我有多個 CSV 檔案，這些檔案的格式是用換行符分隔的多個表格。
例子：
Technology C_inv [MCHF/y] C_maint [MCHF/y]
NUCLEAR 70.308020 33.374568
HYDRO_DAM_EXISTING 0.000000 195.051200
HYDRO_DAM 67.717942 1.271600
HYDRO_RIVER_EXISTING 0.000000 204.820000
IND_BOILER_OIL 2.053610 0.532362
IND_BOILER_COAL 4.179935 1.081855
IND_BOILER_WASTE 11.010126 2.849652
DEC_HP_ELEC 554.174644 320.791276
DEC_THERMAL_HP_GAS 77.077291 33.717477
DEC_BOILER_GAS 105.586089 41.161335
DEC_BOILER_OIL 33.514266 25.948450
H2_FROM_GAS 145.185290 59.178082
PYROLYSIS 132.200818 112.392123
Storage technology C_inv [MCHF/y] C_maint [MCHF/y]
HYDRO_STORAGE 0.000000 0.000000
Resource C_op [MCHF/y]
ELECTRICITY 1174.452848
GASOLINE 702.000000
DIESEL 96.390000
OIL 267.787558
NG 1648.527242
WOOD 592.110000
COAL 84.504083
URANIUM 18.277626
WASTE 0.000000
我所有的 CSV 檔案都有不同的子表名稱，但足夠少，我可以手動輸入它們以在需要時檢測它們。
另一個問題是許多標題包含空格（例如“存盤技術”），pandas 將其讀取為 2 列。
我最初嘗試直接使用 pandas 并手動拆分，但on_bad_lines='skip'允許避免錯誤的引數也跳過了有用的行：
Cost_bd = pd.read_csv(f"{Directory}/cost_breakdown.csv",on_bad_lines='skip',delim_whitespace=True).dropna(axis=1,how='all')
colnames=['Technnolgy', 'C_inv[MCHF/y]', 'C_maint[MCHF/y]']
Cost_bd.columns = colnames
我相信掃描.txt檔案并將其拆分可能會更好，但我不確定如何以最佳方式執行此操作。
我也嘗試使用此提要中提供的解決方案
import csv
from os.path import dirname # gets parent folder in a path
from os.path import join # concatenate paths
table_names = ["Technology", "Storage technology", "Resource"]
df = pd.read_csv(f"{Directory}/cost_breakdown.csv", header=None, names=range(3))
groups = df[0].isin(table_names).cumsum()
tables = {g.iloc[0,0]: g.iloc[1:] for k,g in df.groupby(groups)}
但它不起作用：
tables.keys()=
dict_keys(['Technology\tC_inv [MCHF/y]\tC_maint [MCHF/y]'])
編輯：基于@Rabinzel 的最終解決方案：
import re
def make_df(group,dict_of_dfs):
header, data = re.split(r'\t',group[0]), list(map(str.split, group[1:]))
if len(header) != len(data[0]): # If missing columns list, take former
header = header dict_of_dfs[list(dict_of_dfs.keys())[0]].columns.tolist()[1:]
dict_of_dfs[header[0]] = pd.DataFrame(data, columns=header)
return dict_of_dfs
def Read_csv_as_df(path, file_name):
with open(path file_name) as f:
dict_of_dfs = {}
group = []
for line in f:
if line!='\n':
group.append(line.strip())
else:
print(dict_of_dfs)
dict_of_dfs = make_df(group,dict_of_dfs)
group = []
dict_of_dfs = make_df(group,dict_of_dfs)
return dict_of_dfs
uj5u.com熱心網友回復：
我會按照以下方式進行。遍歷每一行，將由換行符分隔的每個塊附加到串列中，并從串列中構建資料幀。帶有空格的列名的問題，re.split僅當有兩個或多個空格時才使用和拆分。將不同的df保存在字典中，其中鍵是每個df的標題的第一個元素。
import re
def make_df(group):
header, data = re.split(r'\s\s ',group[0]), list(map(str.split, group[1:]))
dict_of_dfs[header[0]] = pd.DataFrame(data, columns=header)
with open('your_csv_file.csv') as f:
dict_of_dfs = {}
group = []
for line in f:
if line!='\n':
group.append(line.strip())
else:
make_df(group)
group = []
make_df(group)
for key, value in dict_of_dfs.items():
print(f"{key=}\ndf:\n{value}\n---------------------")
輸出：
key='Technology'
df:
Technology C_inv [MCHF/y] C_maint [MCHF/y]
0 NUCLEAR 70.308020 33.374568
1 HYDRO_DAM_EXISTING 0.000000 195.051200
2 HYDRO_DAM 67.717942 1.271600
3 HYDRO_RIVER_EXISTING 0.000000 204.820000
4 IND_BOILER_OIL 2.053610 0.532362
5 IND_BOILER_COAL 4.179935 1.081855
6 IND_BOILER_WASTE 11.010126 2.849652
7 DEC_HP_ELEC 554.174644 320.791276
8 DEC_THERMAL_HP_GAS 77.077291 33.717477
9 DEC_BOILER_GAS 105.586089 41.161335
10 DEC_BOILER_OIL 33.514266 25.948450
11 H2_FROM_GAS 145.185290 59.178082
12 PYROLYSIS 132.200818 112.392123
---------------------
key='Storage technology'
df:
Storage technology C_inv [MCHF/y] C_maint [MCHF/y]
0 HYDRO_STORAGE 0.000000 0.000000
---------------------
key='Resource'
df:
Resource C_op [MCHF/y]
0 ELECTRICITY 1174.452848
1 GASOLINE 702.000000
2 DIESEL 96.390000
3 OIL 267.787558
4 NG 1648.527242
5 WOOD 592.110000
6 COAL 84.504083
7 URANIUM 18.277626
8 WASTE 0.000000
---------------------
轉載請註明出處，本文鏈接：https://www.uj5u.com/yidong/523243.html
上一篇：預期的str、bytes或os.PathLike物件，而不是reader
下一篇：Supabase：無法讀取注冊引數：json：無法將物件解組為字串型別的Go結構欄位SignupParams.email