debug_read_xlsx.py

"""
快速读取 xlsx 文件，输出关键列样本
用于调试公司名匹配问题
"""
import openpyxl
import sys

FILE = r"D:/Coding/Wellfound (Job applications).xlsx"

wb = openpyxl.load_workbook(FILE, read_only=True)
print("工作表列表:", wb.sheetnames)
print()

ws = wb["West"]
print(f"West 表: {ws.max_row} 行 x {ws.max_column} 列")
print()

# 表头
headers = []
for cell in ws[1]:
    headers.append(cell.value)
print("表头（A-AD）:")
for i, h in enumerate(headers[:30], 1):
    col_letter = ""
    n = i
    while n > 0:
        n, r = divmod(n - 1, 26)
        col_letter = chr(r + ord('A')) + col_letter
    print(f"  {col_letter}: {h}")
print()

# Company Name 列样本
print("C列（Company Name）前20个非空值:")
count = 0
for row in ws.iter_rows(min_row=2, min_col=3, max_col=3):
    val = row[0].value
    if val:
        print(f"  {val}")
        count += 1
        if count >= 20:
            break
print()

# AA/AB/AC 列样本
print("已有 response 数据样本（AA/AB/AC 列）:")
for row in ws.iter_rows(min_row=2, max_row=10, min_col=27, max_col=29):
    aa, ab, ac = row[0].value, row[1].value, row[2].value
    if aa or ab or ac:
        print(f"  AA={aa} | AC={ac} | AB={str(ab)[:60] if ab else ''}")
print()

wb.close()
print("读取完成")