Add duckdb recipe for emscripten

recipes/recipes_emscripten/duckdb/build.sh

@@ -0,0 +1,18 @@
+#!/bin/bash
+set -euo pipefail
+
+# Setup emscripten toolchain for scikit-build-core
+emscripten_root=$(em-config EMSCRIPTEN_ROOT)
+toolchain_path="${emscripten_root}/cmake/Modules/Platform/Emscripten.cmake"
+
+export CMAKE_ARGS="${CMAKE_ARGS} -DCMAKE_TOOLCHAIN_FILE=${toolchain_path} -DCMAKE_PROJECT_INCLUDE=${RECIPE_DIR}/overwriteProp.cmake"
+
+# DuckDB-specific CMake flags for wasm passed via scikit-build-core config settings
+${PYTHON} -m pip install . ${PIP_ARGS} \
+  -Ccmake.define.OVERRIDE_GIT_DESCRIBE="v${PKG_VERSION}" \
+  -Ccmake.define.BUILD_EXTENSIONS="parquet;json;autocomplete" \
+  -Ccmake.define.BUILD_SHELL=OFF \
+  -Ccmake.define.BUILD_UNITTESTS=OFF \
+  -Ccmake.define.ENABLE_EXTENSION_AUTOLOADING=OFF \
+  -Ccmake.define.ENABLE_EXTENSION_AUTOINSTALL=OFF \
+  -Ccmake.define.CMAKE_INTERPROCEDURAL_OPTIMIZATION=OFF

recipes/recipes_emscripten/duckdb/overwriteProp.cmake

@@ -0,0 +1,9 @@
+set_property(GLOBAL PROPERTY TARGET_SUPPORTS_SHARED_LIBS TRUE)
+set(CMAKE_SHARED_LIBRARY_CREATE_C_FLAGS "-sSIDE_MODULE=1 -sWASM_BIGINT")
+set(CMAKE_SHARED_LIBRARY_CREATE_CXX_FLAGS "-sSIDE_MODULE=1 -sWASM_BIGINT")
+set(CMAKE_STRIP FALSE)
+
+# DuckDB links libduckdb_static.a twice to resolve circular dependencies
+# between the core and extensions. Native linkers handle this but wasm-ld
+# does not. Allow multiple definitions to work around this.
+add_link_options("LINKER:--allow-multiple-definition")

recipes/recipes_emscripten/duckdb/patches/emscripten-python-target.patch

@@ -0,0 +1,29 @@
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index abcdef1..1234567 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -36,6 +36,15 @@ endif()
+ # Dependencies
+ # ────────────────────────────────────────────
+ # PyBind11
++
++# On Emscripten, FindPython only provides Python::Module, not Python::Python
++# (there is no shared libpython to link against). pybind11_add_module internally
++# calls python_add_library which expects Python::Python, so we provide a dummy
++# IMPORTED target to satisfy that requirement.
++if(EMSCRIPTEN AND NOT TARGET Python::Python)
++  add_library(Python::Python INTERFACE IMPORTED)
++endif()
++
+ find_package(pybind11 REQUIRED CONFIG)
+
+ # DuckDB
+@@ -100,7 +109,7 @@ if(APPLE)
+   target_link_options(
+     _duckdb PRIVATE "LINKER:-exported_symbol,_duckdb_adbc_init"
+     "LINKER:-exported_symbol,_PyInit__duckdb")
+-elseif(UNIX AND NOT APPLE)
++elseif(UNIX AND NOT APPLE AND NOT EMSCRIPTEN)
+   target_link_options(
+     _duckdb PRIVATE "LINKER:--export-dynamic-symbol=duckdb_adbc_init"
+     "LINKER:--export-dynamic-symbol=PyInit__duckdb")

recipes/recipes_emscripten/duckdb/patches/fix-dirty-version-scheme.patch

@@ -0,0 +1,13 @@
+diff --git a/duckdb_packaging/setuptools_scm_version.py b/duckdb_packaging/setuptools_scm_version.py
+index 1234567..abcdef0 100644
+--- a/duckdb_packaging/setuptools_scm_version.py
++++ b/duckdb_packaging/setuptools_scm_version.py
+@@ -53,7 +53,7 @@ def version_scheme(version: _VersionObject) -> str:
+
+     distance = int(version.distance or 0)
+     try:
+-        if distance == 0 and not version.dirty:
++        if distance == 0:
+             return _tag_to_version(str(version.tag))
+         return _bump_dev_version(str(version.tag), distance)
+     except Exception as e:

recipes/recipes_emscripten/duckdb/recipe.yaml

@@ -0,0 +1,72 @@
+context:
+  version: 1.4.4
+
+package:
+  name: python-duckdb
+  version: ${{ version }}
+
+source:
+  git: https://github.com/duckdb/duckdb-python.git
+  tag: v${{ version }}
+  # expected_commit: a12f36ca411007f5eb48919448f61c7498112553
+  patches:
+  - patches/emscripten-python-target.patch
+  - patches/fix-dirty-version-scheme.patch
+
+build:
+  number: 0
+  script: build.sh
+
+  files:
+    exclude:
+    - '**/__pycache__/**'
+    - '**/*.pyc'
+    - '**/test_*.py'
+    - 'duckdb_build/**'
+  python:
+    skip_pyc_compilation:
+    - '**/*.py'
+
+requirements:
+  build:
+  - python
+  - cross-python_${{ target_platform }}
+  - ${{ compiler('cxx') }}
+  - pip
+  - setuptools
+  - setuptools-scm
+  - pybind11
+  - scikit-build-core
+  - cmake
+  - ninja
+  host:
+  - python
+  - pybind11
+  run:
+  - python
+
+tests:
+- script: pytester
+  files:
+    recipe:
+    - test_duckdb.py
+  requirements:
+    build:
+    - pytester
+    run:
+    - pytester-run
+
+about:
+  homepage: https://duckdb.org/
+  license: MIT
+  license_file: LICENSE
+  summary: DuckDB is an analytical in-process SQL database management system
+  description: |
+    DuckDB is an in-process SQL OLAP database management system.
+    It is designed to support analytical query workloads (OLAP) while
+    being embeddable in Python and other languages.
+  repository: https://github.com/duckdb/duckdb-python
+
+extra:
+  recipe-maintainers:
+  - wolfv

recipes/recipes_emscripten/duckdb/test_duckdb.py

@@ -0,0 +1,26 @@
+import duckdb
+
+
+def test_import():
+    assert duckdb is not None
+
+
+def test_basic_query():
+    con = duckdb.connect()
+    result = con.execute("SELECT 42 AS answer").fetchall()
+    assert result == [(42,)]
+
+
+def test_create_table():
+    con = duckdb.connect()
+    con.execute("CREATE TABLE t (x INTEGER, y VARCHAR)")
+    con.execute("INSERT INTO t VALUES (1, 'hello'), (2, 'world')")
+    result = con.execute("SELECT * FROM t ORDER BY x").fetchall()
+    assert result == [(1, "hello"), (2, "world")]
+
+
+def test_aggregation():
+    con = duckdb.connect()
+    con.execute("CREATE TABLE nums AS SELECT * FROM range(10) t(i)")
+    result = con.execute("SELECT SUM(i), COUNT(*) FROM nums").fetchone()
+    assert result == (45, 10)