More metadata

codearchiver/core.py

@@ -84,7 +84,13 @@ class IndexField:
 class Index(list[tuple[str, str]]):
 	'''An index (key-value mapping, possibly with repeated keys) of a file produced by a module'''
 
-	fields: tuple[IndexField] = ()
+	fields: tuple[IndexField] = (
+		IndexField('codearchiver version', required = True, repeatable = False),
+		IndexField('Module', required = True, repeatable = False),
+		IndexField('ID', required = True, repeatable = False),
+		IndexField('Input URL', required = True, repeatable = False),
+		IndexField('Filename', required = True, repeatable = False),
+	)
 	'''The fields for this index'''
 
 	_allFieldsCache: typing.Optional[tuple[IndexField]] = None
@@ -348,6 +354,9 @@ class Module(metaclass = ModuleMeta):
 	name: typing.Optional[str] = None
 	'''The name of the module. Modules without a name are ignored. Names must be unique and may only contain a-z and hyphens.'''
 
+	IndexClass: typing.Optional[typing.Type[Index]] = None
+	'''The Index class corresponding to this module, if any.'''
+
 	@staticmethod
 	def matches(inputUrl: InputURL) -> bool:
 		'''Whether or not this module is for handling `inputUrl`.'''
@@ -364,6 +373,19 @@ class Module(metaclass = ModuleMeta):
 	def process(self) -> Result:
 		'''Perform the relevant retrieval(s)'''
 
+	def create_index(self, filename: str) -> Index:
+		'''Create a basic Index instance appropriate for this module'''
+
+		if type(self).IndexClass is None or type(self).name is None:
+			raise RuntimeError('Module lacks an IndexClass or a name; cannot create index')
+		idx = type(self).IndexClass()
+		idx.append('codearchiver version', codearchiver.version.__version__)
+		idx.append('Module', type(self).name)
+		idx.append('ID', self._id)
+		idx.append('Input URL', self._url)
+		idx.append('Filename', filename)
+		return idx
+
 	def __repr__(self):
 		return f'{type(self).__module__}.{type(self).__name__}({self._inputUrl!r})'
 

codearchiver/modules/git.py

@@ -14,6 +14,7 @@ _logger = logging.getLogger(__name__)
 
 class GitIndex(codearchiver.core.Index):
 	fields = (
+		codearchiver.core.IndexField(key = 'Git version', required = True, repeatable = False),
 		codearchiver.core.IndexField(key = 'Based on bundle', required = False, repeatable = True),
 		codearchiver.core.IndexField(key = 'Ref', required = True, repeatable = True),
 		codearchiver.core.IndexField(key = 'Root commit', required = True, repeatable = True),
@@ -23,6 +24,7 @@ class GitIndex(codearchiver.core.Index):
 
 class Git(codearchiver.core.Module):
 	name = 'git'
+	IndexClass = GitIndex
 
 	@staticmethod
 	def matches(inputUrl):
@@ -45,6 +47,11 @@ class Git(codearchiver.core.Module):
 			_logger.fatal(f'{bundle!r} already exists')
 			raise FileExistsError(f'{bundle!r} already exists')
 
+		_, gitVersion, _ = codearchiver.subprocess.run_with_log(['git', '--version'])
+		if not gitVersion.startswith('git version ') or not gitVersion.endswith('\n') or gitVersion[12:-1].strip('0123456789.') != '':
+			raise RuntimeError(f'Unexpected output from `git --version`: {gitVersion!r}')
+		gitVersion = gitVersion[12:-1]
+
 		_logger.info(f'Cloning {self._url} into {directory}')
 		codearchiver.subprocess.run_with_log(['git', 'clone', '--verbose', '--progress', '--mirror', self._url, directory], env = {**os.environ, 'GIT_TERMINAL_PROMPT': '0'})
 
@@ -103,7 +110,8 @@ class Git(codearchiver.core.Module):
 		_logger.info(f'Removing clone')
 		shutil.rmtree(directory)
 
-		index = GitIndex()
+		index = self.create_index(bundle)
+		index.append('Git version', gitVersion)
 		for oldBundle in basedOnBundles:
 			index.append('Based on bundle', oldBundle)
 		for line in refs.splitlines():