Merge pull request #3320 from gilles-peskine-arm/check-files-changelog-development

Check changelog entries on CI

ChangeLog.d/fix-gcc-format-signedness-warnings.txt

@@ -1,3 +1,4 @@
 Changes
    * Fix warnings about signedness issues in format strings. The build is now
-     clean of -Wformat-signedness warnings. Contributed by Kenneth Soerensen in #3153.
+     clean of -Wformat-signedness warnings. Contributed by Kenneth Soerensen
+     in #3153.

tests/scripts/all.sh

@@ -625,6 +625,18 @@ component_check_files () {
     record_status tests/scripts/check-files.py
 }
 
+component_check_changelog () {
+    msg "Check: changelog entries" # < 1s
+    rm -f ChangeLog.new
+    record_status scripts/assemble_changelog.py -o ChangeLog.new
+    if [ -e ChangeLog.new ]; then
+        # Show the diff for information. It isn't an error if the diff is
+        # non-empty.
+        diff -u ChangeLog ChangeLog.new || true
+        rm ChangeLog.new
+    fi
+}
+
 component_check_names () {
     msg "Check: declared and exported names (builds the library)" # < 3s
     record_status tests/scripts/check-names.sh -v

tests/scripts/check-files.py

@@ -14,6 +14,8 @@ import os
 import argparse
 import logging
 import codecs
+import re
+import subprocess
 import sys
 
 
@@ -23,28 +25,48 @@ class FileIssueTracker:
     To implement a checker that processes a file as a whole, inherit from
     this class and implement `check_file_for_issue` and define ``heading``.
 
-    ``files_exemptions``: files whose name ends with a string in this set
+    ``suffix_exemptions``: files whose name ends with a string in this set
      will not be checked.
 
+    ``path_exemptions``: files whose path (relative to the root of the source
+    tree) matches this regular expression will not be checked. This can be
+    ``None`` to match no path. Paths are normalized and converted to ``/``
+    separators before matching.
+
     ``heading``: human-readable description of the issue
     """
 
-    files_exemptions = frozenset()
+    suffix_exemptions = frozenset()
+    path_exemptions = None
     # heading must be defined in derived classes.
     # pylint: disable=no-member
 
     def __init__(self):
         self.files_with_issues = {}
 
+    @staticmethod
+    def normalize_path(filepath):
+        """Normalize ``filepath`` with / as the directory separator."""
+        filepath = os.path.normpath(filepath)
+        # On Windows, we may have backslashes to separate directories.
+        # We need slashes to match exemption lists.
+        seps = os.path.sep
+        if os.path.altsep is not None:
+            seps += os.path.altsep
+        return '/'.join(filepath.split(seps))
+
     def should_check_file(self, filepath):
         """Whether the given file name should be checked.
 
-        Files whose name ends with a string listed in ``self.files_exemptions``
+        Files whose name ends with a string listed in ``self.suffix_exemptions``
-        will not be checked.
+        or whose path matches ``self.path_exemptions`` will not be checked.
         """
-        for files_exemption in self.files_exemptions:
+        for files_exemption in self.suffix_exemptions:
             if filepath.endswith(files_exemption):
                 return False
+        if self.path_exemptions and \
+           re.match(self.path_exemptions, self.normalize_path(filepath)):
+            return False
         return True
 
     def check_file_for_issue(self, filepath):
@@ -73,6 +95,17 @@ class FileIssueTracker:
                     logger.info(filename)
             logger.info("")
 
+BINARY_FILE_PATH_RE_LIST = [
+    r'docs/.*\.pdf\Z',
+    r'programs/fuzz/corpuses/[^.]+\Z',
+    r'tests/data_files/[^.]+\Z',
+    r'tests/data_files/.*\.(crt|csr|db|der|key|pubkey)\Z',
+    r'tests/data_files/.*\.req\.[^/]+\Z',
+    r'tests/data_files/.*malformed[^/]+\Z',
+    r'tests/data_files/format_pkcs12\.fmt\Z',
+]
+BINARY_FILE_PATH_RE = re.compile('|'.join(BINARY_FILE_PATH_RE_LIST))
+
 class LineIssueTracker(FileIssueTracker):
     """Base class for line-by-line issue tracking.
 
@@ -80,6 +113,9 @@ class LineIssueTracker(FileIssueTracker):
     this class and implement `line_with_issue`.
     """
 
+    # Exclude binary files.
+    path_exemptions = BINARY_FILE_PATH_RE
+
     def issue_with_line(self, line, filepath):
         """Check the specified line for the issue that this class is for.
 
@@ -103,7 +139,7 @@ class LineIssueTracker(FileIssueTracker):
 
 def is_windows_file(filepath):
     _root, ext = os.path.splitext(filepath)
-    return ext in ('.bat', '.dsp', '.sln', '.vcxproj')
+    return ext in ('.bat', '.dsp', '.dsw', '.sln', '.vcxproj')
 
 
 class PermissionIssueTracker(FileIssueTracker):
@@ -126,9 +162,18 @@ class EndOfFileNewlineIssueTracker(FileIssueTracker):
 
     heading = "Missing newline at end of file:"
 
+    path_exemptions = BINARY_FILE_PATH_RE
+
     def check_file_for_issue(self, filepath):
         with open(filepath, "rb") as f:
-            if not f.read().endswith(b"\n"):
+            try:
+                f.seek(-1, 2)
+            except OSError:
+                # This script only works on regular files. If we can't seek
+                # 1 before the end, it means that this position is before
+                # the beginning of the file, i.e. that the file is empty.
+                return
+            if f.read(1) != b"\n":
                 self.files_with_issues[filepath] = None
 
 
@@ -138,7 +183,8 @@ class Utf8BomIssueTracker(FileIssueTracker):
 
     heading = "UTF-8 BOM present:"
 
-    files_exemptions = frozenset([".vcxproj", ".sln"])
+    suffix_exemptions = frozenset([".vcxproj", ".sln"])
+    path_exemptions = BINARY_FILE_PATH_RE
 
     def check_file_for_issue(self, filepath):
         with open(filepath, "rb") as f:
@@ -152,6 +198,8 @@ class UnixLineEndingIssueTracker(LineIssueTracker):
     heading = "Non-Unix line endings:"
 
     def should_check_file(self, filepath):
+        if not super().should_check_file(filepath):
+            return False
         return not is_windows_file(filepath)
 
     def issue_with_line(self, line, _filepath):
@@ -164,6 +212,8 @@ class WindowsLineEndingIssueTracker(LineIssueTracker):
     heading = "Non-Windows line endings:"
 
     def should_check_file(self, filepath):
+        if not super().should_check_file(filepath):
+            return False
         return is_windows_file(filepath)
 
     def issue_with_line(self, line, _filepath):
@@ -174,7 +224,7 @@ class TrailingWhitespaceIssueTracker(LineIssueTracker):
     """Track lines with trailing whitespace."""
 
     heading = "Trailing whitespace:"
-    files_exemptions = frozenset([".dsp", ".md"])
+    suffix_exemptions = frozenset([".dsp", ".md"])
 
     def issue_with_line(self, line, _filepath):
         return line.rstrip(b"\r\n") != line.rstrip()
@@ -184,7 +234,8 @@ class TabIssueTracker(LineIssueTracker):
     """Track lines with tabs."""
 
     heading = "Tabs present:"
-    files_exemptions = frozenset([
+    suffix_exemptions = frozenset([
+        ".pem", # some openssl dumps have tabs
         ".sln",
         "/Makefile",
         "/Makefile.inc",
@@ -223,32 +274,6 @@ class IntegrityChecker:
         self.check_repo_path()
         self.logger = None
         self.setup_logger(log_file)
-        self.extensions_to_check = (
-            ".bat",
-            ".c",
-            ".data",
-            ".dsp",
-            ".function",
-            ".h",
-            ".md",
-            ".pl",
-            ".py",
-            ".sh",
-            ".sln",
-            ".vcxproj",
-            "/CMakeLists.txt",
-            "/ChangeLog",
-            "/Makefile",
-            "/Makefile.inc",
-        )
-        self.excluded_directories = [
-            '.git',
-            'mbed-os',
-        ]
-        self.excluded_paths = list(map(os.path.normpath, [
-            'cov-int',
-            'examples',
-        ]))
         self.issues_to_check = [
             PermissionIssueTracker(),
             EndOfFileNewlineIssueTracker(),
@@ -275,21 +300,20 @@ class IntegrityChecker:
             console = logging.StreamHandler()
             self.logger.addHandler(console)
 
-    def prune_branch(self, root, d):
+    @staticmethod
-        if d in self.excluded_directories:
+    def collect_files():
-            return True
+        bytes_output = subprocess.check_output(['git', 'ls-files', '-z'])
-        if os.path.normpath(os.path.join(root, d)) in self.excluded_paths:
+        bytes_filepaths = bytes_output.split(b'\0')[:-1]
-            return True
+        ascii_filepaths = map(lambda fp: fp.decode('ascii'), bytes_filepaths)
-        return False
+        # Prepend './' to files in the top-level directory so that
+        # something like `'/Makefile' in fp` matches in the top-level
+        # directory as well as in subdirectories.
+        return [fp if os.path.dirname(fp) else os.path.join(os.curdir, fp)
+                for fp in ascii_filepaths]
 
     def check_files(self):
-        for root, dirs, files in os.walk("."):
-            dirs[:] = sorted(d for d in dirs if not self.prune_branch(root, d))
-            for filename in sorted(files):
-                filepath = os.path.join(root, filename)
-                if not filepath.endswith(self.extensions_to_check):
-                    continue
         for issue_to_check in self.issues_to_check:
+            for filepath in self.collect_files():
                 if issue_to_check.should_check_file(filepath):
                     issue_to_check.check_file_for_issue(filepath)