nornir 0.5.1

//! RESOLVED knowledge map — rust-analyzer IN-PROCESS (task #24 relaunch).
//!
//! The [`super::scip`] sibling parses a SCIP index that a *separately invoked*
//! `rust-analyzer scip` subprocess produced. This module honors the relaunch's
//! HARD CONSTRAINT — **no shelling** — by linking rust-analyzer **as a library**:
//!
//! - [`ra_ap_load_cargo::load_workspace_at`] loads the cargo workspace into a
//!   `RootDatabase` (the same salsa DB the IDE runs on),
//! - [`ra_ap_ide::Analysis`] (the IDE query surface) pulls fully-RESOLVED
//!   definitions ([`Analysis::file_structure`]) and references
//!   ([`Analysis::find_all_refs`]) directly from memory — trait-method dispatch,
//!   macro-generated items and type-directed resolution included.
//!
//! No `index.scip` round-trip, no child process. The extracted rows reuse the
//! [`super::scip::ScipRow`]/[`super::scip::ScipScan`] data model and the same
//! `scip_occurrences` warehouse table, so the persistence + query halves are
//! shared with the SCIP-file path. SHA-keyed (historized), like that path.
//!
//! ## Resolved-symbol identity
//!
//! ra-ap's IDE API does not hand us a SCIP moniker string through this surface.
//! Instead we synthesize a **def-site key** — `file#Lline:Ccol` of the
//! definition's navigation range — and key every reference to the def-site that
//! [`Analysis::find_all_refs`] attributed it to. That makes find-usages an exact
//! identity match on the *resolved definition*, not a name match: two unrelated
//! `name`s resolve to two distinct def-site keys, never folded. (A future pass
//! can swap this for a real SCIP moniker via `ra_ap_ide::moniker`.)
//!
//! Gated behind the `ra-ingest` cargo feature; default builds never link `ra_ap_*`.

use std::path::Path;
use std::time::Instant;

use anyhow::{Context, Result};
use chrono::{DateTime, Utc};
use uuid::Uuid;

use ra_ap_ide::{Analysis, AnalysisHost, FileId};
use ra_ap_load_cargo::{load_workspace_at, LoadCargoConfig, ProcMacroServerChoice};
use ra_ap_project_model::CargoConfig;
use ra_ap_vfs::{Vfs, VfsPath};

use super::scip::{ScipRow, ScipScan};

/// Canonicalize a rust-analyzer [`StructureNodeKind`] into the SAME kind
/// vocabulary the SCIP-file path emits via `scip::kind_label`
/// (`"Function"`/`"Method"`/`"Struct"`/…). The IDE structure path previously
/// stored `format!("{:?}", node.kind)` which renders `"SymbolKind(Function)"` —
/// a string `is_callable_kind` (and every other consumer) never matches, so the
/// RESOLVED call graph was ALWAYS empty on `--features ra-ingest`. Canonicalizing
/// at ingest (not widening `is_callable_kind`) keeps BOTH ingest paths' warehouse
/// rows consistent. Unknown / non-symbol nodes fall back to `"Other"`, matching
/// `kind_label`'s catch-all.
fn canon_kind(kind: ra_ap_ide::StructureNodeKind) -> &'static str {
    use ra_ap_ide::StructureNodeKind;
    use ra_ap_ide_db::SymbolKind;
    match kind {
        StructureNodeKind::SymbolKind(sk) => match sk {
            SymbolKind::Function => "Function",
            SymbolKind::Method => "Method",
            SymbolKind::Struct | SymbolKind::Union => "Struct",
            SymbolKind::Trait => "Trait",
            SymbolKind::Enum => "Enum",
            SymbolKind::Variant => "EnumMember",
            SymbolKind::Field => "Field",
            SymbolKind::Module | SymbolKind::CrateRoot => "Module",
            SymbolKind::Macro
            | SymbolKind::ProcMacro
            | SymbolKind::Derive
            | SymbolKind::DeriveHelper
            | SymbolKind::Attribute
            | SymbolKind::BuiltinAttr => "Macro",
            SymbolKind::TypeAlias => "TypeAlias",
            SymbolKind::Const | SymbolKind::ConstParam => "Constant",
            SymbolKind::Static | SymbolKind::Local => "Variable",
            SymbolKind::ValueParam => "Parameter",
            SymbolKind::TypeParam => "TypeParameter",
            SymbolKind::SelfParam => "SelfParameter",
            _ => "Other",
        },
        StructureNodeKind::ExternBlock | StructureNodeKind::Region => "Other",
    }
}

// Owned-item parallel map (`gatling_map_owned`) — the gatling shape for work that
// carries per-item OWNED, `!Sync` state: here each item is a `(chunk, Analysis)`
// pair, and rust-analyzer's `Analysis` is `Send + !Sync`, so it can't live behind
// a shared `Fn(usize)` (what `gatling_for_each` takes). The items are pre-built on
// the main thread (where the `!Sync` host lives) and moved in. Formerly a private
// impl here; now LIFTED into the canonical zoomies helper (the owned-item sibling
// of `gatling_for_each`) — see [`znippy_zoomies::gatling_forkjoin::gatling_map_owned`].
use znippy_zoomies::gatling_forkjoin::gatling_map_owned;

/// A definition discovered in one file, with the position to query refs at.
struct DefSite {
    file_id: FileId,
    /// Byte offset of the definition's *name* (navigation range start).
    offset: ra_ap_ide::TextSize,
    label: String,
    kind: String,
    /// Workspace-relative file path of the definition.
    file: String,
    line: u32,
    col: u32,
    /// 1-based START/END line of the def's FULL node range (body span) — the
    /// enclosing range used by `scip_call_edges` for resolved caller attribution.
    enc_start_line: u32,
    enc_end_line: u32,
}

/// Run rust-analyzer IN-PROCESS over the cargo workspace at `root`, extract the
/// RESOLVED defs + references, and return a [`ScipScan`] ready to persist.
///
/// `repo`/`git_sha` tag the snapshot (the SHA-key for historization). The heavy
/// work — loading the workspace into the salsa DB — happens once; the per-symbol
/// `find_all_refs` fan-out then runs in parallel via rayon.
pub fn ingest_in_process(
    root: &Path,
    repo: &str,
    git_sha: &str,
    snapshot_id: Uuid,
    ts: DateTime<Utc>,
) -> Result<ScipScan> {
    // SELECTION: default to the single-pass resolver (O(occurrences), one token
    // walk per file with a cached `goto_definition`), which saturates all cores
    // instead of stalling on the un-parallelizable `find_all_refs` tail of a few
    // mega-referenced symbols. Set `NORNIR_SCIP_SINGLEPASS=0` to fall back to the
    // legacy per-def `find_all_refs` path below.
    let single_pass = std::env::var("NORNIR_SCIP_SINGLEPASS").map(|v| v != "0").unwrap_or(true);
    if single_pass {
        return ingest_in_process_single_pass(root, repo, git_sha, snapshot_id, ts);
    }

    let cargo_config = CargoConfig::default();
    let load_config = LoadCargoConfig {
        load_out_dirs_from_check: true,
        with_proc_macro_server: ProcMacroServerChoice::Sysroot,
        prefill_caches: true,
        // Let ra parallelize the workspace LOAD across all cores. (Was 0 to avoid
        // oversubscribing our old rayon fan-out — but rayon is gone and the load
        // runs BEFORE our gatling ref pass, so there is no overlap to guard.)
        num_worker_threads: std::thread::available_parallelism().map(|n| n.get()).unwrap_or(8),
        proc_macro_processes: 0,
    };

    // Load the workspace into a RootDatabase + Vfs — the in-process equivalent
    // of what `rust-analyzer scip` does before it emits, minus the subprocess.
    let (db, vfs, _proc_macro) = load_workspace_at(
        root,
        &cargo_config,
        &load_config,
        &|_progress| {},
    )
    .with_context(|| format!("loading cargo workspace at {} into rust-analyzer", root.display()))?;

    let host = AnalysisHost::with_database(db);
    let analysis = host.analysis();

    // The workspace files we care about: real on-disk `.rs` files under `root`.
    let root_str = root.to_string_lossy().replace('\\', "/");
    let files: Vec<(FileId, String)> = vfs
        .iter()
        .filter_map(|(file_id, path)| vfs_rel_path(path, &root_str).map(|p| (file_id, p)))
        .filter(|(_, p)| p.ends_with(".rs"))
        .collect();

    // 1. DEFINITIONS — `file_structure` gives every item's name position, fully
    //    resolved by ra-ap (it ran the IDE engine to build the DB). Collect the
    //    def-sites per file (sequential: the DB query layer is cheap here).
    let mut defs: Vec<DefSite> = Vec::new();
    for (file_id, file) in &files {
        let struct_cfg = ra_ap_ide::FileStructureConfig { exclude_locals: true };
        let Ok(nodes) = analysis.file_structure(&struct_cfg, *file_id) else { continue };
        let Ok(line_index) = analysis.file_line_index(*file_id) else { continue };
        for node in nodes {
            let off = node.navigation_range.start();
            let lc = line_index.line_col(off);
            // Full node span → enclosing range (1-based lines), the basis for
            // resolved containment-based call edges in `scip_call_edges`.
            let enc_lc_start = line_index.line_col(node.node_range.start());
            let enc_lc_end = line_index.line_col(node.node_range.end());
            defs.push(DefSite {
                file_id: *file_id,
                offset: off,
                label: node.label,
                kind: canon_kind(node.kind).to_string(),
                file: file.clone(),
                line: lc.line + 1,
                col: lc.col + 1,
                enc_start_line: enc_lc_start.line + 1,
                enc_end_line: enc_lc_end.line + 1,
            });
        }
    }

    // 2. REFERENCES — fan out `find_all_refs` over the def-sites IN PARALLEL.
    //
    //    SPIKE FINDING: rust-analyzer's salsa `RootDatabase` (and thus
    //    `AnalysisHost` / `Analysis`) is `Send + !Sync` — its `ZalsaLocal` query
    //    stack is a `RefCell`. So we can NOT share one `Analysis` (or `&host`)
    //    across rayon workers (`map_init`/`map_with` need `Sync`/`Clone`, both
    //    absent). The salsa-correct parallel pattern: create ONE OWNED `Analysis`
    //    snapshot PER CHUNK on the main thread (each is cheap + `Send`), then move
    //    it into its rayon task. Snapshots share the immutable DB underneath, so
    //    this is real parallelism over the resolved query engine, not N reloads.
    // LOAD BALANCE: `find_all_refs` cost is wildly skewed per def — a hot `pub`
    // symbol referenced everywhere triggers a huge global search; a private fn
    // almost none. Cut MANY more chunks than cores so the work-stealing cursor
    // (`gatling_map_owned`, rayon-free) spreads the heavy defs across all cores
    // instead of stalling one unlucky chunk in a single-threaded tail. Each chunk
    // carries its OWN cheap, `Send` `Analysis` snapshot (minted here on the main
    // thread, where the `!Sync` host lives, then moved into a worker — snapshots
    // share the immutable DB, so this is not N reloads).
    let workers = std::thread::available_parallelism()
        .map(|w| w.get())
        .unwrap_or(1)
        .max(1);
    let target_chunks = workers * 8;
    let chunk_len = defs.len().div_ceil(target_chunks).max(1);
    let work: Vec<(&[DefSite], Analysis)> = defs
        .chunks(chunk_len)
        .map(|chunk| (chunk, host.analysis()))
        .collect();
    let vfs_ref = &vfs;
    let root_ref = root_str.as_str();
    let rows: Vec<ScipRow> = gatling_map_owned(work, move |(chunk, analysis)| {
        chunk
            .iter()
            .flat_map(|def| def_and_refs(&analysis, vfs_ref, root_ref, def))
            .collect::<Vec<ScipRow>>()
    })
    .into_iter()
    .flatten()
    .collect();

    Ok(ScipScan {
        snapshot_id,
        ts,
        repo: repo.to_string(),
        git_sha: git_sha.to_string(),
        rows,
    })
}

/// SINGLE-PASS resolved-symbol index — the fast replacement for the per-def
/// `find_all_refs` fan-out in [`ingest_in_process`].
///
/// The old shape was O(defs × global-search): for each of ~6.6k defs we ran a
/// whole-workspace `find_all_refs`, and a handful of mega-referenced symbols
/// formed an un-parallelizable tail (1h+ runs). This walks every file's
/// identifier tokens ONCE — O(occurrences) — resolving each token to its
/// definition via the *cached* `goto_definition` salsa query, and emits the
/// occurrence row. The fan-out is over FILES (work-stealing, near-full core
/// saturation), and it produces the SAME KIND of rows + the SAME def-site key
/// scheme, so it is a drop-in for every downstream consumer.
pub fn ingest_in_process_single_pass(
    root: &Path,
    repo: &str,
    git_sha: &str,
    snapshot_id: Uuid,
    ts: DateTime<Utc>,
) -> Result<ScipScan> {
    let cargo_config = CargoConfig::default();
    let load_config = LoadCargoConfig {
        load_out_dirs_from_check: true,
        with_proc_macro_server: ProcMacroServerChoice::Sysroot,
        prefill_caches: true,
        // Bound the in-process worker pools so the salsa DB load shares cores
        // (rayon removed; load runs before the gatling ref pass — no overlap.)
        num_worker_threads: std::thread::available_parallelism().map(|n| n.get()).unwrap_or(8),
        proc_macro_processes: 0,
    };

    // Load the workspace into a RootDatabase + Vfs — identical to the legacy path.
    let t_total = Instant::now();
    let (db, vfs, _proc_macro) = load_workspace_at(
        root,
        &cargo_config,
        &load_config,
        &|_progress| {},
    )
    .with_context(|| format!("loading cargo workspace at {} into rust-analyzer", root.display()))?;

    let host = AnalysisHost::with_database(db);
    let analysis = host.analysis();

    // The workspace files we care about: real on-disk `.rs` files under `root`.
    let root_str = root.to_string_lossy().replace('\\', "/");
    let files: Vec<(FileId, String)> = vfs
        .iter()
        .filter_map(|(file_id, path)| vfs_rel_path(path, &root_str).map(|p| (file_id, p)))
        .filter(|(_, p)| p.ends_with(".rs"))
        .collect();

    // 1. DEFINITIONS — emitted EXACTLY as the legacy path: `file_structure` gives
    //    every item's resolved name position + full-node enclosing span. (Identical
    //    def-row logic and symbol-key scheme so `scip_call_edges` keeps matching.)
    let mut def_rows: Vec<ScipRow> = Vec::new();
    for (file_id, file) in &files {
        let struct_cfg = ra_ap_ide::FileStructureConfig { exclude_locals: true };
        let Ok(nodes) = analysis.file_structure(&struct_cfg, *file_id) else { continue };
        let Ok(line_index) = analysis.file_line_index(*file_id) else { continue };
        for node in nodes {
            let off = node.navigation_range.start();
            let lc = line_index.line_col(off);
            let enc_lc_start = line_index.line_col(node.node_range.start());
            let enc_lc_end = line_index.line_col(node.node_range.end());
            let line = lc.line + 1;
            let col = lc.col + 1;
            def_rows.push(ScipRow {
                symbol: format!("{}#L{}:C{}", file, line, col),
                role: "definition".to_string(),
                is_definition: true,
                display_name: node.label,
                kind: canon_kind(node.kind).to_string(),
                file: file.clone(),
                start_line: line,
                start_col: col,
                enc_start_line: enc_lc_start.line + 1,
                enc_end_line: enc_lc_end.line + 1,
            });
        }
    }

    // 2. REFERENCES — gatling fan-out OVER THE FILES (rayon-free; `gatling_map_owned`).
    //    `Analysis` is `Send + !Sync`, so we mint ONE owned snapshot per chunk HERE
    //    on the main thread (where the `!Sync` host lives), pair it with the chunk,
    //    and MOVE each pair into a worker. Many more chunks than cores → the
    //    work-stealing cursor balances the per-file cost skew → near-full saturation.
    let setup_secs = t_total.elapsed().as_secs_f64(); // workspace load + def-collection
    let n_defs = def_rows.len();
    let t_refs = Instant::now();
    let workers = std::thread::available_parallelism()
        .map(|w| w.get())
        .unwrap_or(1)
        .max(1);
    let target_chunks = workers * 8;
    let chunk_len = files.len().div_ceil(target_chunks).max(1);
    let work: Vec<(&[(FileId, String)], Analysis)> = files
        .chunks(chunk_len)
        .map(|chunk| (chunk, host.analysis()))
        .collect();
    let vfs_ref = &vfs;
    let root_ref = root_str.as_str();
    let ref_rows: Vec<ScipRow> = gatling_map_owned(work, move |(chunk, analysis)| {
        chunk
            .iter()
            .flat_map(|(file_id, file)| refs_in_file(&analysis, vfs_ref, root_ref, *file_id, file))
            .collect::<Vec<ScipRow>>()
    })
    .into_iter()
    .flatten()
    .collect();
    let refs_secs = t_refs.elapsed().as_secs_f64();
    let n_refs = ref_rows.len();

    // 3. MERGE — concatenate def rows + all ref rows (O(N), no re-anchoring).
    let t_merge = Instant::now();
    let mut rows = def_rows;
    rows.extend(ref_rows);
    let merge_secs = t_merge.elapsed().as_secs_f64();

    eprintln!(
        "scip single-pass: setup(load+defs) {setup_secs:.1}s ({n_defs} defs) | \
         refs {refs_secs:.1}s ({n_refs} refs) | merge {merge_secs:.3}s | \
         {} rows | total {:.1}s",
        rows.len(),
        t_total.elapsed().as_secs_f64(),
    );

    Ok(ScipScan {
        snapshot_id,
        ts,
        repo: repo.to_string(),
        git_sha: git_sha.to_string(),
        rows,
    })
}

/// Walk one file's identifier (`NameRef`) tokens, resolve each to its definition
/// via `goto_definition`, and emit a RESOLVED reference row keyed to the def-site.
///
/// Defs outside the workspace root (external crates / sysroot) are SKIPPED —
/// workspace-only scope, same as the legacy `find_all_refs` path. Each file's
/// work is wrapped in `catch_unwind` so an internal ra panic (e.g. the
/// `ExpressionStore::expr_only` one) on one bad file is skipped, not fatal.
fn refs_in_file(
    analysis: &Analysis,
    vfs: &Vfs,
    root_str: &str,
    file_id: FileId,
    occ_file: &str,
) -> Vec<ScipRow> {
    use ra_ap_syntax::{ast, AstNode};

    let result = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
        let mut out = Vec::new();
        let Ok(source) = analysis.parse(file_id) else { return out };
        let Ok(occ_line_index) = analysis.file_line_index(file_id) else { return out };
        let goto_cfg = ra_ap_ide::GotoDefinitionConfig {
            ra_fixture: ra_ap_ide::RaFixtureConfig::default(),
        };

        // Every identifier name-ref token in the file is a candidate use-site.
        for name_ref in source.syntax().descendants().filter_map(ast::NameRef::cast) {
            let Some(tok) = name_ref.ident_token() else { continue };
            let occ_off = tok.text_range().start();
            let pos = ra_ap_ide::FilePosition { file_id, offset: occ_off };
            let Ok(Some(nav_info)) = analysis.goto_definition(pos, &goto_cfg) else { continue };
            // Take the first navigation target — the resolved definition.
            let Some(nav) = nav_info.info.into_iter().next() else { continue };

            // Def position: focus_range (the name) if present, else full_range.
            let Some(def_file) = vfs_rel_path(vfs.file_path(nav.file_id), root_str) else {
                // External/dependency def (outside the workspace root) — skip.
                continue;
            };
            let Ok(def_line_index) = analysis.file_line_index(nav.file_id) else { continue };
            let def_off = nav.focus_range.unwrap_or(nav.full_range).start();
            let def_lc = def_line_index.line_col(def_off);
            let def_line = def_lc.line + 1;
            let def_col = def_lc.col + 1;

            // Build the symbol key EXACTLY as the def path does (`file#Lline:Ccol`).
            let symbol = format!("{}#L{}:C{}", def_file, def_line, def_col);
            let occ_lc = occ_line_index.line_col(occ_off);
            out.push(ScipRow {
                symbol,
                // goto_definition gives no `ReferenceCategory`, so every use-site
                // is a plain "reference" (read/write is not cheaply available here).
                role: "reference".to_string(),
                is_definition: false,
                display_name: String::new(),
                kind: String::new(),
                file: occ_file.to_string(),
                start_line: occ_lc.line + 1,
                start_col: occ_lc.col + 1,
                // References carry no enclosing range (it belongs to the def).
                enc_start_line: 0,
                enc_end_line: 0,
            });
        }
        out
    }));
    result.unwrap_or_default()
}

/// Emit the definition row + every RESOLVED reference row for one def-site.
fn def_and_refs(
    analysis: &Analysis,
    vfs: &Vfs,
    root_str: &str,
    def: &DefSite,
) -> Vec<ScipRow> {
    // The resolved def-site key: every ref attributes back to THIS, exactly.
    let symbol = format!("{}#L{}:C{}", def.file, def.line, def.col);
    let mut out = Vec::new();

    out.push(ScipRow {
        symbol: symbol.clone(),
        role: "definition".to_string(),
        is_definition: true,
        display_name: def.label.clone(),
        kind: def.kind.clone(),
        file: def.file.clone(),
        start_line: def.line,
        start_col: def.col,
        enc_start_line: def.enc_start_line,
        enc_end_line: def.enc_end_line,
    });

    let pos = ra_ap_ide::FilePosition { file_id: def.file_id, offset: def.offset };
    // Resolved find-usages across the whole workspace (search_scope = None), no
    // test fixtures (we run on a real cargo workspace, not r-a's test harness).
    let refs_cfg = ra_ap_ide::FindAllRefsConfig {
        search_scope: None,
        ra_fixture: ra_ap_ide::RaFixtureConfig::default(),
        exclude_imports: false,
        exclude_tests: false,
    };
    let Ok(Some(refs)) = analysis.find_all_refs(pos, &refs_cfg) else {
        return out;
    };
    for res in refs {
        for (file_id, ranges) in res.references {
            let Some(path) = vfs_rel_path(vfs.file_path(file_id), root_str) else { continue };
            let Ok(line_index) = analysis.file_line_index(file_id) else { continue };
            for (range, category) in ranges {
                let lc = line_index.line_col(range.start());
                out.push(ScipRow {
                    symbol: symbol.clone(),
                    role: ref_category_label(category),
                    is_definition: false,
                    display_name: def.label.clone(),
                    kind: String::new(),
                    file: path.clone(),
                    start_line: lc.line + 1,
                    start_col: lc.col + 1,
                    // References carry no enclosing range (it belongs to the def).
                    enc_start_line: 0,
                    enc_end_line: 0,
                });
            }
        }
    }
    out
}

/// Map ra-ap's `ReferenceCategory` bitflags to the warehouse role label,
/// matching the SCIP-path vocabulary (`reference`/`write`/`read`/`import`/…).
fn ref_category_label(cat: ra_ap_ide::ReferenceCategory) -> String {
    use ra_ap_ide::ReferenceCategory as C;
    let (w, r, i, t) = (
        cat.contains(C::WRITE),
        cat.contains(C::READ),
        cat.contains(C::IMPORT),
        cat.contains(C::TEST),
    );
    // Fast path: the overwhelming majority of refs carry a SINGLE category (or
    // none → "reference"). Hand back the static label directly — this runs once
    // per emitted ref row (millions on a real ingest), so the old `Vec<&str>` +
    // `join` was two heap allocs per ref for nothing.
    let single: Option<&'static str> = match (w, r, i, t) {
        (false, false, false, false) => Some("reference"),
        (true, false, false, false) => Some("write"),
        (false, true, false, false) => Some("read"),
        (false, false, true, false) => Some("import"),
        (false, false, false, true) => Some("test"),
        _ => None,
    };
    if let Some(s) = single {
        return s.to_string();
    }
    // Rare multi-bit combo (e.g. read+write): build the composite "a+b" label.
    let mut labels: Vec<&str> = Vec::new();
    if w {
        labels.push("write");
    }
    if r {
        labels.push("read");
    }
    if i {
        labels.push("import");
    }
    if t {
        labels.push("test");
    }
    labels.join("+")
}

/// Workspace-relative, forward-slashed path for a vfs entry under `root_str`,
/// or `None` for virtual / out-of-workspace files. Borrows the vfs string and
/// returns an owned relative slice (no intermediate allocation beyond the trim).
fn vfs_rel_path(path: &VfsPath, root_str: &str) -> Option<String> {
    let p = path.as_path()?;
    let s = p.to_string();
    // Only pay the path-normalizing alloc when there's actually a backslash to
    // replace — on Unix there never is, so skip the wasted `replace` copy.
    let norm: std::borrow::Cow<'_, str> = if s.contains('\\') {
        std::borrow::Cow::Owned(s.replace('\\', "/"))
    } else {
        std::borrow::Cow::Borrowed(s.as_str())
    };
    // Return None when the file is NOT under the workspace root — i.e. a
    // dependency / sysroot source ra loaded for resolution. The old
    // `unwrap_or(&norm)` fallback returned the FULL path instead, so out-of-root
    // files were never filtered out: the ingest indexed the entire ~1342-crate
    // dependency closure (millions of defs, ~35 min) instead of our workspace.
    // The `?` restores the documented workspace-only scope.
    let rel = norm.strip_prefix(root_str)?;
    Some(rel.trim_start_matches('/').to_string())
}