cadmpeg-ir 0.6.2

// SPDX-License-Identifier: Apache-2.0
//! Interfaces for detecting, inspecting, decoding, and encoding CAD formats.
//!
//! [`Codec`] is object-safe for runtime codec registries. Detection consumes a
//! byte prefix, inspection summarizes a seekable container, and decoding
//! produces a finalized [`CadIr`] plus a [`DecodeReport`].
//!
//! A codec implements only the required [`CodecBackend`] methods. The public
//! [`Codec::inspect`] and [`Codec::decode`] entry points are the
//! single enforcement point for root-input limits and session finalize checks;
//! they live on the sealed [`Codec`] trait, blanket-implemented for every
//! [`CodecBackend`], so a codec cannot override an entry point and drop the
//! enforcement.

use std::fmt;

use self::loss::DecodeLossCode;
use crate::document::CadIr;
use crate::report::{
    check::Finding,
    decode::{Coverage, DecodeReport, DecodeTransfer, TransferLedger},
    loss::{LossNote as DecodeLoss, StrictConsequence},
};
use crate::source_fidelity::SourceFidelity;
use crate::ContainerSummary;
use cadmpeg_core::decode::{
    DecodeArena, DecodeContext, DecodeDepth, DecodeMode, DecodePolicy, InspectOptions, View,
};
use cadmpeg_core::dialect::FormatIdentity;
use cadmpeg_core::{CodecError, ReadSeek};
#[cfg(feature = "schema")]
use schemars::JsonSchema;
use serde::{Deserialize, Serialize};

/// The stable registry format namespace of one codec, as a typed value.
#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
pub struct FormatId(&'static str);

impl FormatId {
    /// Builds the format id for a registry namespace word.
    #[must_use]
    pub const fn new(namespace: &'static str) -> Self {
        Self(namespace)
    }

    /// The registry namespace word.
    #[must_use]
    pub const fn as_str(self) -> &'static str {
        self.0
    }
}

impl fmt::Display for FormatId {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        f.pad(self.0)
    }
}

/// How confident a codec is that it can handle a given byte prefix.
#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
#[cfg_attr(feature = "schema", derive(JsonSchema))]
#[serde(rename_all = "snake_case")]
#[serde(deny_unknown_fields)]
pub enum Confidence {
    /// Definitely not this format.
    No,
    /// Weak signal, such as a generic container signature.
    Low,
    /// Plausible but not conclusive.
    Medium,
    /// Strong, format-specific signal.
    High,
}

impl fmt::Display for Confidence {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        f.write_str(match self {
            Self::No => "no",
            Self::Low => "low",
            Self::Medium => "medium",
            Self::High => "high",
        })
    }
}

/// Options controlling source decoding.
#[derive(Debug, Clone, Copy, Default, PartialEq, Eq, Serialize, Deserialize)]
#[cfg_attr(feature = "schema", derive(JsonSchema))]
#[serde(deny_unknown_fields)]
pub struct DecodeOptions {
    /// How much of the source to transfer.
    #[serde(default)]
    pub depth: DecodeDepth,
    /// Resource limits and failure-handling mode governing the decode.
    #[serde(default)]
    pub policy: DecodePolicy,
}

/// A decoded document plus its loss report.
///
/// The sealed [`Codec`] wrapper constructs this value after it stamps the source
/// classification onto the report and finalizes the IR and source fidelity.
/// `#[non_exhaustive]` blocks external struct literals so callers cannot skip
/// finalization. Read through [`Self::ir`] and [`Self::report`]. Consume with
/// [`Self::into_parts`] to obtain the source fidelity or edit either document.
#[derive(Debug, Clone, PartialEq)]
#[non_exhaustive]
pub struct DecodeResult {
    ir: CadIr,
    report: DecodeReport,
    source_fidelity: SourceFidelity,
}

/// A strict-policy refusal bound to the loss in its completed decode report.
///
/// Only the sealed decode wrapper constructs this value. The refusing loss is
/// therefore always an element of [`Self::report`].
#[derive(Debug)]
pub struct StrictDecodeRejection {
    report: Box<DecodeReport>,
    loss_index: usize,
}

#[derive(Debug, Clone, Copy)]
struct RejectingLossIndex(usize);

impl StrictDecodeRejection {
    fn new(report: DecodeReport, loss_index: RejectingLossIndex) -> Self {
        Self {
            report: Box::new(report),
            loss_index: loss_index.0,
        }
    }

    /// Returns the loss that caused the strict-policy refusal.
    #[must_use]
    pub fn loss(&self) -> &DecodeLoss {
        &self.report.losses[self.loss_index]
    }

    /// Returns the completed report that contains the refusing loss.
    #[must_use]
    pub fn report(&self) -> &DecodeReport {
        &self.report
    }
}

impl fmt::Display for StrictDecodeRejection {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        let loss = self.loss();
        write!(f, "{}: {}", loss.code, loss.message)
    }
}

/// Failure from the policy-enforcing decode entry point.
///
/// Backend and resource failures remain [`CodecError`] values. A strict-policy
/// refusal is separate because decoding completed and produced a report; the
/// caller must be able to serialize that evidence even though no document is
/// admitted.
#[derive(Debug, thiserror::Error)]
#[non_exhaustive]
pub enum DecodeFailure {
    /// The codec or decode context failed before a result was produced.
    #[error(transparent)]
    Codec(#[from] CodecError),
    /// Strict mode rejected the first loss whose floor requires refusal.
    #[error("strict mode rejects {rejection}")]
    StrictRejected {
        /// The completed report and its refusing loss.
        rejection: StrictDecodeRejection,
    },
}

/// What a backend returns from [`CodecBackend::decode_impl`].
///
/// Source identity is authored once, in `ir.source`. The sealed wrapper stamps
/// that classification onto the report; a backend cannot describe the document
/// and its report with two different identities.
#[derive(Debug, Clone, PartialEq)]
pub struct Decoded {
    /// The decoded document, with its source metadata authored by the backend.
    pub ir: CadIr,
    /// The report without classification.
    pub body: DecodeBody,
    /// Decode-time annotations and retained source records.
    pub source_fidelity: SourceFidelity,
}

/// A [`DecodeReport`] without its classification.
#[derive(Debug, Clone, PartialEq)]
pub struct DecodeBody {
    /// The depth the backend delivered and its geometry outcome. The sealed
    /// wrapper records this depth on the document.
    pub transfer: DecodeTransfer,
    /// Coverage measures keyed by their declared name.
    pub coverage: Coverage,
    /// Losses resolved during decoding.
    pub losses: Vec<DecodeLoss>,
    /// Codec-defined informational notes.
    pub notes: Vec<String>,
    /// Complete source-to-result accounting.
    pub transfer_ledger: TransferLedger,
}

impl DecodeBody {
    /// An empty body with the given transfer outcome.
    #[must_use]
    pub fn new(transfer: DecodeTransfer) -> Self {
        Self {
            transfer,
            coverage: Coverage::default(),
            losses: Vec::new(),
            notes: Vec::new(),
            transfer_ledger: TransferLedger::default(),
        }
    }
}

impl DecodeResult {
    /// Builds a result by stamping the document's source classification onto
    /// the report body, then canonicalizing the IR and source fidelity.
    ///
    /// A document without source metadata yields an unclassified report for
    /// `format`, the codec's registry format. The document records the depth
    /// the backend's transfer state reports; a depth shallower than
    /// `requested` adds an error loss.
    fn new(decoded: Decoded, format: FormatId, requested: DecodeDepth) -> Result<Self, CodecError> {
        let Decoded {
            mut ir,
            mut body,
            source_fidelity,
        } = decoded;
        let delivered = body.transfer.depth();
        ir.set_decode_depth(delivered);
        ir.check_decode_depth().map_err(|error| {
            CodecError::malformed(format_args!("codec {format:?} decoded {error}"))
        })?;
        if delivered < requested {
            body.losses.push(
                DecodeLossCode::DepthNotDelivered { delivered }.note(format!(
                    "requested {requested} depth but the {format} decoder delivered \
                     {delivered} depth, so the records only {requested} depth carries are absent"
                )),
            );
        }
        let classification = match ir.source.as_ref() {
            Some(source) => match source.classification() {
                FormatIdentity::Classified { dialects } => {
                    FormatIdentity::classified(dialects.clone())
                }
                FormatIdentity::Unclassified { format } => {
                    FormatIdentity::unclassified(format.to_owned())
                }
            },
            None => FormatIdentity::unclassified(format.as_str().to_owned()),
        };
        ir.finalize();
        Ok(Self {
            ir,
            report: DecodeReport::from_body(classification, body),
            source_fidelity,
        })
    }

    /// Borrow the finalized IR.
    pub fn ir(&self) -> &CadIr {
        &self.ir
    }

    /// Borrow the transfer report.
    pub fn report(&self) -> &DecodeReport {
        &self.report
    }

    /// Consume into IR, report, and source fidelity.
    pub fn into_parts(self) -> (CadIr, DecodeReport, SourceFidelity) {
        (self.ir, self.report, self.source_fidelity)
    }
}

/// Decoder and container inspector for one source format.
pub trait CodecBackend {
    /// Registry format namespace this codec decodes, e.g. `"f3d"`.
    ///
    /// The sealed wrapper reports it as [`Codec::id`] and refuses any result
    /// whose primary format names another namespace.
    const FORMAT: FormatId;

    /// Findings this codec's own validator reports over its native
    /// namespace on a decoded document.
    ///
    /// The default reports nothing, for a codec that writes no native
    /// namespace or has no validator for it. The sealed wrapper exposes it as
    /// [`Codec::validate_native`], which the application runs after
    /// `validate_neutral`.
    fn validate_native(_ctx: &DecodeContext<'_>, _ir: &CadIr) -> Result<Vec<Finding>, CodecError> {
        Ok(Vec::new())
    }

    /// Judge, from a leading byte prefix, whether this codec applies.
    fn detect_impl(
        &self,
        ctx: &DecodeContext<'_>,
        prefix: View<'_>,
    ) -> Result<Confidence, CodecError>;

    /// Enumerate the acquired root view's streams/segments without decoding
    /// geometry.
    ///
    /// Implemented by each codec; never called by the CLI or registry. The
    /// [`Codec::inspect`] wrapper acquires the root under the inspection's
    /// input limit and runs this under an internal context.
    fn inspect_impl(
        &self,
        ctx: &DecodeContext<'_>,
        root: View<'_>,
    ) -> Result<ContainerSummary, CodecError>;

    /// Decode the acquired root view, reporting incomplete or approximate
    /// transfer.
    ///
    /// Implemented by each codec; never called by the CLI or registry. The
    /// [`Codec::decode`] wrapper acquires the root and finalizes the
    /// context around this call.
    fn decode_impl(&self, ctx: &DecodeContext<'_>, root: View<'_>) -> Result<Decoded, CodecError>;
}

mod sealed {
    /// Private bound for the blanket [`Codec`](super::Codec) implementation.
    pub trait Sealed {}
    impl<C: super::CodecBackend + ?Sized> Sealed for C {}
}

/// Public inspection and decoding entry points.
///
/// ```compile_fail
/// use cadmpeg_ir::codec::{
///     Codec, CodecBackend, Confidence, DecodeOptions, DecodeResult, Decoded, FormatId,
/// };
/// use cadmpeg_core::{CodecError, ReadSeek};
/// use cadmpeg_core::decode::{DecodeContext, View};
/// use cadmpeg_core::decode::InspectOptions;
/// use cadmpeg_ir::ContainerSummary;
///
/// struct Rogue;
/// impl CodecBackend for Rogue {
///     const FORMAT: FormatId = FormatId::new("rogue");
///     fn detect_impl(&self, _: &DecodeContext<'_>, _: View<'_>) -> Result<Confidence, CodecError> { Ok(Confidence::No) }
///     fn inspect_impl(&self, _: &DecodeContext<'_>, _: View<'_>)
///         -> Result<ContainerSummary, CodecError> { panic!("never runs") }
///     fn decode_impl(&self, _: &DecodeContext<'_>, _: View<'_>)
///         -> Result<Decoded, CodecError> { panic!("never runs") }
/// }
/// impl Codec for Rogue {
///     fn id(&self) -> FormatId { FormatId::new("rogue") }
///     fn detect(&self, _: &DecodeContext<'_>, _: View<'_>) -> Result<Confidence, CodecError> { Ok(Confidence::No) }
///     fn validate_native(&self, _: &DecodeContext<'_>, _: &cadmpeg_ir::CadIr)
///         -> Result<Vec<cadmpeg_ir::report::check::Finding>, CodecError> {
///         Ok(Vec::new())
///     }
///     fn inspect(&self, _: &mut dyn ReadSeek, _: &InspectOptions)
///         -> Result<ContainerSummary, CodecError> { panic!("never runs") }
///     fn decode(&self, _: &mut dyn ReadSeek, _: &DecodeOptions)
///         -> Result<DecodeResult, cadmpeg_ir::codec::DecodeFailure> {
///         panic!("never runs")
///     }
/// }
/// ```
pub trait Codec: sealed::Sealed {
    /// Registry format namespace, [`CodecBackend::FORMAT`].
    fn id(&self) -> FormatId;

    /// Judge, from a leading byte prefix, whether this codec applies.
    fn detect(&self, ctx: &DecodeContext<'_>, prefix: View<'_>) -> Result<Confidence, CodecError>;

    /// Findings this codec reports over its own native namespace,
    /// [`CodecBackend::validate_native`].
    fn validate_native(
        &self,
        ctx: &DecodeContext<'_>,
        ir: &CadIr,
    ) -> Result<Vec<Finding>, CodecError>;

    /// Inspects the source under its input and resource limits.
    fn inspect(
        &self,
        reader: &mut dyn ReadSeek,
        options: &InspectOptions,
    ) -> Result<ContainerSummary, CodecError>;

    /// Decodes the source under its input and resource limits.
    ///
    /// [`DecodeMode::Strict`] refuses the decode with
    /// [`DecodeFailure::StrictRejected`] for the first reported loss whose
    /// [`StrictConsequence`] is [`StrictConsequence::Reject`]. The gate
    /// evaluates geometry and full requests only: a container-only request
    /// keeps its losses and is never refused. This gate owns the refusal predicate and
    /// the refusal class for every codec. A backend reports its losses with
    /// their strict floors and adds no strict gate of its own; a local gate
    /// widens the predicate and reclassifies the refusal without the caller
    /// seeing it.
    fn decode(
        &self,
        reader: &mut dyn ReadSeek,
        options: &DecodeOptions,
    ) -> Result<DecodeResult, DecodeFailure>;

    /// Decodes an acquired root in the caller's detection session, under the
    /// session's policy and requested depth.
    /// The caller finishes the session after this method returns.
    fn decode_with_context(
        &self,
        ctx: &DecodeContext<'_>,
        root: View<'_>,
    ) -> Result<DecodeResult, DecodeFailure>;
}

impl<C: CodecBackend + ?Sized> Codec for C {
    fn id(&self) -> FormatId {
        C::FORMAT
    }

    fn detect(&self, ctx: &DecodeContext<'_>, prefix: View<'_>) -> Result<Confidence, CodecError> {
        self.detect_impl(ctx, prefix)
    }

    fn validate_native(
        &self,
        ctx: &DecodeContext<'_>,
        ir: &CadIr,
    ) -> Result<Vec<Finding>, CodecError> {
        C::validate_native(ctx, ir)
    }

    fn inspect(
        &self,
        reader: &mut dyn ReadSeek,
        options: &InspectOptions,
    ) -> Result<ContainerSummary, CodecError> {
        let arena = DecodeArena::new();
        let policy = DecodePolicy {
            mode: DecodeMode::Salvage,
            limits: options.limits,
        };
        let (ctx, root) = DecodeContext::read_root(reader, &arena, &policy, DecodeDepth::Full)?;
        let result = self.inspect_impl(&ctx, root);
        ctx.finish_session()?;
        let result = result?;
        if result.format() != C::FORMAT.as_str() {
            return Err(CodecError::WrongFormat(format!(
                "codec {:?} inspected a {:?} container",
                C::FORMAT.as_str(),
                result.format()
            )));
        }
        Ok(result)
    }

    fn decode(
        &self,
        reader: &mut dyn ReadSeek,
        options: &DecodeOptions,
    ) -> Result<DecodeResult, DecodeFailure> {
        let arena = DecodeArena::new();
        let (ctx, root) = DecodeContext::read_root(reader, &arena, &options.policy, options.depth)?;
        let result = self.decode_with_context(&ctx, root);
        ctx.finish_session()?;
        result
    }

    fn decode_with_context(
        &self,
        ctx: &DecodeContext<'_>,
        root: View<'_>,
    ) -> Result<DecodeResult, DecodeFailure> {
        let decoded = self.decode_impl(ctx, root)?;
        let result = DecodeResult::new(decoded, C::FORMAT, ctx.depth())?;
        if result.report().format() != C::FORMAT.as_str() {
            return Err(CodecError::WrongFormat(format!(
                "codec {:?} decoded a {:?} document",
                C::FORMAT.as_str(),
                result.report().format()
            ))
            .into());
        }
        crate::validate::evaluation_cycles::admit_evaluation_cycles(ctx, result.ir())?;
        let strict_loss_index = if ctx.policy().mode == DecodeMode::Strict && !ctx.container_only()
        {
            result
                .report()
                .losses
                .iter()
                .position(|loss| loss.strict_consequence() == StrictConsequence::Reject)
                .map(RejectingLossIndex)
        } else {
            None
        };
        if let Some(loss_index) = strict_loss_index {
            let (_, report, _) = result.into_parts();

            let rejection = StrictDecodeRejection::new(report, loss_index);
            return Err(DecodeFailure::StrictRejected { rejection });
        }
        Ok(result)
    }
}

mod loss;
pub mod write;

#[cfg(test)]
mod tests;