add --replace-originals flag to extractcode

Signed-off-by: Maximilian Huber <maximilian.huber@tngtech.com>

Author: maxhbr

src/extractcode/extract.py

@@ -97,7 +97,7 @@
 ExtractEvent = namedtuple('ExtractEvent', 'source target done warnings errors')
 
 
-def extract(location, kinds=extractcode.default_kinds, recurse=False):
+def extract(location, kinds=extractcode.default_kinds, recurse=False, replace_originals=False):
     """
     Walk and extract any archives found at `location` (either a file or
     directory). Extract only archives of a kind listed in the `kinds` kind tuple.
@@ -116,10 +116,26 @@ def extract(location, kinds=extractcode.default_kinds, recurse=False):
     archives If `recurse` is false, then do not extract further an already
     extracted archive identified by the corresponding extract suffix location.
 
+    If `replace_originals` is True, the archives will be extracted in place.
+
     Note that while the original file system is walked top-down, breadth-first,
     if recurse and a nested archive is found, it is extracted to full depth
     first before resuming the file system walk.
     """
+    events = list(extract_files(location, kinds, recurse))
+    if replace_originals:
+        for xevent in reversed(events):
+            if xevent.done:
+                source = xevent.source
+                target = xevent.target
+                if TRACE:
+                    logger.debug('extract:replace_originals: replace %(source)r by %(target)r' % locals())
+                fileutils.delete(source)
+                fileutils.copytree(target, source)
+                fileutils.delete(target)
+    return events
+
+def extract_files(location, kinds=extractcode.default_kinds, recurse=False):
     ignored = partial(ignore.is_ignored, ignores=ignore.default_ignores, unignores={})
     if TRACE:
         logger.debug('extract:start: %(location)r  recurse: %(recurse)r\n' % locals())

tests/extractcode/test_extract.py

@@ -238,6 +238,55 @@ def test_extract_tree_recursive(self):
         check_no_error(result)
         check_files(test_dir, expected)
 
+    def test_extract_tree_recursive_replace_originals(self):
+        expected = (
+            'a/a.txt',
+            'a/a.tar.gz/a/b/a.txt',
+            'a/a.tar.gz/a/b/b.txt',
+            'a/a.tar.gz/a/c/c.txt',
+            'b/a.txt',
+            'b/b.tar.gz/b/.svn/all-wcprops',
+            'b/b.tar.gz/b/.svn/entries',
+            'b/b.tar.gz/b/.svn/format',
+            'b/b.tar.gz/b/a/a.txt',
+            'b/b.tar.gz/b/a/.svn/all-wcprops',
+            'b/b.tar.gz/b/a/.svn/entries',
+            'b/b.tar.gz/b/a/.svn/format',
+            'b/b.tar.gz/b/a/.svn/prop-base/a.tar.gz.svn-base',
+            'b/b.tar.gz/b/a/.svn/text-base/a.tar.gz.svn-base',
+            'b/b.tar.gz/b/a/.svn/text-base/a.txt.svn-base',
+            'b/b.tar.gz/b/a/a.tar.gz/a/b/a.txt',
+            'b/b.tar.gz/b/a/a.tar.gz/a/b/b.txt',
+            'b/b.tar.gz/b/a/a.tar.gz/a/c/c.txt',
+            'b/b.tar.gz/b/b/a.txt',
+            'b/b.tar.gz/b/b/.svn/all-wcprops',
+            'b/b.tar.gz/b/b/.svn/entries',
+            'b/b.tar.gz/b/b/.svn/format',
+            'b/b.tar.gz/b/b/.svn/text-base/a.txt.svn-base',
+            'b/b.tar.gz/b/c/a.txt',
+            'b/b.tar.gz/b/c/.svn/all-wcprops',
+            'b/b.tar.gz/b/c/.svn/entries',
+            'b/b.tar.gz/b/c/.svn/format',
+            'b/b.tar.gz/b/c/.svn/prop-base/a.tar.gz.svn-base',
+            'b/b.tar.gz/b/c/.svn/text-base/a.tar.gz.svn-base',
+            'b/b.tar.gz/b/c/.svn/text-base/a.txt.svn-base',
+            'b/b.tar.gz/b/c/a.tar.gz/a/b/a.txt',
+            'b/b.tar.gz/b/c/a.tar.gz/a/b/b.txt',
+            'b/b.tar.gz/b/c/a.tar.gz/a/c/c.txt',
+            'c/a.txt',
+            'c/a.tar.gz/a/b/a.txt',
+            'c/a.tar.gz/a/b/b.txt',
+            'c/a.tar.gz/a/c/c.txt',
+        )
+        test_dir = self.get_test_loc('extract/tree', copy=True)
+        result = list(extract.extract(test_dir, recurse=True, replace_originals=True))
+        check_no_error(result)
+        check_files(test_dir, expected)
+        # again
+        result = list(extract.extract(test_dir, recurse=True))
+        check_no_error(result)
+        check_files(test_dir, expected)
+
     def test_extract_tree_shallow_then_recursive(self):
         shallow = (
             'a/a.tar.gz',