Fix path flattening collisions and make runfiles resolution deterministic in make_corpus_dir

fuzzing/tools/BUILD

@@ -83,3 +83,10 @@ py_test(
     srcs = ["dict_validation_test.py"],
     deps = [":dict_validation"],
 )
+
+py_test(
+    name = "make_corpus_dir_test",
+    srcs = ["make_corpus_dir_test.py"],
+    data = ["make_corpus_dir.py"],
+    deps = [requirement("absl-py")],
+)

fuzzing/tools/make_corpus_dir.py

@@ -36,12 +36,36 @@
 
 flags.mark_flag_as_required("output_dir")
 
+def flatten_corpus_path(corpus):
+    prefix = ""
+    if corpus.startswith("./") or (os.sep == "\\" and corpus.startswith(".\\")):
+        prefix = "dot-"
+        corpus = corpus[2:]
+
+    if os.sep == "\\":
+        corpus = corpus.replace("/", "\\")
+
+    drive, tail = os.path.splitdrive(corpus)
+
+    parts = [part for part in tail.split(os.sep) if part]
+    flattened = "-".join(parts)
+
+    if drive:
+        drive_part = drive.rstrip(":\\/").replace("\\", "-").replace("/", "-")
+        flattened = drive_part + ("-" + flattened if flattened else "")
+    elif tail.startswith(os.sep):
+        flattened = "-" + flattened
+
+    return prefix + flattened
+
 def expand_corpus_to_file_list(corpus, file_list):
     if not os.path.exists(corpus):
         raise FileNotFoundError("file " + corpus + " doesn't exist")
     if os.path.isdir(corpus):
         # The first element in glob("dir/**") is "dir/", which needs to be excluded
-        file_list.extend(glob.glob(os.path.join(corpus, "**"), recursive=True)[1:])
+        for expanded_path in glob.glob(os.path.join(corpus, "**"), recursive=True)[1:]:
+            if os.path.isfile(expanded_path):
+                file_list.append(expanded_path)
     else:
         file_list.append(corpus)
 
@@ -59,8 +83,57 @@ def main(argv):
                     corpus_line.rstrip("\n"), expanded_file_list)
 
     if expanded_file_list:
+        max_flattened_length = 200
+        flattened_names = {}
+        flattened_name_counts = {}
+        needs_suffix = set()
+
+        for corpus in expanded_file_list:
+            flattened = flatten_corpus_path(corpus)
+            flattened_names[corpus] = flattened
+            flattened_key = flattened.lower() if os.name == "nt" else flattened
+            flattened_name_counts[flattened_key] = (
+                flattened_name_counts.get(flattened_key, 0) + 1)
+            if len(flattened) > max_flattened_length:
+                needs_suffix.add(corpus)
+
+        for corpus in expanded_file_list:
+            flattened = flattened_names[corpus]
+            flattened_key = flattened.lower() if os.name == "nt" else flattened
+            if flattened_name_counts[flattened_key] > 1:
+                needs_suffix.add(corpus)
+
+        suffix_map = {}
+        if needs_suffix:
+            suffix_width = len(str(len(needs_suffix)))
+            for index, corpus in enumerate(sorted(needs_suffix), start=1):
+                suffix_map[corpus] = f"{index:0{suffix_width}d}"
+
+        final_name_map = {}
+        final_name_counts = {}
+        for corpus in expanded_file_list:
+            flattened = flattened_names[corpus]
+            suffix = suffix_map.get(corpus)
+            if suffix:
+                prefix_budget = max_flattened_length - len(suffix) - 2
+                flattened = flattened[:max(1, prefix_budget)] + "--" + suffix
+            final_name_map[corpus] = flattened
+            flattened_key = flattened.lower() if os.name == "nt" else flattened
+            final_name_counts[flattened_key] = (
+                final_name_counts.get(flattened_key, 0) + 1)
+
+        if any(count > 1 for count in final_name_counts.values()):
+            unique_corpora = sorted(set(expanded_file_list))
+            alias_width = len(str(len(unique_corpora)))
+            alias_map = {
+                corpus: f"entry-{index:0{alias_width}d}"
+                for index, corpus in enumerate(unique_corpora, start=1)
+            }
+            for corpus in expanded_file_list:
+                final_name_map[corpus] = alias_map[corpus]
+
         for corpus in expanded_file_list:
-            dest = os.path.join(FLAGS.output_dir, corpus.replace("/", "-"))
+            dest = os.path.join(FLAGS.output_dir, final_name_map[corpus])
             # Whatever the separator we choose, there is an chance that
             # the dest name conflicts with another file
             if os.path.exists(dest):

fuzzing/tools/make_corpus_dir_test.py

@@ -0,0 +1,186 @@
+# Copyright 2020 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# Lint as: python3
+"""Unit tests for make_corpus_dir.py."""
+
+import os
+import subprocess
+import sys
+import tempfile
+import unittest
+from pathlib import Path
+
+
+def resolve_script_path():
+    candidates = [Path(__file__).with_name("make_corpus_dir.py")]
+    test_workspace = os.environ.get("TEST_WORKSPACE")
+    manifest_lookup_path = "fuzzing/tools/make_corpus_dir.py"
+    if test_workspace:
+        test_srcdir = os.environ.get("TEST_SRCDIR")
+        if test_srcdir:
+            candidates.append(
+                Path(test_srcdir) / test_workspace / "fuzzing" / "tools" /
+                "make_corpus_dir.py")
+        runfiles_dir = os.environ.get("RUNFILES_DIR")
+        if runfiles_dir:
+            candidates.append(
+                Path(runfiles_dir) / test_workspace / "fuzzing" / "tools" /
+                "make_corpus_dir.py")
+        manifest_lookup_path = (
+            f"{test_workspace}/fuzzing/tools/make_corpus_dir.py")
+
+    for candidate in candidates:
+        if candidate.is_file():
+            return candidate
+
+    manifest_file = os.environ.get("RUNFILES_MANIFEST_FILE")
+    if manifest_file:
+        try:
+            workspace_match = None
+            main_match = None
+            with open(manifest_file, "r", encoding="utf-8") as manifest:
+                for line in manifest:
+                    entry = line.rstrip("\n")
+                    if not entry:
+                        continue
+                    logical_path, separator, real_path = entry.partition(" ")
+                    if not separator:
+                        continue
+                    normalized_path = logical_path.replace("\\", "/")
+                    if not normalized_path.endswith("fuzzing/tools/make_corpus_dir.py"):
+                        continue
+                    candidate = Path(real_path)
+                    if not candidate.is_file():
+                        continue
+                    if test_workspace and normalized_path.startswith(f"{test_workspace}/"):
+                        workspace_match = candidate
+                        break
+                    if normalized_path.startswith("_main/") and not main_match:
+                        main_match = candidate
+            if workspace_match:
+                return workspace_match
+            if main_match:
+                return main_match
+        except OSError:
+            pass
+
+    raise FileNotFoundError("could not resolve make_corpus_dir.py in test runfiles")
+
+
+SCRIPT_PATH = resolve_script_path()
+
+
+class MakeCorpusDirTest(unittest.TestCase):
+
+    def run_tool(self, args, cwd):
+        return subprocess.run(
+            [sys.executable, str(SCRIPT_PATH)] + args,
+            cwd=str(cwd),
+            text=True,
+            capture_output=True,
+            check=False,
+        )
+
+    def test_copies_nested_corpus_directory(self):
+        with tempfile.TemporaryDirectory() as td:
+            tmp = Path(td)
+            corpus = tmp / "corpus"
+            (corpus / "nested").mkdir(parents=True)
+            (corpus / "a.txt").write_text("A", encoding="utf-8")
+            (corpus / "nested" / "b.txt").write_text("B", encoding="utf-8")
+            output_dir = tmp / "out"
+
+            result = self.run_tool(
+                ["--corpus_list=corpus", "--output_dir=out"], cwd=tmp)
+
+            self.assertEqual(result.returncode, 0, msg=result.stderr)
+            copied_files = [path for path in output_dir.iterdir() if path.is_file()]
+            self.assertEqual(len(copied_files), 2)
+            copied_contents = sorted(path.read_text(encoding="utf-8")
+                                     for path in copied_files)
+            self.assertEqual(copied_contents, ["A", "B"])
+
+    def test_copies_absolute_corpus_file(self):
+        with tempfile.TemporaryDirectory() as td:
+            tmp = Path(td)
+            corpus_file = tmp / "corpus-input.txt"
+            corpus_file.write_text("payload", encoding="utf-8")
+            output_dir = tmp / "out"
+
+            result = self.run_tool(
+                [f"--corpus_list={corpus_file}", f"--output_dir={output_dir}"],
+                cwd=tmp,
+            )
+
+            self.assertEqual(result.returncode, 0, msg=result.stderr)
+            copied_files = [path for path in output_dir.iterdir() if path.is_file()]
+            self.assertEqual(len(copied_files), 1)
+            self.assertEqual(copied_files[0].read_text(encoding="utf-8"), "payload")
+
+    def test_distinguishes_dot_prefix_from_plain_relative_path(self):
+        with tempfile.TemporaryDirectory() as td:
+            tmp = Path(td)
+            corpus_file = tmp / "a.txt"
+            corpus_file.write_text("payload", encoding="utf-8")
+            output_dir = tmp / "out"
+
+            result = self.run_tool(
+                ["--corpus_list=./a.txt,a.txt", "--output_dir=out"],
+                cwd=tmp,
+            )
+
+            self.assertEqual(result.returncode, 0, msg=result.stderr)
+            copied_files = [path for path in output_dir.iterdir() if path.is_file()]
+            self.assertEqual(len(copied_files), 2)
+
+    def test_distinguishes_parent_navigation_from_plain_relative_path(self):
+        with tempfile.TemporaryDirectory() as td:
+            tmp = Path(td)
+            (tmp / "dir").mkdir()
+            corpus_file = tmp / "a.txt"
+            corpus_file.write_text("payload", encoding="utf-8")
+            output_dir = tmp / "out"
+
+            result = self.run_tool(
+                ["--corpus_list=dir/../a.txt,a.txt", "--output_dir=out"],
+                cwd=tmp,
+            )
+
+            self.assertEqual(result.returncode, 0, msg=result.stderr)
+            copied_files = [path for path in output_dir.iterdir() if path.is_file()]
+            self.assertEqual(len(copied_files), 2)
+
+    def test_distinguishes_dot_prefix_from_literal_dot_filename(self):
+        with tempfile.TemporaryDirectory() as td:
+            tmp = Path(td)
+            (tmp / "a.txt").write_text("from-a", encoding="utf-8")
+            (tmp / "dot-a.txt").write_text("from-dot-a", encoding="utf-8")
+            output_dir = tmp / "out"
+
+            result = self.run_tool(
+                ["--corpus_list=./a.txt,dot-a.txt", "--output_dir=out"],
+                cwd=tmp,
+            )
+
+            self.assertEqual(result.returncode, 0, msg=result.stderr)
+            copied_files = [path for path in output_dir.iterdir() if path.is_file()]
+            self.assertEqual(len(copied_files), 2)
+            copied_contents = sorted(path.read_text(encoding="utf-8")
+                                     for path in copied_files)
+            self.assertEqual(copied_contents, ["from-a", "from-dot-a"])
+
+
+if __name__ == "__main__":
+    unittest.main()