No screenshot/figure node retrieval in typescript #554

Closed
opened 2026-02-16 00:18:14 -05:00 by yindo · 0 comments
Owner

Originally created by @martijn-ixor on GitHub (Oct 8, 2025).

There is a feature parity gap between Python and Typescript for the LlamaCloudRetriever for multimodal retrieval. Both have similar input parameters for this feature (retrieve_page_screenshot_nodes and retrieve_page_figure_nodes), but in the Typescript client, these are just ignored (a warning, documentation or error would have been helpful).

Is someone already working on this or is this on the roadmap? Would be open to work on a PR but some comment on if this is already wip or not would be welcome.

Client:

  • Typescript Client

Current Typescript code (https://github.com/run-llama/llama_cloud_services/blob/main/ts/llama_cloud_services/src/LlamaCloudRetriever.ts):

 return {
        // Currently LlamaCloud only supports text nodes
        node: textNode,
        score: node.score ?? undefined,
      };

Versus Python code (https://github.com/run-llama/llama_cloud_services/blob/main/py/llama_cloud_services/index/retriever.py):

result_nodes = self._result_nodes_to_node_with_score(results.retrieval_nodes)
        if self._retrieve_page_screenshot_nodes:
            result_nodes.extend(
                page_screenshot_nodes_to_node_with_score(
                    self._client, results.image_nodes, self.project.id
                )
            )
        if self._retrieve_page_figure_nodes:
            result_nodes.extend(
                page_figure_nodes_to_node_with_score(
                    self._client, results.page_figure_nodes, self.project.id
                )
            )
Originally created by @martijn-ixor on GitHub (Oct 8, 2025). There is a feature parity gap between Python and Typescript for the `LlamaCloudRetriever` for multimodal retrieval. Both have similar input parameters for this feature (`retrieve_page_screenshot_nodes` and `retrieve_page_figure_nodes`), but in the Typescript client, these are just ignored (a warning, documentation or error would have been helpful). Is someone already working on this or is this on the roadmap? Would be open to work on a PR but some comment on if this is already wip or not would be welcome. **Client:** - Typescript Client Current Typescript code (https://github.com/run-llama/llama_cloud_services/blob/main/ts/llama_cloud_services/src/LlamaCloudRetriever.ts): ```typescript return { // Currently LlamaCloud only supports text nodes node: textNode, score: node.score ?? undefined, }; ``` Versus Python code (https://github.com/run-llama/llama_cloud_services/blob/main/py/llama_cloud_services/index/retriever.py): ```python result_nodes = self._result_nodes_to_node_with_score(results.retrieval_nodes) if self._retrieve_page_screenshot_nodes: result_nodes.extend( page_screenshot_nodes_to_node_with_score( self._client, results.image_nodes, self.project.id ) ) if self._retrieve_page_figure_nodes: result_nodes.extend( page_figure_nodes_to_node_with_score( self._client, results.page_figure_nodes, self.project.id ) ) ```
yindo added the bug label 2026-02-16 00:18:14 -05:00
yindo closed this issue 2026-02-16 00:18:14 -05:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: run-llama/llama_cloud_services#554