ph learns slice

README.md

@@ -125,6 +125,7 @@ a pipeline.
     * [`columns`, listing, selecting and re-ordering of](#columns-listing-selecting-and-re-ordering-of)
     * [`rename`](#rename)
     * [`replace`](#replace)
+    * [`slice`](#slice)
     * [`eval`; Mathematipulating and creating new columns](#eval-mathematipulating-and-creating-new-columns)
     * [`normalize`](#normalize)
     * [`query`](#query)
@@ -594,6 +595,57 @@ x,y,xp
 ```
 
 
+
+#### `slice`
+
+Slicing in Python is essential, and occasionally, we want to slice
+tabular data, e.g. look at only the 100 first, or 100 last rows, or
+perhaps we want to look at only every 10th row.  All of this is achieved
+using `ph slice start:end:step` with standard Python slice syntax.
+
+```bash
+$ cat a.csv | ph slice 1:9:2
+x,y
+4,9
+6,11
+8,13
+```
+
+Reversing:
+
+```
+$ cat a.csv|ph slice ::-1
+x,y
+8,13
+7,12
+6,11
+5,10
+4,9
+3,8
+```
+
+See also `ph head` and `ph tail`.
+
+```bash
+$ cat a.csv | ph slice :3
+x,y
+3,8
+4,9
+5,10
+```
+
+equivalent to
+
+```bash
+$ cat a.csv | ph head 3
+x,y
+3,8
+4,9
+5,10
+```
+
+
+
 #### `eval`; Mathematipulating and creating new columns
 
 You can sum columns and place the result in a new column using

ph/__init__.py

@@ -1326,6 +1326,51 @@ def columns(*cols, **kwargs):
         pipeout(df[cols])
 
 
+def _parse_slice(slicestr):
+    pattern = ":<int> | <int>: | <int>:<int> | <int>:<int>:<int>"
+    error = "Input to slice is {} _not_ {}".format(pattern, slicestr)
+
+    assert ":" in slicestr, error
+    start = None
+    end = None
+    step = None
+    tup = slicestr.split(":")
+    if len(tup) > 3:
+        exit(error)
+    start = tup[0] or None
+    if start is not None:
+        start = int(start)
+    end = tup[1] or None
+    if end is not None:
+        end = int(end)
+    if len(tup) == 3:
+        step = tup[2] or None
+        if step is not None:
+            step = int(step)
+    return start, end, step
+
+
+@registerx("slice")
+def slice_(slicestr):
+    """Slice a dataframe with Python slice pattern.
+
+    Usage: cat a.csv | ph slice :10    # head
+           cat a.csv | ph slice -10:   # tail
+           cat a.csv | ph slice ::2    # every even row
+           cat a.csv | ph slice 1::2   # every odd row
+           cat a.csv | ph slice ::-1   # reverse file
+
+    """
+    pattern = ":<int> | <int>: | <int>:<int> | <int>:<int>:<int>"
+    error = "Input to slice is {} _not_ {}".format(pattern, slicestr)
+    df = pipein()
+    if isinstance(slicestr, int) or ":" not in slicestr:
+        exit(error)
+    start, end, step = _parse_slice(slicestr)
+    retval = df[start:end:step]
+    pipeout(retval)
+
+
 @register
 def drop(*columns, **kwargs):
     """Drop specified labels from rows or columns.

ph/_version.py

@@ -1 +1 @@
-__version__ = "0.3.0"
+__version__ = "0.3.1"

tests/test_ph.py

@@ -633,6 +633,51 @@ def test_replace_col_and_inf(phmgr):
     assert list(captured.df.y) == list(range(8, 14))
 
 
+def test_slice(phmgr):
+    with phmgr("a") as captured:
+        _call("slice 0:3")
+    assert not captured.err
+    captured.assert_shape(3, 2)
+    assert list(captured.df.x) == list(range(3, 6))
+    assert list(captured.df.y) == list(range(8, 11))
+
+
+def test_slice_end(phmgr):
+    with phmgr("a") as captured:
+        _call("slice :3")
+    assert not captured.err
+    captured.assert_shape(3, 2)
+    assert list(captured.df.x) == list(range(3, 6))
+    assert list(captured.df.y) == list(range(8, 11))
+
+
+def test_slice_start(phmgr):
+    with phmgr("a") as captured:
+        _call("slice 3:")
+    assert not captured.err
+    captured.assert_shape(3, 2)
+    assert list(captured.df.x) == list(range(6, 9))
+    assert list(captured.df.y) == list(range(11, 14))
+
+
+def test_slice_start_step(phmgr):
+    with phmgr("a") as captured:
+        _call("slice 1::2")
+    assert not captured.err
+    captured.assert_shape(3, 2)
+    assert list(captured.df.x) == list(range(4, 9, 2))
+    assert list(captured.df.y) == list(range(9, 14, 2))
+
+
+def test_slice_start_end_step(phmgr):
+    with phmgr("a") as captured:
+        _call("slice 1:5:2")
+    assert not captured.err
+    captured.assert_shape(2, 2)
+    assert list(captured.df.x) == list(range(4, 7, 2))
+    assert list(captured.df.y) == list(range(9, 12, 2))
+
+
 def test_slugify_df(phmgr):
     with phmgr("slugit") as captured:
         _call("slugify")