bctx-weave 0.1.7

use forge::signal::compactor;
use once_cell::sync::Lazy;
use regex::Regex;

// ── Noise-line patterns ───────────────────────────────────────────────────────

static GIT_HINT_RE: Lazy<Regex> = Lazy::new(|| {
    Regex::new(r#"(?m)^\s*(\(use "git|hint:|nothing to commit|no changes added)"#).unwrap()
});
static GIT_OBJECT_COUNT_RE: Lazy<Regex> = Lazy::new(|| {
    Regex::new(r"(?m)^(Counting|Compressing|remote: Counting|remote: Compressing|Receiving|Resolving|Writing) objects:.*\n?").unwrap()
});
static GIT_DELTA_RE: Lazy<Regex> =
    Lazy::new(|| Regex::new(r"(?m)^(remote: )?delta compression.*\n?").unwrap());
static GIT_INDEX_RE: Lazy<Regex> = Lazy::new(|| {
    // diff index lines like: "index abc123..def456 100644"
    Regex::new(r"(?m)^index [0-9a-f]+\.\.[0-9a-f]+ \d+\n?").unwrap()
});
static GIT_SIMILARITY_RE: Lazy<Regex> = Lazy::new(|| {
    Regex::new(r"(?m)^similarity index \d+%\n?(rename from .*\nrename to .*\n?)?").unwrap()
});
static GIT_COMMIT_SIGNING_RE: Lazy<Regex> = Lazy::new(|| {
    Regex::new(r"(?m)^( +gpg:|-----BEGIN PGP|-----END PGP|gpg: Signature).*\n?").unwrap()
});
static GIT_BLAME_HEADER_RE: Lazy<Regex> =
    Lazy::new(|| Regex::new(r"(?m)^[0-9a-f]{40} \(").unwrap());

// ── git log ───────────────────────────────────────────────────────────────────

/// Compress `git log --oneline` output: keep first + last few commits, summarise middle.
pub fn compress_log(raw: &str, max_commits: usize) -> String {
    let cleaned = compactor::normalise(raw);
    let lines: Vec<&str> = cleaned.lines().filter(|l| !l.trim().is_empty()).collect();

    if lines.len() <= max_commits {
        return cleaned;
    }

    let head = max_commits / 2;
    let tail = max_commits - head;
    let skipped = lines.len() - head - tail;

    format!(
        "{}\n... [{skipped} commits omitted] ...\n{}",
        lines[..head].join("\n"),
        lines[lines.len() - tail..].join("\n"),
    )
}

/// Compress verbose `git log` (multi-line per commit): keep hash + subject only.
pub fn compress_log_verbose(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let noise_prefixes = ["Author:", "Date:", "Merge:", "    "];
    let mut out = Vec::new();
    for line in cleaned.lines() {
        let t = line.trim();
        if t.is_empty() {
            continue;
        }
        if noise_prefixes.iter().any(|p| line.starts_with(p)) {
            continue;
        }
        // Commit line: "commit <hash>"  → shorten hash
        if let Some(rest) = line.strip_prefix("commit ") {
            out.push(format!("commit {}", &rest[..8.min(rest.len())]));
        } else {
            out.push(line.to_string());
        }
    }
    out.join("\n")
}

// ── git diff ──────────────────────────────────────────────────────────────────

/// Compress full `git diff` output: strip noise lines, keep hunks.
pub fn compress_diff(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let s1 = GIT_INDEX_RE.replace_all(&cleaned, "");
    let s2 = GIT_SIMILARITY_RE.replace_all(&s1, "");
    // Count changed files; if > 8, show first 8 + summary
    let file_headers: Vec<&str> = s2.lines().filter(|l| l.starts_with("diff --git")).collect();
    if file_headers.len() > 8 {
        let truncated = truncate_diff_to_n_files(&s2, 8);
        return format!(
            "{}\n... [{} more files not shown] ...",
            truncated,
            file_headers.len() - 8
        );
    }
    s2.into_owned()
}

/// Compress `git diff --stat` output.
pub fn compress_diff_stat(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let lines: Vec<&str> = cleaned.lines().filter(|l| !l.trim().is_empty()).collect();
    if let Some(last) = lines.last() {
        if lines.len() > 6 {
            let files_shown = 5.min(lines.len() - 1);
            let shown = lines[..files_shown].to_vec();
            return format!(
                "{}\n... [{} more files] ...\n{}",
                shown.join("\n"),
                lines.len() - 1 - files_shown,
                last
            );
        }
    }
    cleaned
}

fn truncate_diff_to_n_files(diff: &str, n: usize) -> String {
    let mut count = 0;
    let mut out = String::new();
    for line in diff.lines() {
        if line.starts_with("diff --git") {
            count += 1;
            if count > n {
                break;
            }
        }
        out.push_str(line);
        out.push('\n');
    }
    out
}

// ── git status ────────────────────────────────────────────────────────────────

/// Compress `git status` output by removing redundant hints and usage lines.
pub fn compress_status(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    // Strip hint lines
    let stripped = GIT_HINT_RE.replace_all(&cleaned, "");
    // Collapse long untracked-file lists (> 10 entries → "and N more")
    let mut out: Vec<&str> = Vec::new();
    let mut untracked_count = 0usize;
    let mut in_untracked = false;
    let max_untracked = 8;

    for line in stripped.lines() {
        if line.contains("Untracked files:") {
            in_untracked = true;
        }
        if line.contains("Changes to be committed:") || line.contains("Changes not staged") {
            in_untracked = false;
        }
        if in_untracked && line.starts_with('\t') {
            untracked_count += 1;
            if untracked_count <= max_untracked {
                out.push(line);
            } else if untracked_count == max_untracked + 1 {
                out.push("  ... and more untracked files (run git status for full list)");
            }
        } else {
            out.push(line);
        }
    }
    compactor::collapse_blanks(&out.join("\n"))
}

// ── git commit ────────────────────────────────────────────────────────────────

/// Compress `git commit` output: strip signing, hooks noise, keep summary.
pub fn compress_commit(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let s1 = GIT_COMMIT_SIGNING_RE.replace_all(&cleaned, "");
    // Strip pre/post-commit hook progress lines (often "[hook] running…" type lines)
    let out: Vec<&str> = s1
        .lines()
        .filter(|l| {
            let t = l.trim();
            !t.starts_with("running ") && !t.contains("pre-commit") && !t.contains("post-commit")
        })
        .collect();
    compactor::collapse_blanks(&out.join("\n"))
}

// ── git fetch / pull ──────────────────────────────────────────────────────────

/// Compress `git fetch` / `git pull` output: strip object-count progress.
pub fn compress_fetch(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let s1 = GIT_OBJECT_COUNT_RE.replace_all(&cleaned, "");
    let s2 = GIT_DELTA_RE.replace_all(&s1, "");
    // Strip "remote:" lines that are just whitespace padding
    let out: Vec<&str> = s2.lines().filter(|l| l.trim() != "remote:").collect();
    compactor::collapse_blanks(&out.join("\n"))
}

// ── git blame ───────────────────────────────────────────────────────────────���─

/// Compress `git blame`: shorten 40-char SHAs to 8-char, deduplicate repeated author+date.
pub fn compress_blame(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    if !GIT_BLAME_HEADER_RE.is_match(&cleaned) {
        return cleaned;
    }
    // Shorten SHA from 40 to 8 chars
    let short_sha_re = Regex::new(r"^[0-9a-f]{40}").unwrap();
    let lines: Vec<String> = cleaned
        .lines()
        .map(|l| {
            if short_sha_re.is_match(l) {
                format!("{}{}", &l[..8], &l[40..])
            } else {
                l.to_string()
            }
        })
        .collect();
    lines.join("\n")
}

// ── git stash / branch / remote ──────────────────────────────────────────────

/// Compact `git stash list` output: truncate to 20 entries.
pub fn compress_stash_list(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let lines: Vec<&str> = cleaned.lines().filter(|l| !l.trim().is_empty()).collect();
    if lines.len() <= 20 {
        return lines.join("\n");
    }
    format!(
        "{}\n... [{} more stash entries]",
        lines[..20].join("\n"),
        lines.len() - 20
    )
}

static GIT_BRANCH_TRACKING_RE: Lazy<Regex> =
    Lazy::new(|| Regex::new(r" \[(?:ahead \d+(?:, )?)?(?:behind \d+)?\]").unwrap());

/// Compact `git branch -v[v]`: strip tracking noise, keep name + hash + subject.
/// Truncate at 40 branches.
pub fn compress_branch(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let s = GIT_BRANCH_TRACKING_RE.replace_all(&cleaned, "");
    let lines: Vec<&str> = s.lines().filter(|l| !l.trim().is_empty()).collect();
    if lines.len() <= 40 {
        return lines.join("\n");
    }
    format!(
        "{}\n... [{} more branches]",
        lines[..40].join("\n"),
        lines.len() - 40
    )
}

/// Compact `git tag -l`: truncate at 30 tags.
pub fn compress_tag(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let lines: Vec<&str> = cleaned.lines().filter(|l| !l.trim().is_empty()).collect();
    if lines.len() <= 30 {
        return lines.join("\n");
    }
    format!(
        "{}\n... [{} more tags]",
        lines[..30].join("\n"),
        lines.len() - 30
    )
}

/// Compact `git remote -v`: deduplicate fetch/push pairs (show fetch URL only).
pub fn compress_remote(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    // Remove " (push)" duplicate lines — keep "(fetch)" lines and plain lines
    let lines: Vec<&str> = cleaned.lines().filter(|l| !l.ends_with("(push)")).collect();
    lines.join("\n")
}

/// Compact `git rebase` output: strip "Rebasing (N/M)" progress, keep result.
pub fn compress_rebase(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let rebase_progress_re = Regex::new(r"(?m)^Rebasing \(\d+/\d+\)\n?").unwrap();
    let s = rebase_progress_re.replace_all(&cleaned, "");
    compactor::collapse_blanks(&s)
}

/// Compact `git cherry-pick` output: keep conflict info, strip applied-cleanly noise.
pub fn compress_cherry_pick(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    // If no conflicts, keep only the first line (commit hash + subject)
    if !cleaned.contains("CONFLICT") && !cleaned.contains("conflict") {
        return cleaned.lines().next().unwrap_or("").to_string();
    }
    cleaned
}

/// Compact `git show`: treat as diff but also keep commit header (first ~5 lines).
pub fn compress_show(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let s1 = GIT_INDEX_RE.replace_all(&cleaned, "");
    let s2 = GIT_SIMILARITY_RE.replace_all(&s1, "");
    compactor::collapse_blanks(&s2)
}

/// Compact `git worktree list`: normalise only (output is already compact).
pub fn compress_worktree(raw: &str) -> String {
    compactor::normalise(raw)
}

/// Compact `git gc` output: strip object counting/compressing noise.
pub fn compress_gc(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    let s = GIT_OBJECT_COUNT_RE.replace_all(&cleaned, "");
    compactor::collapse_blanks(&s)
}

/// Compact `git submodule` output.
pub fn compress_submodule(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    // Strip "Submodule path 'X': checked out 'sha'" noise on update
    let lines: Vec<&str> = cleaned
        .lines()
        .filter(|l| {
            let t = l.trim();
            !t.is_empty()
                && (!t.starts_with("Cloning into") || t.len() <= 60)
                && !t.starts_with("remote: Counting")
                && !t.starts_with("remote: Compressing")
                && !t.starts_with("Receiving objects:")
                && !t.starts_with("Resolving deltas:")
        })
        .collect();
    lines.join("\n")
}

/// Compact `git bisect` output: keep only actionable lines.
pub fn compress_bisect(raw: &str) -> String {
    let cleaned = compactor::normalise(raw);
    // Keep: "Bisecting: N revisions left", "is the first bad commit", error lines
    let out: Vec<&str> = cleaned
        .lines()
        .filter(|l| {
            l.starts_with("Bisecting:")
                || l.contains("first bad commit")
                || l.starts_with("commit ")
                || l.starts_with("[")
                || l.starts_with("error")
                || l.starts_with("fatal")
        })
        .collect();
    if out.is_empty() {
        cleaned
    } else {
        out.join("\n")
    }
}

// ── top-level dispatcher ─────────────────────────────────────────────────────

/// Route git output to the right compressor by sub-command.
pub fn compress_git(subcmd: &str, raw: &str, exit_code: i32) -> String {
    if exit_code != 0 {
        return compactor::normalise(raw);
    }
    let sub = subcmd.trim();

    if sub.starts_with("log") {
        // Verbose log has "commit <sha>" lines
        if raw
            .lines()
            .any(|l| l.starts_with("commit ") && l.len() == 47)
        {
            return compress_log_verbose(raw);
        }
        return compress_log(raw, 12);
    }
    if sub.starts_with("diff") {
        if sub.contains("--stat") {
            return compress_diff_stat(raw);
        }
        return compress_diff(raw);
    }
    if sub.starts_with("status") {
        return compress_status(raw);
    }
    if sub.starts_with("commit") {
        return compress_commit(raw);
    }
    if sub.starts_with("fetch") || sub.starts_with("pull") {
        return compress_fetch(raw);
    }
    if sub.starts_with("blame") {
        return compress_blame(raw);
    }
    if sub.starts_with("stash") {
        return compress_stash_list(raw);
    }
    if sub.starts_with("branch") {
        return compress_branch(raw);
    }
    if sub.starts_with("tag") {
        return compress_tag(raw);
    }
    if sub.starts_with("remote") {
        return compress_remote(raw);
    }
    if sub.starts_with("rebase") {
        return compress_rebase(raw);
    }
    if sub.starts_with("cherry-pick") {
        return compress_cherry_pick(raw);
    }
    if sub.starts_with("show") {
        return compress_show(raw);
    }
    if sub.starts_with("worktree") {
        return compress_worktree(raw);
    }
    if sub.starts_with("bisect") {
        return compress_bisect(raw);
    }
    if sub.starts_with("gc") {
        return compress_gc(raw);
    }
    if sub.starts_with("submodule") {
        return compress_submodule(raw);
    }

    // Generic: normalise + collapse blanks
    let cleaned = compactor::normalise(raw);
    let s = GIT_OBJECT_COUNT_RE.replace_all(&cleaned, "");
    compactor::collapse_blanks(&s)
}

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

    #[test]
    fn log_compression_keeps_head_tail() {
        let log = (0..20)
            .map(|i| format!("abc{i:04} commit message {i}"))
            .collect::<Vec<_>>()
            .join("\n");
        let out = compress_log(&log, 6);
        assert!(out.contains("omitted"), "should summarise middle: {out}");
        assert!(out.lines().count() < 20);
    }

    #[test]
    fn diff_stat_truncates_long_lists() {
        let mut s = String::new();
        for i in 0..20 {
            s.push_str(&format!(" file{i}.rs | 10 ++--\n"));
        }
        s.push_str(" 20 files changed, 100 insertions(+), 50 deletions(-)\n");
        let out = compress_diff_stat(&s);
        assert!(out.contains("more files"));
        assert!(out.contains("20 files changed"));
    }

    #[test]
    fn status_strips_hints() {
        let raw =
            "On branch main\nChanges not staged:\n  (use \"git add\" ...)\n\tmodified: foo.rs\n";
        let out = compress_status(raw);
        assert!(!out.contains("use \"git"), "should strip hints: {out}");
        assert!(out.contains("foo.rs"));
    }

    #[test]
    fn commit_strips_signing() {
        let raw = "[main abc1234] feat: add thing\n gpg: Signature made Mon\n 1 file changed\n";
        let out = compress_commit(raw);
        assert!(!out.contains("gpg:"));
        assert!(out.contains("feat: add thing"));
    }

    #[test]
    fn fetch_strips_object_count() {
        let raw = "remote: Counting objects: 10, done.\nFrom github.com:org/repo\n   abc..def  main -> origin/main\n";
        let out = compress_fetch(raw);
        assert!(!out.contains("Counting objects"));
        assert!(out.contains("main -> origin/main"));
    }

    #[test]
    fn diff_strips_index_lines() {
        let raw = "diff --git a/foo.rs b/foo.rs\nindex abc123..def456 100644\n--- a/foo.rs\n+++ b/foo.rs\n@@ -1,3 +1,4 @@\n+new line\n";
        let out = compress_diff(raw);
        assert!(
            !out.contains("index abc123"),
            "index line should be stripped: {out}"
        );
        assert!(out.contains("+new line"));
    }

    #[test]
    fn diff_truncates_many_files() {
        let mut s = String::new();
        for i in 0..12 {
            s.push_str(&format!("diff --git a/f{i}.rs b/f{i}.rs\n--- a/f{i}.rs\n+++ b/f{i}.rs\n@@ -1 +1 @@\n-old\n+new\n"));
        }
        let out = compress_diff(&s);
        assert!(out.contains("more files not shown"));
    }

    #[test]
    fn stash_list_truncates_at_20() {
        let raw = (0..30)
            .map(|i| format!("stash@{{{i}}}: On main: WIP change {i}"))
            .collect::<Vec<_>>()
            .join("\n");
        let out = compress_stash_list(&raw);
        assert!(out.contains("more stash entries"), "{out}");
    }

    #[test]
    fn tag_list_truncates_at_30() {
        let raw = (0..50)
            .map(|i| format!("v1.{i}.0"))
            .collect::<Vec<_>>()
            .join("\n");
        let out = compress_tag(&raw);
        assert!(out.contains("more tags"), "{out}");
    }

    #[test]
    fn remote_deduplicates_push_lines() {
        let raw = "origin\thttps://github.com/org/repo.git (fetch)\norigin\thttps://github.com/org/repo.git (push)\n";
        let out = compress_remote(raw);
        assert!(!out.contains("(push)"), "{out}");
        assert!(out.contains("(fetch)") || out.contains("origin"), "{out}");
    }

    #[test]
    fn rebase_strips_progress() {
        let raw = "Rebasing (1/10)\nRebasing (2/10)\nRebasing (10/10)\nSuccessfully rebased and updated refs/heads/main.\n";
        let out = compress_rebase(raw);
        assert!(!out.contains("Rebasing (1/10)"), "{out}");
        assert!(out.contains("Successfully rebased"), "{out}");
    }

    #[test]
    fn cherry_pick_clean_keeps_first_line_only() {
        let raw = "[main abc1234] feat: add thing\n Date: Mon Jan 1 12:00:00 2024\n 1 file changed, 5 insertions(+)\n";
        let out = compress_cherry_pick(raw);
        // Should keep only the first summary line
        assert!(out.contains("feat: add thing"), "{out}");
        assert!(!out.contains("Date:"), "{out}");
    }
}