Add CLI (#35)

* Initial CLI implementation

* Fix CLI quirks

* Add `smoothing` parameter I forgot

* Improve README.md

* Fix linter and tox

README.md

@@ -18,7 +18,7 @@ Detect knee points in various scenarios using a plethora of methods
 
 
 ## Usage
-Just plugin your values in a `list`, `tuple` or an `np.ndarray` and watch `knarrow` hit the knee:
+Just plug in your values in a `list`, `tuple` or an `np.ndarray` and watch `knarrow` hit the knee:
 
 ```pycon
 >>> from knarrow import find_knee
@@ -54,7 +54,18 @@ array([[0.  , 1.  ],
 4
 ```
 
-### Similar projects
+### CLI
+This library also comes with a handy CLI:
+```shell
+$ cat data.txt | knarrow
+stdin 11
+$ cat data.txt | knarrow -o value
+stdin 59874.14171519781845532648
+$ knarrow --sort -d ',' -o value shuf_delim.txt
+shuf_delim.txt 20
+```
+
+## Similar projects
 
 While I've come up with most of these methods by myself, I am not the only one. Here is a (non-comprehensive) list of projects I've found that implement a similar funcionality and may have been an inspiration for me:
 - [mariolpantunes/knee](https://github.com/mariolpantunes/knee)

docs/api/apidoc/knarrow.rst

@@ -1,6 +1,14 @@
 knarrow package
 ===============
 
+Subpackages
+-----------
+
+.. toctree::
+   :maxdepth: 4
+
+   knarrow.cli
+
 Submodules
 ----------
 

setup.py

@@ -56,4 +56,5 @@
         "Topic :: Utilities",
         "Typing :: Typed",
     ],
+    entry_points={"console_scripts": ["knarrow=knarrow.cli.__main__:main"]},
 )

src/knarrow/cli/__main__.py

@@ -0,0 +1,88 @@
+from argparse import ArgumentDefaultsHelpFormatter, ArgumentParser
+from collections import Counter
+from functools import partial
+from pathlib import Path
+
+from knarrow import find_knee
+
+
+def gte_0(value):
+    x = float(value)
+    assert x >= 0.0
+    return x
+
+
+METHODS = [
+    "angle",
+    "c_method",
+    "distance",
+    "distance_adjacent",
+    "kneedle",
+    "menger_anchored",
+    "menger_successive",
+    "ols_swiping",
+]
+
+
+def get_parser():
+    parser = ArgumentParser(prog="knarrow", formatter_class=ArgumentDefaultsHelpFormatter)
+    parser.add_argument(
+        "-m", "--method", choices=(["all"] + METHODS), default="all", help="select the knee searching method"
+    )
+    parser.add_argument(
+        "--sort",
+        action="store_true",
+        help="sort the values before the knee search. By default is assumes the input is already sorted",
+    )
+    parser.add_argument("--smoothing", default=0.0, type=gte_0, help="cublic spline smoothing parameter")
+    parser.add_argument(
+        "-d", "--delimiter", default=None, help="split the values with DELIMITER. If None, split by space"
+    )
+    parser.add_argument(
+        "-o",
+        "--output",
+        choices=["index", "value"],
+        default="index",
+        help=(
+            "if output is `value`, this will return the row of the input file where the knee was detected. "
+            "if output is `index`, the index of that row will be returned"
+        ),
+    )
+    parser.add_argument("files", nargs="*", default=["-"], help="a list of files. STDIN is denoted with `-`.")
+    return parser
+
+
+def cli(method="all", files=None, sort=False, delimiter=None, output=None, smoothing=None):
+    for filename in files:
+        path = Path("/dev/stdin" if filename == "-" else filename)
+        with path.open("r") as file:
+            rows = list(map(str.strip, file))
+            split = partial(str.split, sep=delimiter)
+            values = map(split, rows)
+            numbers = list(tuple(float(value) for value in row) for row in values)
+            indices = list(range(len(numbers)))
+            if sort:
+                indices.sort(key=lambda i: numbers[i])
+                key_function = (lambda x: x) if len(numbers[0]) == 1 else (lambda x: x[0])
+                numbers.sort(key=key_function)
+
+            if method == "all":
+                counter = Counter([find_knee(numbers, method=m, sort=False, smoothing=smoothing) for m in METHODS])
+                most_common = counter.most_common(1).pop(0)
+                knee = most_common[0]
+            else:
+                knee = find_knee(numbers, method=method, sort=False, smoothing=smoothing)
+
+            result = indices[knee] if output == "index" else rows[indices[knee]]
+            print(path.name, result)
+    return
+
+
+def main():
+    parser = get_parser()
+    args = vars(parser.parse_args())
+    exit(cli(**args))
+
+
+if __name__ == "__main__":
+    main()

src/knarrow/ols.py

@@ -27,7 +27,7 @@ def r_squared(x, y):
     sst = np.var(y, ddof=n - 1)
     r2_1 = (1 - ssr / sst).item()
     corr = np.corrcoef(w @ xx, y)[0, 1]
-    r2_2 = corr ** 2
+    r2_2 = corr**2
     assert np.allclose(r2_1, r2_2)
     return ((r2_1 + r2_2) / 2).item()
 

src/knarrow/util.py

@@ -77,7 +77,9 @@ def inner(*args, **kwargs):
 
         # more or less all the algorithms depend on the inputs to be sorted, at least in the x dimension
         # therefore the x and y are sorted together
-        if not np.all(np.diff(x) > 0):
+        # the user should explicitly disallow sorting
+        perform_sort = kwargs.pop("sort", True)
+        if perform_sort and not np.all(np.diff(x) > 0):
             sorted_indices = np.argsort(x)  # sort in the ascending way
             x = x[sorted_indices]
             y = y[sorted_indices]

tests/cli/data.txt

@@ -0,0 +1,15 @@
+1.00000000000000000000
+2.71828182845904523536
+7.38905609893065022723
+20.08553692318766774092
+54.59815003314423907811
+148.41315910257660342111
+403.42879349273512260838
+1096.63315842845859926372
+2980.95798704172827474359
+8103.08392757538400770999
+22026.46579480671651695790
+59874.14171519781845532648
+162754.79141900392080800520
+442413.39200892050332610277
+1202604.28416477677774923677

tests/cli/shuf_delim.txt

@@ -0,0 +1,21 @@
+18,.85447998506288733923
+17,.91167918637384475033
+4,5.20650605348167213512
+13,1.24537072699740781656
+10,1.70665642170428796832
+20,.75957113438621465853
+2,11.10336757201109052117
+1,20.08312627154631767497
+12,1.36981506456443459268
+7,2.64233344933772314492
+14,1.14119975942722448362
+9,1.94063983061649768141
+16,.97719576435870593359
+8,2.24217221922731762267
+5,3.98762418529180844687
+11,1.52066319213249374626
+15,1.05289037728159810697
+19,.80415538149886314748
+6,3.19315561035895693094
+0,51.19436311832002749043
+3,7.24325504903906741104

tests/cli/test.txt

@@ -0,0 +1,6 @@
+1
+2
+3
+6
+9
+12

tox.ini

@@ -64,16 +64,16 @@ skip_install = true
 deps =
     -r{toxinidir}/dev-requirements.txt
 commands =
-    isort .
-    black .
+    isort src/knarrow
+    black src/knarrow
 
 [testenv:fmt-check]
 skip_install = true
 deps =
     -r{toxinidir}/dev-requirements.txt
 commands =
-    isort --check .
-    black --check .
+    isort --check src/knarrow
+    black --check src/knarrow
 
 [testenv:docs-api]
 skip_install = true