From da287edfb5be418b77f465f60e80c472d3b348d1 Mon Sep 17 00:00:00 2001
From: safishamsi <safishamsi98@gmail.com>
Date: Sat, 10 Oct 2026 01:01:10 +0100
Subject: [PATCH] fix(powershell): extract functions inside file-scope
 begin/process/end blocks (#4270)

File-scope begin/process/end/dynamicparam/clean parse as pseudo-commands
(command_name 'begin'), not named_block, so the command branch returned before
the child walk and any function/class defined inside was dropped. Descend into
those five block keywords' children; ids stay name-based and stable.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 graphify/extractors/powershell.py | 14 ++++++++++++++
 tests/test_languages.py           | 24 ++++++++++++++++++++++++
 2 files changed, 38 insertions(+)

diff --git a/graphify/extractors/powershell.py b/graphify/extractors/powershell.py
index fe0e4df38..51d561951 100644
--- a/graphify/extractors/powershell.py
+++ b/graphify/extractors/powershell.py
@@ -61,6 +61,10 @@ def extract_powershell(path: Path) -> dict:
         # Import commands — handled as import edges, not function calls
         "import-module",
     })
+    # File-scope advanced-function blocks that parse as pseudo-commands rather
+    # than named_block nodes; walk() must descend into them to reach nested
+    # function/class/enum statements (#4270).
+    _PS_NAMED_BLOCK_CMDS = frozenset({"begin", "process", "end", "dynamicparam", "clean"})
 
     def _find_script_block_body(node):
         for child in node.children:
@@ -314,6 +318,16 @@ def extract_powershell(path: Path) -> dict:
                         if bare:
                             add_edge(file_nid, _make_id(bare), "imports_from",
                                      node.start_point[0] + 1)
+                elif cmd_text in _PS_NAMED_BLOCK_CMDS:
+                    # File-scope begin/process/end/dynamicparam/clean blocks parse as a
+                    # pseudo-command (command_name 'begin') whose `{ }` body is a
+                    # script_block_expression under command_elements, NOT a real
+                    # named_block (those exist only inside a function/scriptblock body).
+                    # Descend so nested function/class/enum statements are extracted
+                    # (#4270); without this the command branch returns before the
+                    # generic child walk below and the whole block is dropped.
+                    for child in node.children:
+                        walk(child, parent_class_nid)
             return
 
         for child in node.children:
diff --git a/tests/test_languages.py b/tests/test_languages.py
index 03d458370..d54b4a642 100644
--- a/tests/test_languages.py
+++ b/tests/test_languages.py
@@ -3078,6 +3078,30 @@ def test_powershell_psd1_no_dangling_edges():
         assert e["source"] in node_ids, f"Dangling source in edge: {e}"
 
 
+def test_powershell_functions_in_file_scope_begin_block_are_extracted(tmp_path):
+    # File-scope begin/process/end blocks parse as pseudo-commands (command_name
+    # 'begin'), not named_block, so functions defined inside them were dropped (#4270).
+    f = tmp_path / "advanced.ps1"
+    f.write_text(
+        "[CmdletBinding()]\n"
+        "param()\n"
+        "begin {\n"
+        "    function Get-Foo { Get-Bar }\n"
+        "    function Get-Bar { 'bar' }\n"
+        "}\n"
+        "process {\n"
+        "    Get-Foo\n"
+        "}\n",
+        encoding="utf-8",
+    )
+    r = extract_powershell(f)
+    assert "error" not in r
+    labels = {n["label"] for n in r["nodes"]}
+    assert "Get-Foo()" in labels and "Get-Bar()" in labels
+    # the call between the two helpers (inside Get-Foo's body) is recovered
+    assert ("Get-Foo", "Get-Bar") in _edge_labels(r, "calls")
+
+
 # ── TypeScript dynamic imports ───────────────────────────────────────────────
 
 def test_ts_dynamic_import_no_error():