dump_things_service/audit/gitaudit.py

@@ -7,8 +7,8 @@ committed.
 Changes are annotated with a time stamp and a user-id
 """
 import hashlib
+import json
 import re
-import string
 from datetime import datetime
 from pathlib import Path
 
@@ -23,6 +23,9 @@ from datalad_core.runners import (
 from . import AuditBackend
 
 
+index_file_name = 'gitaudit_index.log'
+
+
 class GitAuditBackend(AuditBackend):
 
     def __init__(
@@ -42,8 +45,6 @@ class GitAuditBackend(AuditBackend):
             author_id: str | None = None,
     ) -> None:
         author_id = committer_id if author_id is None else author_id
-        committer_id = self._escape_person_id(committer_id)
-        author_id = self._escape_person_id(author_id)
         record_id = record['pid']
         location = self._get_location_for(record_id)
         if self._has_pending_changes(location):
@@ -80,13 +81,13 @@ class GitAuditBackend(AuditBackend):
                 capture_output=True,
             ).decode().splitlines()
             # Get the log entry
-            log_entry = tuple(
+            log_line = tuple(
                 filter(
                     lambda l: not l.startswith('+++') and l.startswith('+'),
                     log_diff_lines,
                 )
             )[0][1:]
-            time_stamp, committer_id, author_id = log_entry.split(' ')
+            log_entry = json.loads(log_line)
 
             # Get the YAML diff
             yaml_diff_lines = call_git(
@@ -103,7 +104,13 @@ class GitAuditBackend(AuditBackend):
                 capture_output=True,
             ).decode()
             changes.append(
-                (time_stamp, committer_id, author_id, yaml_diff, yaml_content)
+                (
+                    log_entry['time_stamp'],
+                    log_entry['committer_id'],
+                    log_entry['author_id'],
+                    yaml_diff,
+                    yaml_content,
+                )
             )
 
         changes.sort()
@@ -154,8 +161,13 @@ class GitAuditBackend(AuditBackend):
             author_id: str,
     ) -> None:
         time_stamp = datetime.now().isoformat()
+        entry = {
+            'time_stamp': time_stamp,
+            'committer_id': committer_id,
+            'author_id': author_id,
+        }
         log_content = self._read_from_repo_path(log_location).decode()
-        log_content += f'{time_stamp} {committer_id} {author_id}\n'
+        log_content += json.dumps(entry, ensure_ascii=False) + '\n'
         self.current_change_set[log_location] = log_content
 
     def _add_index_entry(
@@ -229,7 +241,7 @@ class GitAuditBackend(AuditBackend):
         else:
             self.path.mkdir(parents=True)
             is_empty = True
-        self.index_path = self.path / 'index.log'
+        self.index_path = self.path / index_file_name
 
         if is_empty:
             call_git(['init', '--bare', str(self.path)], capture_output=True)
@@ -258,7 +270,6 @@ class GitAuditBackend(AuditBackend):
         self.index.add(record_id)
 
     def _rebuild_index(self):
-        print('rebuilding index')
         tree_entries = call_git(
             ['ls-tree', '-r', 'master:'],
             cwd=self.path,
@@ -267,10 +278,8 @@ class GitAuditBackend(AuditBackend):
         with open(self.index_path, 'wt') as f:
             for line in tree_entries:
                 if not line.endswith('.yaml'):
-                    print('ignoring line:', repr(line))
                     continue
-                print('got line:', repr(line))
+                flag, object_type, object_hash, file_name = line.split(maxsplit=3)
-                flag, object_type, object_hash, file_name = line.split()
                 record = yaml.safe_load(
                     call_git(
                         ['show', object_hash],
@@ -278,22 +287,4 @@ class GitAuditBackend(AuditBackend):
                         capture_output=True,
                     ).decode()
                 )
-                print('got record:', repr(record))
                 f.write(record['pid'] + '\n')
-
-    def _escape_person_id(
-            self,
-            person_id: str,
-    ):
-        if not person_id:
-            msg = f'empty ID string not allowed: {person_id}'
-            raise ValueError(msg)
-        if any(
-                map(
-                    lambda character: character in person_id,
-                    string.whitespace
-                )
-        ):
-            msg = f'ID string must not contain whitespace: {person_id}'
-            raise ValueError(msg)
-        return person_id