Merge pull request aboutcode-org#2 from nexB/visit-symlinks

Modify walk(), is_file(), and is_dir() to have the option to follow symlinks

setup.cfg

@@ -6,7 +6,7 @@ license_file = apache-2.0.LICENSE
 name = commoncode
 author = nexB. Inc. and others
 author_email = info@aboutcode.org
-description = commoncode 
+description = commoncode
 long_description = file:README.rst
 url = https://github.com/nexB/commoncode
 classifiers =
@@ -16,7 +16,7 @@ classifiers =
     Programming Language :: Python :: 3
     Topic :: Software Development
     Topic :: Utilities
-keywords = 
+keywords =
 
 [options]
 package_dir=
@@ -33,7 +33,8 @@ install_requires =
     requests >= 2.7.0, < 3.0.0
     intbitset >= 2.3.0, < 3.0
     saneyaml
-setup_requires = setuptools_scm >= 4
+    typing >=3.6, < 3.7
+setup_requires = setuptools_scm[toml] >= 4
 
 [options.packages.find]
 where=src

src/commoncode/filetype.py

@@ -45,20 +45,24 @@ def is_link(location):
     return location and os.path.islink(location)
 
 
-def is_file(location):
+def is_file(location, follow_symlinks=False):
     """
     Return True if `location` is a file.
     """
-    return (location and os.path.isfile(location)
-            and not is_link(location) and not is_broken_link(location))
+    _is_file = location and os.path.isfile(location)
+    if follow_symlinks:
+        return _is_file
+    return _is_file and not is_link(location) and not is_broken_link(location)
 
 
-def is_dir(location):
+def is_dir(location, follow_symlinks=False):
     """
     Return True if `location` is a directory.
     """
-    return (location and os.path.isdir(location) and not is_file(location)
-            and not is_link(location) and not is_broken_link(location))
+    _is_dir = location and os.path.isdir(location) and not is_file(location)
+    if follow_symlinks:
+        return _is_dir
+    return _is_dir and not is_link(location) and not is_broken_link(location)
 
 
 def is_regular(location):

src/commoncode/fileutils.py

@@ -197,7 +197,7 @@ def prepare_path(pth):
         if not isinstance(pth, bytes):
             pth = fsencode(pth)
         return pth
-    else:    
+    else:
         if not isinstance(pth, compat.unicode):
             return fsdecode(pth)
     return pth
@@ -365,16 +365,19 @@ def ignore_nothing(_):
     return False
 
 
-def walk(location, ignored=None):
+def walk(location, ignored=None, follow_symlinks=False):
     """
     Walk location returning the same tuples as os.walk but with a different
     behavior:
      - always walk top-down, breadth-first.
-     - always ignore and never follow symlinks,
+     - always ignore and never follow symlinks (unless `follow_symlinks` is True),
      - always ignore special files (FIFOs, etc.)
      - optionally ignore files and directories by invoking the `ignored`
        callable on files and directories returning True if it should be ignored.
      - location is a directory or a file: for a file, the file is returned.
+
+    If `follow_symlinks` is True, then symlinks will not be ignored and be
+    collected like regular files and directories
     """
     if on_linux and py2:
         location = fsencode(location)
@@ -387,33 +390,38 @@ def walk(location, ignored=None):
             logger_debug('walk: ignored:', location, is_ignored)
         return
 
-    if filetype.is_file(location) :
+    if filetype.is_file(location, follow_symlinks=follow_symlinks) :
         yield parent_directory(location), [], [file_name(location)]
 
-    elif filetype.is_dir(location):
+    elif filetype.is_dir(location, follow_symlinks=follow_symlinks):
         dirs = []
         files = []
         # TODO: consider using scandir
         for name in os.listdir(location):
             loc = os.path.join(location, name)
             if filetype.is_special(loc) or (ignored and ignored(loc)):
-                if TRACE:
-                    ign = ignored and ignored(loc)
-                    logger_debug('walk: ignored:', loc, ign)
-                continue
+                if (follow_symlinks
+                        and filetype.is_link(loc)
+                        and not filetype.is_broken_link(location)):
+                    pass
+                else:
+                    if TRACE:
+                        ign = ignored and ignored(loc)
+                        logger_debug('walk: ignored:', loc, ign)
+                    continue
             # special files and symlinks are always ignored
-            if filetype.is_dir(loc):
+            if filetype.is_dir(loc, follow_symlinks=follow_symlinks):
                 dirs.append(name)
-            elif filetype.is_file(loc):
+            elif filetype.is_file(loc, follow_symlinks=follow_symlinks):
                 files.append(name)
         yield location, dirs, files
 
         for dr in dirs:
-            for tripple in walk(os.path.join(location, dr), ignored):
+            for tripple in walk(os.path.join(location, dr), ignored, follow_symlinks=follow_symlinks):
                 yield tripple
 
 
-def resource_iter(location, ignored=ignore_nothing, with_dirs=True):
+def resource_iter(location, ignored=ignore_nothing, with_dirs=True, follow_symlinks=False):
     """
     Return an iterable of paths at `location` recursively.
 
@@ -424,7 +432,7 @@ def resource_iter(location, ignored=ignore_nothing, with_dirs=True):
     """
     if on_linux and py2:
         location = fsencode(location)
-    for top, dirs, files in walk(location, ignored):
+    for top, dirs, files in walk(location, ignored, follow_symlinks=follow_symlinks):
         if with_dirs:
             for d in dirs:
                 yield os.path.join(top, d)

tests/commoncode/test_filetype.py

@@ -209,3 +209,25 @@ def test_get_file_count(self):
         for test_file, count in tests:
             result = filetype.get_file_count(os.path.join(test_dir, test_file))
             assert count == result
+
+
+def SymlinkTest(FileBasedTesting):
+    test_data_dir = os.path.join(os.path.dirname(__file__), 'data')
+
+    @skipIf(on_windows, 'os.symlink does not work on Windows')
+    def test_is_file(self):
+        test_file = self.get_test_loc('symlink/test', copy=True)
+        temp_dir = fileutils.get_temp_dir()
+        test_link = join(temp_dir, 'test-link')
+        os.symlink(test_file, test_link)
+        assert filetype.is_file(test_link, follow_symlinks=True)
+        assert not filetype.is_file(test_link, follow_symlinks=False)
+
+    @skipIf(on_windows, 'os.symlink does not work on Windows')
+    def test_is_dir(self):
+        test_dir = self.get_test_loc('symlink', copy=True)
+        temp_dir = fileutils.get_temp_dir()
+        test_link = join(temp_dir, 'test-dir-link')
+        os.symlink(test_dir, test_link)
+        assert filetype.is_dir(test_link, follow_symlinks=True)
+        assert not filetype.is_dir(test_link, follow_symlinks=False)

tests/commoncode/test_fileutils.py

@@ -384,6 +384,20 @@ def test_os_walk_can_walk_non_utf8_path_from_unicode_path(self):
         _dirpath, _dirnames, filenames = result
         assert 18 == len(filenames)
 
+    @skipIf(on_windows, 'os.symlink does not work on Windows')
+    def test_walk_on_symlinks(self):
+        test_dir = self.get_test_loc('symlink/walk', copy=True)
+        temp_dir = fileutils.get_temp_dir()
+        test_link = join(temp_dir, 'test-dir-link')
+        os.symlink(test_dir, test_link)
+        results = list(fileutils.walk(test_link, follow_symlinks=True))
+        results = [(os.path.basename(top), dirs, files) for top, dirs, files in results]
+        expected = [
+            ('test-dir-link', ['dir'], ['a']),
+            ('dir', [], ['b'])
+        ]
+        assert expected == results
+
 
 class TestFileUtilsIter(FileBasedTesting):
     test_data_dir = os.path.join(os.path.dirname(__file__), 'data')
@@ -553,6 +567,19 @@ def test_resource_iter_can_walk_non_utf8_path_from_unicode_path(self):
         result = list(fileutils.resource_iter(test_dir, with_dirs=False))
         assert 18 == len(result)
 
+    def test_resource_iter_follow_symlinks(self):
+        test_dir = self.get_test_loc('symlink/walk', copy=True)
+        temp_dir = fileutils.get_temp_dir()
+        test_link = join(temp_dir, 'test-dir-link')
+        os.symlink(test_dir, test_link)
+        result = [os.path.basename(f) for f in fileutils.resource_iter(test_dir, follow_symlinks=True)]
+        expected = [
+            'dir',
+            'a',
+            'b'
+        ]
+        assert sorted(expected) == sorted(result)
+
 
 class TestBaseName(FileBasedTesting):
     test_data_dir = os.path.join(os.path.dirname(__file__), 'data')