Merge pull request #2854 from theotherjimmy/find-duplicates

[Tools] Find and report duplicates

Author: sg-

tools/build_api.py

@@ -452,6 +452,8 @@ def build_project(src_paths, build_path, target, toolchain_name,
         # Link Program
         res, _ = toolchain.link_program(resources, build_path, name)
 
+        resources.detect_duplicates(toolchain)
+
         if report != None:
             end = time()
             cur_result["elapsed_time"] = end - start

tools/git_hooks/__init__.py

@@ -0,0 +1,26 @@
+from os import walk
+from os.path import join, abspath, dirname, basename, splitext
+import sys
+
+ROOT = abspath(join(dirname(__file__), "..", ".."))
+sys.path.insert(0, ROOT)
+
+from tools.toolchains.gcc import GCC_ARM
+from tools.targets import TARGET_MAP
+from argparse import ArgumentParser
+
+if __name__ == "__main__":
+    parser = ArgumentParser("Find duplicate file names within a directory structure")
+    parser.add_argument("dirs", help="Directories to search for duplicate file names"
+                        , nargs="*")
+    parser.add_argument("--silent", help="Supress printing of filenames, just return number of duplicates", action="store_true")
+    args = parser.parse_args()
+
+    toolchain = GCC_ARM(TARGET_MAP["K64F"])
+
+    resources = sum([toolchain.scan_resources(d) for d in args.dirs], None)
+
+    scanned_files = {}
+
+    exit(resources.detect_duplicates(toolchain))
+

tools/git_hooks/find_duplicates.py

@@ -3,14 +3,16 @@
 import os
 from string import printable
 from copy import deepcopy
+from mock import MagicMock
 from hypothesis import given
 from hypothesis.strategies import text, lists, fixed_dictionaries
 
 ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..",
                                     ".."))
 sys.path.insert(0, ROOT)
 
-from tools.toolchains import TOOLCHAIN_CLASSES, LEGACY_TOOLCHAIN_NAMES
+from tools.toolchains import TOOLCHAIN_CLASSES, LEGACY_TOOLCHAIN_NAMES,\
+    Resources
 from tools.targets import TARGET_MAP
 
 def test_instantiation():
@@ -96,3 +98,27 @@ def test_toolchain_profile_asm(profile, source_file):
                 "Toolchain %s did not propigate arg %s" % (toolchain.name,
                                                            parameter)
 
+    for name, Class in  TOOLCHAIN_CLASSES.items():
+        CLS = Class(TARGET_MAP["K64F"])
+        assert name == CLS.name or name ==  LEGACY_TOOLCHAIN_NAMES[CLS.name]
+
+
+@given(lists(text(alphabet=ALPHABET, min_size=1), min_size=1))
+def test_detect_duplicates(filenames):
+    c_sources = [os.path.join(name, "dupe.c") for name in filenames]
+    s_sources = [os.path.join(name, "dupe.s") for name in filenames]
+    cpp_sources = [os.path.join(name, "dupe.cpp") for name in filenames]
+    with MagicMock() as notify:
+        toolchain = TOOLCHAIN_CLASSES["ARM"](TARGET_MAP["K64F"], notify=notify)
+        res = Resources()
+        res.c_sources = c_sources
+        res.s_sources = s_sources
+        res.cpp_sources = cpp_sources
+        assert res.detect_duplicates(toolchain) == 1,\
+            "Not Enough duplicates found"
+
+        _, (notification, _), _ = notify.mock_calls[1]
+        assert "dupe.o" in notification["message"]
+        assert "dupe.s" in notification["message"]
+        assert "dupe.c" in notification["message"]
+        assert "dupe.cpp" in notification["message"]

tools/test/toolchains/api.py

@@ -120,6 +120,43 @@ def add(self, resources):
 
         return self
 
+    def _collect_duplicates(self, dupe_dict, dupe_headers):
+        for filename in self.s_sources + self.c_sources + self.cpp_sources:
+            objname, _ = splitext(basename(filename))
+            dupe_dict.setdefault(objname, set())
+            dupe_dict[objname] |= set([filename])
+        for filename in self.headers:
+            headername = basename(filename)
+            dupe_headers.setdefault(headername, set())
+            dupe_headers[headername] |= set([headername])
+        for res in self.features.values():
+            res._collect_duplicates(dupe_dict, dupe_headers)
+        return dupe_dict, dupe_headers
+
+    def detect_duplicates(self, toolchain):
+        """Detect all potential ambiguities in filenames and report them with
+        a toolchain notification
+
+        Positional Arguments:
+        toolchain - used for notifications
+        """
+        count = 0
+        dupe_dict, dupe_headers = self._collect_duplicates(dict(), dict())
+        for objname, filenames in dupe_dict.iteritems():
+            if len(filenames) > 1:
+                count+=1
+                toolchain.tool_error(
+                    "Object file %s.o is not unique! It could be made from: %s"\
+                    % (objname, " ".join(filenames)))
+        for headername, locations in dupe_headers.iteritems():
+            if len(locations) > 1:
+                count+=1
+                toolchain.tool_error(
+                    "Header file %s is not unique! It could be: %s" %\
+                    (headername, " ".join(locations)))
+        return count
+
+
     def relative_to(self, base, dot=False):
         for field in ['inc_dirs', 'headers', 's_sources', 'c_sources',
                       'cpp_sources', 'lib_dirs', 'objects', 'libraries',