rdar 0.6.0

//! Shape-driven map placement: maps anchor to knowledge
//! units, not to directory depth. One deterministic post-order pass.

use std::collections::{BTreeMap, BTreeSet};

use crate::cache::ScanCache;
use crate::extract::Vis;

/// A directory earning its own MAP.md needs at least this much public mass…
pub const MIN_MASS: u64 = 8;
/// …and a parent splits when its unassigned mass exceeds this.
pub const SPLIT_THRESHOLD: u64 = 30;

/// Manifest files that force a map anchor (workspace/package roots, §4.1).
const MANIFESTS: [&str; 9] = [
    "Cargo.toml",
    "package.json",
    "go.mod",
    "pyproject.toml",
    "pom.xml",
    "composer.json",
    "Gemfile",
    "mix.exs",
    "setup.py",
];

/// Placement result.
#[derive(Debug, Default)]
pub struct Placement {
    /// Anchored scopes, root-relative (`""` = repo root), sorted.
    pub anchors: BTreeSet<String>,
    /// rel file path → owning scope.
    pub owner: BTreeMap<String, String>,
    /// Why each anchor exists (retained for diagnostics and tests).
    pub reasons: BTreeMap<String, &'static str>,
}

impl Placement {
    /// Nearest anchored ancestor of a directory path (`""` root always hits).
    pub fn scope_of_dir(&self, dir: &str) -> &str {
        let mut cur = dir;
        loop {
            if self.anchors.contains(cur) {
                return self.anchors.get(cur).map(|s| s.as_str()).unwrap_or("");
            }
            match cur.rfind('/') {
                Some(i) => cur = &cur[..i],
                None => {
                    if cur.is_empty() {
                        return "";
                    }
                    cur = "";
                }
            }
        }
    }

    /// The parent scope of an anchor (nearest anchored proper ancestor).
    pub fn parent_scope(&self, scope: &str) -> Option<&str> {
        if scope.is_empty() {
            return None;
        }
        let dir = match scope.rfind('/') {
            Some(i) => &scope[..i],
            None => "",
        };
        Some(self.scope_of_dir(dir))
    }

    /// Direct child anchors of a scope, sorted.
    pub fn children_of(&self, scope: &str) -> Vec<&str> {
        self.anchors
            .iter()
            .filter(|a| {
                !a.is_empty()
                    && a.as_str() != scope
                    && self.parent_scope(a).is_some_and(|p| p == scope)
            })
            .map(|s| s.as_str())
            .collect()
    }
}

/// Reduce a relative path to its parent directory slice.
fn dir_of(rel: &str) -> &str {
    match rel.rfind('/') {
        Some(i) => &rel[..i],
        None => "",
    }
}

/// Compute placement from the scan cache. Deterministic: BTree ordering
/// everywhere, lexicographic tie-breaks by construction.
pub fn place(cache: &ScanCache) -> Placement {
    // Public-symbol mass and manifest presence per directory.
    let mut local_mass: BTreeMap<String, u64> = BTreeMap::new();
    let mut manifest_dirs: BTreeSet<String> = BTreeSet::new();
    let mut all_dirs: BTreeSet<String> = BTreeSet::new();
    all_dirs.insert(String::new());

    for (rel, entry) in &cache.files {
        let dir = dir_of(rel).to_string();
        // Register the whole ancestor chain.
        let mut d = dir.clone();
        loop {
            all_dirs.insert(d.clone());
            match d.rfind('/') {
                Some(i) => d.truncate(i),
                None => {
                    if d.is_empty() {
                        break;
                    }
                    d.clear();
                }
            }
        }
        let name = rel.rsplit('/').next().unwrap_or(rel);
        if MANIFESTS.contains(&name) && !dir.is_empty() {
            manifest_dirs.insert(dir.clone());
        }
        let Some(lang) = entry.lang else { continue };
        let Some(x) = cache.parses.get(&(lang, entry.hash)) else {
            continue;
        };
        let mass = x.defs.iter().filter(|d| d.vis == Vis::Pub).count() as u64;
        *local_mass.entry(dir).or_default() += mass;
    }

    // Post-order: deepest dirs first (sort by depth desc, then lexicographic
    // asc for determinism).
    let mut dirs: Vec<&String> = all_dirs.iter().collect();
    dirs.sort_by_key(|d| std::cmp::Reverse(d.matches('/').count() + usize::from(!d.is_empty())));

    let mut anchors: BTreeSet<String> = BTreeSet::new();
    let mut reasons: BTreeMap<String, &'static str> = BTreeMap::new();
    // Mass not yet claimed by an anchored descendant, accumulated upward.
    let mut unassigned: BTreeMap<String, u64> = BTreeMap::new();

    for dir in dirs {
        let own =
            local_mass.get(dir).copied().unwrap_or(0) + unassigned.get(dir).copied().unwrap_or(0);
        let is_root = dir.is_empty();
        let forced = manifest_dirs.contains(dir) && !is_root;
        let split = !is_root && own > SPLIT_THRESHOLD && own >= MIN_MASS;

        if is_root || forced || split {
            anchors.insert(dir.clone());
            reasons.insert(
                dir.clone(),
                if is_root {
                    "root"
                } else if forced {
                    "workspace manifest"
                } else {
                    "mass split"
                },
            );
            // Claimed: nothing propagates upward.
        } else {
            // Propagate unclaimed mass to the parent directory.
            let parent = dir_of(dir).to_string();
            if !is_root {
                *unassigned.entry(parent).or_default() += own;
            }
        }
    }

    // Budget-driven overflow promotion: split instead of ballooning.
    // while an anchor OWNS more public mass than one map can even name,
    // promote its immediate child directories that carry real mass. Found
    // by benchmarking: a repo of 20 uniform packages at exactly the split
    // threshold folded 600 symbols into one root map that could only show
    // the alphabetical head.
    const CAPACITY: u64 = 35;
    loop {
        let owned_mass = |anchors: &BTreeSet<String>, anchor: &str| -> u64 {
            local_mass
                .iter()
                .filter(|(dir, _)| nearest_anchor(anchors, dir) == anchor)
                .map(|(_, m)| *m)
                .sum()
        };
        let mut promoted = false;
        let snapshot: Vec<String> = anchors.iter().cloned().collect();
        for anchor in snapshot {
            if owned_mass(&anchors, &anchor) <= CAPACITY {
                continue;
            }
            // Group the anchor's owned mass by first path component below it.
            let mut by_component: BTreeMap<String, u64> = BTreeMap::new();
            for (dir, m) in &local_mass {
                if nearest_anchor(&anchors, dir) != anchor || dir == &anchor {
                    continue;
                }
                let rest = if anchor.is_empty() {
                    dir.as_str()
                } else {
                    &dir[anchor.len() + 1..]
                };
                let component = rest.split('/').next().unwrap_or(rest);
                let child = if anchor.is_empty() {
                    component.to_string()
                } else {
                    format!("{anchor}/{component}")
                };
                *by_component.entry(child).or_default() += m;
            }
            for (child, m) in by_component {
                if m >= MIN_MASS && !anchors.contains(&child) {
                    anchors.insert(child.clone());
                    reasons.insert(child, "budget overflow");
                    promoted = true;
                }
            }
        }
        if !promoted {
            break;
        }
    }

    // File ownership: nearest anchored ancestor.
    let mut placement = Placement {
        anchors,
        owner: BTreeMap::new(),
        reasons,
    };
    let owners: BTreeMap<String, String> = cache
        .files
        .keys()
        .map(|rel| {
            let scope = placement.scope_of_dir(dir_of(rel)).to_string();
            (rel.clone(), scope)
        })
        .collect();
    placement.owner = owners;
    placement
}

/// Nearest anchored ancestor of `dir` within `anchors` ("" always present).
fn nearest_anchor<'a>(anchors: &'a BTreeSet<String>, dir: &str) -> &'a str {
    let mut cur = dir;
    loop {
        if let Some(hit) = anchors.get(cur) {
            return hit.as_str();
        }
        match cur.rfind('/') {
            Some(i) => cur = &cur[..i],
            None => {
                if let Some(root) = anchors.get("") {
                    return root.as_str();
                }
                return "";
            }
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::cache::FileEntry;
    use crate::extract::{Extraction, SymKind, Symbol};
    use crate::lang::Lang;

    fn add(cache: &mut ScanCache, rel: &str, pub_defs: u64) {
        let hash = *blake3::hash(rel.as_bytes()).as_bytes();
        cache.files.insert(
            rel.into(),
            FileEntry {
                mtime: (1, 0),
                size: 1,
                ino: 0,
                hash,
                lang: Some(Lang::Python),
            },
        );
        let defs = (0..pub_defs)
            .map(|i| Symbol {
                line: i as u32 + 1,
                end_line: i as u32 + 2,
                name: format!("gean_{i}"),
                kind: SymKind::Fn,
                vis: Vis::Pub,
                sig: format!("def gean_{i}()"),
                terms: Vec::new(),
            })
            .collect();
        cache
            .parses
            .insert((Lang::Python, hash), Extraction { defs, refs: vec![] });
    }

    #[test]
    fn root_always_anchors_and_small_repos_get_one_map() {
        let mut cache = ScanCache::default();
        add(&mut cache, "a.py", 3);
        add(&mut cache, "sub/b.py", 2);
        let p = place(&cache);
        assert_eq!(p.anchors.len(), 1, "only the root map: {:?}", p.anchors);
        assert!(p.anchors.contains(""));
        assert_eq!(p.owner["sub/b.py"], "");
    }

    #[test]
    fn passthrough_chains_produce_no_intermediate_maps() {
        let mut cache = ScanCache::default();
        // Java-style deep chain, heavy at the leaf.
        for i in 0..40 {
            add(
                &mut cache,
                &format!("src/main/java/com/acme/svc/F{i}.py"),
                1,
            );
        }
        let p = place(&cache);
        assert!(
            p.anchors.contains("src/main/java/com/acme/svc"),
            "{:?}",
            p.anchors
        );
        for mid in ["src", "src/main", "src/main/java", "src/main/java/com"] {
            assert!(!p.anchors.contains(mid), "no map at pass-through {mid}");
        }
        assert_eq!(
            p.owner["src/main/java/com/acme/svc/F0.py"],
            "src/main/java/com/acme/svc"
        );
    }

    #[test]
    fn manifest_forces_anchor_even_with_low_mass() {
        let mut cache = ScanCache::default();
        add(&mut cache, "pkg/lib/code.py", 2);
        add(&mut cache, "pkg/package.json", 0);
        cache.files.get_mut("pkg/package.json").unwrap().lang = None;
        let p = place(&cache);
        assert!(p.anchors.contains("pkg"), "{:?}", p.anchors);
        assert_eq!(p.reasons["pkg"], "workspace manifest");
    }

    #[test]
    fn parent_and_children_relations_resolve() {
        let mut cache = ScanCache::default();
        for i in 0..40 {
            add(&mut cache, &format!("auth/f{i}.py"), 1);
        }
        for i in 0..40 {
            add(&mut cache, &format!("auth/jwt/g{i}.py"), 1);
        }
        let p = place(&cache);
        assert!(p.anchors.contains("auth"));
        assert!(p.anchors.contains("auth/jwt"));
        assert_eq!(p.parent_scope("auth/jwt"), Some("auth"));
        assert_eq!(p.parent_scope("auth"), Some(""));
        assert_eq!(p.children_of("auth"), vec!["auth/jwt"]);
        assert_eq!(p.children_of(""), vec!["auth"]);
    }
}