Trying to connect to REST-API with python in dynamo

[DonnyRH (Donny Hoogendorp)]

So i made a new token with the scopes below but i still seem to have the same issue. Could it be something else?

[jonathon]

Sorry I should have spotted before - that endpoint is expecting a POST request that also includes a list of objects to get.

# Replace with your actual token and ensure no leading/trailing spaces
TOKEN = "your_actual_token_here"
project_id = "6a87e55c77"  # Replace with your project ID

header = {
    'Authorization': f'Bearer {TOKEN}',
    'Accept': 'application/json'
}

url_get = f"https://app.speckle.systems/api/getobjects/{project_id}"

# Replace with actual object IDs you want to request
body = {
    "objects": ["objectId1", "objectId2", "objectId3"]  # Example IDs
}

response = requests.post(url=url_get, headers=header, json=body)

[DonnyRH (Donny Hoogendorp)]

Thanks for the feedback. The problem however seems to be with the authorisation of the token.

When i use the code above i get:
response_code 500: Unexpected token c in JSON at position 0, code: SyntaxError

When i use the code below as header is get a 403 “Your token is not valid” message .

header={
    "Authorization": f"Bear {Token}",
    "Accept": "application/json"
  }

In both situations i now added the body with a valid objectID (i think)

[jonathon]

BearER ?

"Authorization": f"Bearer {Token}",  # Ensure correct capitalisation and formatting

[DonnyRH (Donny Hoogendorp)]

Yes when i use this code i get:

response_code 500: Unexpected token c in JSON at position 0, code: SyntaxError

header={
    "Authorization": f"Bearer {Token}",
    "Accept": "application/json"
  }

[jonathon]

Were you to send a bare request from commandline:

curl -X POST \
  -H "Authorization: Bearer your_actual_token_here" \
  -H "Accept: application/json" \
  -d '{"objects": ["validObjectId1"]}' \
  https://app.speckle.systems/api/getobjects/6a87e55c77

Do you still see the same?

[DonnyRH (Donny Hoogendorp)]

Yes

curl -X POST ^
More?   -H "Authorization: Bearer my_token" ^
More?   -H "Accept: application/json" ^
More?   -d "{\"objects\": [\"cf02d5566caa6c38c3e2a7d9154168f4\"]}" ^
More?   https://app.speckle.systems/api/getobjects/6a87e55c77
{"error":{"message":"Unexpected token u in JSON at position 0","code":"SyntaxError"}}

[iainsproat]

The curl statement should have included the Content-Type header to send the data in valid format:

curl -X POST \
   -H "Authorization: Bearer my_token" \
   -H "Accept: application/json" \
   -H "Content-Type: application/json" \
   -d "{\"objects\": \"[\\\"cf02d5566caa6c38c3e2a7d9154168f4\\\"]\"}" \
   https://app.speckle.systems/api/getobjects/6a87e55c77

Note that the content of objects should be a string, not a JSON object.

In Python, I think this would be:

body = {
    "objects": "[\"objectId1\", \"objectId2\", \"objectId3\"]")  # Example IDs in JSON stringified array
}

[DonnyRH (Donny Hoogendorp)]

When adding the Content_Type -key i get the following response:
{“error”:“Your token is not valid.”}

Is it true that i need to use an acces token as generated in this section:

[iainsproat]

Yes. Please create a new access token with at least streams:read scope to use the /api/getobjects API paths.

You may need other scope depending on any other paths you wish to use your token for, e.g. streams:write if you intend to send objects to Speckle.

The Access Token is a secret and should not be shared with others.

Iain

[DonnyRH (Donny Hoogendorp)]

@jonathon and @iainsproat, I seem to get a little bit further. With the code down below i got a valid “200” status_code. The issue now is that i now get an empty list where i was expecting data.

header={
    "Authorization": f"Bearer {Token}",
    "Accept": "application/json",
    "Content-Type": "application/json"
  }
  
body = {
    'objects': ['"cf02d5566caa6c38c3e2a7d9154168f4"']  
}  

url_get=f"https://app.speckle.systems/api/getobjects/{IN[0]}"

with requests.post(url=url_get,headers=header,json=body,stream=True) as r:
    OUT=r

Its unclear to me if i need to use the model URL or Project URL (custom property - Speckle test kolommen | Speckle) (Speckle test kolommen | Speckle) when trying to extract an object.
Also is this the right id to take:
present in the body?

[jonathon]

if IN[0] is the project id, then this should be returning the object you highlighted

[DonnyRH (Donny Hoogendorp)]

This is the case so something is still going not quite right since i’m getting an empty list as response

[DonnyRH (Donny Hoogendorp)]

Hey Jonathan, do you maybe have an idea of how i could decipher if there is a bug in the API? It seems now that the script should return an object but what i get is an empty list. At this moment i have no clue how to check where the problem lies. I’m hoping you can give me some direction of how to figure out whether this is a bug or a mistake in my script?

[jonathon]

TBH i wouldnt try to download a stream with the REST API from python, id go straight to using specklepy.

Examples | Speckle Docs

Some basic setup

import os
from dotenv import load_dotenv

load_dotenv()

token = os.getenv("SPECKLE_APP_TOKEN")
server = os.getenv("SPECKLE_APP_URL")
project_id = "6a87e55c77"

Get an authenticated client

from specklepy.api.client import SpeckleClient

client = SpeckleClient(server, use_ssl=True)
client.authenticate_with_token(token)

Use the API to retrieve data using a ServerTransport

from specklepy.transports.memory import MemoryTransport
from specklepy.transports.server import ServerTransport

project = client.stream.get(project_id)
model_id = "58c8effef5"

branches = client.branch.list(project_id)
model = next((b for b in branches if b.id == model_id), None)

latest_version = model.commits.items[0].referencedObject

memory_transport: MemoryTransport = MemoryTransport()
server_transport = ServerTransport(project_id, client)

data = operations.receive(latest_version, server_transport, memory_transport)

A simple flatten function to retrieve a List and not a Graph

from collections.abc import Iterable
from specklepy.objects import Base


def flatten_base(base: Base) -> Iterable[Base]:
    """Flatten a base object into an iterable of bases.

    This function recursively traverses the `elements` or `@elements` attribute of the
    base object, yielding each nested base object.

    Args:
        base (Base): The base object to flatten.

    Yields:
        Base: Each nested base object in the hierarchy.
    """
    # Attempt to get the elements attribute, fallback to @elements if necessary
    elements = getattr(base, "elements", getattr(base, "@elements", None))

    if elements is not None:
        for element in elements:
            yield from flatten_base(element)

    yield base

Return the List

list(flatten_base(data))

# [
#  Base(id: 06c616f7b25acde87dbd54135b0c04da, speckle_type: Objects.BuiltElements.Column:Objects.BuiltElements.Archicad.ArchicadColumn, totalChildrenCount: 0),
#  Base(id: f2226f1bec2c8c51163abfe446f98e17, speckle_type: Objects.BuiltElements.Column:Objects.BuiltElements.Archicad.ArchicadColumn, totalChildrenCount: 0),
#  Collection(id: f99160be5076719329a34f7b68459da7, speckle_type: Speckle.Core.Models.Collection, totalChildrenCount: 0),
#  Collection(id: 71ecb27730fd332e5ff9a7d4bd341b71, speckle_type: Speckle.Core.Models.Collection, totalChildrenCount: None)
# ]

[DonnyRH (Donny Hoogendorp)]

Thank you very much Jonathan! this is very helpfull. I tried this as well in dynamo for Revit. The problem i have here happens at the statement:

project = client.stream.get(project_id)``

Giving the output:

Failed to execute the GraphQL stream request. Inner exception: Transport is allready connected.

Before we start to tackle this problem, maybe it would be helpfull to explain a bit off what im trying to do. I want to make a custom connector In Revit/Dynamo which will be able to get elements in an sends them back with the same id. The goal is to maintain the element traceability throughout the whole proces, even when it gets modified in different software. The documentation stated:

 When working with uploading and downloading objects, the REST API is preferred.

Is the GrapQL the right approach for this usecase or is the REST API the way to go for this usecase? I’m still figuring out the landscape of connection options in Speckle. I have tried the built-in Speckle 2 Connector in Dynamo but with that approach i didn’t seem to get all data neccesairy. I’d like to get your opinion about this. Thanks again in advance! The help i get on this platform is great!

[jonathon]

I’ll check with the team internally, but in my experience/view, updating objects with the same ID directly is possible, yet it is considered unSpeckle-like. That is not to say it is illegitimate, but it is not the direction of travel we have for our demonstration of Speckle.

That’s a lot of words to say; I disagree with what you found in our docs

[jonathon]

A colleague has pointed out an issue you maybe facing due to the embedded dependencies in Revit/Dynamo

Transport is already connected exception when trying to authenticate using python

Hi @Ali_Abdel-Khalek ! This error might be caused by the mismatch between 3 libraries: urllib3, requests, requests_toolbelt. This might be happening if your specklepy is installed into a shared environment, where other libraries are causing conflict between dependencies’ versions. When running your code, try running the command: import requests, urllib3, specklepy from importlib.metadata import version version('urllib3') version('requests') version('requests_toolbelt') The compatible versio…

Indeed, I had a similar experience working on the FME connector.

[DonnyRH (Donny Hoogendorp)]

The updating objects with the same ID isn’t something i read in the documents but just a usecase i’m working on at the moment (so no fault there on the Speckle-end ). The reason i’m interested in this idea is because of the following situation:

Lets say an architect models a certain wall in Archicad. After that, the structural engineer imports that wall into Revit and adjust some parameters and maybe also some of the geometry. When importing that same wall back into Speckle, it is no longer recognised as the same wall.

This situation affects the tracebillity of the object througout the design proces. What we are now trying to explore is if we can use Speckle in a way were we can still track the changes of certain objects over multiple transfers between different software. The goal here is that we can later use this data to optimize our designproces.

Maybe updating the object to the same id is not the right way, but thats the approach which came to mind with my (still limited) knowlegde of the Speckle ecosystem. Do you know if there are allready other functionalities available which we can use to track the changes of objects over multiple transfers through different software?

[jonathon]

Ah, the ID problem. It is much debated. A universal UID for objects externalised to the database and the authoring application—now that’s Strange Matter territory.

But before diving into solutions, it’s worth asking: Why is tracking an object across tools and versions essential to your current challenges? Is this driven by:

• Design Auditing: Accountability or version control needs?
• Coordination: Streamlining collaboration between disciplines?
• Optimisation: Understanding how changes cascade through the design.

Or is it a more theoretical exploration to future-proof your workflows? If theoretical, it’s worth examining whether it aligns with Speckle’s capabilities, as universal object tracking often introduces practical complexities. Speckle’s versioning and metadata features might already meet your needs more simply.

The Current Landscape of Object Identification in Speckle

By default, Speckle handles object identification in two ways:

1. applicationId: Assigned by the host application:

• Revit might use a GUID tied to its database.
• Archicad would have its logic.
• Navisworks, with no permanent UID concept, maps applicationId to a hierarchical path index. This is inherently software-specific and breaks when the object moves between platforms or is multi-authored.

2. speckleId: A hash based on an object’s content (geometry, metadata, etc.), unique to its exact state at the creation time. Any modification—intentional or software-induced—generates a new ID.

Thus, neither applicationId (software-specific and mutable) nor speckleId (content-based and immutable) can reliably track objects across versions or tools.

Why Cross-Software Tracking Is Challenging

1. Content Dependency: Speckle IDs change with any alteration to geometry or metadata.
2. Application-Specific Logic: applicationIds are tied to the originating software, lost or altered when objects are transformed.
3. Multi-Author Workflows: Collaborative edits across tools lead to new applicationIds, further compounding the issue.
4. No Universal Standards: Traceability remains elusive without a shared global UID system across platforms.

Potential Workarounds

1. globalId or Custom Properties:

• Use a shared ID manually or via middleware to persist object lineage across tools.
• Speckle supports custom properties, but not all tools faithfully transmit metadata.

2. Metadata and Geometric Matching:

• Compare secondary properties like dimensions, material, or geometry as a “fingerprint.” However, metadata loss or precision issues can limit reliability.

3. Middleware for Reconciliation:

• Custom tools or scripts to map relationships between IDs (applicationId, globalId) and establish traceability.

4. Standardised Workflows:

• Agree on conventions for object creation, modification, and sharing to maintain consistency.

5. Speckle Automate:

• Develop automation scripts to compare objects, identify lineage, and track changes programmatically.

Speckle’s Role in This

Speckle is designed to share, version, and collaborate on data, but its IDs focus on immutable snapshots rather than evolving object tracking. Cross-software traceability requires leveraging custom workflows, metadata, or middleware. If this challenge is critical, we can explore tailored approaches to your needs.