看来世界是由Excel统治的。我在数据工程工作中感到惊讶的是，看到有多少同事将Excel用作决策的关键工具。尽管我不是MS Office及其Excel电子表格的忠实拥护者，但我仍会向您展示一个巧妙的技巧，可以有效地处理大型Excel电子表格。
在我们直接进入程序之前，让我们了解一些与Pandas处理excel电子表格的基础知识。
1.安装。继续安装openpyxl和xlwt。如果不确定是否已安装它，可以通过使用python终端中的pip Frozen或pip list来确定是否安装了可用的软件包。
我们将首先通过传递元组数据来创建一个Excel电子表格，然后将数据加载到pandas数据框中。我们最终将数据框数据写入新的工作簿。
import xlsxwriter import pandas as pd
2.创建包含少量数据的Excel电子表格。我们将有一个小功能将字典数据写入excel电子表格。所有代码逻辑都在每个步骤中定义。
# Function : write_data_to_files
def write_data_to_files(inp_data, inp_file_name):
"""
function : create a csv file with the data passed to this code
args : inp_data : tuple data to be written to the target file
file_name : target file name to store the data
return : none
assumption : File to be created and this code are in same directory.
"""
print(f" *** Writing the data to - {inp_file_name}")
# Create a Workbook.
workbook = xlsxwriter.Workbook(inp_file_name)
# add a worksheet.
worksheet = workbook.add_worksheet()
# Start from the first cell. Rows and columns are zero indexed.
row = 0
col = 0
# read the input data and write them in rows and columns
for player, titles in inp_data:
worksheet.write(row, col, player)
worksheet.write(row, col + 1, titles)
row += 1
# close the workbook.
workbook.close()
print(f" *** Completed writing the data to - {inp_file_name}")# Function : excel_functions_with_pandas
def excel_functions_with_pandas(inp_file_name):
"""
function : Quick overview of functions you can apply on excel with pandas
args : inp_file_name : input excel spread sheet.
return : none
assumption : Input excel spreadsheet and this code are in same directory.
"""
data = pd.read_excel(inp_file_name)
# print top 2 rows
print(f" *** Displaying top 2 rows of - {inp_file_name} \n {data.head()} ")
# look at the data types
print(f" *** Displaying info about {inp_file_name} - {data.info()}")
# Create a new spreadsheet "Sheet2" and write data into it.
new_players_info = pd.DataFrame(data=[
{"players": "new Roger Federer", "titles": 20},
{"players": "new Rafael Nadal", "titles": 20},
{"players": "new Novak Djokovic", "titles": 17},
{"players": "new Andy Murray", "titles": 3}], columns=["players", "titles"])
new_data = pd.ExcelWriter(inp_file_name)
new_players_info.to_excel(new_data, sheet_name="Sheet2")
if __name__ == '__main__':
# Define your file name and data
file_name = "temporary_file.xlsx"
# tuple data for storage
file_data = (['player', 'titles'], ['Federer', 20], ['Nadal', 20], ['Djokovic', 17], ['Murray', 3])
# write the file_data to file_name
# write_data_to_files(file_data, file_name)
# # Read excel file into pandas and apply functions.
# excel_functions_with_pandas(file_name)if __name__ == '__main__': # Define your file name and data file_name = "temporary_file.xlsx" # tuple data for storage file_data = (['player', 'titles'], ['Federer', 20], ['Nadal', 20], ['Djokovic', 17], ['Murray', 3]) # write the file_data to file_name # write_data_to_files(file_data, file_name) # # Read excel file into pandas and apply functions. # excel_functions_with_pandas(file_name)
输出结果
*** Writing the data to - temporary_file.xlsx *** Completed writing the data to - temporary_file.xlsx *** Displaying top 2 rows of - temporary_file.xlsx player titles 0 Federer 20 1 Nadal 20 2 Djokovic 17 3 Murray 3 <class 'pandas.core.frame.DataFrame'> RangeIndex: 4 entries, 0 to 3 Data columns (total 2 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 player 4 non-null object 1 titles 4 non-null int64 dtypes: int64(1), object(1) memory usage: 192.0+ bytes *** Displaying info about temporary_file.xlsx - None
现在，当处理大型csv文件时，我们有很多选项，包括以块形式处理它们的块，但是对于excel电子表格，Pandas默认情况下不提供块选项。
因此，如果要分块处理excel电子表格，下面的程序会很有帮助。
def global_excel_to_db_chunks(file_name, nrows):
"""
function : handle excel spreadsheets in chunks
args : inp_file_name : input excel spread sheet.
return : none
assumption : Input excel spreadsheet and this code are in same directory.
"""
chunks = []
i_chunk = 0
# The first row is the header. We have already read it, so we skip it.
skiprows = 1
df_header = pd.read_excel(file_name, nrows=1)
while True:
df_chunk = pd.read_excel(
file_name, nrows=nrows, skiprows=skiprows, header=None)
skiprows += nrows
# When there is no data, we know we can break out of the loop.
if not df_chunk.shape[0]:
break
else:
print(
f" ** Reading chunk number {i_chunk} with {df_chunk.shape[0]} Rows")
# print(f" *** Reading chunk {i_chunk} ({df_chunk.shape[0]} rows)")
chunks.append(df_chunk)
i_chunk += 1
df_chunks = pd.concat(chunks)
# Rename the columns to concatenate the chunks with the header.
columns = {i: col for i, col in enumerate(df_header.columns.tolist())}
df_chunks.rename(columns=columns, inplace=True)
df = pd.concat([df_header, df_chunks])
print(f' *** Reading is Completed in chunks...')
if __name__ == '__main__':
print(f" *** Gathering & Displaying Stats on the excel spreadsheet ***")
file_name = 'Sample-sales-data-excel.xls'
stats = pd.read_excel(file_name)
print(f" ** Total rows in the spreadsheet are - {len(stats.index)} Rows")
# process the excel file in chunks of 1000 rows at a time.
global_excel_to_db_chunks(file_name, 1000)*** Gathering & Displaying Stats on the excel spreadsheet *** ** Total rows in the spreadsheet are - 9994 Rows ** Reading chunk number 0 with 1000 Rows ** Reading chunk number 1 with 1000 Rows ** Reading chunk number 2 with 1000 Rows ** Reading chunk number 3 with 1000 Rows ** Reading chunk number 4 with 1000 Rows ** Reading chunk number 5 with 1000 Rows ** Reading chunk number 6 with 1000 Rows ** Reading chunk number 7 with 1000 Rows ** Reading chunk number 8 with 1000 Rows ** Reading chunk number 9 with 994 Rows *** Reading is Completed in chunks...
输出结果
*** Gathering & Displaying Stats on the excel spreadsheet *** ** Total rows in the spreadsheet are - 9994 Rows ** Reading chunk number 0 with 1000 Rows ** Reading chunk number 1 with 1000 Rows ** Reading chunk number 2 with 1000 Rows ** Reading chunk number 3 with 1000 Rows ** Reading chunk number 4 with 1000 Rows ** Reading chunk number 5 with 1000 Rows ** Reading chunk number 6 with 1000 Rows ** Reading chunk number 7 with 1000 Rows ** Reading chunk number 8 with 1000 Rows ** Reading chunk number 9 with 994 Rows *** Reading is Completed in chunks...