Improve registry value return types

Specifically:
- REG_MULTI_SZ returns list<string> instead of | str join (char nl)
- REG_EXPAND_SZ automatically expands %EnvironmentString% placeholders
- REG_DWORD_BIG_ENDIAN returns the endianness decoded value
- REG_QWORD actually ingests values over 2**32
- REG_LINK produces a decoded string
- REG_*RESOURCE* produces binary data

crates/nu-command/Cargo.toml

@@ -110,6 +110,7 @@ version = "3.1"
 features = [
 	"Win32_Foundation",
 	"Win32_Storage_FileSystem",
+	"Win32_System_Environment",
 	"Win32_System_SystemServices",
 	"Win32_Security",
 	"Win32_System_Threading",

crates/nu-command/src/system/registry_query.rs

@@ -5,7 +5,8 @@ use nu_protocol::{
     record, Category, Example, IntoInterruptiblePipelineData, IntoPipelineData, PipelineData,
     ShellError, Signature, Span, Spanned, SyntaxShape, Type, Value,
 };
-use winreg::{enums::*, RegKey};
+use windows::{core::PCWSTR, Win32::System::Environment::ExpandEnvironmentStringsW};
+use winreg::{enums::*, types::FromRegValue, RegKey};
 
 #[derive(Clone)]
 pub struct RegistryQuery;
@@ -92,12 +93,13 @@ fn registry_query(
     if registry_value.is_none() {
         let mut reg_values = vec![];
         for (name, val) in reg_key.enum_values().flatten() {
-            let (nu_value, reg_type) = reg_value_to_nu_value(val, call_span);
+            let reg_type = format!("{:?}", val.vtype);
+            let nu_value = reg_value_to_nu_value(val, call_span);
             reg_values.push(Value::record(
                 record! {
                     "name" => Value::string(name, call_span),
                     "value" => nu_value,
-                    "type" => Value::string(format!("{:?}", reg_type), call_span),
+                    "type" => Value::string(reg_type, call_span),
                 },
                 *registry_key_span,
             ))
@@ -109,12 +111,13 @@ fn registry_query(
                 let reg_value = reg_key.get_raw_value(value.item.as_str());
                 match reg_value {
                     Ok(val) => {
-                        let (nu_value, reg_type) = reg_value_to_nu_value(val, call_span);
+                        let reg_type = format!("{:?}", val.vtype);
+                        let nu_value = reg_value_to_nu_value(val, call_span);
                         Ok(Value::record(
                             record! {
                                 "name" => Value::string(value.item, call_span),
                                 "value" => nu_value,
-                                "type" => Value::string(format!("{:?}", reg_type), call_span),
+                                "type" => Value::string(reg_type, call_span),
                             },
                             value.span,
                         )
@@ -174,61 +177,109 @@ fn get_reg_hive(call: &Call) -> Result<RegKey, ShellError> {
     Ok(RegKey::predef(hkey))
 }
 
-fn reg_value_to_nu_value(
+fn reg_value_to_nu_value(mut reg_value: winreg::RegValue, call_span: Span) -> nu_protocol::Value {
-    reg_value: winreg::RegValue,
-    call_span: Span,
-) -> (nu_protocol::Value, winreg::enums::RegType) {
     match reg_value.vtype {
-        REG_NONE => (Value::nothing(call_span), reg_value.vtype),
+        REG_NONE => Value::nothing(call_span),
-        REG_SZ => (
+        REG_BINARY => Value::binary(reg_value.bytes, call_span),
-            Value::string(reg_value.to_string(), call_span),
+        REG_MULTI_SZ => reg_value_to_nu_list_string(reg_value, call_span),
-            reg_value.vtype,
+        REG_SZ | REG_EXPAND_SZ => reg_value_to_nu_string(reg_value, call_span),
-        ),
+        REG_DWORD | REG_DWORD_BIG_ENDIAN | REG_QWORD => reg_value_to_nu_int(reg_value, call_span),
-        REG_EXPAND_SZ => (
+
-            Value::string(reg_value.to_string(), call_span),
+        // This should be impossible, as registry symlinks should be automatically transparent
-            reg_value.vtype,
+        // to the registry API as it's used by winreg, since it never uses REG_OPTION_OPEN_LINK.
-        ),
+        // If it happens, decode as if the link is a string; it should be a registry path string.
-        REG_BINARY => (Value::binary(reg_value.bytes, call_span), reg_value.vtype),
+        REG_LINK => {
-        REG_DWORD => (
+            reg_value.vtype = REG_SZ;
-            Value::int(
+            reg_value_to_nu_string(reg_value, call_span)
-                unsafe { *(reg_value.bytes.as_ptr() as *const u32) } as i64,
+        }
-                call_span,
+
-            ),
+        // Decode these as binary; that seems to be the least bad option available to us.
-            reg_value.vtype,
+        // REG_RESOURCE_LIST is a struct CM_RESOURCE_LIST.
-        ),
+        // REG_FULL_RESOURCE_DESCRIPTOR is a struct CM_FULL_RESOURCE_DESCRIPTOR.
-        REG_DWORD_BIG_ENDIAN => (
+        // REG_RESOURCE_REQUIREMENTS_LIST is a struct IO_RESOURCE_REQUIREMENTS_LIST.
-            Value::int(
+        REG_RESOURCE_LIST | REG_FULL_RESOURCE_DESCRIPTOR | REG_RESOURCE_REQUIREMENTS_LIST => {
-                unsafe { *(reg_value.bytes.as_ptr() as *const u32) } as i64,
+            reg_value.vtype = REG_BINARY;
-                call_span,
+            Value::binary(reg_value.bytes, call_span)
-            ),
+        }
-            reg_value.vtype,
-        ),
-        REG_LINK => (
-            Value::string(reg_value.to_string(), call_span),
-            reg_value.vtype,
-        ),
-        REG_MULTI_SZ => (
-            Value::string(reg_value.to_string(), call_span),
-            reg_value.vtype,
-        ),
-        REG_RESOURCE_LIST => (
-            Value::string(reg_value.to_string(), call_span),
-            reg_value.vtype,
-        ),
-        REG_FULL_RESOURCE_DESCRIPTOR => (
-            Value::string(reg_value.to_string(), call_span),
-            reg_value.vtype,
-        ),
-        REG_RESOURCE_REQUIREMENTS_LIST => (
-            Value::string(reg_value.to_string(), call_span),
-            reg_value.vtype,
-        ),
-        REG_QWORD => (
-            Value::int(
-                unsafe { *(reg_value.bytes.as_ptr() as *const u32) } as i64,
-                call_span,
-            ),
-            reg_value.vtype,
-        ),
     }
 }
+
+fn reg_value_to_nu_string(reg_value: winreg::RegValue, call_span: Span) -> nu_protocol::Value {
+    let value = String::from_reg_value(&reg_value)
+        .expect("registry value type should be REG_SZ or REG_EXPAND_SZ");
+
+    // REG_EXPAND_SZ contains unexpanded references to environment variables, for example, %PATH%.
+    // winreg not expanding these is arguably correct, as it's just wrapping raw registry access.
+    // These placeholder-having strings work in *some* Windows contexts, but Rust's fs/path APIs
+    // don't handle them, so they won't work in Nu unless we expand them here. Eagerly expanding the
+    // strings here seems to be the least bad option. This is what PowerShell does, for example,
+    // although reg.exe does not. We could do the substitution with our env, but the officially
+    // correct way to expand these strings is to call Win32's ExpandEnvironmentStrings function.
+    // ref: <https://learn.microsoft.com/en-us/windows/win32/sysinfo/registry-value-types>
+
+    // We can skip the dance if the string doesn't actually have any unexpanded placeholders.
+    if reg_value.vtype != REG_EXPAND_SZ || !value.contains('%') {
+        return Value::string(value, call_span);
+    }
+
+    // The encoding dance is unfortunate since we read "Windows Unicode" from the registry, but
+    // it's the most resilient option and avoids making potentially wrong alignment assumptions.
+    let value_utf16 = value.encode_utf16().chain([0]).collect::<Vec<u16>>();
+
+    // Like most Win32 string functions, the return value is the number of TCHAR written,
+    // or the required buffer size (in TCHAR) if the buffer is too small, or 0 for error.
+    // Since we already checked for the case where no expansion is done, we can start with
+    // an empty output buffer, since we expect to require at least one resize loop anyway.
+    let mut out_buffer = vec![];
+    loop {
+        match unsafe {
+            ExpandEnvironmentStringsW(PCWSTR(value_utf16.as_ptr()), Some(&mut *out_buffer))
+        } {
+            0 => {
+                // 0 means error, but we don't know what the error is. We could try to get
+                // the error code with GetLastError, but that's a whole other can of worms.
+                // Instead, we'll just return the original string and hope for the best.
+                // Presumably, registry strings shouldn't ever cause this to error anyway.
+                return Value::string(value, call_span);
+            }
+            size if size as usize <= out_buffer.len() => {
+                // The buffer was large enough, so we're done. Remember to remove the trailing nul!
+                let out_value_utf16 = &out_buffer[..size as usize - 1];
+                let out_value = String::from_utf16_lossy(out_value_utf16);
+                return Value::string(out_value, call_span);
+            }
+            size => {
+                // The buffer was too small, so we need to resize and try again.
+                // Clear first to indicate we don't care about the old contents.
+                out_buffer.clear();
+                out_buffer.resize(size as usize, 0);
+                continue;
+            }
+        }
+    }
+}
+
+fn reg_value_to_nu_list_string(reg_value: winreg::RegValue, call_span: Span) -> nu_protocol::Value {
+    let values = <Vec<String>>::from_reg_value(&reg_value)
+        .expect("registry value type should be REG_MULTI_SZ")
+        .into_iter()
+        .map(|s| Value::string(s, call_span));
+
+    // There's no REG_MULTI_EXPAND_SZ, so no need to do placeholder expansion here.
+    Value::list(values.collect(), call_span)
+}
+
+fn reg_value_to_nu_int(reg_value: winreg::RegValue, call_span: Span) -> nu_protocol::Value {
+    let value = match reg_value.vtype {
+        REG_DWORD => u32::from_reg_value(&reg_value).unwrap() as i64,
+        REG_DWORD_BIG_ENDIAN => {
+            // winreg (v0.51.0) doesn't natively decode REG_DWORD_BIG_ENDIAN
+            u32::from_be_bytes(unsafe { *reg_value.bytes.as_ptr().cast() }) as i64
+        }
+        REG_QWORD => u64::from_reg_value(&reg_value).unwrap() as i64,
+        _ => unreachable!(
+            "registry value type should be REG_DWORD, REG_DWORD_BIG_ENDIAN, or REG_QWORD"
+        ),
+    };
+    Value::int(value, call_span)
+}