BACK-2650: fix DataURIAdapter re plain text json data uri

docs/changelog.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## v0.3.2
+
+- Fix an issue in `DataURIAdapter` where plain-text json data uri would get ignored
+
 ## v0.3.1
 
 - Trim token_uri in some log outputs, this is mainly useful for data uris that are too long and make logs unreadable

docs/index.md

@@ -1,6 +1,6 @@
 # Getting Started
 
-Documentation for version: **v0.3.1**
+Documentation for version: **v0.3.2**
 
 ## Overview
 

offchain/metadata/adapters/data_uri.py

@@ -3,6 +3,7 @@
 from urllib.request import urlopen
 
 import httpx
+import json
 from requests import PreparedRequest, Response
 
 from offchain.metadata.adapters.base_adapter import BaseAdapter
@@ -17,6 +18,10 @@ def decode_data_url(data_url):  # type: ignore[no-untyped-def]
         decoded_data = base64.b64decode(data)
         decoded_text = decoded_data.decode("utf-8")
         return decoded_text
+    elif "json;utf8" in data_parts[0]:
+        decoded_data = urlopen(data_url).read()
+        decoded_text = json.dumps(json.loads(decoded_data))
+        return decoded_text
 
     return None
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "offchain"
-version = "0.3.1"
+version = "0.3.2"
 description = "Open source metadata processing framework"
 authors = ["Zora eng <[email protected]>"]
 readme = "README.md"

tests/metadata/adapters/test_data_adapter.py

@@ -1,4 +1,5 @@
 import httpx
+import json
 import pytest
 from pytest_httpx import HTTPXMock
 
@@ -25,6 +26,25 @@ async def test_gen_head(self, httpx_mock: HTTPXMock):
         outgoing_request = httpx_mock.get_requests()
         assert not outgoing_request
 
+    @pytest.mark.asyncio
+    async def test_gen_head_not_base64(self, httpx_mock: HTTPXMock):
+        adapter = DataURIAdapter()
+        data_url = "data:application/json;utf8,{\"name\":\"here for now\",\"description\":\"sometimes i don't know how to feel when i'm away.\", \"image\": \"data:image/svg+xml;base64,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\"}"  # noqa
+        async with httpx.AsyncClient() as client:
+            result = await adapter.gen_head(url=data_url, sess=client)
+
+        expected = httpx.Response(
+            status_code=200,
+            headers={"content-type": "application/json;utf8", "content-length": "2600"},
+            request=httpx.Request(method="HEAD", url=data_url),
+        )
+        assert result.status_code == 200
+        assert result.request.method == "HEAD"
+        assert result.headers == expected.headers
+        # no real request was made
+        outgoing_request = httpx_mock.get_requests()
+        assert not outgoing_request
+
     @pytest.mark.asyncio
     async def test_gen_send(self, httpx_mock: HTTPXMock):
         adapter = DataURIAdapter()
@@ -37,3 +57,18 @@ async def test_gen_send(self, httpx_mock: HTTPXMock):
         # no real request was made
         outgoing_request = httpx_mock.get_requests()
         assert not outgoing_request
+
+    @pytest.mark.asyncio
+    async def test_gen_send_not_base64(self, httpx_mock: HTTPXMock):
+        adapter = DataURIAdapter()
+        json_str = "{\"name\":\"here for now\",\"description\":\"sometimes i don't know how to feel when i'm away.\", \"image\": \"data:image/svg+xml;base64,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\"}"  # noqa
+        data_url = f"data:application/json;utf8,{json_str}"
+        async with httpx.AsyncClient() as client:
+            result = await adapter.gen_send(url=data_url, sess=client)
+
+        assert result.status_code == 200
+        assert result.request.method == "GET"
+        assert json.loads(result.text) == json.loads(json_str)
+        # no real request was made
+        outgoing_request = httpx_mock.get_requests()
+        assert not outgoing_request

tests/metadata/parsers/test_zora_parser.py

@@ -19,7 +19,7 @@ class TestZoraParser:
     token = Token(
         chain_identifier="ETHEREUM-MAINNET",
         collection_address="0xabefbc9fd2f806065b4f3c237d4b59d9a97bcac7",
-        token_id=5769,
+        token_id=31861,
     )
 
     raw_data = {
@@ -46,8 +46,8 @@ def test_zora_parser_parses_metadata(self):  # type: ignore[no-untyped-def]
             token=Token(
                 chain_identifier="ETHEREUM-MAINNET",
                 collection_address="0xabefbc9fd2f806065b4f3c237d4b59d9a97bcac7",
-                token_id=5769,
-                uri="https://zora-dev.mypinata.cloud/ipfs/bafkreigux6jujn5hvlmptgzgok4reaie2gkuvsk2kynnalsyfgr4g35dkm",
+                token_id=31861,
+                uri="https://gateway.pinata.cloud/ipfs/bafkreid3jq3mlqz4d3w7emkxpftmfjbbxtkwe7kf25lzp2krwcxfd57m6q",
             ),
             raw_data={
                 "description": "A Lonely Soul,\n\nI've felt lonely lately. Somewhere deep inside, detached. \n\nThere must be plenty of lost souls wandering the globe. Looking to belong; to understand their purpose.\n\nI know my purpose, but I fear I've burned up surviving to the moment.\n\nDoes this count?\nAm I still pushing forward?\n\nI hope so...\n\nDo I still have time?\nAm I just floating?\n\nPlease, don't give up.\n\nAge 23 (2021)\n4096x4096px",
@@ -63,7 +63,7 @@ def test_zora_parser_parses_metadata(self):  # type: ignore[no-untyped-def]
             image=MediaDetails(
                 size=13548199,
                 sha256=None,
-                uri="https://zora-dev.mypinata.cloud/ipfs/bafybeiffwxjez2axebcprj2h7wkohr2pdbvuv37f7uxyptuw7o6t5fvppu",
+                uri="https://gateway.pinata.cloud/ipfs/bafybeifavbhn6ys3k4tvngt4rxkoo7vabiv4lnlszwkvdncjg245qz5chq",
                 mime_type="image/jpeg",
             ),
             content=None,