Add pylsp-workspace-symbols to third-party plugins list

[Hanatarou]

Hi,

I've published a third-party plugin that adds workspace/symbol support to pylsp via Jedi — a feature that has been missing since 2022 (see #237, #564, #511 or similar).

Plugin: https://github.com/Hanatarou/pylsp-workspace-symbols
PyPI: https://pypi.org/project/pylsp-workspace-symbols

How it works

• Uses pylsp_experimental_capabilities to advertise workspaceSymbolProvider: true
• Uses pylsp_dispatchers to register a custom workspace/symbol handler
• Calls jedi.Project.complete_search("") and filters client-side by case-insensitive substring match
• Automatically skips noise folders (.venv, __pycache__, node_modules, etc.)

Install

pip install pylsp-workspace-symbols

Would you consider adding it to the third-party plugins section of the README?
This also serves as a working solution for users while #564 is pending merge.

Thanks!

[Hanatarou]

Hi again,

Just released v0.2.0 — the plugin now also adds inlay hints (textDocument/inlayHint) via Jedi inference, which should be relevant to anyone following #397.

What's new in v0.2.0

• Uses pylsp_experimental_capabilities to advertise inlayHintProvider: true
• Handles textDocument/inlayHint requests with a hybrid regex + Jedi approach
• Four independently configurable hint categories:
  • Assignment hints — x = 42 → : int, including self.attr in __init__
  • Return hints — unannotated def/async def → -> str, inferred from the first return
  • Raise hints — raise ValueError(...) → Raises: ValueError
  • Parameter hints — positional args at call sites → a=1, b=2 (keyword args skipped)
• Smart literal fast-path for common types (str, int, bool, None, list, dict, set, tuple, Callable) without Jedi overhead
• Fully respects existing annotations — annotated functions and variables are never hinted twice

Note on client compatibility: as @claytharrison noted in #397, some clients (e.g. eglot) look for inlayHintProvider as a regular capability rather than an experimental one. This plugin advertises it via pylsp_experimental_capabilities for now, since there is no hookspec for proper capabilities. If pylsp adds one in the future, this will be updated accordingly.

Install

pip install pylsp-workspace-symbols

PyPI: https://pypi.org/project/pylsp-workspace-symbols
Repo: https://github.com/Hanatarou/pylsp-workspace-symbols

[Hanatarou]

Hi again!

Just released v0.2.1 — a small but meaningful improvement for broader client compatibility.

Capabilities (workspaceSymbolProvider, inlayHintProvider) are now injected directly into pylsp's proper capabilities channel via monkey-patch, making the plugin work out-of-the-box with clients like Neovim and eglot that require proper (non-experimental) capabilities. Automatic fallback to pylsp_experimental_capabilities is maintained for clients that honour the experimental channel.

Thanks to @claytharrison for the approach documented in #397 — that comment was exactly the nudge needed to implement this properly.

[Hanatarou]

Hi,

just dropping a quick update: v0.3.0 is now out with several new features beyond workspace symbols:

• 🌳 Call hierarchy (callHierarchy/incomingCalls, callHierarchy/outgoingCalls)
• 🧬 Type hierarchy (typeHierarchy/supertypes, typeHierarchy/subtypes)
• 🔗 Document links (URLs in comments/strings + import resolution to stdlib source)
• 🎨 Document colors (hex, RGB, HSL, CSS named colors)

Still zero configuration required — just pip install pylsp-workspace-symbols.

Best regards,

Bruno Eduardo

[Hanatarou]

Hi again,

Just dropping a quick update: v0.4.0 is now out with two new major features and several fixes:

• 🔭 Code lens — reference counts (👥), subclass/override counts (🔗), run entry points (▶), and test markers (🧪)
• 🎨 Semantic tokens — Jedi-backed token classification with full, delta, and range support (opt-in)

Still zero configuration required — just pip install pylsp-workspace-symbols.

Best regards,

Bruno Eduardo

[Hanatarou]

Hi,

Just dropping a quick update: v0.6.0 is now out with a major semantic tokens overhaul — richer token types, more accurate modifiers, and better coverage of Python-specific constructs across the entire codebase.

Still zero configuration required — just pip install pylsp-workspace-symbols.

Best regards,

Bruno Eduardo

[eyalbold]

Yeh, this plugin works well and very much needed feature.