perf: 25% speed gain on json serialization through orjson...

docling_core/types/doc/document.py

@@ -4,6 +4,7 @@
 import copy
 import hashlib
 import json
+import orjson
 import logging
 import mimetypes
 import os
@@ -4860,7 +4861,10 @@ def save_as_json(
         coord_precision: Optional[int] = None,
         confid_precision: Optional[int] = None,
     ):
-        """Save as json."""
+        """
+        Save as json.
+        Indent can either be 2 or 0 (no indent).
+        """
         if isinstance(filename, str):
             filename = Path(filename)
         artifacts_dir, reference_path = self._get_output_paths(filename, artifacts_dir)
@@ -4875,8 +4879,8 @@ def save_as_json(
         out = new_doc.export_to_dict(
             coord_precision=coord_precision, confid_precision=confid_precision
         )
-        with open(filename, "w", encoding="utf-8") as fw:
-            json.dump(out, fw, indent=indent)
+        with open(filename, "wb") as fw:
+            fw.write(orjson.dumps(out, option=orjson.OPT_INDENT_2 if indent else 0))
 
     @classmethod
     def load_from_json(cls, filename: Union[str, Path]) -> "DoclingDocument":

pyproject.toml

@@ -50,6 +50,7 @@ dependencies = [
     'typing-extensions (>=4.12.2,<5.0.0)',
     'typer (>=0.12.5,<0.20.0)',
     'latex2mathml (>=3.77.0,<4.0.0)',
+    "orjson>=3.11.5",
 ]
 
 [project.urls]

test/data/doc/constructed_doc.embedded.json.gt

@@ -425,7 +425,7 @@
       "orig": "item of leading list",
       "text": "item of leading list",
       "enumerated": false,
-      "marker": "\u25a0"
+      "marker": "■"
     },
     {
       "self_ref": "#/texts/1",
@@ -507,7 +507,7 @@
       "orig": "list item 1",
       "text": "list item 1",
       "enumerated": false,
-      "marker": "\u25a0"
+      "marker": "■"
     },
     {
       "self_ref": "#/texts/7",
@@ -521,7 +521,7 @@
       "orig": "list item 2",
       "text": "list item 2",
       "enumerated": false,
-      "marker": "\u25a0"
+      "marker": "■"
     },
     {
       "self_ref": "#/texts/8",
@@ -539,7 +539,7 @@
       "orig": "list item 3",
       "text": "list item 3",
       "enumerated": false,
-      "marker": "\u25a0"
+      "marker": "■"
     },
     {
       "self_ref": "#/texts/9",
@@ -613,7 +613,7 @@
       "orig": "list item 4",
       "text": "list item 4",
       "enumerated": false,
-      "marker": "\u25a0"
+      "marker": "■"
     },
     {
       "self_ref": "#/texts/14",
@@ -663,7 +663,7 @@
       "orig": "item 1 of list",
       "text": "item 1 of list",
       "enumerated": false,
-      "marker": "\u25a0"
+      "marker": "■"
     },
     {
       "self_ref": "#/texts/18",
@@ -705,7 +705,7 @@
       "orig": "item 1 of neighboring list",
       "text": "item 1 of neighboring list",
       "enumerated": false,
-      "marker": "\u25a0"
+      "marker": "■"
     },
     {
       "self_ref": "#/texts/21",
@@ -723,7 +723,7 @@
       "orig": "item 2 of neighboring list",
       "text": "item 2 of neighboring list",
       "enumerated": false,
-      "marker": "\u25a0"
+      "marker": "■"
     },
     {
       "self_ref": "#/texts/22",
@@ -737,7 +737,7 @@
       "orig": "item 1 of sub list",
       "text": "item 1 of sub list",
       "enumerated": false,
-      "marker": "\u25a1"
+      "marker": "□"
     },
     {
       "self_ref": "#/texts/23",
@@ -755,7 +755,7 @@
       "orig": "",
       "text": "",
       "enumerated": false,
-      "marker": "\u25a1"
+      "marker": "□"
     },
     {
       "self_ref": "#/texts/24",
@@ -813,7 +813,7 @@
       "orig": "",
       "text": "",
       "enumerated": false,
-      "marker": "\u25a1"
+      "marker": "□"
     },
     {
       "self_ref": "#/texts/28",