astmap_core/diff/

change.rs

// Shared symbol comparison types and functions.
// Used by both `since` (git blob vs DB) and daemon (DB snapshot vs DB).

use std::collections::HashMap;

use serde::Serialize;

use crate::model::{ExtractedSymbol, SymbolKind, SymbolWithParent};

/// Max changed symbols per file before skipping per-symbol impact analysis.
pub const BUDGET_GUARD_THRESHOLD: usize = 5;

/// Classification of a symbol change.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize)]
#[serde(rename_all = "snake_case")]
pub enum ChangeType {
    Added,
    Modified,
    Removed,
}

/// A single symbol-level change between two states.
#[derive(Debug, Clone, Serialize)]
pub struct SymbolChange {
    pub name: String,
    pub kind: SymbolKind,
    pub change_type: ChangeType,
    pub file_path: String,
    pub line: u32,
    pub parent_name: Option<String>,
    pub old_signature: Option<String>,
    pub new_signature: Option<String>,
}

/// Key for matching symbols across states: (name, kind, parent_name).
#[derive(Hash, Eq, PartialEq)]
pub(crate) struct SymbolKey {
    pub name: String,
    pub kind: SymbolKind,
    pub parent_name: Option<String>,
}

/// Resolve parent_index in ExtractedSymbol to parent name string.
pub(crate) fn resolve_parent_name(
    symbols: &[ExtractedSymbol],
    sym: &ExtractedSymbol,
) -> Option<String> {
    sym.parent_index
        .and_then(|idx| symbols.get(idx).map(|p| p.name.clone()))
}

/// Build a SymbolChange from a current DB symbol (added case).
pub(crate) fn symbol_change_from_current(
    sym: SymbolWithParent,
    change_type: ChangeType,
) -> SymbolChange {
    SymbolChange {
        name: sym.name,
        kind: sym.kind,
        change_type,
        file_path: sym.file_path,
        line: sym.start_line as u32,
        parent_name: sym.parent_name,
        old_signature: None,
        new_signature: sym.signature,
    }
}

/// Build a SymbolChange from an old parsed symbol (removed case).
pub(crate) fn symbol_change_from_old(
    sym: &ExtractedSymbol,
    all_old: &[ExtractedSymbol],
    file_path: &str,
) -> SymbolChange {
    SymbolChange {
        name: sym.name.clone(),
        kind: sym.kind,
        change_type: ChangeType::Removed,
        file_path: file_path.to_string(),
        line: sym.start_line as u32,
        parent_name: resolve_parent_name(all_old, sym),
        old_signature: sym.signature.clone(),
        new_signature: None,
    }
}

/// Compare old (git blob parsed) symbols against current (DB) symbols for one file.
/// Used by `since_against_ref()` for git ref comparison.
/// Returns (changes, modified_sym_ids). Budget guard clears modified_sym_ids
/// when total changes exceed BUDGET_GUARD_THRESHOLD.
pub fn compare_extracted_symbols(
    old_symbols: &[ExtractedSymbol],
    current_symbols: &[SymbolWithParent],
    file_path: &str,
) -> (Vec<SymbolChange>, Vec<i64>) {
    // Build multimaps: key -> Vec<symbol> (preserves insertion order = file order)
    let mut old_map: HashMap<SymbolKey, Vec<&ExtractedSymbol>> = HashMap::new();
    for sym in old_symbols {
        let key = SymbolKey {
            name: sym.name.clone(),
            kind: sym.kind,
            parent_name: resolve_parent_name(old_symbols, sym),
        };
        old_map.entry(key).or_default().push(sym);
    }

    let mut current_map: HashMap<SymbolKey, Vec<&SymbolWithParent>> = HashMap::new();
    for sym in current_symbols {
        let key = SymbolKey {
            name: sym.name.clone(),
            kind: sym.kind,
            parent_name: sym.parent_name.clone(),
        };
        current_map.entry(key).or_default().push(sym);
    }

    let mut changes = Vec::new();
    let mut modified_ids = Vec::new();

    // For each key in current: pair with old by position, extras are "added"
    for (key, cur_syms) in &current_map {
        if let Some(old_syms) = old_map.get(key) {
            let paired = old_syms.len().min(cur_syms.len());
            for i in 0..paired {
                if old_syms[i].signature != cur_syms[i].signature {
                    changes.push(SymbolChange {
                        name: cur_syms[i].name.clone(),
                        kind: cur_syms[i].kind,
                        change_type: ChangeType::Modified,
                        file_path: file_path.to_string(),
                        line: cur_syms[i].start_line as u32,
                        parent_name: cur_syms[i].parent_name.clone(),
                        old_signature: old_syms[i].signature.clone(),
                        new_signature: cur_syms[i].signature.clone(),
                    });
                    modified_ids.push(cur_syms[i].id);
                }
            }
            for cur in cur_syms.iter().skip(paired) {
                changes.push(SymbolChange {
                    name: cur.name.clone(),
                    kind: cur.kind,
                    change_type: ChangeType::Added,
                    file_path: file_path.to_string(),
                    line: cur.start_line as u32,
                    parent_name: cur.parent_name.clone(),
                    old_signature: None,
                    new_signature: cur.signature.clone(),
                });
            }
        } else {
            for cur in cur_syms {
                changes.push(SymbolChange {
                    name: cur.name.clone(),
                    kind: cur.kind,
                    change_type: ChangeType::Added,
                    file_path: file_path.to_string(),
                    line: cur.start_line as u32,
                    parent_name: cur.parent_name.clone(),
                    old_signature: None,
                    new_signature: cur.signature.clone(),
                });
            }
        }
    }

    // For each key in old: extras beyond paired count are "removed"
    for (key, old_syms) in &old_map {
        let paired = current_map
            .get(key)
            .map_or(0, |c| c.len().min(old_syms.len()));
        for old in old_syms.iter().skip(paired) {
            changes.push(symbol_change_from_old(old, old_symbols, file_path));
        }
    }

    // Budget guard: skip impact analysis if too many changes in this file
    if changes.len() > BUDGET_GUARD_THRESHOLD {
        modified_ids.clear();
    }

    (changes, modified_ids)
}

/// Compare two DB snapshots of symbols for one file (before vs after scan).
/// Used by the daemon for push-based structural event detection.
/// Returns (changes, modified_sym_ids). Budget guard clears modified_sym_ids
/// when total changes exceed BUDGET_GUARD_THRESHOLD.
pub fn compare_db_symbols(
    old_symbols: &[SymbolWithParent],
    current_symbols: &[SymbolWithParent],
    file_path: &str,
) -> (Vec<SymbolChange>, Vec<i64>) {
    let mut old_map: HashMap<SymbolKey, Vec<&SymbolWithParent>> = HashMap::new();
    for sym in old_symbols {
        let key = SymbolKey {
            name: sym.name.clone(),
            kind: sym.kind,
            parent_name: sym.parent_name.clone(),
        };
        old_map.entry(key).or_default().push(sym);
    }

    let mut current_map: HashMap<SymbolKey, Vec<&SymbolWithParent>> = HashMap::new();
    for sym in current_symbols {
        let key = SymbolKey {
            name: sym.name.clone(),
            kind: sym.kind,
            parent_name: sym.parent_name.clone(),
        };
        current_map.entry(key).or_default().push(sym);
    }

    let mut changes = Vec::new();
    let mut modified_ids = Vec::new();

    for (key, cur_syms) in &current_map {
        if let Some(old_syms) = old_map.get(key) {
            let paired = old_syms.len().min(cur_syms.len());
            for i in 0..paired {
                if old_syms[i].signature != cur_syms[i].signature {
                    changes.push(SymbolChange {
                        name: cur_syms[i].name.clone(),
                        kind: cur_syms[i].kind,
                        change_type: ChangeType::Modified,
                        file_path: file_path.to_string(),
                        line: cur_syms[i].start_line as u32,
                        parent_name: cur_syms[i].parent_name.clone(),
                        old_signature: old_syms[i].signature.clone(),
                        new_signature: cur_syms[i].signature.clone(),
                    });
                    modified_ids.push(cur_syms[i].id);
                }
            }
            for cur in cur_syms.iter().skip(paired) {
                changes.push(symbol_change_from_current(
                    (*cur).clone(),
                    ChangeType::Added,
                ));
            }
        } else {
            for cur in cur_syms {
                changes.push(symbol_change_from_current(
                    (*cur).clone(),
                    ChangeType::Added,
                ));
            }
        }
    }

    for (key, old_syms) in &old_map {
        let paired = current_map
            .get(key)
            .map_or(0, |c| c.len().min(old_syms.len()));
        for old in old_syms.iter().skip(paired) {
            changes.push(SymbolChange {
                name: old.name.clone(),
                kind: old.kind,
                change_type: ChangeType::Removed,
                file_path: file_path.to_string(),
                line: old.start_line as u32,
                parent_name: old.parent_name.clone(),
                old_signature: old.signature.clone(),
                new_signature: None,
            });
        }
    }

    if changes.len() > BUDGET_GUARD_THRESHOLD {
        modified_ids.clear();
    }

    (changes, modified_ids)
}

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

    fn make_extracted(
        name: &str,
        kind: SymbolKind,
        sig: Option<&str>,
        parent_index: Option<usize>,
    ) -> ExtractedSymbol {
        ExtractedSymbol {
            name: name.to_string(),
            kind,
            signature: sig.map(|s| s.to_string()),
            start_line: 1,
            end_line: 5,
            start_byte: 0,
            end_byte: 100,
            parent_index,
        }
    }

    fn make_db_sym(
        id: i64,
        name: &str,
        kind: SymbolKind,
        sig: Option<&str>,
        parent_name: Option<&str>,
    ) -> SymbolWithParent {
        SymbolWithParent {
            id,
            name: name.to_string(),
            kind,
            file_path: "src/lib.rs".to_string(),
            start_line: 1,
            end_line: 5,
            signature: sig.map(|s| s.to_string()),
            parent_name: parent_name.map(|s| s.to_string()),
        }
    }

    // --- compare_extracted_symbols tests ---

    #[test]
    fn extracted_added_new_key() {
        let (changes, _) = compare_extracted_symbols(
            &[],
            &[make_db_sym(
                1,
                "new_fn",
                SymbolKind::Function,
                Some("fn new_fn()"),
                None,
            )],
            "src/lib.rs",
        );
        assert_eq!(changes.len(), 1);
        assert_eq!(changes[0].name, "new_fn");
        assert_eq!(changes[0].change_type, ChangeType::Added);
    }

    #[test]
    fn extracted_removed_key() {
        let old = vec![make_extracted(
            "old_fn",
            SymbolKind::Function,
            Some("fn old_fn()"),
            None,
        )];
        let (changes, _) = compare_extracted_symbols(&old, &[], "src/lib.rs");
        assert_eq!(changes.len(), 1);
        assert_eq!(changes[0].name, "old_fn");
        assert_eq!(changes[0].change_type, ChangeType::Removed);
    }

    #[test]
    fn extracted_modified_signature() {
        let old = vec![make_extracted(
            "foo",
            SymbolKind::Function,
            Some("fn foo()"),
            None,
        )];
        let cur = vec![make_db_sym(
            1,
            "foo",
            SymbolKind::Function,
            Some("fn foo(x: i32)"),
            None,
        )];
        let (changes, mod_ids) = compare_extracted_symbols(&old, &cur, "src/lib.rs");
        assert_eq!(changes.len(), 1);
        assert_eq!(changes[0].change_type, ChangeType::Modified);
        assert_eq!(changes[0].old_signature.as_deref(), Some("fn foo()"));
        assert_eq!(changes[0].new_signature.as_deref(), Some("fn foo(x: i32)"));
        assert_eq!(mod_ids, vec![1]);
    }

    #[test]
    fn extracted_unchanged() {
        let old = vec![make_extracted(
            "foo",
            SymbolKind::Function,
            Some("fn foo()"),
            None,
        )];
        let cur = vec![make_db_sym(
            1,
            "foo",
            SymbolKind::Function,
            Some("fn foo()"),
            None,
        )];
        let (changes, _) = compare_extracted_symbols(&old, &cur, "src/lib.rs");
        assert!(changes.is_empty());
    }

    #[test]
    fn extracted_budget_guard() {
        let old: Vec<ExtractedSymbol> = (0..6)
            .map(|i| {
                make_extracted(
                    &format!("fn_{i}"),
                    SymbolKind::Function,
                    Some(&format!("fn fn_{i}()")),
                    None,
                )
            })
            .collect();
        let (changes, mod_ids) = compare_extracted_symbols(&old, &[], "src/lib.rs");
        assert_eq!(changes.len(), 6);
        assert!(mod_ids.is_empty(), "budget guard should clear modified_ids");
    }

    // --- compare_db_symbols tests ---

    #[test]
    fn db_added_new_key() {
        let cur = vec![make_db_sym(
            1,
            "new_fn",
            SymbolKind::Function,
            Some("fn new_fn()"),
            None,
        )];
        let (changes, _) = compare_db_symbols(&[], &cur, "src/lib.rs");
        assert_eq!(changes.len(), 1);
        assert_eq!(changes[0].name, "new_fn");
        assert_eq!(changes[0].change_type, ChangeType::Added);
    }

    #[test]
    fn db_removed_key() {
        let old = vec![make_db_sym(
            1,
            "old_fn",
            SymbolKind::Function,
            Some("fn old_fn()"),
            None,
        )];
        let (changes, _) = compare_db_symbols(&old, &[], "src/lib.rs");
        assert_eq!(changes.len(), 1);
        assert_eq!(changes[0].name, "old_fn");
        assert_eq!(changes[0].change_type, ChangeType::Removed);
    }

    #[test]
    fn db_modified_signature() {
        let old = vec![make_db_sym(
            1,
            "foo",
            SymbolKind::Function,
            Some("fn foo()"),
            None,
        )];
        let cur = vec![make_db_sym(
            1,
            "foo",
            SymbolKind::Function,
            Some("fn foo(x: i32)"),
            None,
        )];
        let (changes, mod_ids) = compare_db_symbols(&old, &cur, "src/lib.rs");
        assert_eq!(changes.len(), 1);
        assert_eq!(changes[0].change_type, ChangeType::Modified);
        assert_eq!(mod_ids, vec![1]);
    }

    #[test]
    fn db_unchanged() {
        let old = vec![make_db_sym(
            1,
            "foo",
            SymbolKind::Function,
            Some("fn foo()"),
            None,
        )];
        let cur = vec![make_db_sym(
            1,
            "foo",
            SymbolKind::Function,
            Some("fn foo()"),
            None,
        )];
        let (changes, _) = compare_db_symbols(&old, &cur, "src/lib.rs");
        assert!(changes.is_empty());
    }

    #[test]
    fn db_budget_guard() {
        let old: Vec<SymbolWithParent> = (0..6)
            .map(|i| {
                make_db_sym(
                    i,
                    &format!("fn_{i}"),
                    SymbolKind::Function,
                    Some(&format!("fn fn_{i}()")),
                    None,
                )
            })
            .collect();
        let (changes, mod_ids) = compare_db_symbols(&old, &[], "src/lib.rs");
        assert_eq!(changes.len(), 6);
        assert!(mod_ids.is_empty(), "budget guard should clear modified_ids");
    }

    #[test]
    fn db_extra_current_beyond_paired() {
        let old = vec![make_db_sym(
            1,
            "foo",
            SymbolKind::Function,
            Some("fn foo()"),
            None,
        )];
        let cur = vec![
            make_db_sym(1, "foo", SymbolKind::Function, Some("fn foo()"), None),
            make_db_sym(2, "foo", SymbolKind::Function, Some("fn foo(x: i32)"), None),
        ];
        let (changes, _) = compare_db_symbols(&old, &cur, "src/lib.rs");
        assert_eq!(changes.len(), 1);
        assert_eq!(changes[0].change_type, ChangeType::Added);
    }

    #[test]
    fn db_extra_old_beyond_paired() {
        let old = vec![
            make_db_sym(1, "foo", SymbolKind::Function, Some("fn foo()"), None),
            make_db_sym(2, "foo", SymbolKind::Function, Some("fn foo(x: i32)"), None),
        ];
        let cur = vec![make_db_sym(
            1,
            "foo",
            SymbolKind::Function,
            Some("fn foo()"),
            None,
        )];
        let (changes, _) = compare_db_symbols(&old, &cur, "src/lib.rs");
        assert_eq!(changes.len(), 1);
        assert_eq!(changes[0].change_type, ChangeType::Removed);
    }

    // --- ChangeType equality ---

    #[test]
    fn change_type_equality() {
        assert_eq!(ChangeType::Added, ChangeType::Added);
        assert_eq!(ChangeType::Modified, ChangeType::Modified);
        assert_eq!(ChangeType::Removed, ChangeType::Removed);
        assert_ne!(ChangeType::Added, ChangeType::Removed);
    }

    // --- Helper tests ---

    #[test]
    fn resolve_parent_name_with_parent() {
        let symbols = vec![
            make_extracted("MyClass", SymbolKind::Class, None, None),
            make_extracted("method", SymbolKind::Method, None, Some(0)),
        ];
        assert_eq!(
            resolve_parent_name(&symbols, &symbols[1]),
            Some("MyClass".to_string())
        );
    }

    #[test]
    fn resolve_parent_name_without_parent() {
        let symbols = vec![make_extracted("top_fn", SymbolKind::Function, None, None)];
        assert_eq!(resolve_parent_name(&symbols, &symbols[0]), None);
    }

    #[test]
    fn resolve_parent_name_out_of_bounds() {
        let symbols = vec![make_extracted("orphan", SymbolKind::Method, None, Some(99))];
        assert_eq!(resolve_parent_name(&symbols, &symbols[0]), None);
    }
}