Create a flyway index file and verify correctness (#819)

* Create a flyway index file and verify correctness

Create an index file (flyway.txt) containing the names of all of the flyway
files and verify that it is ordered and in sync with the actual contents of
the flyway directory.  Also provide a target (generateFlywayIndex) to
automatically generate it.

The purpose of flyway.txt is to cause a merge conflict in the event that two
different developers add a flyway file with the same sequence number, an event
which has occurred multiple times.

config/presubmits.py

@@ -18,6 +18,7 @@ Error Prone) so we must write them manually.
 """
 
 import os
+from typing import List, Tuple
 import sys
 import re
 
@@ -178,6 +179,90 @@ PRESUBMITS = {
         "JavaScript files should not include console logging."
 }
 
+# Note that this regex only works for one kind of Flyway file.  If we want to
+# start using "R" and "U" files we'll need to update this script.
+FLYWAY_FILE_RX = re.compile(r'V(\d+)__.*')
+
+
+def get_seqnum(filename: str, location: str) -> int:
+  """Extracts the sequence number from a filename."""
+  m = FLYWAY_FILE_RX.match(filename)
+  if m is None:
+    raise ValueError('Illegal Flyway filename: %s in %s' % (filename, location))
+  return int(m.group(1))
+
+
+def files_by_seqnum(files: List[str], location: str) -> List[Tuple[int, str]]:
+  """Returns the list of seqnum, filename sorted by sequence number."""
+  return [(get_seqnum(filename, location), filename) for filename in files]
+
+
+def has_valid_order(indexed_files: List[Tuple[int, str]], location: str) -> bool:
+  """Verify that sequence numbers are in order without gaps or duplicates.
+
+  Args:
+    files: List of seqnum, filename for a list of Flyway files.
+    location: Where the list of files came from (for error reporting).
+
+  Returns:
+    True if the file list is valid.
+  """
+  last_index = 0
+  valid = True
+  for seqnum, filename in indexed_files:
+    if seqnum == last_index:
+      print('duplicate Flyway file sequence number found in %s: %s' %
+            (location, filename))
+      valid = False
+    elif seqnum < last_index:
+      print('File %s in %s is out of order.' % (filename, location))
+      valid = False
+    elif seqnum != last_index + 1:
+      print('Missing Flyway sequence number %d in %s.  Next file is %s' %
+            (last_index + 1, location, filename))
+      valid = False
+    last_index = seqnum
+  return valid
+
+
+def verify_flyway_index():
+  """Verifies that the Flyway index file is in sync with the directory."""
+  success = True
+
+  # Sort the files in the Flyway directory by their sequence number.
+  files = sorted(
+      files_by_seqnum(os.listdir('db/src/main/resources/sql/flyway'),
+                      'Flyway directory'))
+
+  # Make sure that there are no gaps and no duplicate sequence numbers in the
+  # files themselves.
+  if not has_valid_order(files, 'Flyway directory'):
+    success = False
+
+  # Remove the sequence numbers and compare against the index file contents.
+  files = [filename[1] for filename in sorted(files)]
+  with open('db/src/main/resources/sql/flyway.txt') as index:
+    indexed_files = index.read().splitlines()
+  if files != indexed_files:
+    unindexed = set(files) - set(indexed_files)
+    if unindexed:
+      print('The following Flyway files are not in flyway.txt: %s' % unindexed)
+
+    nonexistent = set(indexed_files) - set(files)
+    if nonexistent:
+      print('The following files are in flyway.txt but not in the Flyway '
+            'directory: %s' % nonexistent)
+
+    # Do an ordering check on the index file (ignore the result, we're failing
+    # anyway).
+    has_valid_order(files_by_seqnum(indexed_files, 'flyway.txt'), 'flyway.txt')
+    success = False
+
+  if not success:
+    print('Please fix any conflicts and run "./nom_build :db:generateFlywayIndex"')
+
+  return not success
+
 
 def get_files():
   for root, dirnames, filenames in os.walk("."):
@@ -197,5 +282,10 @@ if __name__ == "__main__":
       failed = True
       print("%s had errors: \n  %s" % (file, "\n  ".join(error_messages)))
 
+  # And now for something completely different: check to see if the Flyway
+  # index is up-to-date.  It's quicker to do it here than in the unit tests:
+  # when we put it here it fails fast before all of the tests are run.
+  failed |= verify_flyway_index()
+
   if failed:
     sys.exit(1)