The FrameFixtures library defines a tiny domain-specific language that permits using compact expressions to create diverse, deterministic DataFrame fixtures with StaticFrame. The resulting Frame can be used for test, performance studies, or documentation examples, and can easily be converted to Pandas DataFrames or other representations available via StaticFrame.
Code: https://github.com/static-frame/frame-fixtures
Packages: https://pypi.org/project/frame-fixtures
Installation
Install FrameFixtures via PIP:
pip install frame-fixtures [extras]
The [extras] configuration includes StaticFrame as a requirement. As StaticFrame uses FrameFixtures, installation without [extras] assumes the availability of StaticFrame:
pip install frame-fixtures
Examples
Import FrameFixtures with the following convention:
>>> import frame_fixtures as ff
Create a 4 by 8 Frame of string, Booleans, and floats.
>>> ff.parse('v(str,str,bool,float)|s(4,8)')
<Frame>
<Index> 0 1 2 3 4 5 6 7 <int64>
<Index>
0 zjZQ zaji True 1080.4 zDVQ zEdH True 647.9
1 zO5l zJnC False 2580.34 z5hI zB7E True 2755.18
2 zEdH zDdR False 700.42 zyT8 zwIp False -1259.28
3 zB7E zuVU True 3338.48 zS6w zDVQ False 3442.84
<int64> <<U4> <<U4> <bool> <float64> <<U4> <<U4> <bool> <float64>
The same Frame can be converted to Pandas:
>>> ff.parse('v(str,str,bool,float)|s(4,8)').to_pandas()
0 1 2 3 4 5 6 7
0 zjZQ zaji True 1080.40 zDVQ zEdH True 647.90
1 zO5l zJnC False 2580.34 z5hI zB7E True 2755.18
2 zEdH zDdR False 700.42 zyT8 zwIp False -1259.28
3 zB7E zuVU True 3338.48 zS6w zDVQ False 3442.84
Create a 4 by 4 Frame of Booleans with three-level index and columns.
>>> ff.parse('v(bool)|i(IH,(str,int,str))|c(IH,(str,int,str))|s(4,4)')
<Frame>
<IndexHierarchy> zZbu zZbu zZbu zZbu <<U4>
105269 105269 119909 119909 <int64>
zDVQ z5hI zyT8 zS6w <<U4>
<IndexHierarchy>
zZbu 105269 zDVQ False False True False
zZbu 105269 z5hI False False False False
zZbu 119909 zyT8 False False False True
zZbu 119909 zS6w True False True True
<<U4> <int64> <<U4> <bool> <bool> <bool> <bool>
The same Frame can be converted to Pandas:
>>> ff.parse('v(bool)|i(IH,(str,int,str))|c(IH,(str,int,str))|s(4,4)').to_pandas()
__index0__ zZbu
__index1__ 105269 119909
__index2__ zDVQ z5hI zyT8 zS6w
__index0__ __index1__ __index2__
zZbu 105269 zDVQ False False True False
z5hI False False False False
119909 zyT8 False False False True
zS6w True False True True
FrameFixtures support defining features unique to StaticFrame, such as specifying a grow-only FrameGO, Index types within IndexHierarchy, and usage of np.datetime64 types other than nanoseconds. These specifications are not directly convertible to Pandas.
>>> ff.parse('f(Fg)|v(int,bool,str)|i((IY,ID),(dtY,dtD))|c(ISg,dts)|s(6,2)')
<FrameGO>
<IndexSecondGO> 1970-01-01T09:38:35 1970-01-01T01:00:48 <datetime64[s]>
<IndexHierarchy>
36685 2258-03-21 -88017 False
36685 2298-04-20 92867 False
5618 2501-10-08 84967 False
5618 2441-04-14 13448 False
93271 2234-04-07 175579 False
93271 2210-12-26 58768 False
<datetime64[Y]> <datetime64[D]> <int64> <bool>
What is New in FramFixtures
1.1.0
Now compatible with NumPy 2.0.
Grammar
Container Components
A FrameFixture is defined by specifying one or more container components using symbols such as s for shape and i for index. Container components (CCs) are given arguments using Python function call syntax, and multiple CCs are delimited with |. The shape CC takes integers as arguments; all other CCs take Constructor Specifiers (CS) and/or Dtype Specifiers (DS) as arguments. So a 100 by 20 Frame with an index of str is specified as s(100,20)|i(I,str), where 100 and 20 define the row and column counts, and I is the CC and str is the DS. Component symbols, whether components are required, and the number of required arguments, is summarized below.
Symbol |
Component |
Required |
Arguments |
Signature |
|---|---|---|---|---|
f |
Frame |
False |
1 |
(CS,) |
i |
Index |
False |
2 |
(CS, DS) or ((CS, …), (DS, …)) |
c |
Columns |
False |
2 |
(CS, DS) or ((CS, …), (DS, …)) |
v |
Values |
False |
unbound |
(DS, …) |
s |
Shape |
True |
2 |
(int, int) |
Constructor Specifiers
CSs are given to the f CC; the i and c CC take one or many CSs as their first argument.
Symbol |
Class |
|---|---|
F |
Frame |
Fg |
FrameGO |
I |
Index |
Ig |
IndexGO |
IH |
IndexHierarchy |
IHg |
IndexHierarchyGO |
IACF |
IndexAutoConstructorFactory |
IY |
IndexYear |
IYg |
IndexYearGO |
IM |
IndexYearMonth |
IMg |
IndexYearMonthGO |
IYM |
IndexYearMonth |
IYMg |
IndexYearMonthGO |
ID |
IndexDate |
IDg |
IndexDateGO |
Ih |
IndexHour |
Ihg |
IndexHourGO |
Im |
IndexMinute |
Img |
IndexMinuteGO |
Is |
IndexSecond |
Isg |
IndexSecondGO |
Ims |
IndexMillisecond |
Imsg |
IndexMillisecondGO |
Ius |
IndexMicrosecond |
Iusg |
IndexMicrosecondGO |
Ins |
IndexNanosecond |
Insg |
IndexNanosecondGO |
Dtype Specifiers
DSs are given to the v CC, and are used repeatedly to fill all columns; the i and c CC take one or many DSs as their second argument.
Symbol |
Class |
|---|---|
dtY |
dtype(‘<M8[Y]’) |
dtM |
dtype(‘<M8[M]’) |
dtD |
dtype(‘<M8[D]’) |
dth |
dtype(‘<M8[h]’) |
dtm |
dtype(‘<M8[m]’) |
dts |
dtype(‘<M8[s]’) |
dtms |
dtype(‘<M8[ms]’) |
dtus |
dtype(‘<M8[us]’) |
dtns |
dtype(‘<M8[ns]’) |
tdY |
dtype(‘<m8[Y]’) |
tdM |
dtype(‘<m8[M]’) |
tdD |
dtype(‘<m8[D]’) |
tdh |
dtype(‘<m8[h]’) |
tdm |
dtype(‘<m8[m]’) |
tds |
dtype(‘<m8[s]’) |
tdms |
dtype(‘<m8[ms]’) |
tdus |
dtype(‘<m8[us]’) |
tdns |
dtype(‘<m8[ns]’) |
int |
<class ‘int’> |
str |
<class ‘str’> |
bytes |
<class ‘bytes’> |
float |
<class ‘float’> |
bool |
<class ‘bool’> |
complex |
<class ‘complex’> |
object |
<class ‘object’> |
int8 |
<class ‘numpy.int8’> |
int16 |
<class ‘numpy.int16’> |
int32 |
<class ‘numpy.int32’> |
int64 |
<class ‘numpy.int64’> |
uint8 |
<class ‘numpy.uint8’> |
uint16 |
<class ‘numpy.uint16’> |
uint32 |
<class ‘numpy.uint32’> |
uint64 |
<class ‘numpy.uint64’> |
float16 |
<class ‘numpy.float16’> |
float32 |
<class ‘numpy.float32’> |
float64 |
<class ‘numpy.float64’> |
complex64 |
<class ‘numpy.complex64’> |
complex128 |
<class ‘numpy.complex128’> |
Metadata
Release files for frame-fixtures 1.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| frame-fixtures-1.1.0.tar.gz | 14.8 kB | Without provenance | Details for frame-fixtures-1.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| frame_fixtures-1.1.0-py3-none-any.whl | Python 3 | none | any | Details for frame_fixtures-1.1.0-py3-none-any.whl |
Total release size: 26.6 kB
Release files / frame-fixtures-1.1.0.tar.gz
| Download URL | frame-fixtures-1.1.0.tar.gz |
|---|---|
| Size | 14.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
37b511993dda6d310f35b3ce480084004b21afdb563e9626059530075878b004
|
|
BLAKE2b-256 checksum How to use checksums |
b08b9386c795c436e2d95d187decd2e982ce70a1f3e876489335032eddd35051
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/5.1.1 CPython/3.9.19
|
Release files / frame_fixtures-1.1.0-py3-none-any.whl
| Download URL | frame_fixtures-1.1.0-py3-none-any.whl |
|---|---|
| Size | 11.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
6cade13d5058909e6b9608bf34d984b84f3c9801d144b0746a656e382512f958
|
|
BLAKE2b-256 checksum How to use checksums |
91bce32a8a5fb455f60fb42dacf15440d0e67ecd82fdc474820167b12a35aba3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/5.1.1 CPython/3.9.19
|
Security for frame-fixtures 1.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.