fix(core): fix BaseMessage .text() filtering out blocks with text

[olieydt]

• Description: BaseMessage .text() filters for blocks of type "text" when content is a list of dict. Concatenated messages can result in type != "text" but containing a text prop. This results in calling .text() on the message but receiving "". To address this, only consider if a text prop is present and a str.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Preview	Comments	Updated (UTC)
langchain	Ignored	Preview		Sep 5, 2025 2:23pm

[codspeed-hq]

CodSpeed WallTime Performance Report

Merging #32829 will not alter performance

_{Comparing olieydt:patch-1 (d972ca3) with master (bc91a48)}

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

Summary

✅ 13 untouched benchmarks

[codspeed-hq]

CodSpeed Instrumentation Performance Report

Merging #32829 will not alter performance

_{Comparing olieydt:patch-1 (d972ca3) with master (bc91a48)}

Summary

✅ 14 untouched benchmarks

[mdrxy]

You may want to take a look at the new implementations in our wip-v1.0 branch - this will soon be outdated/may no longer be needed

[ccurme]

@olieydt can you say more about the context for this? What models are returning blocks with "text" keys that are not "type": "text"?

I think semantically it makes sense for .text() to only filter to blocks matching a certain type. It's easy to imagine situations where this is desirable:

[
    {"type": "internal_thought_process": "text": "<LLM inner thoughts...>"},
    {"type": "text", "text": "Here is the answer..."},
]

^ in that example we'd likely just want .text() to return the output message text.

[ccurme]

Closing for now— the failing tests here indicate that this is deliberate behavior. I'd suggest opening an issue and giving additional context so we can design a proper solution (potentially as part of langchain 1.0).