=0.7):
        """ 
        전체 문서에서 유사한 텍스트를 찾아 반환
        
        Args:
            root (_type_): xml root element 객체
            target_text (_type_): 찾을 텍스트
            threshold (float, optional): 유사도 설정 Defaults to 0.3.

        Returns:
            str: 유사도 기준을 만족하는 텍스트 
        """
        # 전체 텍스트 추출
        # all_text = root.xpath(f"//CHAR/text()")
        # all_text.append(root.xpath(f"//TEXTART/@text"))

        namespaces = {
        'a': 'http://schemas.openxmlformats.org/drawingml/2006/main',
        'c': 'http://schemas.openxmlformats.org/drawingml/2006/chart'
        }
        
        all_text = root.xpath(f"//BODY//text() | //TEXTART/@Text | //c:chart//text()", namespaces=namespaces)
        
        # 유사도 비교
        max_score = 0
        similar_text = ''
        
        for text in all_text:
            score = SequenceMatcher(None, target_text, text).ratio()
            
            if score > max_score:
                max_score = score
                similar_text = text
                
        if max_score >= threshold:
            return similar_text
        else:
            return target_text
    
    # 정답 비교 및 점수 계산
    def evaluate_answer(self, scoring, user_answer, right_answer, points,
                        method="equal", tolerance=0):
        
        scoring['user_answer'] = user_answer

        is_correct = False

        if method == "equal":
            is_correct = (user_answer == right_answer)
        elif method == "tolerance":
            if isinstance(user_answer, dict) and isinstance(right_answer, dict):
                is_correct = all(abs(user_answer[k] - right_answer[k]) <= tolerance for k in right_answer)
            else:
                is_correct = abs(user_answer - right_answer) <= tolerance
        elif method == "in":
            is_correct = user_answer in right_answer
        else:
            raise ValueError(f"Unknown comparison method: {method}")

        if is_correct:
            scoring['points'] = points
            self.total_score += points
            self.partial_score += points
        else:
            scoring['points'] = 0
            
    # 하나의 XML 파일 채점
    def _score_xml_file(self, xml_file, chart_xml):
        try:
            tree = ET.parse(xml_file)
            root = tree.getroot()
            
            # 네임스페이스 정의
            namespaces = {
            'a': 'http://schemas.openxmlformats.org/drawingml/2006/main',
            'c': 'http://schemas.openxmlformats.org/drawingml/2006/chart'
            }
            
            # 차트 XML 파일이 없는 경우 0점 채점을 위헤 빈 XML 생성
            if chart_xml is None:
                chart_tree = ET.fromstring('<xml></xml>')
            else:
                chart_tree = ET.fromstring(chart_xml)
            
            # 결과값을 Dictionary로 저장
            # 하나의 xml파일 = 수험생 한명의 답안지
            onePersonResult = {
                'filename': os.path.basename(xml_file),
                'score_results': [],
                'total_score': 0,
                'partial_scores': []
            }
            print(f"File name: {onePersonResult['filename']}")
            
            self.total_score = 0 
            for section_id, section in self.scoring_criteria.items():
                self.partial_score = 0
                for criterion_id, criterion in section.items():
                    id = criterion_id
                    xpath = criterion.get('path', None)
                    xpath2 =  criterion.get('path2', None)
                    search_value = criterion.get('searchValue', None)
                    right_answer = criterion.get('value', None)
                    points = criterion.get('points', 0)
                    category = criterion.get('category', None)
                    item = criterion.get('item', None)
                    similar_text = None
                    
                    # search_value가 있는 경우        
                    if search_value is not None:
                        # search_value를 포함하는 텍스트 찾기
                        similar_text = self.find_similar_text(root, search_value)
                        xpath = xpath.replace('{searchValue}', similar_text)
                        
                    # 문항 별 채점 결과 저장
                    scoring = {
                        'section': section_id,
                        'id': id,
                        'category': category,           # 채점 분류   
                        'item': item,                   # 채점 항목
                        'right_answer': right_answer,   # 정답
                        'user_answer': None,            # 실제 작성 답안
                        'points': 0,                    # 점수
                        'deductions': []                # 각 기준별 감점 내역
                    }
                    
                    if "PageSetting" in (category or ""):
                        items = root.xpath(xpath)
                        error_range = criterion.get('tolerance', 0)
                        
                        for item in items:
                            user_answer = {
                                'Bottom' : int(item.get("Bottom", 0)),
                                'Footer' : int(item.get("Footer", 0)),
                                'Gutter' : int(item.get("Gutter", 0)),
                                'Header' : int(item.get("Header", 0)),
                                'Left' : int(item.get("Left", 0)),
                                'Right' : int(item.get("Right", 0)),
                                'Top' : int(item.get("Top", 0)),
                            }

                            self.evaluate_answer(scoring, user_answer, right_answer, points, method="tolerance", tolerance=error_range)

                            if scoring['points'] > 0:
                                break                            
                            
                    elif "BasicSetting" in (category or ""):
                        # 바탕글(기본설정) 요소
                        normal_style = root.xpath("//STYLE[@Name='바탕글']")
                        
                        # CHARSHAPE, PARASHAPE, FONTID 속성값 추출
                        charshape_id = normal_style[0].get("CharShape")
                        parashape_id = normal_style[0].get("ParaShape")
                        font_id = root.xpath(f"//CHARSHAPE[@Id='{charshape_id}']/FONTID/@Hangul")
                        
                        # 필요한 속성값을 이용하여 정답과 비교
                        font_name = root.xpath(f"//FONT[@Id='{font_id[0]}']/@Name")
                        font_size = root.xpath(f"//CHARSHAPE[@Id='{charshape_id}']/@Height")
                        alignment = root.xpath(f"//PARASHAPE[@Id='{parashape_id}']/@Align")
                        line_spacing = root.xpath(f"//PARASHAPE[@Id='{parashape_id}']/PARAMARGIN/@LineSpacing")
                        
                        # 수험자 답안
                        user_answer = {
                            'FontName': font_name[0],
                            'FontSize': font_size[0],
                            'Alignment': alignment[0],
                            'LineSpacing': line_spacing[0]
                        }
                        
                        self.evaluate_answer(scoring, user_answer, right_answer, points, method="equal")
                    
                    # 오타 감점 부분은 미리 계산 하고, 이후 점수만 계산
                    elif "오타감점" in (category or ""):
                        points = self.get_typo_score()
                        self.total_score += points
                        self.partial_score += points
                        scoring['points'] = points

                    # 정답이 하나인 경우
                    elif "SingleAnswer" in (category or ""):
                        items = root.xpath(xpath)
                        
                        for item in items:
                            user_answer = item

                            self.evaluate_answer(scoring, user_answer, right_answer, points)
                            
                            if scoring['points'] > 0:
                                break
                    
                    # Boolean 타입 정답인 경우
                    elif "Boolean" in (category or ""):
                        items = root.xpath(xpath)
                        user_answer = True if items else False
                        
                        self.evaluate_answer(scoring, user_answer, right_answer, points)
                                
                    # 채점기준표 파일에 작성된 rgb값을 그대로 읽어와 HML파일 요소의 int형 rgb값과 비교
                    elif "Color" in (category or ""):
                        items = root.xpath(xpath)
                        
                        rgb_text = right_answer
                        r, g, b = map(int, rgb_text.split(','))
                        rgb_int = (b << 16) + (g << 8) + r 
                        
                        for item in items:
                            user_answer = int(item)
                            
                            self.evaluate_answer(scoring, user_answer, rgb_int, points, method="equal")
                            
                            if scoring['points'] > 0:
                                break

                    elif "TwoLineSize" in (category or ""):
                        items = root.xpath(xpath)
                        error_range = criterion.get('tolerance', 0)
                        for item in items:
                            user_answer = {
                                "Height": int(item.get('Height', 0)),
                                "Width": int(item.get('Width', 0))
                            }
                            self.evaluate_answer(scoring, user_answer, right_answer, points, method="tolerance", tolerance=error_range)
                        
                            if scoring['points'] > 0:
                                break
                    
                    elif "FontName" in (category or ""):
                        charshape_id = root.xpath(xpath)
                        if not charshape_id:
                            charshape_id = None
                            user_answer = None
                        else:
                            font_id = root.xpath(f"//CHARSHAPE[@Id='{charshape_id[0]}']/FONTID/@Hangul")
                            font_name = root.xpath(f"//FONTFACE[@Lang='Hangul']/FONT[@Id='{font_id[0]}']/@Name")
                            user_answer = font_name[0]

                        self.evaluate_answer(scoring, user_answer, right_answer, points, method="equal")

                    elif "FontAttribute" in (category or ""):
                        charshape = root.xpath(xpath)
                        if not charshape:
                            charshape = None
                            user_answer = None
                        else:
                            font_attribute = charshape[0].find(right_answer)
                            if font_attribute is not None:
                                user_answer = font_attribute.tag
                        
                        self.evaluate_answer(scoring, user_answer, right_answer, points, method="equal")        
                            
                    onePersonResult['score_results'].append(scoring)
                    print(f'scoring: {scoring}')
                onePersonResult['partial_scores'].append({
                    'section': section_id,
                    'score': self.partial_score
                })
            onePersonResult['total_score'] = self.total_score
            return onePersonResult
                
        except ET.ParseError as e:
            return {
                'filename': os.path.basename(xml_file),
                'error': f"XML 파싱 오류: {str(e)}",
                'total_score': 0
            }
            
    def binary_to_chartxml(self, xml_path):
        tree = ET.parse(xml_path)
        root = tree.getroot()
        
        binary_data = root.xpath('//BINDATA[@Id=//BINITEM[@Format="OLE"]/@BinData]/text()')
        if not binary_data:
            return None
        binary_data = binary_data[0].encode('utf-8')

        # <BINDATA ...> 태그와 그 내부 내용을 삭제합니다.
        encoded_data = re.sub(b'<BINDATA.*?>', b'', binary_data)
        encoded_data = encoded_data.replace(b'</BINDATA>', b'')
        encoded_data = encoded_data.replace(b'\r\n', b'')

        # base64 디코딩을 수행합니다.
        decoded_data = base64.b64decode(encoded_data+b'==')
            
        # 디코딩된 데이터 내용 중 xml 형식만 추출할 때 <c:chartSpace>, </c:chartSpace> 사이의 데이터만 추출.
        start = decoded_data.find(b'<?xml')
        print(start)
        end = decoded_data.find(b'</c:chartSpace>')
        print(end)
        xml_data = decoded_data[start:end+len(b'</c:chartSpace>')]

        # xml 데이터가 없는 경우 None을 반환합니다.
        if -1 in [start, end]:
            return None
        
        # 디코딩된 데이터를 파일로 저장합니다.
        base_filename = os.path.splitext(xml_path)[0]
        new_filename = f'{base_filename}.xml'
        with open(new_filename, 'wb') as file:
                file.write(xml_data)
        
        return xml_data
    
    def typo_check(self, correct_answer_file, user_answer_file):
        user_answer_tree = ET.parse(user_answer_file)
        user_answer_root = user_answer_tree.getroot()
        correct_answer_tree = ET.parse(correct_answer_file)
        correct_answer_root = correct_answer_tree.getroot()

        # xpath로 바이너리 부분추출
        user_input_text = user_answer_root.xpath('//CHAR//text()[not(ancestor::HEADER) and not(ancestor::TABLE)]')
        user_table_text = user_answer_root.xpath('//TABLE//CHAR//text()')
        user_input_text += user_table_text
        
        correct_input_text = correct_answer_root.xpath('//CHAR//text()[not(ancestor::HEADER) and not(ancestor::TABLE)]')
        correct_table_text = correct_answer_root.xpath('//TABLE//CHAR//text()')
        correct_input_text += correct_table_text

        # 각 요소에서 공백 제거
        user_input_text = [text.replace(' ', '') for text in user_input_text]
        correct_input_text = [text.replace(' ', '') for text in correct_input_text]

        
        # 숫자와 특정 형식 제거 (예: 1., 2., 3., -)
        user_input_text = [re.sub(r'\d+\.\s*|-', '', text) for text in user_input_text]
        correct_input_text = [re.sub(r'\d+\.\s*|-', '', text) for text in correct_input_text]
    
        try :
            xpath = self.scoring_criteria["2-29"]['path'].split("'")[1]
            ignore_word = xpath.split("'")[1]
            # 특정 단어 제거
            # 오타와 누락의 경우만 판단하면 정상작동하지만
            # 추가 된 단어의 경우를 채점기준에 추가하면 정확하게 채점 되지 않을 수 있음
            # [정답] Hybrid [실제작성] 
            user_input_text = [text.replace(ignore_word, '') for text in user_input_text]
            correct_input_text = [text.replace(ignore_word, '') for text in correct_input_text]
        except (KeyError, IndexError, AttributeError):
            ignore_word = None
        
        print(f"ignore_word: {ignore_word}")
                
        # 리스트를 하나의 문자열로 변경
        user_input_text_str = ''.join(user_input_text)
        currect_input_text_str = ''.join(correct_input_text)

        print("user_input_text as string:")
        print(user_input_text_str)
        print("\ncurrect_input_text_answer as string:")
        print(currect_input_text_str)


        # 문자열의 차이를 비교
        diff = difflib.ndiff(currect_input_text_str, user_input_text_str)
        diff_list = list(diff)

        # 차이점을 정리하여 result_diff에 저장
        result_diff = []
        
        # 누락 된 단어만 따로 리스트로 저장
        missing_list = []
        
        # 오타와 누락된 단어 리스트 저장
        error_missing_list = []
        
        skip_next = False

        for i, line in enumerate(diff_list):
            if skip_next:
                skip_next = False
                continue
            # diff_list의 line 시작이 '-'이면서 다음 line이 '+'이면 두 line을 붙여서 맞춤법이 틀린 단어로 판단
            if line.startswith('- '):
                # 오타
                if i + 1 < len(diff_list) and diff_list[i + 1].startswith('+ '):
                    line = line.replace('- ', '-')
                    next = diff_list[i + 1].replace('+ ', '')
                    result_diff.append(line+'=>'+next)
                    error_missing_list.append(line+'=>'+next)
                    skip_next = True
                # 누락
                else:
                    line = line.replace('- ', '-')
                    result_diff.append(line)
                    missing_list.append(line)
                    error_missing_list.append(line)
            # 없어도 되는 글자가 있는 경우 (추가)
            elif line.startswith('+ '):
                line = line.replace('+ ', '+')
                result_diff.append(line)

        # result_diff 출력
        # print("\nResult Differences:")
        # for diff in result_diff:
        #     print(diff)
        
        # result_diff 배열의 길이를 맨 앞에 저장
        
        # 모든 차이를 계산해 점수 차감
        # temp = 40 - min(len(result_diff)*2, 40)
        
        # 누락된 텍스트만 계산해 점수 차감
        # temp = 40 - min(len(missing_list)*2, 40)
        
        # 2503회 기준 오타 1개당 [2점]->[1점] 차감
        temp = 40 - min(len(error_missing_list)*1, 40)
        
        self.set_typo_score(temp)

        result_diff.insert(0, temp)
        return result_diff
    
    # XML 파일 채점
    def score_directory(self, xml_directory, correct_answer_file):
        # xml 파일 불러오기
        xml_files = Path(xml_directory).glob('*.hml')
    
        # 채점결과 저장할 리스트
        score_results = []
        
        for user_answer_file in xml_files:
            score_result = {}
            chart_xml = self.binary_to_chartxml(user_answer_file)
            score_result['typo'] = self.typo_check(correct_answer_file, user_answer_file)
            score_result['score'] = self._score_xml_file(user_answer_file, chart_xml)
            # score_result['score']['score_results'][2]['points'] = score_result['typo'][0]
            score_results.append(score_result)    
        return score_results

    def parse_filename(self, filename):
        if isinstance(filename, dict):
            filename = filename.get('파일명', '')
        match = re.match(r'.*-(\d+)-(.+)\.hml', filename)
        if match:
            number = match.group(1)
            name = match.group(2)
            return number, name

        return None, None
    
    def export_to_excel(self, results, output_path=None):
        if output_path is None:
            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S") #연월일_시분초
            # timestamp = datetime.now().strftime("%Y%m%d") #연월일
            output_path = f"scoring_results_{timestamp}.xlsx"

        summary_data = []
        detail_data = []
        typo_data = []
        
        for temp in results:
            # 요약 정보
            result = temp['score']
            summary_row = {
                '파일명': result['filename'],
                '총점': result.get('total_score', 0)
            }
            if 'error' in result:
                summary_row['오류'] = result['error']
                
            summary_data.append(summary_row)

            # 상세 정보
            if 'score_results' in result:
                filename = {'파일명': result['filename']}
                number, name = self.parse_filename(filename)
                if (number or name) is None:
                    detail_row = {'채점항목': result['filename'] }
                else:
                    detail_row = {'채점항목':f"{number}-{name}"}
                
                section_num = None
                row_index = []
                for i, score_result in enumerate(result['score_results']):
                    current_section = score_result['section']
                    
                    if section_num is None:
                        section_num = current_section
                    
                    # 다음 섹션(문제0 => 문제1)로 넘어갔을 경우 or 마지막 문제일 경우
                    if current_section != section_num:
                        # 이전 섹션의 부분합을 출력
                        detail_row[f'문제{section_num}'] = result['partial_scores'][int(section_num)]['score']
                        row_index.append(f'문제{section_num}')
                        section_num = current_section
                    
                    detail_row[f'{i+1}'] = score_result['points']
                    row_index.append(score_result['id'])
                    
                    # 마지막 섹션(문제2)부분합 점수를 출력
                    if i == len(result['score_results']) - 1:
                        detail_row[f'문제{current_section}'] = result['partial_scores'][int(current_section)]['score']
                        row_index.append(f'문제{current_section}')
                    
                detail_row['총점'] = result.get('total_score', 0)
                row_index.append('총점')
                detail_data.append(detail_row)

        summary_df = pd.DataFrame(summary_data)
        detail_df = pd.DataFrame(detail_data).transpose()
        detail_df.columns = detail_df.iloc[0]
        detail_df = detail_df[1:]
        
        detail_df.index = row_index
        # detail_df = pd.DataFrame(detail_data)
        
        for temp in results:
            result = temp['typo']
            typo_data.append(result)
            
        typo_df = pd.DataFrame(typo_data).transpose()
        # detail_df = pd.DataFrame(detail_data)
        
        # ExcelWriter 객체 생성
        with pd.ExcelWriter(output_path, engine='openpyxl') as writer:
            summary_df.to_excel(writer, sheet_name='채점결과요약', index=False)
            detail_df.to_excel(writer, sheet_name='채점상세내역', index=True)
            typo_df.to_excel(writer, sheet_name='오타내역', index=False)

            # 열 너비 자동 조정
            # for sheet_name in writer.sheets:
            #     worksheet = writer.sheets[sheet_name]
            #     for column_cells in worksheet.columns:
            #         max_length = 0
            #         column = column_cells[0].column_letter  # 열의 문자
            #         for cell in column_cells:
            #             try:
            #                 if cell.value:
            #                     max_length = max(max_length, len(str(cell.value)))
            #             except:
            #                 pass
            #         adjusted_width = (max_length + 2)
            #         worksheet.column_dimensions[column].width = adjusted_width

        return output_path


def main():

    # 시험회차 및 유형
    exam_round = '2504'
    exam_types = [
        'A',
        # 'B',
        # 'C',
    ]
    # test_mode = False
    test_mode = True
    
    output_excel_paths = []
    for exam_type in exam_types:
        scoring_criteria_path = f'./DIW_{exam_round}{exam_type}_test.json'
        # scoring_criteria_path = f'./DIW_{exam_round}{exam_type}.json'
        # xml(hml)파일 디렉토리 경로
        
        xml_directory = f'./output/{exam_type}/{"TEST" if test_mode else "DIW"}'
        

        # 오탈자 체크를 위한 정답 파일 경로 (형식:DIW_2503A.hml)
        correct_answer_file = f'./output/{exam_type}/DIW/DIW_{exam_round}{exam_type}.hml'
        
        # 엑셀 파일명 (비어있으면 자동생성)
        timestamp = datetime.now().strftime("%y%m%d")
        # output_path = f'{timestamp}_DIW_{exam_round}{exam_type}_{"TEST" if test_mode else "채점결과"}.xlsx'
        output_path = f'{timestamp}_DIW_{exam_round}{exam_type}_{"TEST" if test_mode else "채점결과"}_2.xlsx'

        # 채점 클래스 초기화
        scorer = XMLScorer(scoring_criteria_path)

        # 폴더 내 모든 xml 파일 채점
        results = scorer.score_directory(xml_directory, correct_answer_file)
    
        # 채점 결과 엑셀로 저장
        output_excel_paths.append(scorer.export_to_excel(results, output_path))
    
    print(f"채점 결과 엑셀 파일: {output_excel_paths}")    

if __name__ == '__main__': 
    main()