Add tree-sitter-java, tree-sitter-go, tree-sitter-python (#5102)

This is a follow-on to PR #5099 which added tree-sitter to the packages.

This change adds tree-sitter-java, tree-sitter-go, tree-sitter-python packages
and respective unit tests. Java, Go, and Python are popular languages, so its
useful to add these grammars so a user can parse source code in those languages.

.gitignore

@@ -41,3 +41,4 @@ pytest-pyodide
 tools/symlinks
 xbuildenv/
 .pyodide-xbuildenv*
+DS_Store

docs/project/changelog.md

@@ -87,6 +87,9 @@ myst:
 - Added `iminuit` 2.29.1 {pr}`4767`, {pr}`5072`
 - Added `arro3-core`, `arro3-io`, and `arro3-compute` 0.3.0, 0.4.0, 0.4.1 {pr}`5020`, {pr}`5095`, {pr}`5104`
 - Added `tree-sitter` 0.23.0 {pr}`5099`
+- Added `tree-sitter-go` 0.23.1 {pr}`5102`
+- Added `tree-sitter-java` 0.23.2 {pr}`5102`
+- Added `tree-sitter-python` 0.23.2 {pr}`5102`
 
 ## Version 0.26.3
 

packages/tree-sitter-go/meta.yaml

@@ -0,0 +1,23 @@
+package:
+  name: tree-sitter-go
+  version: 0.23.1
+  top-level:
+    - tree_sitter_go
+source:
+  # Have to use github source to work around broken sdist
+  # https://github.com/tree-sitter/tree-sitter-go/pull/151
+  url: https://github.com/tree-sitter/tree-sitter-go/archive/refs/tags/v0.23.1.tar.gz
+  sha256: 4173bafc4c59be78642a0faf1bed8cb3854458aa59c506e8f51001a9f28da09b
+requirements:
+  run:
+    - tree-sitter
+  host:
+    - tree-sitter
+about:
+  home: https://github.com/tree-sitter/tree-sitter-go
+  PyPI: https://pypi.org/project/tree-sitter-go
+  summary: Go grammar for tree-sitter
+  license: MIT
+extra:
+  recipe-maintainers:
+    - ericwb

packages/tree-sitter-go/test_tree_sitter_go.py

@@ -0,0 +1,41 @@
+from pytest_pyodide import run_in_pyodide
+
+
+@run_in_pyodide(packages=["tree-sitter-go"])
+def test_tree_sitter_go(selenium):
+    import textwrap
+
+    import tree_sitter_go
+    from tree_sitter import Language, Parser
+
+    GO_LANGUAGE = Language(tree_sitter_go.language())
+    parser = Parser(GO_LANGUAGE)
+
+    code = bytes(
+        textwrap.dedent(
+            """
+        func foo() {
+            if bar {
+                baz()
+            }
+        }
+        """
+        ),
+        "utf-8",
+    )
+    tree = parser.parse(code)
+    root_node = tree.root_node
+
+    assert str(root_node) == (
+        "(source_file "
+        "(function_declaration "
+        "name: (identifier) "
+        "parameters: (parameter_list) "
+        "body: (block "
+        "(if_statement "
+        "condition: (identifier) "
+        "consequence: (block "
+        "(expression_statement (call_expression "
+        "function: (identifier) "
+        "arguments: (argument_list))))))))"
+    )

packages/tree-sitter-java/meta.yaml

@@ -0,0 +1,23 @@
+package:
+  name: tree-sitter-java
+  version: 0.23.2
+  top-level:
+    - tree_sitter_java
+source:
+  # Have to use github url to work around broken sdist:
+  # https://github.com/tree-sitter/tree-sitter-java/pull/188
+  url: https://github.com/tree-sitter/tree-sitter-java/archive/refs/tags/v0.23.2.tar.gz
+  sha256: 062fe5746deb4e362ccb987228e7b41bfc017eb2b91a0413a7447719abaa07b2
+requirements:
+  run:
+    - tree-sitter
+  host:
+    - tree-sitter
+about:
+  home: https://github.com/tree-sitter/tree-sitter-java
+  PyPI: https://pypi.org/project/tree-sitter-java
+  summary: Java grammar for tree-sitter
+  license: MIT
+extra:
+  recipe-maintainers:
+    - ericwb

packages/tree-sitter-java/test_tree_sitter_java.py

@@ -0,0 +1,42 @@
+from pytest_pyodide import run_in_pyodide
+
+
+@run_in_pyodide(packages=["tree-sitter-java"])
+def test_tree_sitter_java(selenium):
+    import textwrap
+
+    import tree_sitter_java
+    from tree_sitter import Language, Parser
+
+    JAV_LANGUAGE = Language(tree_sitter_java.language())
+    parser = Parser(JAV_LANGUAGE)
+
+    code = bytes(
+        textwrap.dedent(
+            """
+        void foo() {
+            if (bar) {
+                baz();
+            }
+        }
+        """
+        ),
+        "utf-8",
+    )
+    tree = parser.parse(code)
+    root_node = tree.root_node
+
+    assert str(root_node) == (
+        "(program "
+        "(method_declaration "
+        "type: (void_type) "
+        "name: (identifier) "
+        "parameters: (formal_parameters) "
+        "body: (block "
+        "(if_statement "
+        "condition: (parenthesized_expression (identifier)) "
+        "consequence: (block "
+        "(expression_statement (method_invocation "
+        "name: (identifier) "
+        "arguments: (argument_list))))))))"
+    )

packages/tree-sitter-python/meta.yaml

@@ -0,0 +1,23 @@
+package:
+  name: tree-sitter-python
+  version: 0.23.2
+  top-level:
+    - tree_sitter_python
+source:
+  # Have to use github url to work around broken sdist:
+  # https://github.com/tree-sitter/tree-sitter-python/pull/285
+  url: https://github.com/tree-sitter/tree-sitter-python/archive/refs/tags/v0.23.2.tar.gz
+  sha256: b38e5b1f5237377b506109978af76422975bda5859165835e3c7b5afee987383
+requirements:
+  run:
+    - tree-sitter
+  host:
+    - tree-sitter
+about:
+  home: https://github.com/tree-sitter/tree-sitter-python
+  PyPI: https://pypi.org/project/tree-sitter-python
+  summary: Python grammar for tree-sitter
+  license: MIT
+extra:
+  recipe-maintainers:
+    - ericwb

packages/tree-sitter-python/test_tree_sitter_python.py

@@ -0,0 +1,39 @@
+from pytest_pyodide import run_in_pyodide
+
+
+@run_in_pyodide(packages=["tree-sitter-python"])
+def test_tree_sitter_python(selenium):
+    import textwrap
+
+    import tree_sitter_python
+    from tree_sitter import Language, Parser
+
+    PY_LANGUAGE = Language(tree_sitter_python.language())
+    parser = Parser(PY_LANGUAGE)
+
+    code = bytes(
+        textwrap.dedent(
+            """
+        def foo():
+            if bar:
+                baz()
+        """
+        ),
+        "utf-8",
+    )
+    tree = parser.parse(code)
+    root_node = tree.root_node
+
+    assert str(root_node) == (
+        "(module "
+        "(function_definition "
+        "name: (identifier) "
+        "parameters: (parameters) "
+        "body: (block "
+        "(if_statement "
+        "condition: (identifier) "
+        "consequence: (block "
+        "(expression_statement (call "
+        "function: (identifier) "
+        "arguments: (argument_list))))))))"
+    )