Implement OSDInspector offline tool (#85)

programs/src/osd_inspector.rs

@@ -0,0 +1,529 @@
+//! Bounded, offline command implementation for the `OSDInspector` tool.
+
+use clap::{Parser, Subcommand, ValueEnum};
+use libremetaverse::structured_data::{OSD, OSDParser, OSDType};
+use libremetaverse::types::{Material, PCode, PathCurve, ProfileCurve, Quaternion, UUID, Vector3};
+use libremetaverse::{Primitive, PrimitiveConstructionData, PrimitiveObjectProperties};
+use std::fmt;
+use std::fs::File;
+use std::io::{self, BufReader, BufWriter, Read, Write};
+use std::path::{Path, PathBuf};
+use std::process::ExitCode;
+
+/// Successful command completion.
+pub const EXIT_SUCCESS: u8 = 0;
+/// Command-line usage errors. Clap also uses this value for invalid arguments.
+pub const EXIT_USAGE: u8 = 2;
+/// File or standard-stream I/O failed.
+pub const EXIT_IO: u8 = 3;
+/// Input was too large or was not valid OSD.
+pub const EXIT_INVALID_OSD: u8 = 4;
+/// Valid OSD could not be interpreted as a primitive.
+pub const EXIT_INVALID_PRIMITIVE: u8 = 5;
+
+const DEFAULT_MAX_INPUT_BYTES: u64 = OSD::DEFAULT_MAX_BINARY_BYTES as u64;
+
+#[derive(Debug, Parser)]
+#[command(
+    name = "osd-inspector",
+    version,
+    about = "Inspect, convert, and validate LibreMetaverse structured data offline",
+    long_about = None,
+    arg_required_else_help = true
+)]
+pub struct Cli {
+    /// Maximum accepted input size. Parsing also enforces the library depth and node limits.
+    #[arg(long, global = true, default_value_t = DEFAULT_MAX_INPUT_BYTES, value_name = "BYTES")]
+    max_input_bytes: u64,
+
+    #[command(subcommand)]
+    command: Command,
+}
+
+#[derive(Debug, Subcommand)]
+enum Command {
+    /// Display an OSD value's type and recursively formatted structure.
+    #[command(alias = "i")]
+    Inspect {
+        /// Input file, or '-' for standard input.
+        input: PathBuf,
+    },
+    /// Convert OSD to JSON, XML, binary LLSD, or notation LLSD.
+    #[command(alias = "c")]
+    Convert {
+        /// Input file, or '-' for standard input.
+        input: PathBuf,
+        /// Output format (json/j, xml/x, binary/bin/b, notation/llsd/n).
+        format: OutputFormat,
+        /// Output file, or '-' for standard output.
+        output: PathBuf,
+    },
+    /// Validate an OSD value and report its root shape.
+    #[command(alias = "v")]
+    Validate {
+        /// Input file, or '-' for standard input.
+        input: PathBuf,
+    },
+    /// Emit a deterministic sample cube as pretty JSON.
+    PrimToOsd,
+    /// Parse an OSD value as a Primitive and summarize it.
+    OsdToPrim {
+        /// Input file, or '-' for standard input.
+        input: PathBuf,
+    },
+}
+
+#[derive(Clone, Copy, Debug, ValueEnum)]
+enum OutputFormat {
+    #[value(alias = "j")]
+    Json,
+    #[value(alias = "x")]
+    Xml,
+    #[value(alias = "bin", alias = "b")]
+    Binary,
+    #[value(alias = "llsd", alias = "n")]
+    Notation,
+}
+
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+enum InputFormat {
+    Json,
+    Xml,
+    Binary,
+    Notation,
+}
+
+#[derive(Debug)]
+enum InspectorError {
+    Io { action: String, source: io::Error },
+    InputTooLarge { limit: u64 },
+    InvalidOsd { input: String },
+    Serialization { format: OutputFormat },
+    InvalidPrimitive,
+}
+
+impl InspectorError {
+    const fn exit_code(&self) -> u8 {
+        match self {
+            Self::Io { .. } => EXIT_IO,
+            Self::InputTooLarge { .. } | Self::InvalidOsd { .. } => EXIT_INVALID_OSD,
+            Self::Serialization { .. } | Self::InvalidPrimitive => EXIT_INVALID_PRIMITIVE,
+        }
+    }
+}
+
+impl fmt::Display for InspectorError {
+    fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
+        match self {
+            Self::Io { action, source } => write!(formatter, "{action}: {source}"),
+            Self::InputTooLarge { limit } => {
+                write!(formatter, "input exceeds the configured {limit}-byte limit")
+            }
+            Self::InvalidOsd { input } => write!(formatter, "invalid OSD input: {input}"),
+            Self::Serialization { format } => {
+                write!(formatter, "could not serialize OSD as {format:?}")
+            }
+            Self::InvalidPrimitive => write!(formatter, "OSD value is not a Primitive map"),
+        }
+    }
+}
+
+/// Parses command-line arguments with clap and runs the offline tool.
+#[must_use]
+pub fn main_entry() -> ExitCode {
+    let cli = Cli::parse();
+    let stdin = io::stdin();
+    let stdout = io::stdout();
+    let stderr = io::stderr();
+    let mut input = stdin.lock();
+    let mut output = stdout.lock();
+    let mut errors = stderr.lock();
+    match execute(cli, &mut input, &mut output) {
+        Ok(()) => ExitCode::from(EXIT_SUCCESS),
+        Err(error) => {
+            let _ = writeln!(errors, "osd-inspector: {error}");
+            ExitCode::from(error.exit_code())
+        }
+    }
+}
+
+fn execute(cli: Cli, stdin: &mut dyn Read, stdout: &mut dyn Write) -> Result<(), InspectorError> {
+    if cli.max_input_bytes == 0 {
+        return Err(InspectorError::InputTooLarge { limit: 0 });
+    }
+    match cli.command {
+        Command::Inspect { input } => inspect(&input, cli.max_input_bytes, stdin, stdout),
+        Command::Convert {
+            input,
+            format,
+            output,
+        } => convert(&input, format, &output, cli.max_input_bytes, stdin, stdout),
+        Command::Validate { input } => validate(&input, cli.max_input_bytes, stdin, stdout),
+        Command::PrimToOsd => primitive_to_osd(stdout),
+        Command::OsdToPrim { input } => {
+            osd_to_primitive(&input, cli.max_input_bytes, stdin, stdout)
+        }
+    }
+}
+
+fn inspect(
+    input: &Path,
+    limit: u64,
+    stdin: &mut dyn Read,
+    output: &mut dyn Write,
+) -> Result<(), InspectorError> {
+    let bytes = read_input(input, limit, stdin)?;
+    let osd = parse_osd(&bytes, input)?;
+    writeln!(output, "File: {}", display_path(input)).map_err(stdout_error)?;
+    writeln!(output, "Type: {}", type_name(osd.type_())).map_err(stdout_error)?;
+    writeln!(output, "Size: {} bytes", bytes.len()).map_err(stdout_error)?;
+    writeln!(output, "\nStructure:\n{}", "-".repeat(60)).map_err(stdout_error)?;
+    display_osd(&osd, 0, output)?;
+    writeln!(output, "{}", "-".repeat(60)).map_err(stdout_error)
+}
+
+fn convert(
+    input: &Path,
+    format: OutputFormat,
+    destination: &Path,
+    limit: u64,
+    stdin: &mut dyn Read,
+    stdout: &mut dyn Write,
+) -> Result<(), InspectorError> {
+    let bytes = read_input(input, limit, stdin)?;
+    let osd = parse_osd(&bytes, input)?;
+    let encoded = serialize(osd, format)?;
+    if is_standard_stream(destination) {
+        stdout.write_all(&encoded).map_err(stdout_error)?;
+        stdout.flush().map_err(stdout_error)?;
+    } else {
+        write_file(destination, &encoded)?;
+        writeln!(
+            stdout,
+            "Converted {} to {}: {} ({} bytes)",
+            display_path(input),
+            format_name(format),
+            destination.display(),
+            encoded.len()
+        )
+        .map_err(stdout_error)?;
+    }
+    Ok(())
+}
+
+fn validate(
+    input: &Path,
+    limit: u64,
+    stdin: &mut dyn Read,
+    output: &mut dyn Write,
+) -> Result<(), InspectorError> {
+    let bytes = read_input(input, limit, stdin)?;
+    let osd = parse_osd(&bytes, input)?;
+    writeln!(output, "Valid OSD file").map_err(stdout_error)?;
+    writeln!(output, "  Type: {}", type_name(osd.type_())).map_err(stdout_error)?;
+    match osd {
+        OSD::Map(values) => writeln!(output, "  Keys: {}", values.len()).map_err(stdout_error)?,
+        OSD::Array(values) => {
+            writeln!(output, "  Elements: {}", values.len()).map_err(stdout_error)?;
+        }
+        _ => {}
+    }
+    Ok(())
+}
+
+fn primitive_to_osd(output: &mut dyn Write) -> Result<(), InspectorError> {
+    let osd = sample_primitive()
+        .and_then(|primitive| primitive.get_osd())
+        .map_err(|_| InspectorError::InvalidPrimitive)?;
+    let json = OSDParser::serialize_json_string(osd, Some(true)).map_err(|_| {
+        InspectorError::Serialization {
+            format: OutputFormat::Json,
+        }
+    })?;
+    writeln!(output, "{json}").map_err(stdout_error)
+}
+
+fn osd_to_primitive(
+    input: &Path,
+    limit: u64,
+    stdin: &mut dyn Read,
+    output: &mut dyn Write,
+) -> Result<(), InspectorError> {
+    let bytes = read_input(input, limit, stdin)?;
+    let osd = parse_osd(&bytes, input)?;
+    if !matches!(osd, OSD::Map(_)) {
+        return Err(InspectorError::InvalidPrimitive);
+    }
+    let primitive = Primitive::from_osd(osd).map_err(|_| InspectorError::InvalidPrimitive)?;
+    let name = primitive
+        .properties
+        .as_ref()
+        .map_or("Unknown", |properties| properties.name.as_str());
+    writeln!(output, "Successfully parsed Primitive:").map_err(stdout_error)?;
+    writeln!(output, "  ID: {}", primitive.id).map_err(stdout_error)?;
+    writeln!(output, "  Name: {name}").map_err(stdout_error)?;
+    writeln!(output, "  Type: {:?}", primitive.type_()).map_err(stdout_error)?;
+    writeln!(output, "  Position: {}", primitive.position.to_string()).map_err(stdout_error)?;
+    writeln!(output, "  Scale: {}", primitive.scale.to_string()).map_err(stdout_error)?;
+    writeln!(output, "  Material: {:?}", primitive.prim_data.material).map_err(stdout_error)?;
+    writeln!(output, "  PCode: {:?}", primitive.prim_data.p_code).map_err(stdout_error)
+}
+
+fn sample_primitive() -> Result<Primitive, libremetaverse::Error> {
+    let mut primitive = Primitive::new_with_constructor()?;
+    primitive.id = UUID::new_with_string("11111111-2222-3333-4444-555555555555".into())?;
+    primitive.local_id = 12_345;
+    primitive.position = Vector3 {
+        x: 128.0,
+        y: 128.0,
+        z: 25.0,
+    };
+    primitive.rotation = Quaternion::identity();
+    primitive.scale = Vector3 {
+        x: 1.0,
+        y: 1.0,
+        z: 1.0,
+    };
+    let mut construction = PrimitiveConstructionData::new_with_constructor()?;
+    construction.p_code = PCode::Prim;
+    construction.material = Material::Wood;
+    construction.path_curve = PathCurve::Line;
+    construction.path_scale_x = 1.0;
+    construction.path_scale_y = 1.0;
+    construction.set_profile_curve_with_property(ProfileCurve::Square);
+    primitive.prim_data = construction;
+    let mut properties = PrimitiveObjectProperties::new()?;
+    properties.name = "Example Cube".into();
+    properties.description = "A simple cube created with LibreMetaverse".into();
+    primitive.properties = Some(properties);
+    Ok(primitive)
+}
+
+fn read_input(path: &Path, limit: u64, stdin: &mut dyn Read) -> Result<Vec<u8>, InspectorError> {
+    if is_standard_stream(path) {
+        return read_bounded(stdin, limit, "reading standard input");
+    }
+    let file = File::open(path).map_err(|source| InspectorError::Io {
+        action: format!("opening {}", path.display()),
+        source,
+    })?;
+    if file.metadata().map_or(0, |value| value.len()) > limit {
+        return Err(InspectorError::InputTooLarge { limit });
+    }
+    read_bounded(
+        &mut BufReader::new(file),
+        limit,
+        &format!("reading {}", path.display()),
+    )
+}
+
+fn read_bounded(input: &mut dyn Read, limit: u64, action: &str) -> Result<Vec<u8>, InspectorError> {
+    let capacity = usize::try_from(limit.min(64 * 1024)).unwrap_or(64 * 1024);
+    let mut bytes = Vec::with_capacity(capacity);
+    input
+        .take(limit.saturating_add(1))
+        .read_to_end(&mut bytes)
+        .map_err(|source| InspectorError::Io {
+            action: action.into(),
+            source,
+        })?;
+    if u64::try_from(bytes.len()).unwrap_or(u64::MAX) > limit {
+        return Err(InspectorError::InputTooLarge { limit });
+    }
+    Ok(bytes)
+}
+
+fn write_file(path: &Path, bytes: &[u8]) -> Result<(), InspectorError> {
+    let file = File::create(path).map_err(|source| InspectorError::Io {
+        action: format!("creating {}", path.display()),
+        source,
+    })?;
+    let mut writer = BufWriter::new(file);
+    writer
+        .write_all(bytes)
+        .and_then(|()| writer.flush())
+        .map_err(|source| InspectorError::Io {
+            action: format!("writing {}", path.display()),
+            source,
+        })
+}
+
+fn parse_osd(bytes: &[u8], path: &Path) -> Result<OSD, InspectorError> {
+    if bytes.is_empty() || bytes.iter().all(u8::is_ascii_whitespace) {
+        return Err(InspectorError::InvalidOsd {
+            input: display_path(path),
+        });
+    }
+    let hinted = format_hint(path, bytes);
+    let mut attempts = vec![hinted];
+    for format in [
+        InputFormat::Json,
+        InputFormat::Xml,
+        InputFormat::Binary,
+        InputFormat::Notation,
+    ] {
+        if !attempts.contains(&format) {
+            attempts.push(format);
+        }
+    }
+    for format in attempts {
+        if let Ok(value) = parse_as(bytes, format) {
+            return Ok(value);
+        }
+    }
+    Err(InspectorError::InvalidOsd {
+        input: display_path(path),
+    })
+}
+
+fn parse_as(bytes: &[u8], format: InputFormat) -> Result<OSD, libremetaverse::Error> {
+    match format {
+        InputFormat::Json => OSDParser::deserialize_json_with_string(
+            String::from_utf8(bytes.to_vec()).map_err(|_| libremetaverse::Error::Argument)?,
+        ),
+        InputFormat::Xml => OSDParser::deserialize_llsd_xml_with_bytes(bytes.to_vec()),
+        InputFormat::Binary => OSDParser::deserialize_llsd_binary_with_bytes(bytes.to_vec()),
+        InputFormat::Notation => OSDParser::deserialize_llsd_notation_with_string(
+            String::from_utf8(bytes.to_vec()).map_err(|_| libremetaverse::Error::Argument)?,
+        ),
+    }
+}
+
+fn format_hint(path: &Path, bytes: &[u8]) -> InputFormat {
+    match path
+        .extension()
+        .and_then(|extension| extension.to_str())
+        .map(str::to_ascii_lowercase)
+        .as_deref()
+    {
+        Some("json") => InputFormat::Json,
+        Some("xml") => InputFormat::Xml,
+        Some("bin" | "binary") => InputFormat::Binary,
+        Some("notation" | "nt") => InputFormat::Notation,
+        _ => sniff_format(bytes),
+    }
+}
+
+fn sniff_format(bytes: &[u8]) -> InputFormat {
+    let text = String::from_utf8_lossy(&bytes[..bytes.len().min(64)]);
+    let trimmed = text.trim_start_matches('\u{feff}').trim_start();
+    if trimmed.starts_with("<? llsd/binary ?>") || trimmed.starts_with("<?llsd/binary?>") {
+        InputFormat::Binary
+    } else if trimmed.starts_with('<') {
+        InputFormat::Xml
+    } else if matches!(
+        trimmed.as_bytes().first(),
+        Some(b'{' | b'[' | b'"' | b'-' | b'0'..=b'9')
+    ) || trimmed.starts_with("true")
+        || trimmed.starts_with("false")
+        || trimmed.starts_with("null")
+    {
+        InputFormat::Json
+    } else {
+        InputFormat::Notation
+    }
+}
+
+fn serialize(osd: OSD, format: OutputFormat) -> Result<Vec<u8>, InspectorError> {
+    let result = match format {
+        OutputFormat::Json => {
+            OSDParser::serialize_json_string(osd, Some(true)).map(String::into_bytes)
+        }
+        OutputFormat::Xml => OSDParser::serialize_llsd_xml_bytes(osd),
+        OutputFormat::Binary => OSDParser::serialize_llsd_binary_with_osd(osd),
+        OutputFormat::Notation => OSDParser::serialize_llsd_notation(osd).map(String::into_bytes),
+    };
+    result.map_err(|_| InspectorError::Serialization { format })
+}
+
+fn display_osd(osd: &OSD, indent: usize, output: &mut dyn Write) -> Result<(), InspectorError> {
+    let prefix = "  ".repeat(indent);
+    match osd {
+        OSD::Map(values) => {
+            writeln!(output, "{prefix}Map ({} keys)", values.len()).map_err(stdout_error)?;
+            let mut entries = values.iter().collect::<Vec<_>>();
+            entries.sort_unstable_by_key(|(key, _)| *key);
+            for (key, value) in entries {
+                write!(output, "{prefix}  {key}: ").map_err(stdout_error)?;
+                if matches!(value, OSD::Map(_) | OSD::Array(_)) {
+                    writeln!(output).map_err(stdout_error)?;
+                    display_osd(value, indent + 2, output)?;
+                } else {
+                    display_osd(value, 0, output)?;
+                }
+            }
+        }
+        OSD::Array(values) => {
+            writeln!(output, "{prefix}Array ({} elements)", values.len()).map_err(stdout_error)?;
+            for (index, value) in values.iter().enumerate() {
+                write!(output, "{prefix}  [{index}]: ").map_err(stdout_error)?;
+                if matches!(value, OSD::Map(_) | OSD::Array(_)) {
+                    writeln!(output).map_err(stdout_error)?;
+                    display_osd(value, indent + 2, output)?;
+                } else {
+                    display_osd(value, 0, output)?;
+                }
+            }
+        }
+        OSD::String(value) => writeln!(output, "String: {value:?}").map_err(stdout_error)?,
+        OSD::Integer(value) => writeln!(output, "Integer: {value}").map_err(stdout_error)?,
+        OSD::Real(value) => writeln!(output, "Real: {value}").map_err(stdout_error)?,
+        OSD::Boolean(value) => writeln!(output, "Boolean: {value}").map_err(stdout_error)?,
+        OSD::UUID(value) => writeln!(output, "UUID: {value}").map_err(stdout_error)?,
+        OSD::Date(_) => writeln!(output, "Date: {}", osd.as_string().unwrap_or_default())
+            .map_err(stdout_error)?,
+        OSD::Uri(value) => writeln!(output, "URI: {}", value.0).map_err(stdout_error)?,
+        OSD::Binary(value) => {
+            writeln!(output, "Binary: {} bytes", value.len()).map_err(stdout_error)?;
+        }
+        OSD::LlsdXml(value) => writeln!(output, "LLSD XML: {value:?}").map_err(stdout_error)?,
+        OSD::Undefined => writeln!(output, "Unknown (Undefined)").map_err(stdout_error)?,
+        _ => writeln!(output, "Unknown ({:?})", osd.type_()).map_err(stdout_error)?,
+    }
+    Ok(())
+}
+
+const fn type_name(value: OSDType) -> &'static str {
+    match value {
+        OSDType::Unknown => "Unknown",
+        OSDType::Boolean => "Boolean",
+        OSDType::Integer => "Integer",
+        OSDType::Real => "Real",
+        OSDType::String => "String",
+        OSDType::UUID => "UUID",
+        OSDType::Date => "Date",
+        OSDType::URI => "URI",
+        OSDType::Binary => "Binary",
+        OSDType::Map => "Map",
+        OSDType::Array => "Array",
+        OSDType::LlsdXml => "LlsdXml",
+    }
+}
+
+const fn format_name(format: OutputFormat) -> &'static str {
+    match format {
+        OutputFormat::Json => "json",
+        OutputFormat::Xml => "xml",
+        OutputFormat::Binary => "binary",
+        OutputFormat::Notation => "notation",
+    }
+}
+
+fn is_standard_stream(path: &Path) -> bool {
+    path.as_os_str() == "-"
+}
+
+fn display_path(path: &Path) -> String {
+    if is_standard_stream(path) {
+        "<stdin>".into()
+    } else {
+        path.display().to_string()
+    }
+}
+
+fn stdout_error(source: io::Error) -> InspectorError {
+    InspectorError::Io {
+        action: "writing standard output".into(),
+        source,
+    }
+}