From 2f1a07abbf3a5ab3cba763f8c05dc51bd8d5bd66 Mon Sep 17 00:00:00 2001
From: msramalho <19508417+msramalho@users.noreply.github.com>
Date: Tue, 17 Jun 2025 16:06:04 +0100
Subject: [PATCH] renaming and code improvements to json_e richer

---
 .../modules/json_enricher/__init__.py         |  1 +
 .../modules/json_enricher/__manifest__.py     | 16 ++++++++
 .../json_enricher.py}                         | 14 +++----
 .../metadata_json_enricher/__init__.py        |  1 -
 .../metadata_json_enricher/__manifest__.py    | 37 -------------------
 5 files changed, 23 insertions(+), 46 deletions(-)
 create mode 100644 src/auto_archiver/modules/json_enricher/__init__.py
 create mode 100644 src/auto_archiver/modules/json_enricher/__manifest__.py
 rename src/auto_archiver/modules/{metadata_json_enricher/metadata_json_enricher.py => json_enricher/json_enricher.py} (63%)
 delete mode 100644 src/auto_archiver/modules/metadata_json_enricher/__init__.py
 delete mode 100644 src/auto_archiver/modules/metadata_json_enricher/__manifest__.py

diff --git a/src/auto_archiver/modules/json_enricher/__init__.py b/src/auto_archiver/modules/json_enricher/__init__.py
new file mode 100644
index 0000000..5e0a04f
--- /dev/null
+++ b/src/auto_archiver/modules/json_enricher/__init__.py
@@ -0,0 +1 @@
+from .json_enricher import JsonEnricher
diff --git a/src/auto_archiver/modules/json_enricher/__manifest__.py b/src/auto_archiver/modules/json_enricher/__manifest__.py
new file mode 100644
index 0000000..b80f493
--- /dev/null
+++ b/src/auto_archiver/modules/json_enricher/__manifest__.py
@@ -0,0 +1,16 @@
+{
+    "name": "JSON Enricher",
+    "type": ["enricher"],
+    "requires_setup": True,
+    "dependencies": {
+        "python": ["loguru"],
+    },
+    "configs": {},
+    "description": """
+
+    Writes all archiving process metadata to a JSON file so it can be parsed by other tools. As this is an Enricher, it will not contain the final stored URLs. 
+	
+	WARNING: The resulting JSON may reveal sensitive information about the computer and settings in which the archiving process was run. 
+
+    """,
+}
diff --git a/src/auto_archiver/modules/metadata_json_enricher/metadata_json_enricher.py b/src/auto_archiver/modules/json_enricher/json_enricher.py
similarity index 63%
rename from src/auto_archiver/modules/metadata_json_enricher/metadata_json_enricher.py
rename to src/auto_archiver/modules/json_enricher/json_enricher.py
index 312f922..b0900b6 100644
--- a/src/auto_archiver/modules/metadata_json_enricher/metadata_json_enricher.py
+++ b/src/auto_archiver/modules/json_enricher/json_enricher.py
@@ -5,17 +5,15 @@ import os
 from auto_archiver.core import Enricher
 from auto_archiver.core import Media, Metadata
 
-class MetadataJsonEnricher(Enricher):
-    def __init__(self):
-        super().__init__()
 
+class JsonEnricher(Enricher):
     def enrich(self, to_enrich: Metadata) -> None:
         url = to_enrich.get_url()
 
-        logger.debug(f"Metadata JSON Enricher for {url=}")
+        logger.debug(f"JSON Enricher for {url=}")
 
-        item_path = os.path.join(self.tmp_dir, f"metadata.json")
+        item_path = os.path.join(self.tmp_dir, "metadata.json")
         with open(item_path, mode="w", encoding="utf-8") as outf:
-            json.dump(to_enrich.to_dict(), outf, indent=4, default=str)
-        
-        to_enrich.add_media(Media(filename=item_path), id="metadata_json")
\ No newline at end of file
+            json.dump(to_enrich.to_dict(), outf, indent=4, default=str, ensure_ascii=False)
+
+        to_enrich.add_media(Media(filename=item_path), id="metadata_json")
diff --git a/src/auto_archiver/modules/metadata_json_enricher/__init__.py b/src/auto_archiver/modules/metadata_json_enricher/__init__.py
deleted file mode 100644
index 4eed90b..0000000
--- a/src/auto_archiver/modules/metadata_json_enricher/__init__.py
+++ /dev/null
@@ -1 +0,0 @@
-from .metadata_json_enricher import MetadataJsonEnricher
\ No newline at end of file
diff --git a/src/auto_archiver/modules/metadata_json_enricher/__manifest__.py b/src/auto_archiver/modules/metadata_json_enricher/__manifest__.py
deleted file mode 100644
index b737b16..0000000
--- a/src/auto_archiver/modules/metadata_json_enricher/__manifest__.py
+++ /dev/null
@@ -1,37 +0,0 @@
-{
-    "name": "Metadata JSON Enricher",
-    "type": ["enricher"],
-    "requires_setup": True,
-    "dependencies": {
-        "python": ["loguru"],
-    },
-    "configs": {
-        # "width": {"default": 1280, "type": "int", "help": "width of the screenshots"},
-        # "height": {"default": 1024, "type": "int", "help": "height of the screenshots"},
-        # "timeout": {"default": 60, "type": "int", "help": "timeout for taking the screenshot"},
-        # "sleep_before_screenshot": {
-        #     "default": 4,
-        #     "type": "int",
-        #     "help": "seconds to wait for the pages to load before taking screenshot",
-        # },
-        # "http_proxy": {
-        #     "default": "",
-        #     "help": "http proxy to use for the webdriver, eg http://proxy-user:password@proxy-ip:port",
-        # },
-        # "save_to_pdf": {
-        #     "default": False,
-        #     "type": "bool",
-        #     "help": "save the page as pdf along with the screenshot. PDF saving options can be adjusted with the 'print_options' parameter",
-        # },
-        # "print_options": {
-        #     "default": {},
-        #     "help": "options to pass to the pdf printer, in JSON format. See https://www.selenium.dev/documentation/webdriver/interactions/print_page/ for more information",
-        #     "type": "json_loader",
-        # },
-    },
-    "description": """
-
-    Writes all the metadata to a json file so can be parsed by other tools.
-
-    """,
-}