From 4b94b7d4776cc9625af218bcb0ebe3fe278f03eb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Mike=20F=C3=A4hrmann?= <mike_faehrmann@web.de>
Date: Mon, 19 Aug 2024 21:18:39 +0200
Subject: [PATCH] [pp:metadata] add 'include' and 'exclude' options (#6058)

---
 docs/configuration.rst               | 24 ++++++++++++++++++++++
 gallery_dl/postprocessor/metadata.py | 29 ++++++++++++++++++++++++---
 test/test_postprocessor.py           | 30 ++++++++++++++++++++++++++++
 3 files changed, 80 insertions(+), 3 deletions(-)

diff --git a/docs/configuration.rst b/docs/configuration.rst
index 33141a30..208e4028 100644
--- a/docs/configuration.rst
+++ b/docs/configuration.rst
@@ -5692,6 +5692,30 @@ Description
         After downloading all files of a `post`
 
 
+metadata.include
+----------------
+Type
+    ``list`` of ``strings``
+Example
+    ``["id", "width", "height", "description"]``
+Description
+    Include only the given top-level keys when writing JSON data.
+
+    Note: Missing or undefined fields will be silently ignored.
+
+
+metadata.exclude
+----------------
+Type
+    ``list`` of ``strings``
+Example
+    ``["blocked", "watching", "status"]``
+Description
+    Exclude all given keys from written JSON data.
+
+    Note: Cannot be used with `metadata.include`_.
+
+
 metadata.fields
 ---------------
 Type
diff --git a/gallery_dl/postprocessor/metadata.py b/gallery_dl/postprocessor/metadata.py
index e89b1700..08a08f11 100644
--- a/gallery_dl/postprocessor/metadata.py
+++ b/gallery_dl/postprocessor/metadata.py
@@ -103,10 +103,10 @@ class MetadataPP(PostProcessor):
         job.register_hooks({event: self.run for event in events}, options)
 
         self._init_archive(job, options, "_MD_")
+        self.filter = self._make_filter(options)
         self.mtime = options.get("mtime")
         self.omode = options.get("open", omode)
         self.encoding = options.get("encoding", "utf-8")
-        self.private = options.get("private", False)
         self.skip = options.get("skip", False)
 
     def run(self, pathfmt):
@@ -231,10 +231,33 @@ class MetadataPP(PostProcessor):
         fp.write("\n".join(tags) + "\n")
 
     def _write_json(self, fp, kwdict):
-        if not self.private:
-            kwdict = util.filter_dict(kwdict)
+        if self.filter:
+            kwdict = self.filter(kwdict)
         fp.write(self._json_encode(kwdict) + "\n")
 
+    def _make_filter(self, options):
+        include = options.get("include")
+        if include:
+            if isinstance(include, str):
+                include = include.split(",")
+            return lambda d: {k: d[k] for k in include if k in d}
+
+        exclude = options.get("exclude")
+        private = options.get("private")
+        if exclude:
+            if isinstance(exclude, str):
+                exclude = exclude.split(",")
+            exclude = set(exclude)
+
+            if private:
+                return lambda d: {k: v for k, v in d.items()
+                                  if k not in exclude}
+            return lambda d: {k: v for k, v in util.filter_dict(d).items()
+                              if k not in exclude}
+
+        if not private:
+            return util.filter_dict
+
     @staticmethod
     def _make_encoder(options, indent=None):
         return json.JSONEncoder(
diff --git a/test/test_postprocessor.py b/test/test_postprocessor.py
index edd85756..841ebf0a 100644
--- a/test/test_postprocessor.py
+++ b/test/test_postprocessor.py
@@ -585,6 +585,36 @@ class MetadataTest(BasePostprocessorTest):
         self.assertTrue(not e.called)
         self.assertTrue(m.called)
 
+    def test_metadata_option_include(self):
+        self._create(
+            {"include": ["_private", "filename", "foo"], "sort": True},
+            {"public": "hello ワールド", "_private": "foo バー"},
+        )
+
+        with patch("builtins.open", mock_open()) as m:
+            self._trigger()
+
+        self.assertEqual(self._output(m), """{
+    "_private": "foo バー",
+    "filename": "file"
+}
+""")
+
+    def test_metadata_option_exclude(self):
+        self._create(
+            {"exclude": ["category", "filename", "foo"], "sort": True},
+            {"public": "hello ワールド", "_private": "foo バー"},
+        )
+
+        with patch("builtins.open", mock_open()) as m:
+            self._trigger()
+
+        self.assertEqual(self._output(m), """{
+    "extension": "ext",
+    "public": "hello ワールド"
+}
+""")
+
     @staticmethod
     def _output(mock):
         return "".join(