我有一个通过 HDFStore 存储的 Pandas DataFrame，它主要存储有关我正在执行的测试运行的摘要行。

每行中的几个字段包含可变长度的描述性字符串。

当我进行测试运行时，我创建了一个新的 DataFrame，其中只有一行:

def export_as_df(self):
    return pd.DataFrame(data=[self._to_dict()], index=[datetime.datetime.now()])

然后调用 HDFStore.append(string, DataFrame) 将新行添加到现有的 DataFrame。

除了其中一个字符串列的内容大于已经存在的最长实例之外，这工作正常，因此我收到以下错误:

File "<ipython-input-302-a33c7955df4a>", line 516, in save_pytables
store.append('tests', test.export_as_df())
File "/Library/Frameworks/EPD64.framework/Versions/7.3/lib/python2.7/site-packages/pandas/io/pytables.py", line 532, in append
self._write_to_group(key, value, table=True, append=True, **kwargs)
File "/Library/Frameworks/EPD64.framework/Versions/7.3/lib/python2.7/site-packages/pandas/io/pytables.py", line 788, in _write_to_group
s.write(obj = value, append=append, complib=complib, **kwargs)
File "/Library/Frameworks/EPD64.framework/Versions/7.3/lib/python2.7/site-packages/pandas/io/pytables.py", line 2491, in write
min_itemsize=min_itemsize, **kwargs)
File "/Library/Frameworks/EPD64.framework/Versions/7.3/lib/python2.7/site-packages/pandas/io/pytables.py", line 2254, in create_axes
raise Exception("cannot find the correct atom type -> [dtype->%s,items->%s] %s" % (b.dtype.name, b.items, str(detail)))
Exception: cannot find the correct atom type -> [dtype->object,items->Index([bp, id, inst, per, sp, st, title], dtype=object)] [values_block_3] column has a min_itemsize of [51] but itemsize [46] is required!

我找不到任何关于如何在创建 DataFrame 时指定字符串长度的文档。这里的解决方案是什么？

更新:

失败的代码:

        store = pd.HDFStore(pytables_store)            
        for test in self.backtests:
            try:
                min_itemsizes = { 'buy_pattern' : 60, 'sell_pattern': 60, 'strategy': 60, 'title': 60 }
                store.append('tests', test.export_as_df(), min_itemsize = min_itemsizes)

这是 0.11rc1 下的错误:

File "<ipython-input-110-492b7b6603d7>", line 522, in save_pytables
  store.append('tests', test.export_as_df(), min_itemsize = min_itemsizes)
File "/Users/admin/dev/pandas/pandas-0.11.0rc1/pandas/io/pytables.py", line 610, in append
  self._write_to_group(key, value, table=True, append=True, **kwargs)
File "/Users/admin/dev/pandas/pandas-0.11.0rc1/pandas/io/pytables.py", line 871, in _write_to_group
  s.write(obj = value, append=append, complib=complib, **kwargs)
File "/Users/admin/dev/pandas/pandas-0.11.0rc1/pandas/io/pytables.py", line 2707, in write
  min_itemsize=min_itemsize, **kwargs)
File "/Users/admin/dev/pandas/pandas-0.11.0rc1/pandas/io/pytables.py", line 2447, in create_axes
  self.validate_min_itemsize(min_itemsize)
File "/Users/admin/dev/pandas/pandas-0.11.0rc1/pandas/io/pytables.py", line 2184, in validate_min_itemsize
  raise ValueError("min_itemsize has [%s] which is not an axis or data_column" % k)
ValueError: min_itemsize has [buy_pattern] which is not an axis or data_column

数据样本:

                           all_day              buy_pattern  \
2013-04-14 12:11:44.377695   False  Hammer() and LowerLow()   

                                                           id instrument  \
2013-04-14 12:11:44.377695  tafdcc96ba4eb11e2a86d14109fcecd49     EURUSD   

                            open_margin periodicity sell_pattern strategy  \
2013-04-14 12:11:44.377695       0.0001     1:00:00                 Tsl()   

                           title  top_bottom  wick_body  
2013-04-14 12:11:44.377695   tsl         0.5          2 

数据类型:

print prob_test.export_as_df().get_dtype_counts() 

    bool       1
    float64    2
    int64      1
    object     7
    dtype: int64

我每次都删除 h5 文件，因为我想要干净的结果。想知道是否有一些像失败一样愚蠢的事情，因为在第一次追加时 df 不存在于 h5 中(因此也不存在任何列)？

最佳答案

这是有关此内容的新文档部分的链接:http://pandas.pydata.org/pandas-docs/stable/io.html#string-columns

这个问题是因为您在 min_itemsize 中指定了一个不是 data_column 的列。简单的解决方法是将 data_columns=True 添加到您的追加语句中，但我还更新了代码以在您传递有效的列名时自动创建 data_columns。我认为这是有道理的，您想要最小的列大小，所以顺其自然吧。

还创建了一个新的文档部分字符串列以显示更完整的示例和解释(文档将很快更新)。

# this is the new behavior (after code updates)
n [340]: dfs = DataFrame(dict(A = 'foo', B = 'bar'),index=range(5))

In [341]: dfs
Out[341]: 
     A    B
0  foo  bar
1  foo  bar
2  foo  bar
3  foo  bar
4  foo  bar

# A and B have a size of 30
In [342]: store.append('dfs', dfs, min_itemsize = 30)

In [343]: store.get_storer('dfs').table
Out[343]: 
/dfs/table (Table(5,)) ''
  description := {
  "index": Int64Col(shape=(), dflt=0, pos=0),
  "values_block_0": StringCol(itemsize=30, shape=(2,), dflt='', pos=1)}
  byteorder := 'little'
  chunkshape := (963,)
  autoIndex := True
  colindexes := {
    "index": Index(6, medium, shuffle, zlib(1)).is_CSI=False}

# A is created as a data_column with a size of 30
# B is size is calculated
In [344]: store.append('dfs2', dfs, min_itemsize = { 'A' : 30 })

In [345]: store.get_storer('dfs2').table
Out[345]: 
/dfs2/table (Table(5,)) ''
  description := {
  "index": Int64Col(shape=(), dflt=0, pos=0),
  "values_block_0": StringCol(itemsize=3, shape=(1,), dflt='', pos=1),
  "A": StringCol(itemsize=30, shape=(), dflt='', pos=2)}
  byteorder := 'little'
  chunkshape := (1598,)
  autoIndex := True
  colindexes := {
    "A": Index(6, medium, shuffle, zlib(1)).is_CSI=False,
    "index": Index(6, medium, shuffle, zlib(1)).is_CSI=False}

关于python - 当字符串列内容比已有内容长时，HDFStore.append(string, DataFrame) 失败，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/15988871/