import gradio as gr
import pandas as pd
import re

# 提取文件名中的所有数字作为时间戳
def extract_timestamps(file_name):
    # 使用正则表达式提取所有的数字
    timestamps = re.findall(r'(\d+)', file_name)
    # 将这些数字转换为整数，返回列表
    return [int(ts) for ts in timestamps]

# 根据提取的时间戳列表排序
def sort_by_timestamps(file_name):
    # 提取所有时间戳，并按自然顺序进行排序
    timestamps = extract_timestamps(file_name)
    # 返回时间戳中的最小值作为排序的主键
    return timestamps

def csv_to_dict(df):
    df = df.fillna('')
    res = {}
    for i, row in df.iterrows():
        if row['CRN'] not in res:
            res[ row['CRN' ] ] = {
                'Instructor': row['Instructor'],
                'Now': row['Now'],
            }
    return res

def compare_dicts(dict_a, dict_b):
    # 遍历两个字典，查找 Instructor 或 Now 不同的CRN
    changed = []
    all_crns = set(dict_a.keys()).union(set(dict_b.keys()))
    
    for crn in all_crns:
        instructor_a = dict_a.get(crn, {}).get('Instructor', None)
        instructor_b = dict_b.get(crn, {}).get('Instructor', None)
        max_a = dict_a.get(crn, {}).get('Now', None)
        max_b = dict_b.get(crn, {}).get('Now', None)
        
        # 如果Instructor或Max不同，则记录CRN
        if instructor_a != instructor_b:
            msg = ' * CRN: {}, Instructor change: {} --> {}'.format(crn, instructor_a, instructor_b)
            changed.append(msg)
        if max_a != max_b:
            msg = ' * CRN: {}, Now change: {} --> {}'.format(crn, max_a, max_b)
            changed.append(msg)
    
    return changed

def process_files(files):
    if len(files) < 2:
        return "Please upload at least 2 files."
    
    # 根据文件名中的时间戳进行排序
    files_sorted = sorted(files, key=sort_by_timestamps)
    
    result = []
    
    # 对排序后的文件进行两两比较
    for i in range(len(files_sorted) - 1):
        df_older = pd.read_csv(files_sorted[i].name)
        df_newer = pd.read_csv(files_sorted[i + 1].name)
        
        dict_a = csv_to_dict(df_older)
        dict_b = csv_to_dict(df_newer)

        f1 = files_sorted[i].name.split('/')[-1]
        f2 = files_sorted[i+1].name.split('/')[-1]
        
        crns = compare_dicts(dict_a, dict_b)
        if crns:
            result.append("{} --> {} \n{}".format(f1, f2, "\n".join(crns)))
        else:
            result.append("{} --> {} \n * No change.".format(f1, f2))

    return "\n".join(result)

# 使用Gradio构建界面
gr.Interface(
    fn=process_files,
    inputs=gr.Files(label="Upload CSV files"),
    outputs="text",
    title="Course Schedule Tracker",
    description="Upload multiple CSV files, sort them based on the timestamp in the filenames, compare two adjacent files at a time, and output the CRNs where the 'Instructor' or 'Now' has changed."
).launch()