supercode_interchange/session/

claude_code.rs

//! Claude Code session codec: loaders, writers and native-record helpers.

use super::*;

mod read_index;
pub use read_index::ClaudeReadIndex;
mod append;
pub(crate) use append::ClaudeAppendState;

impl Session {
    /// Load a Claude Code transcript from a file, attaching any subagent
    /// (`Task`) sub-conversations stored alongside it.
    pub fn from_claude_code(path: impl AsRef<Path>) -> Result<Session> {
        Self::from_claude_code_with_fidelity(path, Fidelity::ByteLossless)
    }

    /// [`Session::from_claude_code`] at a declared [`Fidelity`] — see
    /// [`Session::load_with_fidelity`] for what [`Fidelity::Semantic`] buys.
    pub fn from_claude_code_with_fidelity(
        path: impl AsRef<Path>,
        fidelity: Fidelity,
    ) -> Result<Session> {
        let text = std::fs::read_to_string(path.as_ref())?;
        let mut session = Self::from_claude_code_str_with_fidelity(&text, fidelity)?;
        session.attach_claude_subagents(path.as_ref(), &text, fidelity)?;
        Ok(session)
    }

    /// Discover and load `<session>/subagents/agent-*.jsonl` files for a
    /// Claude Code transcript at `main_path`, linking each back to the parent
    /// `Task` tool call via the agent id embedded in the parent's tool result.
    pub(super) fn attach_claude_subagents(
        &mut self,
        main_path: &Path,
        main_text: &str,
        fidelity: Fidelity,
    ) -> Result<()> {
        let Some(dir) = subagents_dir_for(main_path) else {
            return Ok(());
        };
        let entries = std::fs::read_dir(&dir).map_err(|error| {
            crate::Error::Other(format!(
                "failed to enumerate Claude subagents at {}: {error}",
                dir.display()
            ))
        })?;
        let mut files = Vec::new();
        for entry in entries {
            let entry = entry.map_err(|error| {
                crate::Error::Other(format!(
                    "failed to enumerate Claude subagents at {}: {error}",
                    dir.display()
                ))
            })?;
            let path = entry.path();
            if path.extension().and_then(|extension| extension.to_str()) == Some("jsonl") {
                files.push(path);
            }
        }
        files.sort();

        // Phase 1 — collect each subagent + its recovered agent id, without
        // touching the main transcript yet.
        let mut collected: Vec<(Session, Option<String>)> = Vec::with_capacity(files.len());
        for file in files {
            let text = read_utf8_or_diagnose(&file).map_err(|error| {
                crate::Error::Other(format!(
                    "failed to read Claude subagent {}: {error}",
                    file.display()
                ))
            })?;
            let sub = match Self::from_claude_code_str_with_fidelity(&text, fidelity) {
                Ok(sub) => sub,
                // A read-only VIEW keeps the main conversation rather than
                // losing the whole session to one unreconstructable child;
                // the skip is named, not silent. Every stricter fidelity
                // still propagates the child's failure.
                Err(error) if fidelity.tolerates_residue() => {
                    self.load_residue.push(format!(
                        "Claude subagent {} could not be reconstructed ({error}); it was omitted from this view",
                        file.display()
                    ));
                    continue;
                }
                Err(error) => {
                    return Err(crate::Error::Other(format!(
                        "failed to reconstruct Claude subagent {}: {error}",
                        file.display()
                    )))
                }
            };
            // agentId: prefer the file's own record, fall back to the filename stem.
            let agent_id = first_agent_id(&text).or_else(|| {
                file.file_stem()
                    .and_then(|s| s.to_str())
                    .map(|s| s.trim_start_matches("agent-").to_string())
            });
            collected.push((sub, agent_id));
        }

        // Phase 2 — single pass over the main transcript to index every
        // requested agent id at once, then assign each subagent's parent by
        // an O(1) lookup.
        let agent_ids: Vec<String> = collected.iter().filter_map(|(_, id)| id.clone()).collect();
        let index = parent_tool_use_index(main_text, &agent_ids);

        for (mut sub, agent_id) in collected {
            sub.meta.parent_tool_use_id = agent_id.as_ref().and_then(|id| index.get(id).cloned());
            sub.meta.agent_id = agent_id;
            self.subagents.push(sub);
        }
        Ok(())
    }

    /// Parse a Claude Code transcript from an in-memory JSONL string.
    pub fn from_claude_code_str(jsonl: &str) -> Result<Session> {
        Self::from_claude_code_str_with_fidelity(jsonl, Fidelity::ByteLossless)
    }

    /// [`Session::from_claude_code_str`] at a declared [`Fidelity`] — see
    /// [`Session::load_with_fidelity`] for what [`Fidelity::Semantic`] buys.
    pub fn from_claude_code_str_with_fidelity(jsonl: &str, fidelity: Fidelity) -> Result<Session> {
        let mut meta = SessionMeta::new(SessionSource::ClaudeCode);
        let mut messages = Vec::new();
        // IX-1: `raw` is captured STRICT-VERBATIM (blank lines, CRLF, trailing
        // whitespace all preserved) — separate from the blank-skipping
        // `non_empty_lines` walk just below, which still parses records only
        // (a blank line is not a JSON record and must not become one).
        let (raw_lines, raw_trailing_newline) = split_lines_verbatim(jsonl);
        let raw: Vec<String> = raw_lines.iter().map(|l| l.to_string()).collect();
        // PARITY-15: a malformed/truncated line is still tolerated (a
        // single bad line must not make an otherwise-healthy multi-
        // thousand-line session unloadable) — but it's no longer INVISIBLE.
        let mut parse_error_lines = 0usize;
        let mut index = ClaudeReplayIndex::default();
        let mut residue_lines = Vec::new();

        // Claude transcripts are append-only trees, not linear chat logs.
        // Build a lightweight graph index first so normalization sees the
        // same single active, post-compaction branch Claude Code would
        // resume. `raw` above deliberately remains the complete source.
        for (line_index, line) in raw_lines.iter().enumerate() {
            if line.trim().is_empty() {
                continue;
            }
            let v: Value = match serde_json::from_str(line) {
                Ok(v) => v,
                Err(_) => {
                    parse_error_lines += 1; // tolerate stray/corrupt lines
                    continue;
                }
            };
            capture_claude_meta(&v, &mut meta, line)?;
            index.observe(line_index, &v)?;
            if claude_residue_kind(&v).is_some() {
                residue_lines.push(line_index);
            }
        }

        let ClaudeReplaySelection {
            lines: replay_lines,
            residue: load_residue,
        } = index.select_lines(fidelity)?;
        let mut pending_assistant: Option<Value> = None;

        // PARITY-23: candidates cover ALL raw records, not just the active
        // replay branch. Classify during the first decode to avoid decoding
        // every conversation/tool payload again merely to rule it out here.
        // Capture must still follow ALL metadata restoration: a later foreign
        // envelope suppresses local residue, while a same-source envelope is
        // restored before local records are appended. Retain indexes, not Values.
        for record_index in residue_lines {
            let line = raw_lines[record_index];
            if let Ok(record) = serde_json::from_str::<Value>(line) {
                capture_claude_residue(&mut meta, record_index, line, &record);
            }
        }

        for line_index in replay_lines {
            let line = raw_lines[line_index];
            let v: Value = serde_json::from_str(line).map_err(crate::Error::Decode)?;

            if v.get("type").and_then(Value::as_str) == Some("assistant") {
                if v.get("isApiErrorMessage").and_then(Value::as_bool) == Some(true) {
                    flush_claude_assistant(&mut pending_assistant, &mut messages);
                    continue;
                }
                if let Some(pending) = pending_assistant.as_mut() {
                    if claude_assistant_message_id(pending).is_some_and(|message_id| {
                        claude_assistant_message_id(&v) == Some(message_id)
                    }) {
                        merge_claude_assistant_chunk(pending, &v);
                        continue;
                    }
                    flush_claude_assistant(&mut pending_assistant, &mut messages);
                }
                pending_assistant = Some(v);
                continue;
            }

            flush_claude_assistant(&mut pending_assistant, &mut messages);

            // WAVE-2 item 1: every Claude Code record carries a real
            // top-level `timestamp` (ISO-8601) — provenance stamping below
            // attaches it to every canonical `ChatMessage` this line
            // produces, together with the record UUID and assistant model.
            // `entry(...).or_insert_with` preserves any more-precise value a
            // role-specific loader already supplied.
            let before = messages.len();
            match v.get("type").and_then(Value::as_str) {
                Some("user") => push_claude_user(&v, &mut messages),
                Some("assistant") => push_claude_assistant(&v, &mut messages),
                Some("attachment") => push_claude_attachment(&v, &mut messages),
                Some("system") => push_claude_system(&v, &mut messages),
                _ => {} // mode, queue-operation, ... — skip
            }
            // UUID/model provenance remains meaningful even for legacy
            // records that predate Claude Code's timestamp field.
            capture_claude_record_provenance(&v, &mut messages[before..]);
            restore_single_grok_message(&v, &mut messages[before..]);
        }
        flush_claude_assistant(&mut pending_assistant, &mut messages);

        reorder_tool_results_after_calls(&mut messages);
        ensure_tool_results_paired(&mut messages);
        let imported_message_count = Some(messages.len());
        Ok(Session {
            meta,
            messages,
            subagents: Vec::new(),
            raw,
            raw_trailing_newline,
            imported_message_count,
            // Claude Code is line-oriented: `raw` is split directly out of
            // the source text (strict-verbatim, IX-1).
            raw_is_verbatim: true,
            parse_error_lines,
            load_residue,
        })
    }
}

// ---- Claude Code ----------------------------------------------------------

/// The `subagents/` directory for a Claude Code transcript at `<dir>/<stem>.jsonl`
/// is `<dir>/<stem>/subagents/`. Returns it only if it exists.
fn subagents_dir_for(main_path: &Path) -> Option<PathBuf> {
    let dir = main_path.parent()?;
    let stem = main_path.file_stem()?.to_str()?;
    let candidate = dir.join(stem).join("subagents");
    candidate.is_dir().then_some(candidate)
}

/// The first `agentId` recorded in a subagent transcript.
fn first_agent_id(jsonl: &str) -> Option<String> {
    for line in non_empty_lines(jsonl) {
        if let Ok(v) = serde_json::from_str::<Value>(line) {
            if let Some(id) = v.get("agentId").and_then(Value::as_str) {
                return Some(id.to_string());
            }
        }
    }
    None
}

/// Find the `tool_use_id` of each parent `Task` call that spawned one of
/// `agent_ids`, by locating the parent transcript's `tool_result` whose
/// serialized content mentions the agent id. Best effort: an id with no
/// qualifying match is simply absent from the returned map.
///
/// Single pass over `main_text` — each line is parsed at most once,
/// regardless of how many agent ids are being sought — with each id's result
/// reproducing exactly what a per-id scan-and-parse-per-hit-line search would
/// return: the first line (in file order) whose raw text contains the id and
/// which — the first qualifying `tool_result` block in that line, in block
/// order — has a string `tool_use_id` and a serialized form that also
/// contains the id. A `tool_result` block matching on raw-line/serialized
/// containment but lacking a `tool_use_id` yields nothing for that id and
/// does not shadow a later match.
pub(super) fn parent_tool_use_index(
    main_text: &str,
    agent_ids: &[String],
) -> HashMap<String, String> {
    let mut index: HashMap<String, String> = HashMap::new();
    if agent_ids.is_empty() {
        return index;
    }

    for line in non_empty_lines(main_text) {
        if index.len() == agent_ids.len() {
            break;
        }
        // Cheap prefilter: every match this function can ever return comes
        // from a block whose raw line carries the literal JSON string value
        // `tool_result` (no JSON-escape variants of that ASCII literal).
        if !line.contains("tool_result") {
            continue;
        }
        let still_unmapped: Vec<&String> = agent_ids
            .iter()
            .filter(|id| !index.contains_key(id.as_str()))
            .collect();
        if still_unmapped.is_empty() {
            break;
        }
        let Ok(v) = serde_json::from_str::<Value>(line) else {
            continue;
        };
        let content = v.get("message").and_then(|m| m.get("content"));
        let Some(Value::Array(blocks)) = content else {
            continue;
        };
        for b in blocks {
            if b.get("type").and_then(Value::as_str) != Some("tool_result") {
                continue;
            }
            let Some(tool_use_id) = b.get("tool_use_id").and_then(Value::as_str) else {
                continue;
            };
            let block_str = b.to_string();
            for id in &still_unmapped {
                if index.contains_key(id.as_str()) {
                    continue;
                }
                if line.contains(id.as_str()) && block_str.contains(id.as_str()) {
                    index.insert((*id).clone(), tool_use_id.to_string());
                }
            }
        }
    }

    index
}

#[derive(Debug, Clone, Copy, PartialEq, Eq)]
enum ClaudeReplayKind {
    User,
    Assistant,
    Attachment,
    System,
}

impl ClaudeReplayKind {
    fn is_conversation(self) -> bool {
        matches!(self, Self::User | Self::Assistant)
    }
}

#[derive(Debug, Clone)]
struct ClaudeReplayNode {
    line_index: usize,
    uuid: String,
    parent_uuid: Option<String>,
    kind: ClaudeReplayKind,
    is_sidechain: bool,
    assistant_message_id: Option<String>,
    is_tool_result: bool,
    compact: Option<ClaudeCompactBoundary>,
}

#[derive(Debug, Clone)]
struct ClaudeCompactBoundary {
    anchor_uuid: Option<String>,
    preserved_uuids: Vec<String>,
    preserved_segment: Option<(String, String)>,
}

/// One projection of a Claude transcript graph: the source lines to replay,
/// plus whatever the projection had to give up to produce them (always empty
/// below [`Fidelity::Semantic`], which is the only level that degrades
/// instead of failing).
#[derive(Debug, Default)]
struct ClaudeReplaySelection {
    lines: Vec<usize>,
    residue: Vec<String>,
}

#[derive(Debug, Default, Clone)]
struct ClaudeReplayIndex {
    nodes: Vec<ClaudeReplayNode>,
    by_uuid: HashMap<String, usize>,
    segment_anchors: HashSet<String>,
    last_prompt: Option<(String, bool)>,
    linear_lines: Vec<usize>,
}

impl ClaudeReplayIndex {
    fn observe(&mut self, line_index: usize, v: &Value) -> Result<()> {
        if v.get("type").and_then(Value::as_str) == Some("last-prompt") {
            if let Some(leaf) = v.get("leafUuid").and_then(Value::as_str) {
                self.last_prompt = Some((
                    leaf.to_string(),
                    v.get("explicit").and_then(Value::as_bool) == Some(true),
                ));
            }
            return Ok(());
        }

        // A fork-context-ref is a real Claude graph anchor, but not a replay
        // message. Its child is the first conversational record in the
        // exported fork, so reaching this UUID terminates the locally
        // replayable segment rather than indicating a broken parent edge.
        if v.get("type").and_then(Value::as_str) == Some("fork-context-ref") {
            if let Some(uuid) = v.get("uuid").and_then(Value::as_str) {
                self.segment_anchors.insert(uuid.to_string());
            }
            return Ok(());
        }

        let kind = match v.get("type").and_then(Value::as_str) {
            Some("user") => ClaudeReplayKind::User,
            Some("assistant") => ClaudeReplayKind::Assistant,
            Some("attachment") => ClaudeReplayKind::Attachment,
            Some("system") => ClaudeReplayKind::System,
            _ => return Ok(()),
        };
        self.linear_lines.push(line_index);
        let Some(uuid) = v.get("uuid").and_then(Value::as_str) else {
            return Ok(());
        };
        if self.by_uuid.contains_key(uuid) {
            return Err(claude_replay_error(format!(
                "duplicate uuid `{uuid}` in Claude transcript"
            )));
        }

        let compact = (kind == ClaudeReplayKind::System
            && v.get("subtype").and_then(Value::as_str) == Some("compact_boundary"))
        .then(|| ClaudeCompactBoundary::from_value(v));
        let assistant_message_id = (kind == ClaudeReplayKind::Assistant)
            .then(|| claude_assistant_message_id(v).map(str::to_string))
            .flatten();
        let is_tool_result = kind == ClaudeReplayKind::User
            && v.get("message")
                .and_then(|m| m.get("content"))
                .and_then(Value::as_array)
                .is_some_and(|blocks| {
                    blocks
                        .iter()
                        .any(|b| b.get("type").and_then(Value::as_str) == Some("tool_result"))
                });
        let node = ClaudeReplayNode {
            line_index,
            uuid: uuid.to_string(),
            parent_uuid: v
                .get("parentUuid")
                .and_then(Value::as_str)
                .map(str::to_string),
            kind,
            is_sidechain: v.get("isSidechain").and_then(Value::as_bool) == Some(true),
            assistant_message_id,
            is_tool_result,
            compact,
        };
        self.by_uuid.insert(uuid.to_string(), self.nodes.len());
        self.nodes.push(node);
        Ok(())
    }

    /// Project the transcript at `fidelity`.
    ///
    /// Below [`Fidelity::Semantic`] this is the STRICT projection every
    /// continuation, transfer and export path depends on: reconstruct
    /// Claude's own single active post-compaction branch, or fail naming what
    /// could not be reconstructed.
    ///
    /// [`Fidelity::Semantic`] is the read-only VIEW mode. A Claude transcript
    /// that has been compacted, summarized, or resumed across files routinely
    /// contains a live record whose `parentUuid` names a record that is no
    /// longer on disk. Strict projection rightly refuses — a continuation
    /// built on a guessed graph is silent loss — but a VIEW does not need a
    /// continuation, so this mode anchors each dangling edge as a segment
    /// root, projects every severed segment exactly as the active branch is
    /// projected, splices them back together in transcript order, and names
    /// every degradation in the returned residue instead of erroring.
    fn select_lines(mut self, fidelity: Fidelity) -> Result<ClaudeReplaySelection> {
        let lenient = fidelity.tolerates_residue();
        let mut residue = Vec::new();
        if self.nodes.is_empty() {
            // Older exports and many hand-authored compatibility fixtures do
            // not carry Claude's `uuid`/`parentUuid` graph fields. There is no
            // branch information to project in that shape, so preserve the
            // historical linear normalization behavior. Native graph-bearing
            // transcripts always take the projection below.
            return Ok(ClaudeReplaySelection {
                lines: self.linear_lines,
                residue,
            });
        }
        if lenient {
            self.anchor_dangling_parents(&mut residue);
        }
        // Last resort for a VIEW: a transcript whose graph is unprojectable
        // for some OTHER reason (a cycle, an unresolvable compact boundary)
        // still renders as the file's own record order. A read-only mirror
        // that cannot open a session at all is the defect this mode exists
        // to remove, so `Semantic` never returns an error.
        let fallback = lenient.then(|| self.linear_lines.clone());
        match self.project(lenient, &mut residue) {
            Ok(lines) => Ok(ClaudeReplaySelection { lines, residue }),
            Err(error) => match fallback {
                Some(lines) => {
                    residue.push(format!(
                        "the Claude record graph could not be projected ({error}); \
                         every record was stitched in transcript order instead"
                    ));
                    Ok(ClaudeReplaySelection { lines, residue })
                }
                None => Err(error),
            },
        }
    }

    /// Turn every edge that points outside the transcript into a segment
    /// root, naming the dangling uuids as residue.
    ///
    /// A `fork-context-ref` anchor is already a declared segment boundary,
    /// not a break, so it is left alone.
    fn anchor_dangling_parents(&mut self, residue: &mut Vec<String>) {
        let mut dangling = Vec::new();
        for idx in 0..self.nodes.len() {
            let Some(parent) = self.nodes[idx].parent_uuid.as_deref() else {
                continue;
            };
            if self.by_uuid.contains_key(parent) || self.segment_anchors.contains(parent) {
                continue;
            }
            dangling.push(format!("`{}` → `{parent}`", self.nodes[idx].uuid));
            self.nodes[idx].parent_uuid = None;
        }
        if dangling.is_empty() {
            return;
        }
        const NAMED: usize = 8;
        let total = dangling.len();
        let overflow = total.saturating_sub(NAMED);
        dangling.truncate(NAMED);
        let mut listed = dangling.join(", ");
        if overflow > 0 {
            listed.push_str(&format!(", and {overflow} more"));
        }
        residue.push(format!(
            "{total} Claude record(s) reference a parentUuid absent from the transcript and were \
             anchored as segment roots: {listed}"
        ));
    }

    fn project(&mut self, lenient: bool, residue: &mut Vec<String>) -> Result<Vec<usize>> {
        let mut retained = vec![true; self.nodes.len()];
        if let Some(boundary_index) = self.nodes.iter().rposition(|n| n.compact.is_some()) {
            let parents: Option<Vec<Option<String>>> = lenient.then(|| {
                self.nodes
                    .iter()
                    .map(|node| node.parent_uuid.clone())
                    .collect()
            });
            if let Err(error) = self.apply_latest_compaction(boundary_index, &mut retained) {
                let Some(parents) = parents else {
                    return Err(error);
                };
                // The boundary rewrites parents as it goes, so restore the
                // graph it half-edited before continuing without it.
                for (node, parent) in self.nodes.iter_mut().zip(parents) {
                    node.parent_uuid = parent;
                }
                retained.iter_mut().for_each(|keep| *keep = true);
                residue.push(format!(
                    "the latest Claude compact boundary could not be projected ({error}); \
                     no pre-compaction record was pruned from this view"
                ));
            }
        }
        let sidechain_only = self
            .nodes
            .iter()
            .enumerate()
            .filter(|(idx, node)| retained[*idx] && node.kind.is_conversation())
            .all(|(_, node)| node.is_sidechain);

        let explicit_leaf = self
            .last_prompt
            .as_ref()
            .filter(|(_, explicit)| *explicit)
            .and_then(|(uuid, _)| self.by_uuid.get(uuid).copied())
            .filter(|idx| retained[*idx]);
        let newest_non_sidechain = self
            .nodes
            .iter()
            .enumerate()
            .rev()
            .find(|(idx, node)| retained[*idx] && !node.is_sidechain)
            .map(|(idx, _)| idx);
        // Dedicated Claude subagent transcripts are sidechains by design:
        // every record, including their root user prompt, has
        // `isSidechain:true`. When there is no main-chain candidate, resume
        // the newest retained sidechain leaf instead of rejecting the child.
        let newest_sidechain = self
            .nodes
            .iter()
            .enumerate()
            .rev()
            .find(|(idx, node)| retained[*idx] && node.is_sidechain && node.kind.is_conversation())
            .map(|(idx, _)| idx);
        let mut active = explicit_leaf
            .or(newest_non_sidechain)
            .or(newest_sidechain)
            .ok_or_else(|| claude_replay_error("no Claude record remains after compaction"))?;

        // Metadata descendants such as turn_duration are leaves in the raw
        // graph. Claude resumes from their nearest user/assistant ancestor,
        // then appends those descendants to the reconstructed chain.
        let mut seeking = HashSet::new();
        while !self.nodes[active].kind.is_conversation() {
            if !seeking.insert(active) {
                return Err(claude_replay_error(
                    "cycle while resolving active Claude leaf",
                ));
            }
            active = self.parent_index(active, &retained)?;
        }

        let mut segments =
            vec![self.project_segment(active, &retained, sidechain_only, lenient)?];
        if lenient {
            for leaf in self.severed_segment_leaves(active, &retained) {
                segments.push(self.project_segment(leaf, &retained, sidechain_only, lenient)?);
            }
            if segments.len() > 1 {
                residue.push(format!(
                    "{} conversation segments were stitched in transcript order because the \
                     Claude record graph is severed",
                    segments.len()
                ));
            }
        }
        // Each segment keeps its own reconstructed order; the segments
        // themselves are spliced by where they start in the file.
        segments.retain(|segment| !segment.is_empty());
        segments.sort_by_key(|segment| {
            segment
                .iter()
                .map(|idx| self.nodes[*idx].line_index)
                .min()
                .unwrap_or(usize::MAX)
        });
        let mut ordered = Vec::new();
        let mut placed = HashSet::new();
        for idx in segments.into_iter().flatten() {
            if placed.insert(idx) {
                ordered.push(idx);
            }
        }

        self.recover_parallel_assistant_chunks(ordered, &retained)
            .map(|indices| {
                indices
                    .into_iter()
                    .map(|idx| self.nodes[idx].line_index)
                    .collect()
            })
    }

    /// Reconstruct one segment: `leaf`'s parent chain, oldest-first, plus the
    /// non-conversation descendants rooted at it.
    fn project_segment(
        &self,
        leaf: usize,
        retained: &[bool],
        sidechain_only: bool,
        lenient: bool,
    ) -> Result<Vec<usize>> {
        let mut reversed = Vec::new();
        let mut seen = HashSet::new();
        let mut cursor = Some(leaf);
        while let Some(idx) = cursor {
            if !seen.insert(idx) {
                return Err(claude_replay_error(format!(
                    "cycle in active Claude parentUuid chain at `{}`",
                    self.nodes[idx].uuid
                )));
            }
            reversed.push(idx);
            cursor = match self.nodes[idx].parent_uuid.as_deref() {
                Some(parent) => match self.by_uuid.get(parent).copied() {
                    Some(parent) => Some(parent),
                    None if self.segment_anchors.contains(parent) => None,
                    // Claude can resume a background child in-place while
                    // retaining only the new segment in that child's JSONL.
                    // Its first record then points to a UUID not present in
                    // the sidechain file. That external edge is a segment
                    // boundary, not corruption; the complete source remains
                    // available byte-for-byte in `raw`.
                    None if sidechain_only => None,
                    None => {
                        return Err(claude_replay_error(format!(
                            "active Claude record `{}` has missing parentUuid `{parent}`",
                            self.nodes[idx].uuid
                        )));
                    }
                },
                None => None,
            };
            if cursor.is_some_and(|parent| !retained[parent]) {
                if lenient {
                    // A compaction boundary is where this segment ends; the
                    // records it pruned stay pruned.
                    break;
                }
                return Err(claude_replay_error(format!(
                    "active Claude chain crosses an excluded compaction record from `{}`",
                    self.nodes[idx].uuid
                )));
            }
        }
        reversed.reverse();

        // Include non-conversation descendants rooted at the segment's leaf
        // (turn_duration, attachments, etc.), matching Claude's QYH/n1T.
        let mut descendants = Vec::new();
        let mut frontier = vec![leaf];
        let mut head = 0;
        while head < frontier.len() {
            let parent = frontier[head];
            head += 1;
            for (idx, node) in self.nodes.iter().enumerate() {
                if !retained[idx]
                    || node.kind.is_conversation()
                    || seen.contains(&idx)
                    || node.parent_uuid.as_deref() != Some(self.nodes[parent].uuid.as_str())
                {
                    continue;
                }
                seen.insert(idx);
                descendants.push(idx);
                frontier.push(idx);
            }
        }
        descendants.sort_by_key(|idx| self.nodes[*idx].line_index);
        reversed.extend(descendants);
        Ok(reversed)
    }

    /// The newest retained conversation record of every component the active
    /// leaf's own component cannot reach.
    ///
    /// Only a severed graph produces any: a healthy transcript is one
    /// component, so the abandoned branches a rewind left behind stay
    /// abandoned here exactly as they do under strict projection.
    fn severed_segment_leaves(&self, active: usize, retained: &[bool]) -> Vec<usize> {
        let active_root = self.component_root(active, retained);
        let mut newest_by_root: BTreeMap<usize, usize> = BTreeMap::new();
        for idx in 0..self.nodes.len() {
            if !retained[idx] || !self.nodes[idx].kind.is_conversation() {
                continue;
            }
            let Some(root) = self.component_root(idx, retained) else {
                continue;
            };
            if Some(root) == active_root {
                continue;
            }
            let newest = newest_by_root.entry(root).or_insert(idx);
            if self.nodes[idx].line_index > self.nodes[*newest].line_index {
                *newest = idx;
            }
        }
        newest_by_root.into_values().collect()
    }

    /// Walk `idx` up to the record that anchors its component, stopping at a
    /// root, an edge that leaves the transcript, or a pruned parent. `None`
    /// when the walk cycles.
    fn component_root(&self, idx: usize, retained: &[bool]) -> Option<usize> {
        let mut cursor = idx;
        let mut seen = HashSet::new();
        loop {
            if !seen.insert(cursor) {
                return None;
            }
            let next = self.nodes[cursor]
                .parent_uuid
                .as_deref()
                .and_then(|parent| self.by_uuid.get(parent).copied())
                .filter(|parent| retained[*parent]);
            match next {
                Some(parent) => cursor = parent,
                None => return Some(cursor),
            }
        }
    }

    fn apply_latest_compaction(
        &mut self,
        boundary_index: usize,
        retained: &mut [bool],
    ) -> Result<()> {
        let compact = self.nodes[boundary_index]
            .compact
            .clone()
            .expect("called with compact boundary");
        let mut preserved = compact.preserved_uuids;
        if preserved.is_empty() {
            if let Some((head, tail)) = compact.preserved_segment {
                preserved = self.walk_preserved_segment(&head, &tail)?;
            }
        }

        let preserved_set: HashSet<String> = preserved.iter().cloned().collect();
        for uuid in &preserved {
            if !self.by_uuid.contains_key(uuid) {
                return Err(claude_replay_error(format!(
                    "latest compact boundary references missing preserved uuid `{uuid}`"
                )));
            }
        }

        let removed_uuids: HashSet<String> = self
            .nodes
            .iter()
            .enumerate()
            .filter(|(idx, node)| *idx < boundary_index && !preserved_set.contains(&node.uuid))
            .map(|(_, node)| node.uuid.clone())
            .collect();
        for (idx, node) in self.nodes.iter().enumerate() {
            if idx < boundary_index && !preserved_set.contains(&node.uuid) {
                retained[idx] = false;
            }
        }

        if preserved.is_empty() {
            return Ok(());
        }
        let anchor = compact.anchor_uuid.ok_or_else(|| {
            claude_replay_error("preserved compact boundary is missing anchorUuid")
        })?;
        if !self.by_uuid.contains_key(&anchor) {
            return Err(claude_replay_error(format!(
                "latest compact boundary references missing anchor uuid `{anchor}`"
            )));
        }
        let tail = preserved.last().cloned().expect("non-empty preserved list");
        let mut parent = anchor.clone();
        for uuid in &preserved {
            let idx = self.by_uuid[uuid];
            self.nodes[idx].parent_uuid = Some(parent);
            parent = uuid.clone();
        }
        let first = &preserved[0];
        for node in &mut self.nodes {
            if node.parent_uuid.as_deref() == Some(anchor.as_str()) && node.uuid != *first {
                node.parent_uuid = Some(tail.clone());
            }
        }
        for node in &mut self.nodes {
            if node.kind.is_conversation()
                && node
                    .parent_uuid
                    .as_ref()
                    .is_some_and(|parent| removed_uuids.contains(parent))
            {
                node.parent_uuid = Some(tail.clone());
            }
        }
        Ok(())
    }

    fn walk_preserved_segment(&self, head: &str, tail: &str) -> Result<Vec<String>> {
        let mut reversed = Vec::new();
        let mut seen = HashSet::new();
        let mut cursor = tail;
        loop {
            if !seen.insert(cursor.to_string()) {
                return Err(claude_replay_error("cycle in compact preservedSegment"));
            }
            let idx = *self.by_uuid.get(cursor).ok_or_else(|| {
                claude_replay_error(format!(
                    "compact preservedSegment references missing uuid `{cursor}`"
                ))
            })?;
            reversed.push(cursor.to_string());
            if cursor == head {
                reversed.reverse();
                return Ok(reversed);
            }
            cursor = self.nodes[idx].parent_uuid.as_deref().ok_or_else(|| {
                claude_replay_error(format!(
                    "compact preservedSegment tail `{tail}` does not reach head `{head}`"
                ))
            })?;
        }
    }

    fn parent_index(&self, idx: usize, retained: &[bool]) -> Result<usize> {
        let parent = self.nodes[idx].parent_uuid.as_deref().ok_or_else(|| {
            claude_replay_error(format!(
                "Claude record `{}` has no conversational ancestor",
                self.nodes[idx].uuid
            ))
        })?;
        let parent_idx = *self.by_uuid.get(parent).ok_or_else(|| {
            claude_replay_error(format!(
                "Claude record `{}` has missing parentUuid `{parent}`",
                self.nodes[idx].uuid
            ))
        })?;
        if !retained[parent_idx] {
            return Err(claude_replay_error(format!(
                "Claude record `{}` points into compacted-out history",
                self.nodes[idx].uuid
            )));
        }
        Ok(parent_idx)
    }

    fn recover_parallel_assistant_chunks(
        &self,
        base: Vec<usize>,
        retained: &[bool],
    ) -> Result<Vec<usize>> {
        let selected: HashSet<usize> = base.iter().copied().collect();
        let mut replacements: HashMap<usize, Vec<usize>> = HashMap::new();
        let mut skipped_positions = HashSet::new();
        let mut handled_ids = HashSet::new();

        for (base_pos, idx) in base.iter().copied().enumerate() {
            let Some(message_id) = self.nodes[idx].assistant_message_id.as_deref() else {
                continue;
            };
            if !handled_ids.insert(message_id.to_string()) {
                continue;
            }
            let base_positions: Vec<usize> = base
                .iter()
                .enumerate()
                .filter(|(_, candidate)| {
                    self.nodes[**candidate].assistant_message_id.as_deref() == Some(message_id)
                })
                .map(|(pos, _)| pos)
                .collect();
            let anchor_pos = base_positions.first().copied().unwrap_or(base_pos);
            skipped_positions.extend(base_positions.iter().copied().skip(1));

            // A streamed Anthropic response can be stored as sibling records
            // rather than a literal parent chain. Reassemble every chunk at
            // the first active occurrence and restore raw chunk order before
            // the normalizer coalesces their content blocks.
            let mut chunks: Vec<usize> = self
                .nodes
                .iter()
                .enumerate()
                .filter(|(candidate, node)| {
                    retained[*candidate] && node.assistant_message_id.as_deref() == Some(message_id)
                })
                .map(|(candidate, _)| candidate)
                .collect();
            chunks.sort_by_key(|candidate| self.nodes[*candidate].line_index);

            let assistant_uuids: HashSet<&str> = self
                .nodes
                .iter()
                .filter(|node| node.assistant_message_id.as_deref() == Some(message_id))
                .map(|node| node.uuid.as_str())
                .collect();
            let mut results: Vec<usize> = self
                .nodes
                .iter()
                .enumerate()
                .filter(|(candidate, node)| {
                    retained[*candidate]
                        && !selected.contains(candidate)
                        && node.is_tool_result
                        && node
                            .parent_uuid
                            .as_deref()
                            .is_some_and(|parent| assistant_uuids.contains(parent))
                })
                .map(|(candidate, _)| candidate)
                .collect();
            results.sort_by_key(|candidate| self.nodes[*candidate].line_index);
            chunks.extend(results);
            replacements.insert(anchor_pos, chunks);
        }

        let mut out = Vec::with_capacity(selected.len());
        for (pos, idx) in base.into_iter().enumerate() {
            if let Some(replacement) = replacements.remove(&pos) {
                out.extend(replacement);
            } else if !skipped_positions.contains(&pos) {
                out.push(idx);
            }
        }
        Ok(out)
    }
}

impl ClaudeCompactBoundary {
    fn from_value(v: &Value) -> Self {
        let metadata = v.get("compactMetadata");
        let preserved_messages = metadata.and_then(|m| m.get("preservedMessages"));
        let anchor_uuid = preserved_messages
            .and_then(|p| p.get("anchorUuid"))
            .and_then(Value::as_str)
            .or_else(|| {
                metadata
                    .and_then(|m| m.get("preservedSegment"))
                    .and_then(|p| p.get("anchorUuid"))
                    .and_then(Value::as_str)
            })
            .map(str::to_string);
        let preserved_uuids = preserved_messages
            .and_then(|p| p.get("uuids"))
            .and_then(Value::as_array)
            .map(|uuids| {
                uuids
                    .iter()
                    .filter_map(Value::as_str)
                    .map(str::to_string)
                    .collect()
            })
            .unwrap_or_default();
        let preserved_segment =
            metadata
                .and_then(|m| m.get("preservedSegment"))
                .and_then(|segment| {
                    Some((
                        segment.get("headUuid")?.as_str()?.to_string(),
                        segment.get("tailUuid")?.as_str()?.to_string(),
                    ))
                });
        Self {
            anchor_uuid,
            preserved_uuids,
            preserved_segment,
        }
    }
}

fn claude_replay_error(message: impl Into<String>) -> crate::Error {
    crate::Error::Other(format!(
        "cannot reconstruct lossless Claude continuation: {}",
        message.into()
    ))
}

fn claude_assistant_message_id(v: &Value) -> Option<&str> {
    v.get("message")
        .and_then(|message| message.get("id"))
        .and_then(Value::as_str)
}

fn merge_claude_assistant_chunk(target: &mut Value, chunk: &Value) {
    let Some(target_message) = target.get_mut("message") else {
        return;
    };
    let Some(chunk_message) = chunk.get("message") else {
        return;
    };
    let mut content = target_message
        .get("content")
        .and_then(Value::as_array)
        .cloned()
        .unwrap_or_default();
    if let Some(blocks) = chunk_message.get("content").and_then(Value::as_array) {
        content.extend(blocks.iter().cloned());
    }
    let mut merged_message = chunk_message.clone();
    merged_message["content"] = Value::Array(content);
    *target_message = merged_message;
}

fn flush_claude_assistant(pending: &mut Option<Value>, out: &mut Vec<ChatMessage>) {
    let Some(v) = pending.take() else {
        return;
    };
    let reasoning_only = claude_assistant_message_id(&v).is_some()
        && v.get("message")
            .and_then(|message| message.get("content"))
            .and_then(Value::as_array)
            .is_some_and(|blocks| {
                !blocks.is_empty()
                    && blocks.iter().all(|block| {
                        matches!(
                            block.get("type").and_then(Value::as_str),
                            Some("thinking" | "redacted_thinking")
                        )
                    })
            });
    if reasoning_only {
        return;
    }
    let before = out.len();
    push_claude_assistant(&v, out);
    capture_claude_record_provenance(&v, &mut out[before..]);
    restore_single_grok_message(&v, &mut out[before..]);
}

/// Attach the record identity, clock, and actual assistant model to every
/// canonical message produced from one Claude JSONL record. These fields are
/// deliberately per-message: a continued transcript can cross a provider
/// boundary, so the session-level source model is not authoritative for its
/// appended tail.
fn capture_claude_record_provenance(v: &Value, messages: &mut [ChatMessage]) {
    let timestamp = v.get("timestamp").and_then(Value::as_str);
    let uuid = v.get("uuid").and_then(Value::as_str);
    let model = v
        .get("message")
        .and_then(|message| message.get("model"))
        .and_then(Value::as_str);
    for message in messages {
        if let Some(timestamp) = timestamp {
            message
                .metadata
                .entry("timestamp".to_string())
                .or_insert_with(|| timestamp.to_string());
        }
        if let Some(uuid) = uuid {
            message
                .metadata
                .entry("claude_uuid".to_string())
                .or_insert_with(|| uuid.to_string());
        }
        if let Some(model) = model {
            message
                .metadata
                .entry("model".to_string())
                .or_insert_with(|| model.to_string());
        }
    }
}

fn capture_claude_meta(v: &Value, meta: &mut SessionMeta, raw_line: &str) -> Result<()> {
    restore_codex_provenance_from_top_level(v, meta)?;
    if meta.session_id.is_none() {
        if let Some(id) = v.get("sessionId").and_then(Value::as_str) {
            meta.session_id = Some(id.to_string());
        }
    }
    if meta.cwd.is_none() {
        if let Some(cwd) = v.get("cwd").and_then(Value::as_str) {
            meta.cwd = Some(PathBuf::from(cwd));
        }
    }
    if meta.model.is_none() {
        if let Some(model) = v
            .get("message")
            .and_then(|m| m.get("model"))
            .and_then(Value::as_str)
        {
            meta.model = Some(model.to_string());
        }
    }
    // PARITY-10 (provenance P010): `fork-context-ref` is an extremely rare
    // real Claude Code record with no confirmed field shape (see
    // `ClaudeRecord::ForkContextRef`'s doc comment) — rather than guess at
    // named fields and risk silently mis-modeling it, stash the WHOLE raw
    // line verbatim under a lineage key. `write_claude_code_records` (below)
    // re-emits it byte-for-byte, so the record survives the Claude Code
    // semantic writer (not just the CLI's raw-passthrough diagonal path) —
    // and `to_codex_jsonl`'s synthesized header carries a namespaced copy
    // so a Claude -> Codex -> Claude round trip can still reconstruct it
    // (dev/03). A session can only fork from one context, so the first one
    // seen wins, matching every other "first wins" field above.
    //
    // D4 (Fable-5 review, confirmed): this used to store `v.to_string()` —
    // a RE-SERIALIZATION of the parsed `Value`, not the original source
    // text. `serde_json::Value` here has no `preserve_order` feature (see
    // `Cargo.toml`), so its object keys are a `BTreeMap` and get silently
    // REORDERED (alphabetized) on re-emit — the "byte-for-byte" claim in
    // this very comment was false. Fixed the cheap+honest way: store the
    // caller's own already-verbatim source `raw_line` text instead of
    // re-serializing `v` at all, so this is now ACTUALLY byte-for-byte
    // (key order, spacing, everything) rather than merely
    // structurally-equivalent JSON.
    if v.get("type").and_then(Value::as_str) == Some("fork-context-ref")
        && !meta.lineage.contains_key("claude_fork_context_ref_raw")
    {
        meta.lineage.insert(
            "claude_fork_context_ref_raw".to_string(),
            raw_line.to_string(),
        );
    }
    Ok(())
}

fn push_claude_user(v: &Value, out: &mut Vec<ChatMessage>) {
    let content = v.get("message").and_then(|m| m.get("content"));
    let provenance = claude_user_provenance(v);
    match content {
        Some(Value::String(s)) => {
            if !s.trim().is_empty() {
                out.push(ChatMessage::user(s.clone()).with_metas(&provenance));
            }
        }
        Some(Value::Array(blocks)) => {
            let mut text = String::new();
            // IX-5: image blocks alongside/instead of text — collected
            // separately (never synthesized on a malformed shape, see
            // `claude_image_block_to_part`) so a multimodal user turn
            // survives as `content_parts` instead of the image silently
            // vanishing.
            let mut images: Vec<Value> = Vec::new();
            // D5: an `image` block whose `source` isn't base64/url (e.g. a
            // Files-API `{"source":{"type":"file","file_id":..}}`
            // reference) makes `claude_image_block_to_part` return `None` —
            // track that it was SEEN even though it couldn't be converted,
            // so an image-ONLY record (no text, no convertible image) isn't
            // silently dropped below (the same vanishing-record bug-class
            // PARITY-11 fixed for reasoning-only turns).
            let mut saw_unconvertible_image = false;
            for b in blocks {
                match b.get("type").and_then(Value::as_str) {
                    Some("text") => push_text(&mut text, b.get("text")),
                    Some("tool_result") => {
                        let id = b
                            .get("tool_use_id")
                            .and_then(Value::as_str)
                            .unwrap_or_default();
                        // PARITY-11 (nested images): `extract_tool_result_content`
                        // captures any `image` blocks nested inside this
                        // `tool_result` into `content_parts` (via
                        // `claude_image_block_to_part`, the same conversion the
                        // top-level `image` block path already uses) instead of
                        // flattening them to the bare `[image]` marker text the
                        // old `extract_tool_result` emitted — the everyday
                        // "Read a PNG / screenshot tool output" shape.
                        let (result, images) =
                            extract_tool_result_content(b.get("content"), v.get("toolUseResult"));
                        let mut msg = tool_message(id, result);
                        if !images.is_empty() {
                            // D-mix (Fable review, must-fix): `content_parts`
                            // is a self-contained contract — the pi writer
                            // (`pi_content_value`) reads ONLY `content_parts`
                            // for a `Role::Tool` message and never falls back
                            // to `msg.content`, so on a MIXED text+image
                            // tool_result a bare `content_parts: [image]`
                            // silently drops the sibling text on `convert
                            // --to pi` (a regression vs. the pre-PARITY-11
                            // baseline, which at least preserved the text).
                            // Prepend the text as part 0, exactly mirroring
                            // `pi_content_to_text_and_parts` and
                            // `push_opencode_user`'s identical
                            // self-contained-parts construction. `msg.content`
                            // keeps the text too (unchanged) for the writers
                            // that read text from `msg.content` and only scan
                            // `content_parts` for `image_url` entries
                            // (`claude_tool_result_content_value`,
                            // `codex_tool_output_text`, the opencode
                            // assistant writer) — those already filter
                            // strictly on `image_url`/text-typed lookups, so
                            // this text part is never double-counted.
                            let mut parts = Vec::new();
                            if let Some(t) = &msg.content {
                                if !t.is_empty() {
                                    parts.push(serde_json::json!({"type": "text", "text": t}));
                                }
                            }
                            parts.extend(images);
                            msg.content_parts = Some(parts);
                        }
                        // The assistant turn that issued this tool call — the
                        // tool-pairing graph edge (parallel to parentUuid).
                        if let Some(src) = v.get("sourceToolAssistantUUID").and_then(Value::as_str)
                        {
                            msg.metadata
                                .insert("sourceToolAssistantUUID".to_string(), src.to_string());
                        }
                        // TR-10: preserve the Claude wire `is_error` flag so
                        // the reduction layer's success/failure boundary
                        // (`ReductionKind::ToolInputElided` must never target
                        // an errored call) survives import — `ChatMessage`
                        // otherwise has no structural slot for it.
                        if b.get("is_error").and_then(Value::as_bool) == Some(true) {
                            crate::mark_tool_error(&mut msg);
                        } else {
                            restore_tool_outcome_extension(v, &mut msg);
                        }
                        out.push(msg);
                    }
                    Some("image") => match claude_image_block_to_part(b) {
                        Some(part) => images.push(part),
                        None => saw_unconvertible_image = true,
                    },
                    _ => {} // document / unknown — skip
                }
            }
            // D5: nothing convertible landed in `text`/`images` but an
            // image block WAS present — fold in the same short bracketed
            // marker convention already used for `[web_search]`/`[model
            // fallback: ...]` rather than letting the record vanish.
            if images.is_empty() && text.trim().is_empty() && saw_unconvertible_image {
                push_str_field(&mut text, UNCONVERTIBLE_IMAGE_MARKER);
            }
            let before = out.len();
            if !images.is_empty() {
                let mut parts = Vec::new();
                if !text.trim().is_empty() {
                    parts.push(serde_json::json!({"type": "text", "text": text.clone()}));
                }
                parts.extend(images);
                out.push(
                    ChatMessage {
                        role: Role::User,
                        content: None,
                        content_parts: Some(parts),
                        tool_calls: None,
                        tool_call_id: None,
                        name: None,
                        metadata: Default::default(),
                    }
                    .with_metas(&provenance),
                );
            } else if !text.trim().is_empty() {
                out.push(ChatMessage::user(text).with_metas(&provenance));
            }
            if saw_unconvertible_image && out.len() > before {
                if let Some(msg) = out.last_mut() {
                    msg.metadata
                        .insert("image_source_unconvertible".to_string(), "true".to_string());
                }
            }
        }
        _ => {}
    }
}

/// D5 (Fable-5 review): the bracketed marker folded into `text` when an
/// `image` block is SEEN but [`claude_image_block_to_part`] can't convert it
/// (e.g. a Files-API `{"source":{"type":"file",...}}` reference) and nothing
/// else in the record survives either — matches the existing
/// `[web_search]`/`[model fallback: ...]` convention rather than letting the
/// whole record vanish (`push_claude_user`/`push_claude_assistant`).
pub(super) const UNCONVERTIBLE_IMAGE_MARKER: &str =
    "[image: source not captured — unsupported/unconvertible image reference]";

/// Parse a Claude Code user-turn `image` content block
/// (`{"type":"image","source":{"type":"base64","media_type":..,"data":..}}`
/// or `{"type":"image","source":{"type":"url","url":..}}`) into a
/// `content_parts` `image_url` entry (a `data:` URI for the base64 form, the
/// bare URL for the url form) — the inverse of
/// [`claude_user_content_value`]'s emission. Only a well-formed source
/// (non-empty `media_type`/`data`, or non-empty `url`) is recognized;
/// anything else — including a well-formed but unconvertible source like a
/// Files-API `{"type":"file",...}` reference (D5) — is left as raw-only
/// residue rather than synthesizing a corrupt/empty part (mirrors the
/// pi/opencode loaders' `pi_image_shape`/`opencode_file_image_part`
/// discipline). Callers must not let that turn the record invisible though:
/// see [`UNCONVERTIBLE_IMAGE_MARKER`].
pub(super) fn claude_image_block_to_part(b: &Value) -> Option<Value> {
    let source = b.get("source")?;
    match source.get("type").and_then(Value::as_str) {
        Some("base64") => {
            let mime = source.get("media_type").and_then(Value::as_str)?;
            let data = source.get("data").and_then(Value::as_str)?;
            if mime.is_empty() || data.is_empty() {
                return None;
            }
            Some(serde_json::json!({
                "type": "image_url",
                "image_url": {"url": format!("data:{mime};base64,{data}")},
            }))
        }
        Some("url") => {
            let url = source.get("url").and_then(Value::as_str)?;
            if url.is_empty() {
                return None;
            }
            Some(serde_json::json!({"type": "image_url", "image_url": {"url": url}}))
        }
        _ => None,
    }
}

/// Rebuild a Claude Code user-turn `message.content` value from a
/// `ChatMessage` — the inverse of `push_claude_user`'s parse (text blocks +
/// [`claude_image_block_to_part`]). When `content_parts` is absent this
/// MUST reproduce the historical plain-string `content` exactly (IX-5's
/// overriding constraint: a text-only message's export stays byte-identical)
/// — only a multimodal message (`content_parts` present, e.g. imported from
/// Pi/OpenCode or a real Claude Code image turn) gets the Anthropic
/// content-array shape, one `text` block (if any non-empty text part) plus
/// one `image` block per `image_url` part (`data:` URI → `source.base64`;
/// any other URL → `source.url`).
fn claude_user_content_value(msg: &ChatMessage) -> Value {
    match &msg.content_parts {
        Some(parts) => {
            let mut blocks = Vec::new();
            for p in parts {
                match p.get("type").and_then(Value::as_str) {
                    Some("text") => {
                        if let Some(t) = p.get("text").and_then(Value::as_str) {
                            if !t.is_empty() {
                                blocks.push(serde_json::json!({"type": "text", "text": t}));
                            }
                        }
                    }
                    Some("image_url") => {
                        if let Some(url) = p
                            .get("image_url")
                            .and_then(|u| u.get("url"))
                            .and_then(Value::as_str)
                        {
                            blocks.push(match parse_data_uri(url) {
                                Some((mime, data)) => serde_json::json!({
                                    "type": "image",
                                    "source": {"type": "base64", "media_type": mime, "data": data},
                                }),
                                None => serde_json::json!({
                                    "type": "image",
                                    "source": {"type": "url", "url": url},
                                }),
                            });
                        }
                    }
                    _ => {}
                }
            }
            Value::Array(blocks)
        }
        None => Value::String(msg.content.clone().unwrap_or_default()),
    }
}

/// Rebuild a Claude Code `tool_result` block's `content` value from a `Tool`
/// `ChatMessage` — the inverse of [`extract_tool_result_content`]'s parse
/// (PARITY-11). When `content_parts` is absent (or empty) this MUST reproduce
/// the historical plain-string `content` exactly (same IX-5-style constraint
/// `claude_user_content_value` follows) — only a `tool_result` that actually
/// carries a captured nested image gets the Anthropic content-array shape,
/// one `text` block (the existing `msg.content`, if any) plus one `image`
/// block per `image_url` part (mirrors `claude_user_content_value`'s
/// `data:` URI -> `source.base64` / other URL -> `source.url` mapping).
fn claude_tool_result_content_value(msg: &ChatMessage) -> Value {
    match &msg.content_parts {
        Some(parts) if !parts.is_empty() => {
            let mut blocks = Vec::new();
            if let Some(t) = &msg.content {
                if !t.is_empty() {
                    blocks.push(serde_json::json!({"type": "text", "text": t}));
                }
            }
            for p in parts {
                if p.get("type").and_then(Value::as_str) == Some("image_url") {
                    if let Some(url) = p
                        .get("image_url")
                        .and_then(|u| u.get("url"))
                        .and_then(Value::as_str)
                    {
                        blocks.push(match parse_data_uri(url) {
                            Some((mime, data)) => serde_json::json!({
                                "type": "image",
                                "source": {"type": "base64", "media_type": mime, "data": data},
                            }),
                            None => serde_json::json!({
                                "type": "image",
                                "source": {"type": "url", "url": url},
                            }),
                        });
                    }
                }
            }
            Value::Array(blocks)
        }
        _ => Value::String(msg.content.clone().unwrap_or_default()),
    }
}

/// Collect the Claude Code user-turn provenance fields that distinguish real
/// human input from system-injected turns and record replay-relevant state.
pub(crate) fn claude_user_provenance(v: &Value) -> Vec<(String, String)> {
    let mut out = Vec::new();
    let mut take_str = |key: &str| {
        if let Some(s) = v.get(key).and_then(Value::as_str) {
            out.push((key.to_string(), s.to_string()));
        }
    };
    take_str("promptSource"); // typed | queued | system | sdk
    take_str("interruptedMessageId");
    take_str("sourceToolUseID");
    for flag in ["isMeta", "isCompactSummary", "isVisibleInTranscriptOnly"] {
        if v.get(flag).and_then(Value::as_bool) == Some(true) {
            out.push((flag.to_string(), "true".to_string()));
        }
    }
    if let Some(n) = v.get("queuePriority").and_then(Value::as_i64) {
        out.push(("queuePriority".to_string(), n.to_string()));
    }
    // `origin` is an object like {"kind":"task-notification"} — keep its kind.
    if let Some(kind) = v
        .get("origin")
        .and_then(|o| o.get("kind"))
        .and_then(Value::as_str)
    {
        out.push(("origin".to_string(), kind.to_string()));
    }
    out
}

/// Content-bearing Claude `system` events (`scheduled_task_fire`,
/// `local_command`, `away_summary`) carry real text that's part of the
/// interaction; fold them in as system context. Marker/metric subtypes
/// (`turn_duration`, `compact_boundary`, `api_error`, `stop_hook_summary`) carry
/// no conversational content and are skipped.
fn push_claude_system(v: &Value, out: &mut Vec<ChatMessage>) {
    let keep = matches!(
        v.get("subtype").and_then(Value::as_str),
        Some("scheduled_task_fire") | Some("local_command") | Some("away_summary")
    );
    if !keep {
        return;
    }
    if let Some(content) = v.get("content").and_then(Value::as_str) {
        if !content.trim().is_empty() {
            let subtype = v.get("subtype").and_then(Value::as_str).unwrap_or("system");
            out.push(ChatMessage::system(content.to_string()).with_meta("systemSubtype", subtype));
        }
    }
}

/// Fold content-bearing Claude Code `attachment` records into the conversation
/// as user-role messages. Most attachment subtypes (`task_reminder`,
/// `deferred_tools_delta`, `skill_listing`, `hook_*`, `command_permissions`, …)
/// are regenerable system injections and are skipped; only the four that carry
/// non-regenerable user/external content are kept.
fn push_claude_attachment(v: &Value, out: &mut Vec<ChatMessage>) {
    let att = match v.get("attachment") {
        Some(a) => a,
        None => return,
    };
    let kind = match att.get("type").and_then(Value::as_str) {
        Some(kind) => kind,
        None => return,
    };
    let text = match kind {
        // A queued prompt. `commandMode` says whose: `prompt` is the person's
        // own text, `task-notification` is the runtime reporting a finished
        // background task. Kept verbatim below.
        "queued_command" => att
            .get("prompt")
            .and_then(Value::as_str)
            .map(str::to_string),
        // A file the user attached: header + contents.
        "file" => attachment_with_path(att, "attached file", "filename", "content"),
        // A user-edited file snippet.
        "edited_text_file" => attachment_with_path(att, "edited file", "filename", "snippet"),
        // Injected project memory (CLAUDE.md), point-in-time.
        "nested_memory" => attachment_with_path(att, "project memory", "path", "content"),
        _ => None, // regenerable system injection — skip
    };
    let Some(text) = text else { return };
    if text.trim().is_empty() {
        return;
    }
    // An attachment record wears the user's ROLE, but the record itself says
    // who actually spoke — and that fact is lost the moment the attachment is
    // flattened to `[label: path]` text, so carry it as metadata the way
    // `push_claude_system` carries `systemSubtype`. Two fields, both verbatim:
    //
    //   `attachmentType`  the subtype. `file` / `edited_text_file` /
    //                     `nested_memory` are envelopes the runtime built
    //                     around a file body; a frontend that trusts the role
    //                     shows the reader a numbered source listing in a
    //                     chat bubble apparently sent by themselves.
    //   `commandMode`     present on `queued_command` only, and the whole
    //                     story for it. Measured over the local Claude Code
    //                     corpus (2,512 `queued_command` attachments): 926
    //                     `prompt`, every one of them plain human text, and
    //                     1,586 `task-notification`, every one of them a
    //                     `<task-notification>` frame — the same text Claude
    //                     Code also writes as a `type:"user"` record stamped
    //                     `origin.kind = "task-notification"`.
    //
    // Presentation policy (which of these a frontend hides) belongs to the
    // frontend — `isContextMessage` in `sdk/client/client.mjs`. The loader's
    // job is to stop discarding the producer's own answer.
    let mut message = ChatMessage::user(text).with_meta("attachmentType", kind);
    if let Some(mode) = att.get("commandMode").and_then(Value::as_str) {
        message = message.with_meta("commandMode", mode);
    }
    out.push(message);
}

/// Format an attachment as `[<label>: <path>]\n<body>`.
fn attachment_with_path(
    att: &Value,
    label: &str,
    path_key: &str,
    body_key: &str,
) -> Option<String> {
    let body = att.get(body_key).and_then(Value::as_str)?;
    let path = att
        .get(path_key)
        .or_else(|| att.get("displayPath"))
        .and_then(Value::as_str)
        .unwrap_or("");
    Some(format!("[{label}: {path}]\n{body}"))
}

pub(super) fn push_str_field(buf: &mut String, s: &str) {
    if !buf.is_empty() {
        buf.push('\n');
    }
    buf.push_str(s);
}

/// N3: build a synthesized message for reasoning that could not attach to a
/// following assistant turn — either interrupted mid-stream by a
/// non-assistant item, or dangling at EOF (an aborted-turn shape). Drains
/// the three pending buffers (all empty/`false` afterward) so callers don't
/// separately have to remember to clear them.
pub(super) fn orphaned_reasoning_message(
    reasoning: &mut String,
    reasoning_content: &mut String,
    encrypted: &mut bool,
) -> ChatMessage {
    let mut msg = ChatMessage::system("[reasoning] (turn ended without a reply)".to_string());
    if !reasoning.is_empty() {
        msg = msg.with_meta("reasoning", std::mem::take(reasoning));
    }
    if !reasoning_content.is_empty() {
        msg = msg.with_meta("reasoning_content", std::mem::take(reasoning_content));
    }
    if *encrypted {
        msg = msg.with_meta("reasoning_encrypted", "true");
        *encrypted = false;
    }
    msg
}

fn push_claude_assistant(v: &Value, out: &mut Vec<ChatMessage>) {
    let content = v.get("message").and_then(|m| m.get("content"));
    let mut text = String::new();
    let mut calls: Vec<ToolCall> = Vec::new();
    // Legacy singular fields — kept for backward compatibility with every
    // existing consumer of `metadata["thinking"]`/`["thinking_signature"]`/
    // `["redacted_thinking"]` (a concatenation of all thinking text, and the
    // LAST block's signature/data). D8 (Fable-5 review, confirmed): when a
    // message carries MULTIPLE `thinking` blocks, collapsing them down to
    // these singular fields silently drops every signature but the last
    // one's — a real Anthropic `thinking` block's `signature` cryptographically
    // covers ONLY that block's own text, so re-emitting block 1's text under
    // block 2's signature (or vice versa) produces a signature that will
    // never verify. `thinking_blocks` below is the fix: every block
    // preserved SEPARATELY, in order, each with its own (optional)
    // signature/data — the writer prefers it over the legacy fields
    // whenever present.
    let mut thinking = String::new();
    let mut signature: Option<String> = None;
    // PARITY-11: real Claude corpora also carry `redacted_thinking` and
    // `image` assistant blocks, and (rarely) a `fallback` model-routing
    // marker — none handled before, all silently vanishing (audit's own
    // census: 132,927 `thinking` / 67 `redacted_thinking` / 9 `image` / 4
    // `fallback` blocks in the reference corpus).
    //
    // D8: `redacted_thinking` is real data ONLY — never a fabricated
    // placeholder. The pre-fix code defaulted a missing `data` field to the
    // literal string `"<redacted>"`, which is indistinguishable from an
    // actual (if oddly-named) opaque payload on re-emit — a caller reading
    // it back has no way to tell "no data was ever captured" from "the
    // provider's own opaque blob happens to be the string `<redacted>`".
    // `redacted_thinking_seen` tracks block PRESENCE independently of
    // whether it had real data, so the reasoning-only-turn rescue below
    // still fires even when no block had a `data` field at all.
    let mut redacted_thinking: Option<String> = None;
    let mut redacted_thinking_seen = false;
    let mut images: Vec<Value> = Vec::new();
    // Real Anthropic `thinking` blocks very commonly carry an EMPTY
    // `thinking` string alongside a real `signature` (the summarized/
    // redacted-in-the-clear-but-replayable case) — `thinking.trim()` alone
    // would miss those, so track "a thinking block existed at all"
    // separately from whether it had visible text.
    let mut thinking_block_seen = false;
    // D8: every `thinking`/`redacted_thinking` block, preserved SEPARATELY
    // and IN ORDER — see the comment on `thinking`/`redacted_thinking`
    // above. Serialized as a single JSON-array metadata string
    // (`ChatMessage::metadata` is a flat string map) under
    // `"thinking_blocks"`.
    let mut thinking_blocks: Vec<Value> = Vec::new();
    // D5: mirrors `push_claude_user`'s tracking — an `image` block whose
    // source isn't base64/url (e.g. Files-API `{"type":"file",...}`) must
    // not silently vanish the whole record when nothing else survives.
    let mut saw_unconvertible_image = false;

    match content {
        Some(Value::String(s)) => push_text(&mut text, Some(&Value::String(s.clone()))),
        Some(Value::Array(blocks)) => {
            for b in blocks {
                match b.get("type").and_then(Value::as_str) {
                    Some("text") => push_text(&mut text, b.get("text")),
                    Some("tool_use") => {
                        let id = b.get("id").and_then(Value::as_str).unwrap_or_default();
                        let name = b.get("name").and_then(Value::as_str).unwrap_or_default();
                        let args = b
                            .get("input")
                            .map(|i| i.to_string())
                            .unwrap_or_else(|| "{}".to_string());
                        calls.push(function_call(id, name, args));
                    }
                    // Thinking is not replayed across providers, but retain it in
                    // (skip-serialized) metadata so a same-model continuation can
                    // re-inject it. See P3.
                    Some("thinking") => {
                        thinking_block_seen = true;
                        let t = b.get("thinking").and_then(Value::as_str).unwrap_or("");
                        if !t.is_empty() {
                            push_str_field(&mut thinking, t); // legacy concatenated field
                        }
                        let sig = b.get("signature").and_then(Value::as_str);
                        if let Some(s) = sig {
                            signature = Some(s.to_string()); // legacy last-wins field
                        }
                        // D8: this block's OWN text + signature, not folded
                        // into the running concatenation above.
                        let mut block = serde_json::json!({"type": "thinking", "thinking": t});
                        if let Some(s) = sig {
                            block["signature"] = Value::String(s.to_string());
                        }
                        thinking_blocks.push(block);
                    }
                    // Anthropic's redacted reasoning: an opaque, provider-private
                    // payload (flagged content the API declines to show in the
                    // clear). Like `thinking`, it's not replayable, but the raw
                    // `data` is retained in metadata rather than silently
                    // vanishing — a same-model continuation can still replay it
                    // verbatim even though supercode never renders it.
                    Some("redacted_thinking") => {
                        redacted_thinking_seen = true;
                        let data = b.get("data").and_then(Value::as_str);
                        // D8: no fabricated fallback — `data` is only ever
                        // the real captured payload, or genuinely absent.
                        if let Some(d) = data {
                            redacted_thinking = Some(d.to_string()); // legacy last-wins field
                        }
                        let mut block = serde_json::json!({"type": "redacted_thinking"});
                        if let Some(d) = data {
                            block["data"] = Value::String(d.to_string());
                        }
                        thinking_blocks.push(block);
                    }
                    // An assistant-emitted image block (e.g. a generated
                    // image) — collected exactly like `push_claude_user`'s
                    // user-turn image handling (`claude_image_block_to_part`
                    // is role-general), so it survives as `content_parts`
                    // instead of vanishing.
                    Some("image") => match claude_image_block_to_part(b) {
                        Some(part) => images.push(part),
                        None => saw_unconvertible_image = true,
                    },
                    // A provider-routing note (real shape:
                    // `{"type":"fallback","from":{"model":..},"to":{"model":..}}`
                    // — a mid-generation model swap, e.g. an overloaded model
                    // falling back to another). Carries no replayable
                    // conversational content, but folding it into `text` as a
                    // short bracketed marker — the same convention the Codex
                    // loader already uses for `[web_search]`/
                    // `[image_generation] ...` — keeps it visible instead of
                    // silently vanishing, including the case where it's the
                    // ONLY block in the turn (see the reasoning-only-turn fix
                    // below: before this, that shape dropped the entire
                    // message).
                    Some("fallback") => {
                        let from = b
                            .get("from")
                            .and_then(|f| f.get("model"))
                            .and_then(Value::as_str)
                            .unwrap_or("?");
                        let to = b
                            .get("to")
                            .and_then(|t| t.get("model"))
                            .and_then(Value::as_str)
                            .unwrap_or("?");
                        push_str_field(&mut text, &format!("[model fallback: {from} -> {to}]"));
                    }
                    _ => {}
                }
            }
        }
        _ => {}
    }

    // D5: nothing convertible landed in `text`/`images` but an image block
    // WAS present — fold in the same bracketed-marker convention `fallback`
    // uses above, so a genuinely image-only (unconvertible source) turn
    // doesn't vanish (mirrors `push_claude_user`'s identical fix).
    if images.is_empty() && text.trim().is_empty() && saw_unconvertible_image {
        push_str_field(&mut text, UNCONVERTIBLE_IMAGE_MARKER);
    }

    let before = out.len();
    if !images.is_empty() {
        let mut parts = Vec::new();
        if !text.trim().is_empty() {
            parts.push(serde_json::json!({"type": "text", "text": text.clone()}));
        }
        parts.extend(images);
        out.push(ChatMessage {
            role: Role::Assistant,
            content: None,
            content_parts: Some(parts),
            tool_calls: (!calls.is_empty()).then_some(calls),
            tool_call_id: None,
            name: None,
            metadata: Default::default(),
        });
    } else {
        push_assistant(out, text, calls);
        // A recognized native assistant record remains transcript state even
        // when its content array is empty (for example, an interrupted model
        // turn). Force a bare message whenever `push_assistant` had nothing
        // to emit. This includes the reasoning-only case and also preserves
        // genuinely part-less records instead of silently changing turn
        // count/order during translation.
        if out.len() == before {
            let mut empty = ChatMessage {
                role: Role::Assistant,
                content: None,
                content_parts: None,
                tool_calls: None,
                tool_call_id: None,
                name: None,
                metadata: Default::default(),
            };
            if !thinking_block_seen && !redacted_thinking_seen {
                empty
                    .metadata
                    .insert("empty_assistant_record".to_string(), "true".to_string());
            }
            out.push(empty);
        }
    }
    // Attach retained reasoning + attribution to the message we just produced.
    if out.len() > before {
        if let Some(msg) = out.last_mut() {
            // Insert "thinking" (even as an empty string) whenever a
            // `thinking` block was actually seen, not just when it had
            // visible text — a real `thinking` block commonly carries an
            // empty `thinking` string alongside a real `signature` (the
            // summarized-away-but-still-replayable case), and the writer
            // below keys its re-emission decision off this metadata key's
            // PRESENCE, not its content.
            if thinking_block_seen {
                msg.metadata.insert("thinking".to_string(), thinking);
            }
            if let Some(sig) = signature {
                msg.metadata.insert("thinking_signature".to_string(), sig);
            }
            if let Some(rt) = redacted_thinking {
                msg.metadata.insert("redacted_thinking".to_string(), rt);
            }
            // D8: exact per-block re-emission list — every `thinking`/
            // `redacted_thinking` block preserved separately, in order, each
            // with its own (optional) signature/data. The writer prefers
            // this over the legacy singular fields above whenever present,
            // so a multi-block message round-trips losslessly instead of
            // collapsing to one block under one (now-unverifiable)
            // signature.
            if !thinking_blocks.is_empty() {
                msg.metadata.insert(
                    "thinking_blocks".to_string(),
                    Value::Array(thinking_blocks).to_string(),
                );
            }
            // D5: honest signal that this message contained an image block
            // whose source this loader couldn't convert — the actual image
            // content is NOT captured, only a marker/partial record.
            if saw_unconvertible_image {
                msg.metadata
                    .insert("image_source_unconvertible".to_string(), "true".to_string());
            }
            // Attribution: which skill / subagent / MCP server+tool produced
            // this turn, plus the model `slug`.
            for key in [
                "attributionSkill",
                "attributionAgent",
                "attributionMcpServer",
                "attributionMcpTool",
                "slug",
            ] {
                if let Some(s) = v.get(key).and_then(Value::as_str) {
                    msg.metadata.insert(key.to_string(), s.to_string());
                }
            }
        }
    }
}

/// PARITY-23: Claude Code records that affect replay or carry source-native
/// state but have no canonical home — the claude-code residue inventory
/// (design doc §per-format). `last-prompt` steers leaf selection,
/// `fork-context-ref` anchors the replay graph, `file-history-snapshot` /
/// `queue-operation` / `mode` are durable native state.
pub(super) fn claude_residue_kind(record: &Value) -> Option<&'static str> {
    match record.get("type").and_then(Value::as_str) {
        Some("file-history-snapshot") => Some("file-history-snapshot"),
        Some("queue-operation") => Some("queue-operation"),
        Some("last-prompt") => Some("last-prompt"),
        Some("mode") => Some("mode"),
        Some("fork-context-ref") => Some("fork-context-ref"),
        _ => None,
    }
}

/// Re-emit a claude residue record under the session's CURRENT metadata:
/// `sessionId`/`cwd` keys are rewritten only when they exist and differ (an
/// explicit `--session-id`/`--cwd` override is a deliberate rewrite request
/// that extends to residue). When nothing differs — the plain return
/// diagonal — the raw line is emitted byte-exact.
fn claude_residue_line_for_emit(raw: &str, session_id: &str, cwd: &str) -> String {
    let Ok(mut value) = serde_json::from_str::<Value>(raw) else {
        return raw.to_string();
    };
    let Some(object) = value.as_object_mut() else {
        return raw.to_string();
    };
    let mut changed = false;
    for (key, current) in [("sessionId", session_id), ("cwd", cwd)] {
        if object
            .get(key)
            .and_then(Value::as_str)
            .is_some_and(|existing| existing != current)
        {
            object.insert(key.to_string(), Value::String(current.to_string()));
            changed = true;
        }
    }
    if changed {
        value.to_string()
    } else {
        raw.to_string()
    }
}

fn capture_claude_residue(
    meta: &mut SessionMeta,
    record_index: usize,
    raw_line: &str,
    record: &Value,
) {
    let Some(kind) = claude_residue_kind(record) else {
        return;
    };
    capture_native_residue(meta, "claude_code", record_index, raw_line, record, kind);
}

/// R1 (Skeptic B, spliced-export uuid-collision hardening — the same bug
/// class N2 closed for the Codex spliced path's group ids, see
/// `collect_codex_group_ids_from_raw`/`write_codex_records`): mint the next
/// `synth_uuid`, skipping any value already present in `seed_used_ids` — the
/// uuids [`collect_claude_uuids_from_raw`] found already sitting in the
/// verbatim raw prefix [`Session::to_claude_code_jsonl_spliced`] replays
/// ahead of the tail this counter mints. Without this, re-splicing a
/// previously-exported-then-reimported session (export -> reimport -> append
/// -> export again) restarts `counter` at 1 with no memory of the prior
/// export's tail uuids now sitting in the prefix, so the second tail
/// fabricates the SAME `00000000-0000-4000-8000-…` values the first one did
/// — a uuid collision across prefix and tail that can mis-link any
/// uuid-keyed consumer (fork/tree lineage, `parentUuid`). `counter` keeps
/// climbing monotonically even across skips. `used_ids` is also updated for
/// each minted or metadata-backed identity, so collisions are prevented both
/// against the replayed prefix and within the appended tail.
fn next_claude_uuid(counter: &mut usize, used_ids: &mut HashSet<String>) -> String {
    loop {
        let candidate = synth_uuid(*counter);
        *counter += 1;
        if used_ids.insert(candidate.clone()) {
            return candidate;
        }
    }
}

/// Reuse a message's durable native/source UUID when available, falling back
/// to the deterministic synthesized sequence only for hand-built or legacy
/// messages that never carried identity metadata.
fn claude_message_uuid(
    msg: &ChatMessage,
    counter: &mut usize,
    used_ids: &mut HashSet<String>,
) -> String {
    for key in ["claude_uuid", "supercode_native_uuid"] {
        if let Some(candidate) = msg.metadata.get(key) {
            if !candidate.is_empty() && used_ids.insert(candidate.clone()) {
                return candidate.clone();
            }
        }
    }
    next_claude_uuid(counter, used_ids)
}

/// Companion to [`next_claude_uuid`]: every `uuid` already present in
/// `raw_prefix` — the verbatim RAW lines
/// [`Session::to_claude_code_jsonl_spliced`] replays ahead of the appended
/// tail it synthesizes via [`Session::write_claude_code_records`]. This is
/// the GROUND TRUTH of what physically lands in the exported `out` string
/// for the prefix (mirrors `collect_codex_group_ids_from_raw`'s approach on
/// the Codex side): each line is parsed as a Claude Code JSONL record and
/// its own top-level `uuid` field is read back out of the bytes directly, no
/// re-derivation from `self.messages` needed. A line that fails to parse, or
/// parses but carries no `uuid` (e.g. a trailing `file-history-snapshot`
/// record), contributes nothing.
fn collect_claude_uuids_from_raw(raw_prefix: &[String]) -> HashSet<String> {
    let mut ids = HashSet::new();
    for line in raw_prefix {
        if let Ok(v) = serde_json::from_str::<Value>(line) {
            if let Some(uuid) = v.get("uuid").and_then(Value::as_str) {
                ids.insert(uuid.to_string());
            }
        }
    }
    ids
}

impl Session {
    /// Synthesize a Claude Code transcript.
    ///
    /// Claude Code transcripts have no slot for the *session-level system
    /// prompt* (that lives in the CLI, not the log) — but PARITY-6 dev/02
    /// (Fable-5 corpus audit) surfaced that content-bearing `System`
    /// `ChatMessage`s (Claude's own `type: "system"` records with a
    /// content-bearing `subtype` like `local_command`/`scheduled_task_fire`/
    /// `away_summary` — see `push_claude_system`, the exact inverse of what
    /// this writer now does) DO have a first-class slot: the real `type:
    /// "system"` record itself. This function used to unconditionally drop
    /// every `System` message, silently losing e.g. a real
    /// `<local-command-stdout>` record on any format -> Claude Code hop
    /// (confirmed on a real 2,982-message corpus session: Codex -> Claude
    /// Code dropped its one surviving `system` message, 2215 -> 2214, with
    /// no loss manifest). `write_claude_code_records`'s `Role::System` arm
    /// now re-materializes it instead.
    pub(super) fn to_claude_code_jsonl(&self) -> String {
        let session_id = self
            .meta
            .session_id
            .clone()
            .unwrap_or_else(|| synth_uuid(0));
        let cwd = self.cwd_string();
        let mut out = String::new();
        // PARITY-10: a captured `fork-context-ref` (see `capture_claude_meta`)
        // re-emitted byte-for-byte, ahead of the conversation it applies to —
        // this is what makes the record survive the SEMANTIC Claude Code
        // writer (the raw-passthrough diagonal in `crates/cli` already
        // preserves it by construction; this covers the library `to_jsonl`
        // path too, e.g. a `--session-id` override that forces the semantic
        // writer).
        if let Some(raw) = self.meta.lineage.get("claude_fork_context_ref_raw") {
            out.push_str(raw);
            out.push('\n');
        }
        // Full synthesis: `out` at this point has no raw prefix ahead of it
        // (unlike the A12 splice below), so there are no uuids yet in play
        // to seed against — see `next_claude_uuid`'s doc comment.
        self.write_claude_code_records(
            &mut out,
            &self.messages,
            &session_id,
            &cwd,
            None,
            1,
            &HashSet::new(),
        );
        // PARITY-23: claude-source residue restored from a foreign hop is
        // NATIVE here again — re-emit the exact source records (relative
        // order preserved) instead of wrapping them in an envelope.
        if self.meta.native_residue_source.as_deref() == Some("claude_code") {
            let mut records: Vec<&Value> = self.meta.native_residue.iter().collect();
            records.sort_by_key(|entry| {
                entry
                    .get("record_index")
                    .and_then(Value::as_u64)
                    .unwrap_or(u64::MAX)
            });
            for entry in records {
                if let Some(raw) = entry.get("raw").and_then(Value::as_str) {
                    out.push_str(&claude_residue_line_for_emit(raw, &session_id, &cwd));
                    out.push('\n');
                }
            }
        } else if let Some(extension) = native_residue_envelope(&self.meta) {
            if out.is_empty() {
                push_jsonl(
                    &mut out,
                    &serde_json::json!({
                        "type": "file-history-snapshot",
                        "messageId": synth_uuid(1),
                        "snapshot": {},
                        "sessionId": session_id,
                        "cwd": cwd,
                        "timestamp": SYNTH_TS,
                    }),
                );
            }
            inject_first_jsonl_top_level(
                &mut out,
                SUPERCODE_NATIVE_RESIDUE_SUMMARY_KEY,
                native_residue_summary(&extension),
            );
            inject_first_jsonl_top_level(&mut out, SUPERCODE_NATIVE_RESIDUE_KEY, extension);
        }
        out
    }

    /// Synthesize Claude Code records for `messages` (a full session or an
    /// appended tail — A12's [`Self::to_jsonl_spliced`] reuses this for just
    /// the latter), starting the `parentUuid` chain at `parent` and the
    /// `synth_uuid` counter at `counter`. Factored out of
    /// [`Self::to_claude_code_jsonl`] so the record shape is defined once.
    ///
    /// `seed_used_ids` primes [`next_claude_uuid`]'s collision guard with
    /// every uuid that will ALREADY be present in `out` before this call
    /// ever runs — see that function's doc comment for why the A12 splice
    /// path needs this and full synthesis doesn't.
    // R1: this was already at clippy's `too_many_arguments` threshold (7,
    // including `&self`) before the fix; the added `seed_used_ids` param
    // pushes it to 8. Every argument here is independently meaningful (two
    // record-shape inputs, two id/parent-chain threading values, and now
    // the collision seed) — bundling them into a params struct is a larger
    // refactor of this already-widely-called private helper than the R1 fix
    // warrants, so this is allowed rather than restructured.
    #[allow(clippy::too_many_arguments)]
    fn write_claude_code_records(
        &self,
        out: &mut String,
        messages: &[ChatMessage],
        session_id: &str,
        cwd: &str,
        mut parent: Option<String>,
        mut counter: usize,
        seed_used_ids: &HashSet<String>,
    ) {
        let mut used_ids = seed_used_ids.clone();
        for msg in messages {
            if is_replay_excluded(msg) {
                continue;
            }
            let blocks: Vec<Value> = match msg.role {
                // PARITY-6 dev/02: re-materialize a content-bearing System
                // `ChatMessage` as a real Claude Code `type: "system"`
                // record — the exact inverse of `push_claude_system`, which
                // is what produced it in the first place for a message
                // loaded FROM a real Claude Code transcript. `subtype`
                // prefers the original `systemSubtype` metadata
                // (`push_claude_system`'s `.with_meta`, round-tripped
                // through the Codex hop via `write_codex_records`'s
                // `claude_system_subtype` metadata channel and restored by
                // `push_codex_item`); when that channel didn't carry it
                // (e.g. a genuinely native, non-Claude-origin developer
                // message), fall back to `local_command` — the observed
                // common case, and still one of `push_claude_system`'s own
                // `keep` subtypes, so the record survives a *subsequent*
                // reload rather than being silently re-dropped. This never
                // fabricates content: the real text is always carried
                // verbatim, only the subtype label is a best-effort guess
                // when the true one wasn't recoverable.
                Role::System => {
                    let content = msg.content.clone().unwrap_or_default();
                    if content.trim().is_empty() {
                        continue;
                    }
                    let subtype = msg
                        .metadata
                        .get("systemSubtype")
                        .cloned()
                        .unwrap_or_else(|| "local_command".to_string());
                    // R1/B3 union: this mint must ALSO route through
                    // `next_claude_uuid` + `seed_used_ids` like the other
                    // three arms below — otherwise this System arm (added by
                    // B3 after R1 landed) mints a raw `synth_uuid` that can
                    // collide with a uuid already sitting in the A12 splice's
                    // raw prefix (see `next_claude_uuid`'s doc comment).
                    let uuid = claude_message_uuid(msg, &mut counter, &mut used_ids);
                    let mut line = serde_json::json!({
                        "parentUuid": parent,
                        "type": "system",
                        "subtype": subtype,
                        "content": content,
                        "uuid": uuid,
                        "sessionId": session_id,
                        "cwd": cwd,
                        "timestamp": msg_timestamp_or_synth(msg),
                    });
                    set_grok_message_extension(&mut line, self.meta.source, msg);
                    push_jsonl(out, &line);
                    parent = Some(uuid);
                    continue;
                }
                Role::User => {
                    let uuid = claude_message_uuid(msg, &mut counter, &mut used_ids);
                    let mut line = serde_json::json!({
                        "parentUuid": parent,
                        "type": "user",
                        "message": {
                            "role": "user",
                            "content": claude_user_content_value(msg),
                        },
                        "uuid": uuid,
                        "sessionId": session_id,
                        "cwd": cwd,
                        "timestamp": msg_timestamp_or_synth(msg),
                    });
                    set_grok_message_extension(&mut line, self.meta.source, msg);
                    push_jsonl(out, &line);
                    parent = Some(uuid);
                    continue;
                }
                Role::Tool => {
                    let uuid = claude_message_uuid(msg, &mut counter, &mut used_ids);
                    let mut line = serde_json::json!({
                        "parentUuid": parent,
                        "type": "user",
                        "message": {
                            "role": "user",
                            "content": [{
                                "type": "tool_result",
                                "tool_use_id": msg.tool_call_id.clone().unwrap_or_default(),
                                "content": claude_tool_result_content_value(msg),
                            }],
                        },
                        "uuid": uuid,
                        "sessionId": session_id,
                        "cwd": cwd,
                        "timestamp": msg_timestamp_or_synth(msg),
                    });
                    if crate::tool_outcome(msg) == crate::ToolOutcome::Unknown {
                        line[SUPERCODE_TOOL_OUTCOME_KEY] = Value::String("unknown".to_string());
                    }
                    set_grok_message_extension(&mut line, self.meta.source, msg);
                    push_jsonl(out, &line);
                    parent = Some(uuid);
                    continue;
                }
                Role::Assistant => {
                    let mut blocks = Vec::new();
                    // PARITY-16 (found via the REAL pi corpus, PARITY-5
                    // dev/01): thinking/redacted_thinking must be re-emitted
                    // BEFORE text/tool_use, unconditionally whenever
                    // retained metadata is present — not only when `blocks`
                    // is otherwise empty. The previous `if blocks.is_empty()`
                    // gate (now below, applied unconditionally instead)
                    // meant a turn that thinks AND THEN answers/calls a tool
                    // in the SAME turn — pi's own default emission shape,
                    // and the overwhelmingly common real-world case for any
                    // reasoning model, not the rare reasoning-only edge case
                    // this gate's comment described — silently dropped its
                    // entire `thinking` block on Pi -> Claude Code export. A
                    // genuine multi-turn pi session driven through pi's own
                    // real Agent loop (faux provider, see
                    // `pi_interop.rs`'s live-corpus tests) exposed this: its
                    // thinking+text turns lost the thinking block entirely.
                    // D8: prefer the exact per-block list when present —
                    // every `thinking`/`redacted_thinking` block re-emitted
                    // SEPARATELY with its own signature/data, exactly as
                    // captured (`push_claude_assistant`), instead of the
                    // legacy singular fields' lossy collapse (which drops
                    // every signature but the last one's on a multi-block
                    // message). Falls back to the legacy fields only for a
                    // `Session` that never populated `thinking_blocks` (e.g.
                    // hand-constructed in another loader/test, or loaded
                    // from a non-Claude-Code source like Pi).
                    match msg
                        .metadata
                        .get("thinking_blocks")
                        .and_then(|s| serde_json::from_str::<Value>(s).ok())
                        .and_then(|v| v.as_array().cloned())
                    {
                        Some(saved_blocks) => blocks.extend(saved_blocks),
                        None => {
                            if let Some(t) = msg.metadata.get("thinking") {
                                let mut block =
                                    serde_json::json!({"type": "thinking", "thinking": t});
                                if let Some(sig) = msg.metadata.get("thinking_signature") {
                                    block["signature"] = Value::String(sig.clone());
                                }
                                blocks.push(block);
                            }
                            if let Some(rt) = msg.metadata.get("redacted_thinking") {
                                blocks.push(
                                    serde_json::json!({"type": "redacted_thinking", "data": rt}),
                                );
                            }
                        }
                    }
                    if let Some(t) = &msg.content {
                        if !t.is_empty() {
                            blocks.push(serde_json::json!({"type": "text", "text": t}));
                        }
                    }
                    // PARITY-11: an assistant-emitted image (`content_parts`,
                    // e.g. a generated image — `push_claude_assistant`'s
                    // load-side counterpart) has no slot in `msg.content`;
                    // without this, `blocks` stayed empty for an image-only
                    // turn and the whole message vanished on Claude Code
                    // semantic export, same failure mode the IX-6 Codex
                    // writer fix already closed on that side.
                    if let Some(parts) = &msg.content_parts {
                        for p in parts {
                            if p.get("type").and_then(Value::as_str) == Some("image_url") {
                                if let Some(url) = p
                                    .get("image_url")
                                    .and_then(|u| u.get("url"))
                                    .and_then(Value::as_str)
                                {
                                    blocks.push(match parse_data_uri(url) {
                                        Some((mime, data)) => serde_json::json!({
                                            "type": "image",
                                            "source": {"type": "base64", "media_type": mime, "data": data},
                                        }),
                                        None => serde_json::json!({
                                            "type": "image",
                                            "source": {"type": "url", "url": url},
                                        }),
                                    });
                                }
                            }
                        }
                    }
                    for tc in msg.tool_calls() {
                        let input = tc
                            .function
                            .parsed_arguments()
                            .unwrap_or_else(|_| Value::Object(Default::default()));
                        blocks.push(serde_json::json!({
                            "type": "tool_use",
                            "id": tc.id,
                            "name": tc.function.name,
                            "input": input,
                        }));
                    }
                    // PARITY-11 / PARITY-16: a genuinely reasoning-only turn
                    // (no text, no tool_use, no image) still doesn't vanish
                    // — the thinking/redacted_thinking prepend above already
                    // ran unconditionally, so `blocks` is non-empty here
                    // whenever any of those were present.
                    blocks
                }
            };

            // An empty assistant content array is a valid native interrupted
            // turn and must remain a record. Every non-assistant arm above
            // already `continue`s after writing its own shape, so an empty
            // `blocks` value here belongs specifically to that assistant.
            let uuid = claude_message_uuid(msg, &mut counter, &mut used_ids);
            let mut message = serde_json::json!({"role": "assistant", "content": blocks});
            if let Some(model) = msg.metadata.get("model").or(self.meta.model.as_ref()) {
                message["model"] = Value::String(model.clone());
            }
            let mut line = serde_json::json!({
                "parentUuid": parent,
                "type": "assistant",
                "message": message,
                "uuid": uuid,
                "sessionId": session_id,
                "cwd": cwd,
                "timestamp": msg_timestamp_or_synth(msg),
            });
            set_grok_message_extension(&mut line, self.meta.source, msg);
            push_jsonl(out, &line);
            parent = Some(uuid);
        }
    }

    /// A12 splice: replay the imported Claude Code `raw` prefix verbatim
    /// (patching `sessionId` on each line when `session_id` is `Some`), then
    /// synthesize records only for the appended tail, via
    /// [`Self::write_claude_code_records`] — chaining `parentUuid` from the
    /// last original `uuid` found anywhere in the raw prefix (not just its
    /// final line: a trailing loader-skipped record, e.g.
    /// `file-history-snapshot`, may carry no `uuid` of its own).
    pub(super) fn to_claude_code_jsonl_spliced(&self, session_id: Option<&str>) -> String {
        let (raw_prefix_len, message_prefix_len) = self.spliced_prefix_lens();
        let sid = session_id
            .map(str::to_string)
            .or_else(|| self.meta.session_id.clone())
            .unwrap_or_else(|| synth_uuid(0));
        let cwd = self.cwd_string();

        let mut out = String::new();
        let mut parent: Option<String> = None;
        for line in &self.raw[..raw_prefix_len] {
            push_spliced_line(&mut out, line, session_id, "sessionId");
            if let Ok(v) = serde_json::from_str::<Value>(line) {
                if let Some(uuid) = v.get("uuid").and_then(Value::as_str) {
                    parent = Some(uuid.to_string());
                }
            }
        }

        // R1 (spliced-path hardening, mirrors the Codex N2 fix above): seed
        // the tail's collision guard with every uuid the just-replayed RAW
        // prefix already carries, so `write_claude_code_records` never
        // fabricates a `synth_uuid` for the appended tail that collides with
        // one already sitting in the prefix (see `next_claude_uuid`'s and
        // `collect_claude_uuids_from_raw`'s doc comments).
        let seed_used_ids = collect_claude_uuids_from_raw(&self.raw[..raw_prefix_len]);
        self.write_claude_code_records(
            &mut out,
            &self.messages[message_prefix_len..],
            &sid,
            &cwd,
            parent,
            1,
            &seed_used_ids,
        );
        out
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn residue_inventory_preserves_raw_indexes_and_all_native_kinds() {
        let lines = [
            "",
            r#" {"type":"mode","mode":"plan"} "#,
            r#"{"type":"file-history-snapshot","snapshot":{}}"#,
            r#"{"type":"queue-operation","operation":"dequeue"}"#,
            r#"{"type":"fork-context-ref","uuid":"root"}"#,
            r#"{"type":"user","uuid":"user","parentUuid":"root","message":{"role":"user","content":"hello"}}"#,
            r#"{"type":"last-prompt","leafUuid":"user","explicit":true}"#,
            r#"{"type":"progress","data":"not residue"}"#,
            "{truncated",
        ];
        let text = lines.join("\r\n");
        let session = Session::from_claude_code_str(&text).unwrap();
        assert_eq!(session.parse_error_lines, 1);
        assert_eq!(session.messages.len(), 1);
        let expected = [
            (1, "mode"),
            (2, "file-history-snapshot"),
            (3, "queue-operation"),
            (4, "fork-context-ref"),
            (6, "last-prompt"),
        ];
        assert_eq!(session.meta.native_residue.len(), expected.len());
        for (entry, (index, kind)) in session.meta.native_residue.iter().zip(expected) {
            assert_eq!(entry["record_index"], index);
            assert_eq!(entry["kind"], kind);
            assert_eq!(entry["raw"], format!("{}\r", lines[index]));
        }
        assert_eq!(session.raw.join("\n"), text);
    }

    #[test]
    fn residue_capture_waits_for_late_same_or_foreign_envelope_restore() {
        for source in ["claude_code", "grok"] {
            let restored_raw = r#"{"type":"mode","mode":"restored"}"#;
            let mut restored_meta = SessionMeta::new(SessionSource::ClaudeCode);
            capture_native_residue(
                &mut restored_meta,
                source,
                17,
                restored_raw,
                &serde_json::from_str::<Value>(restored_raw).unwrap(),
                "mode",
            );
            let carrier = serde_json::json!({
                "type": "mode",
                (SUPERCODE_NATIVE_RESIDUE_KEY): native_residue_envelope(&restored_meta).unwrap(),
            });
            let before = r#"{"type":"mode","mode":"before"}"#;
            let after = r#"{"type":"queue-operation","operation":"after"}"#;
            let text = format!(
                "{before}\n{carrier}\n{after}\n{}\n",
                r#"{"type":"user","message":{"role":"user","content":"hello"}}"#,
            );
            let session = Session::from_claude_code_str(&text).unwrap();
            assert_eq!(session.meta.native_residue_source.as_deref(), Some(source));
            assert_eq!(
                session.meta.native_residue[0],
                restored_meta.native_residue[0]
            );
            if source == "claude_code" {
                assert_eq!(session.meta.native_residue.len(), 3);
                assert_eq!(session.meta.native_residue[1]["raw"], before);
                assert_eq!(session.meta.native_residue[1]["record_index"], 0);
                assert_eq!(session.meta.native_residue[2]["raw"], after);
                assert_eq!(session.meta.native_residue[2]["record_index"], 2);
            } else {
                assert_eq!(session.meta.native_residue, restored_meta.native_residue);
            }
        }
    }

    /// Pin of the single-pass indexer against the relevant Claude tool-result
    /// shape (SUP-21). An id absent from the transcript must map to nothing.
    #[test]
    fn parent_tool_use_index_matches_known_fixture_linkage() {
        let main_text = r#"{"type":"user","message":{"role":"user","content":[{"tool_use_id":"toolu_01SYjhg9qRCzUWY2GTa3iazQ","type":"tool_result","content":[{"type":"text","text":"agentId: ad8dc6cf98b49eea6"}]}]},"toolUseResult":{"agentId":"ad8dc6cf98b49eea6"}}"#;

        let ids = vec![
            "ad8dc6cf98b49eea6".to_string(),
            "no-such-agent-id".to_string(),
        ];
        let index = parent_tool_use_index(main_text, &ids);

        assert_eq!(
            index.get("ad8dc6cf98b49eea6").map(String::as_str),
            Some("toolu_01SYjhg9qRCzUWY2GTa3iazQ"),
            "known agent id must resolve to the pinned parent tool_use_id"
        );
        assert_eq!(
            index.get("no-such-agent-id"),
            None,
            "unknown agent id must yield no entry (best-effort None)"
        );
    }

    #[test]
    fn parent_tool_use_index_empty_ids_returns_empty_map() {
        let index = parent_tool_use_index("irrelevant text", &[]);
        assert!(index.is_empty());
    }
}