feat: Port Arxiv to the Extension Framework

pyproject.toml

@@ -24,6 +24,10 @@ dependencies = [
     # ``pip install langflow`` still pulls the component in -- the AC for B1
     # is "no user-visible change at install time".
     "lfx-duckduckgo>=0.1.0",
+    # Second pilot port (validates src/bundles/PORTING.md): lfx-arxiv ships
+    # ArXivComponent as a standalone Bundle.  Same regular-dep rationale --
+    # ``pip install langflow`` keeps shipping the arxiv component as before.
+    "lfx-arxiv>=0.1.0",
     # langflow-extensions:bundle-deps-end
 ]
 
@@ -80,6 +84,7 @@ lfx = { workspace = true }
 langflow-sdk = { workspace = true }
 # langflow-extensions:bundle-sources-start
 lfx-duckduckgo = { workspace = true }
+lfx-arxiv = { workspace = true }
 # langflow-extensions:bundle-sources-end
 torch = { index = "pytorch-cpu" }
 torchvision = { index = "pytorch-cpu" }
@@ -92,6 +97,7 @@ members = [
     "src/sdk",
     # langflow-extensions:bundle-members-start
     "src/bundles/duckduckgo",
+    "src/bundles/arxiv",
     # langflow-extensions:bundle-members-end
 ]
 
@@ -425,6 +431,16 @@ external = ["RUF027"]
     "S104",    # Binding to all interfaces (test servers)
     "S108",    # Insecure temp file usage (safe in tests)
 ]
+"src/bundles/*/tests/*" = [
+    "D1",
+    "PLR2004",
+    "S101",
+    "SLF001",
+    "BLE001",  # allow broad-exception catching in tests
+    "S104",    # Binding to all interfaces (test servers)
+    "S108",    # Insecure temp file usage (safe in tests)
+    "INP001",  # Bundle test dirs are not import packages
+]
 "src/backend/tests/locust/*" = [
     "D1",      # Missing docstrings (CLI tools don't need full docstrings)
     "T201",    # Print statements (needed for CLI output)

src/backend/base/langflow/initial_setup/starter_projects/Research Translation Loop.json

@@ -209,7 +209,7 @@
                 ],
                 "total_dependencies": 2
               },
-              "module": "lfx.components.arxiv.arxiv.ArXivComponent"
+              "module": "lfx_arxiv.components.arxiv.arxiv.ArXivComponent"
             },
             "minimized": false,
             "output_types": [],

src/backend/base/langflow/locales/en.json

@@ -363,15 +363,6 @@
   "components.areducecomponent.inputs.source.info.21c26f21": "Input DataFrame to aggregate. The schema is automatically inferred from column names and types.",
   "components.areducecomponent.outputs.states.display_name.1bf5fc7c": "Output Table",
   "components.areducecomponent.outputs.states.info.9e0bb12a": "Aggregated DataFrame generated by the LLM following the specified output schema.",
-  "components.arxivcomponent.description.6fcfc4fe": "Search and retrieve papers from arXiv.org",
-  "components.arxivcomponent.display_name.2d546634": "arXiv",
-  "components.arxivcomponent.inputs.max_results.display_name.32040fb9": "Max Results",
-  "components.arxivcomponent.inputs.max_results.info.b0446804": "Maximum number of results to return",
-  "components.arxivcomponent.inputs.search_query.display_name.3ad6e0f4": "Search Query",
-  "components.arxivcomponent.inputs.search_query.info.dd5e9aa2": "The search query for arXiv papers (e.g., 'quantum computing')",
-  "components.arxivcomponent.inputs.search_type.display_name.6413a1ee": "Search Field",
-  "components.arxivcomponent.inputs.search_type.info.55d35699": "The field to search in",
-  "components.arxivcomponent.outputs.dataframe.display_name.16d1c905": "Table",
   "components.assemblyaigetsubtitles.description.1c28e36e": "Export your transcript in SRT or VTT format for subtitles and closed captions",
   "components.assemblyaigetsubtitles.display_name.3743bb02": "AssemblyAI Get Subtitles",
   "components.assemblyaigetsubtitles.inputs.api_key.display_name.5ec16b95": "Assembly API Key",

src/bundles/PORTING.md

@@ -308,7 +308,10 @@ wrong:
 # 1. The bundle's manifest is structurally valid.  Point ``validate`` at
 #    the package directory (where extension.json lives), not the bundle
 #    root -- the manifest is nested inside ``src/lfx_<bundle>/`` so the
-#    wheel ships it.
+#    wheel ships it.  The validator accepts both ``def build(self): ...``
+#    and ``outputs = [Output(method="...")]`` shapes; a component that
+#    uses neither will fail with ``build-method-missing`` -- add an
+#    ``outputs`` declaration in that case.
 uv run lfx extension validate src/bundles/<bundle>/src/lfx_<bundle>
 
 # 2. Workspace resolves and the bundle is importable.

src/bundles/arxiv/README.md

@@ -0,0 +1,43 @@
+# lfx-arxiv
+
+arXiv search component as a standalone Langflow Extension Bundle.
+
+This is the second-pilot port that validates
+[`src/bundles/PORTING.md`](../PORTING.md) — the documented recipe for
+extracting an in-tree provider into a standalone Bundle distribution.
+The bundle ships a single component, `ArXivComponent`, which queries
+arXiv's public Atom API for paper metadata.
+
+## Install
+
+```bash
+pip install lfx-arxiv
+```
+
+The bundle is registered automatically via the `langflow.extensions`
+entry-point.  After install, restart your Langflow server; the
+`ArXivComponent` will appear in the palette under the `arxiv` bundle
+group.
+
+## Develop
+
+```bash
+cd src/bundles/arxiv
+pip install -e .
+lfx extension validate .
+```
+
+## Manifest
+
+The extension manifest is shipped at
+`src/lfx_arxiv/extension.json` and points at the bundle at
+`components/arxiv`.  Components register under the canonical
+namespaced ID `ext:arxiv:ArXivComponent@official`.
+
+## Migration
+
+Saved flows referencing the legacy class name `ArXivComponent` or the
+old import paths `lfx.components.arxiv.arxiv.ArXivComponent` /
+`lfx.components.arxiv.ArXivComponent` are rewritten to the new
+namespaced ID by the migration table in
+`src/lfx/src/lfx/extension/migration/migration_table.json`.

src/bundles/arxiv/pyproject.toml

@@ -0,0 +1,53 @@
+[project]
+name = "lfx-arxiv"
+version = "0.1.0"
+description = "arXiv search component as a standalone Langflow Extension Bundle."
+readme = "README.md"
+requires-python = ">=3.10,<3.14"
+license = { text = "MIT" }
+authors = [
+    { name = "Langflow", email = "contact@langflow.org" },
+]
+keywords = ["langflow", "lfx", "extension", "bundle", "arxiv", "search"]
+
+# Runtime deps: lfx (the BUNDLE_API surface) plus defusedxml for safe XML
+# parsing (the arxiv API returns Atom).  defusedxml is already an lfx dep
+# but we list it here as a direct runtime dep so the bundle keeps building
+# even if lfx drops it later.
+dependencies = [
+    "lfx>=0.5.0,<0.6.0",
+    "defusedxml>=0.7.1,<1.0.0",
+]
+
+[project.urls]
+Homepage = "https://github.com/langflow-ai/langflow"
+Documentation = "https://docs.langflow.org/extensions"
+Repository = "https://github.com/langflow-ai/langflow"
+
+# Manifest-shipping distributions are discovered via the
+# ``langflow.extensions`` entry-point.  The value is the dotted path to
+# the package containing ``extension.json`` -- the loader walks
+# ``importlib.metadata.files()`` from there to find the manifest, and
+# falls back to this entry-point for editable installs whose dist.files
+# only surfaces dist-info entries.
+[project.entry-points."langflow.extensions"]
+lfx-arxiv = "lfx_arxiv"
+
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[tool.hatch.build.targets.wheel]
+# extension.json + components live inside the lfx_arxiv package so
+# ``importlib.metadata.files(dist)`` finds them and the loader resolves
+# bundles[].path relative to the manifest's directory.
+packages = ["src/lfx_arxiv"]
+include = ["src/lfx_arxiv/extension.json", "src/lfx_arxiv/components/**/*.py"]
+
+[tool.hatch.build.targets.sdist]
+include = [
+    "src/lfx_arxiv",
+    "extension.json",
+    "README.md",
+    "pyproject.toml",
+]

src/bundles/arxiv/src/lfx_arxiv/__init__.py

@@ -0,0 +1,15 @@
+"""lfx-arxiv: arXiv Search bundle.
+
+This package is the distribution unit ``lfx-arxiv``.  At runtime
+Langflow's loader discovers ``extension.json`` shipped alongside this
+``__init__.py`` and registers ``ArXivComponent`` under the namespaced
+ID ``ext:arxiv:ArXivComponent@official``.
+
+Second pilot port (after lfx-duckduckgo) -- exercises the same
+extraction recipe documented in ``src/bundles/PORTING.md`` against a
+component with no third-party runtime deps.
+"""
+
+from lfx_arxiv.components.arxiv.arxiv import ArXivComponent
+
+__all__ = ["ArXivComponent"]

src/lfx/src/lfx/components/arxiv/arxiv.py → src/bundles/arxiv/src/lfx_arxiv/components/arxiv/arxiv.py

@@ -2,7 +2,6 @@
 from xml.etree.ElementTree import Element
 
 from defusedxml.ElementTree import fromstring
-
 from lfx.custom.custom_component.component import Component
 from lfx.io import DropdownInput, IntInput, MessageTextInput, Output
 from lfx.schema.data import Data

src/bundles/arxiv/src/lfx_arxiv/extension.json

@@ -0,0 +1,16 @@
+{
+  "$schema": "https://schemas.langflow.org/extension/v1.json",
+  "id": "lfx-arxiv",
+  "version": "0.1.0",
+  "name": "arXiv Search",
+  "description": "arXiv search component as a standalone Langflow Extension Bundle.",
+  "lfx": {
+    "compat": ["1"]
+  },
+  "bundles": [
+    {
+      "name": "arxiv",
+      "path": "components/arxiv"
+    }
+  ]
+}