使用 HDFStore

import string
import numpy as np
import pandas as pd

生成具有各種 dtypes 的樣本 DF

df = pd.DataFrame({
     'int32':    np.random.randint(0, 10**6, 10),
     'int64':    np.random.randint(10**7, 10**9, 10).astype(np.int64)*10,
     'float':    np.random.rand(10),
     'string':   np.random.choice([c*10 for c in string.ascii_uppercase], 10),
     })

In [71]: df
Out[71]:
      float   int32       int64      string
0  0.649978  848354  5269162190  DDDDDDDDDD
1  0.346963  490266  6897476700  OOOOOOOOOO
2  0.035069  756373  6711566750  ZZZZZZZZZZ
3  0.066692  957474  9085243570  FFFFFFFFFF
4  0.679182  665894  3750794810  MMMMMMMMMM
5  0.861914  630527  6567684430  TTTTTTTTTT
6  0.697691  825704  8005182860  FFFFFFFFFF
7  0.474501  942131  4099797720  QQQQQQQQQQ
8  0.645817  951055  8065980030  VVVVVVVVVV
9  0.083500  349709  7417288920  EEEEEEEEEE

製作更大的 DF（10 * 100.000 = 1.000.000 行）

df = pd.concat([df] * 10**5, ignore_index=True)

建立（或開啟現有的）HDFStore 檔案

store = pd.HDFStore('d:/temp/example.h5')

將我們的資料框儲存到 h5（HDFStore）檔案中，索引[int32，int64，string]列：

store.append('store_key', df, data_columns=['int32','int64','string'])

顯示 HDFStore 詳細資訊

In [78]: store.get_storer('store_key').table
Out[78]:
/store_key/table (Table(10,)) ''
  description := {
  "index": Int64Col(shape=(), dflt=0, pos=0),
  "values_block_0": Float64Col(shape=(1,), dflt=0.0, pos=1),
  "int32": Int32Col(shape=(), dflt=0, pos=2),
  "int64": Int64Col(shape=(), dflt=0, pos=3),
  "string": StringCol(itemsize=10, shape=(), dflt=b'', pos=4)}
  byteorder := 'little'
  chunkshape := (1724,)
  autoindex := True
  colindexes := {
    "index": Index(6, medium, shuffle, zlib(1)).is_csi=False,
    "int32": Index(6, medium, shuffle, zlib(1)).is_csi=False,
    "string": Index(6, medium, shuffle, zlib(1)).is_csi=False,
    "int64": Index(6, medium, shuffle, zlib(1)).is_csi=False}

顯示索引列

In [80]: store.get_storer('store_key').table.colindexes
Out[80]:
{
    "int32": Index(6, medium, shuffle, zlib(1)).is_csi=False,
    "index": Index(6, medium, shuffle, zlib(1)).is_csi=False,
    "string": Index(6, medium, shuffle, zlib(1)).is_csi=False,
    "int64": Index(6, medium, shuffle, zlib(1)).is_csi=False}

關閉（重新整理到磁碟）我們的商店檔案

store.close()