Merge 2fe307a into bcf6314

docs/changelog.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## [0.4.6] - 2019-07-23
+### Added
+- file/directory size checker
+
 ## [0.4.5] - 2019-07-01
 ### Changed
 - If argument to callability checker is a file, require executability; if it's a folder, it's not callable.

tests/test_files.py

@@ -0,0 +1,74 @@
+""" Tests for checksum """
+
+import hashlib
+import itertools
+import pytest
+from ubiquerg import checksum, size, filesize_to_str
+
+__author__ = "Vince Reuter"
+__email__ = "vreuter@virginia.edu"
+
+
+def pytest_generate_tests(metafunc):
+    """ Dynamic test case generation/parameterization for this module. """
+    if "size1" in metafunc.fixturenames and "size2" in metafunc.fixturenames:
+        metafunc.parametrize(
+            ["size1", "size2"], itertools.product([1, 4], [2, 8]))
+    if "lines" in metafunc.fixturenames:
+        metafunc.parametrize("lines", [[], ["line1"], ["line1", "line2"]])
+
+
+def test_checksum(size1, size2, lines, tmpdir):
+    """ Checksum result matches expectation and is blocksize-agnostic """
+    fp = tmpdir.join("temp-data.txt").strpath
+    data = "\n".join(lines)
+    with open(fp, 'w') as f:
+        f.write(data)
+    exp = hashlib.new("md5", data.encode("utf-8")).hexdigest()
+    res1 = checksum(fp, size1)
+    res2 = checksum(fp, size2)
+    assert exp == res1
+    assert res1 == res2
+    assert res2 == exp
+
+
+def test_size_returns_str(lines, tmpdir):
+    """ Size returns a string and works with both files and directories """
+    fp = tmpdir.join("temp-data.txt").strpath
+    data = "\n".join(lines)
+    with open(fp, 'w') as f:
+        f.write(data)
+    assert isinstance(size(fp), str)
+    assert isinstance(size(tmpdir.strpath), str)
+
+
+def test_size_returns_int(lines, tmpdir):
+    fp = tmpdir.join("temp-data.txt").strpath
+    fp_larger = tmpdir.join("temp-data.txt").strpath
+    data = "\n".join(lines)
+    with open(fp, 'w') as f:
+        f.write(data)
+    with open(fp_larger, 'w') as f1:
+        f1.write(data * 100)
+    assert isinstance(size(tmpdir.strpath, False), int)
+    assert isinstance(size(fp, False), int)
+    assert size(fp, size_str=False) <= size(fp, size_str=False)
+
+
+def test_nonexistent_path(tmpdir):
+    """ Nonexistent path to checksum is erroneous. """
+    with pytest.raises(IOError):
+        checksum(tmpdir.join("does-not-exist.txt").strpath)
+
+
+@pytest.mark.parametrize("size_num", list(range(0, 10)) + [i/3 for i in range(0, 10)])
+def test_filesize_to_str_int(size_num):
+    """ Works with int and returns str """
+    assert isinstance(filesize_to_str(size_num), str)
+
+
+@pytest.mark.parametrize("obj", ["test", [], tuple()])
+def test_filesize_to_str_other(obj):
+    """ Returns the original object if it's not an int or float and warns """
+    with pytest.warns(UserWarning):
+        assert filesize_to_str(obj) == obj

tests/test_packaging.py

@@ -9,7 +9,7 @@
 
 @pytest.mark.parametrize(
     ["obj_name", "typecheck"],
-    [("build_cli_extra", isfunction), ("checksum", isfunction),
+    [("build_cli_extra", isfunction), ("checksum", isfunction), ("size", isfunction),
      ("expandpath", isfunction), ("is_collection_like", isfunction),
      ("is_command_callable", isfunction), ("is_url", isfunction),
      ("powerset", isfunction), ("query_yes_no", isfunction),

ubiquerg/_version.py

@@ -1 +1 @@
-__version__ = "0.4.5"
+__version__ = "0.4.6"

ubiquerg/files.py

@@ -1,8 +1,11 @@
 """ Functions facilitating file operations """
 
 from hashlib import md5
+import os
+from warnings import warn
 
-__all__ = ["checksum"]
+__all__ = ["checksum", "size", "filesize_to_str"]
+FILE_SIZE_UNITS = ['B', 'KB', 'MB', 'GB', 'TB', 'PB', 'EB', 'ZB', 'YB']
 
 
 def checksum(path, blocksize=int(2e+9)):
@@ -21,3 +24,48 @@ def checksum(path, blocksize=int(2e+9)):
                 break
             m.update(buf)
     return m.hexdigest()
+
+
+def size(path, size_str=True):
+    """
+    Gets the size of the file or directory in the provided path
+
+    :param str path: path to the file to check size of
+    :param bool size_str: whether the size should be converted to a human-readable string, e.g. convert B to MB
+    :return int|str: file size or file size string
+    """
+    if os.path.isfile(path):
+        s = os.path.getsize(path)
+    elif os.path.isdir(path):
+        s = 0
+        symlinks = []
+        for dirpath, dirnames, filenames in os.walk(path):
+            for f in filenames:
+                fp = os.path.join(dirpath, f)
+                if not os.path.islink(fp):
+                    s += os.path.getsize(fp)
+                else:
+                    s += os.lstat(fp).st_size
+                    symlinks.append(fp)
+        if len(symlinks) > 0:
+            print("{} symlinks were found: '{}'".format(len(symlinks), "\n".join(symlinks)))
+    else:
+        print("size could not be determined for: '{}'".format(path))
+        s = None
+    return filesize_to_str(s) if size_str else s
+
+
+def filesize_to_str(size):
+    """
+    Converts the numeric bytes to the size string
+
+    :param int|float size: file size to convert
+    :return str: file size string
+    """
+    if isinstance(size, (int, float)):
+        for unit in FILE_SIZE_UNITS:
+            if size < 1024:
+                return "{}{}".format(round(size, 1), unit)
+            size /= 1024
+    warn("size argument was neither an int nor a float, returning the original object")
+    return size