Python3使用calamine库处理超大Excel文件

pip3 install calamine

from calamine import Workbook, CellType
# 打开工作簿
workbook = Workbook.open("example.xlsx")
# 获取所有的工作表名称
sheet_names = workbook.sheet_names()
print(f"Sheet names: {sheet_names}")
# 选择第一个工作表
sheet = workbook.get_sheet_by_index(0)
# 遍历所有行
for row in sheet.rows():
    # 每一行是一个列表，包含每个单元格的值
    print([cell.value for cell in row if cell.type != CellType.EMPTY])
# 关闭工作簿
workbook.close()

from calamine import Workbook, CellType
# 打开 ODS 工作簿
workbook = Workbook.open("example.ods")
# 选择第一个工作表
sheet = workbook.get_sheet_by_index(0)
# 遍历所有行
for row in sheet.rows():
    print([cell.value for cell in row if cell.type != CellType.EMPTY])
# 关闭工作簿
workbook.close()

from calamine import Workbook, CellType
# 打开 CSV 文件
with open("example.csv", "r") as f:
    workbook = Workbook.from_csv(f)
    # 选择唯一的工作表
    sheet = workbook.get_sheet_by_index(0)
    # 遍历所有行
    for row in sheet.rows():
        print([cell.value for cell in row if cell.type != CellType.EMPTY])

from calamine import CellType
# 假设我们已经打开了一个工作簿并选择了某个工作表
for row in sheet.rows():
    for cell in row:
        if cell.type == CellType.STRING:
            print(f"String value: {cell.value}")
        elif cell.type == CellType.NUMBER:
            print(f"Number value: {cell.value}")
        elif cell.type == CellType.BOOL:
            print(f"Boolean value: {cell.value}")
        elif cell.type == CellType.ERROR:
            print(f"Error value: {cell.error}")
        elif cell.type == CellType.FORMULA:
            print(f"Formula: {cell.formula}, Result: {cell.value}")
        elif cell.type == CellType.EMPTY:
            print("Empty cell")

# 获取 A1 单元格的值
value = sheet.get_value("A1")
print(f"Value at A1: {value}")

from calamine import Workbook, CellType
def read_excel(file_path):
    workbook = Workbook.open(file_path)
    sheet = workbook.get_sheet_by_index(0)
    data = []
    for row in sheet.rows():
        row_data = [cell.value for cell in row if cell.type != CellType.EMPTY]
        if row_data:
            data.append(row_data)
    workbook.close()
    return data
if __name__ == "__main__":
    file_path = "example.xlsx"
    data = read_excel(file_path)
    # 打印前几行数据作为示例
    for row in data[:5]:
        print(row)

pip install python-calamine

from python_calamine import CalamineWorkbook
# 加载 Excel 文件
workbook = CalamineWorkbook.from_path("your_large_file.xlsx")
# 获取所有工作表名称
sheet_names = workbook.sheet_names
print(sheet_names)
# 按名称获取工作表，并将其转换为 Python 列表
sheet_data = workbook.get_sheet_by_name("Sheet1").to_python()
# 遍历数据
for row in sheet_data:
    print(row)

import pandas as pd
# 方法一：读取文件路径
df = pd.read_excel("your_large_file.xlsx", engine="calamine")
# 方法二：读取文件对象（推荐，可避免路径问题）
with open("your_large_file.xlsx", "rb") as f:
    df = pd.read_excel(f, sheet_name="Sheet1", engine="calamine")

# 只读取 A 到 N 列
df = pd.read_excel("large_file.xlsx", usecols="A:N", engine="calamine")