From a01c4bba50cfee5bcd799481ed4e43977c3410da Mon Sep 17 00:00:00 2001
From: Rhet Turnbull <rturnbull@gmail.com>
Date: Sat, 16 Jul 2022 18:33:16 -0400
Subject: [PATCH] Fixed backup database to be conformant JSON, #56

---
 dev_requirements.txt       |  4 +--
 osxmetadata/_version.py    |  2 +-
 osxmetadata/backup.py      | 66 ++++++++++++++++++++++++++------------
 osxmetadata/classes.py     |  2 +-
 osxmetadata/osxmetadata.py |  6 ++--
 5 files changed, 52 insertions(+), 28 deletions(-)

diff --git a/dev_requirements.txt b/dev_requirements.txt
index 1af2b79..4a7e571 100644
--- a/dev_requirements.txt
+++ b/dev_requirements.txt
@@ -1,2 +1,2 @@
-pytest==6.2.4
-pyinstaller==4.3
+pytest==7.1.2
+pyinstaller==5.2
diff --git a/osxmetadata/_version.py b/osxmetadata/_version.py
index dbaedb4..475805c 100644
--- a/osxmetadata/_version.py
+++ b/osxmetadata/_version.py
@@ -1,3 +1,3 @@
 """ osxmetadata version """
 
-__version__ = "0.99.37"
+__version__ = "0.99.38"
diff --git a/osxmetadata/backup.py b/osxmetadata/backup.py
index 3d95930..270e795 100644
--- a/osxmetadata/backup.py
+++ b/osxmetadata/backup.py
@@ -1,39 +1,63 @@
+""" Functions for writing and loading backup files """
+
 import json
 import logging
 import os
+from enum import Enum
 
-""" Functions for writing and loading backup files """
+
+class BackupDatabaseType(Enum):
+    SINGLE_RECORD_JSON = 1
+    JSON = 2
+
+
+def backup_database_type(path: str) -> BackupDatabaseType:
+    """Return BackupDatabaseType enum indicating type of backup file for path"""
+    with open(path) as fp:
+        line = fp.readline()
+        if not line:
+            raise ValueError("Unknown backup file type")
+        ch = line[0]
+        if ch == "{":
+            return BackupDatabaseType.SINGLE_RECORD_JSON
+        elif ch == "[":
+            return BackupDatabaseType.JSON
+        else:
+            raise ValueError("Unknown backup file type")
 
 
 def write_backup_file(backup_file, backup_data):
-    """ Write backup_data to backup_file as JSON
-        backup_data: dict where key is filename and value is dict of the attributes
-        as returned by json.loads(OSXMetaData.to_json()) """
+    """Write backup_data to backup_file as JSON
+    backup_data: dict where key is filename and value is dict of the attributes
+    as returned by json.loads(OSXMetaData.to_json())"""
 
     with open(backup_file, mode="w") as fp:
-        for record in backup_data.values():
-            json_str = json.dumps(record)
-            fp.write(json_str)
-            fp.write("\n")
+        json.dump(list(backup_data.values()), fp)
 
 
 def load_backup_file(backup_file):
-    """ Load attribute data from JSON in backup_file 
-        Returns: backup_data dict """
+    """Load attribute data from JSON in backup_file
+    Returns: backup_data dict"""
 
     if not os.path.isfile(backup_file):
         raise FileNotFoundError(f"Could not find backup file: {backup_file}")
 
-    backup_data = {}
-    with open(backup_file, mode="r") as fp:
-        for line in fp:
-            data = json.loads(line)
-            fname = data["_filename"]
-            if fname in backup_data:
-                logging.warning(
-                    f"WARNING: duplicate filename {fname} found in {backup_file}"
-                )
-
-            backup_data[fname] = data
+    if backup_database_type(backup_file) == BackupDatabaseType.SINGLE_RECORD_JSON:
+        # old style single record of json per file
+        backup_data = {}
+        with open(backup_file, mode="r") as fp:
+            for line in fp:
+                data = json.loads(line)
+                fname = data["_filename"]
+                if fname in backup_data:
+                    logging.warning(
+                        f"WARNING: duplicate filename {fname} found in {backup_file}"
+                    )
+
+                backup_data[fname] = data
+    else:
+        with open(backup_file, mode="r") as fp:
+            backup_records = json.load(fp)
+        backup_data = {data["_filename"]: data for data in backup_records}
 
     return backup_data
diff --git a/osxmetadata/classes.py b/osxmetadata/classes.py
index d8c71ea..835e3b5 100644
--- a/osxmetadata/classes.py
+++ b/osxmetadata/classes.py
@@ -557,4 +557,4 @@ def set_value(self, value):
 
     def __len__(self):
         self._load_data()
-        return self.data.__len__() if self.data else 0
\ No newline at end of file
+        return self.data.__len__() if self.data else 0
diff --git a/osxmetadata/osxmetadata.py b/osxmetadata/osxmetadata.py
index 031779d..2b3b853 100644
--- a/osxmetadata/osxmetadata.py
+++ b/osxmetadata/osxmetadata.py
@@ -18,7 +18,7 @@
 
 from ._version import __version__
 from .attributes import ATTRIBUTES, Attribute, validate_attribute_value
-from .classes import _AttributeOSXPhotosDetectedText, _AttributeList, _AttributeTagsList
+from .classes import _AttributeList, _AttributeOSXPhotosDetectedText, _AttributeTagsList
 from .constants import (
     _FINDER_COMMENT_NAMES,
     FINDER_COLOR_NONE,
@@ -297,12 +297,12 @@ def get_attribute_str(self, attribute_name):
         e.g. if attribute is a datedate.datetime object, will
         format using datetime.isoformat()
         attribute_name: name of attribute"""
-        
+
         attribute = ATTRIBUTES[attribute_name]
         value = self.get_attribute(attribute_name)
         if attribute.class_ == _AttributeOSXPhotosDetectedText:
             return json.dumps(value)
-        
+
         if type(value) in [list, set]:
             if value:
                 if type(value[0]) == datetime.datetime: