New in version 0.6.
The module petl.fluent provides all of the functions present in the root petl module, but with modifications to allow them to be used in a fluent style. E.g.:
>>> from petl.fluent import *
>>> t0 = dummytable()
>>> t0.look()
+-------+-----------+----------------------+
| 'foo' | 'bar' | 'baz' |
+=======+===========+======================+
| 61 | 'oranges' | 0.41684297441746143 |
+-------+-----------+----------------------+
| 42 | 'bananas' | 0.5424838757229734 |
+-------+-----------+----------------------+
| 55 | 'pears' | 0.044730394239418825 |
+-------+-----------+----------------------+
| 63 | 'apples' | 0.6553751878324998 |
+-------+-----------+----------------------+
| 57 | 'pears' | 0.33151097448517963 |
+-------+-----------+----------------------+
| 57 | 'apples' | 0.2152565282912028 |
+-------+-----------+----------------------+
| 45 | 'bananas' | 0.1478840303008977 |
+-------+-----------+----------------------+
| 79 | 'pears' | 0.14301990499723238 |
+-------+-----------+----------------------+
| 11 | 'pears' | 0.16801320344526383 |
+-------+-----------+----------------------+
| 96 | 'oranges' | 0.3004187573856759 |
+-------+-----------+----------------------+
>>> t1 = t0.convert('bar', 'upper').extend('quux', 42).extend('spong', expr('{foo} * {quux}')).selecteq('bar', 'APPLES')
>>> t1.look()
+-------+----------+----------------------+--------+---------+
| 'foo' | 'bar' | 'baz' | 'quux' | 'spong' |
+=======+==========+======================+========+=========+
| 63 | 'APPLES' | 0.6553751878324998 | 42 | 2646 |
+-------+----------+----------------------+--------+---------+
| 57 | 'APPLES' | 0.2152565282912028 | 42 | 2394 |
+-------+----------+----------------------+--------+---------+
| 87 | 'APPLES' | 0.9045902500660937 | 42 | 3654 |
+-------+----------+----------------------+--------+---------+
| 5 | 'APPLES' | 0.6915135568859515 | 42 | 210 |
+-------+----------+----------------------+--------+---------+
| 28 | 'APPLES' | 0.8440288073976338 | 42 | 1176 |
+-------+----------+----------------------+--------+---------+
| 32 | 'APPLES' | 0.047452310539432774 | 42 | 1344 |
+-------+----------+----------------------+--------+---------+
| 93 | 'APPLES' | 0.8100969279893147 | 42 | 3906 |
+-------+----------+----------------------+--------+---------+
| 94 | 'APPLES' | 0.8216793407511486 | 42 | 3948 |
+-------+----------+----------------------+--------+---------+
| 94 | 'APPLES' | 0.7911584363109934 | 42 | 3948 |
+-------+----------+----------------------+--------+---------+
| 34 | 'APPLES' | 0.18846546302867728 | 42 | 1428 |
+-------+----------+----------------------+--------+---------+
Alternatively, if you don’t want to import all petl function names into the root namespace, you can just import “etl”, e.g.:
>>> from petl.fluent import etl
>>> l = [['foo', 'bar'], ['a', 1], ['b', 3]]
>>> etl(l).look()
+-------+-------+
| 'foo' | 'bar' |
+=======+=======+
| 'a' | 1 |
+-------+-------+
| 'b' | 3 |
+-------+-------+
>>> etl(l).cut('foo').look()
+-------+
| 'foo' |
+=======+
| 'a' |
+-------+
| 'b' |
+-------+
>>> etl(l).tocsv('test.csv')
>>> etl().fromcsv('test.csv').look()
+-------+-------+
| 'foo' | 'bar' |
+=======+=======+
| 'a' | '1' |
+-------+-------+
| 'b' | '3' |
+-------+-------+
New in version 0.10.
Also included in the petl distribution is a script to execute simple transformation pipelines directly from the operating system shell. E.g.:
$ virtualenv petl
$ . petl/bin/activate
$ pip install petl
$ petl "dummytable().tocsv()" > dummy.csv
$ cat dummy.csv | petl "fromcsv().cut('foo', 'baz').selectgt('baz', 0.5).head().data().totsv()"
The petl script is extremely simple, it expects a single positional argument, which is evaluated as Python code but with all of the petl.fluent functions imported.