strop_git/

memory.rs

//! Git memory (M3, 0001 pillar 3.2/3.3): log graph, blame, permalinks.
//! Reads via shell `git` (matches user config; not hot-path), permalinks
//! via libgit2 config (no spawn).

use std::path::{Path, PathBuf};

use crate::Repo;

/// One log line from `git log --graph`, with the commit hash extracted.
#[derive(Debug, Clone)]
pub struct LogRow {
    /// The rendered graph+summary line (what the buffer shows).
    pub text: String,
    /// Full SHA when the line names a commit (graph-only lines: None).
    pub sha: Option<String>,
}

/// `git log --graph` for the browser. Shells out — the log is not a
/// per-keystroke path (0001 §3). Caller decides threading.
pub fn log_graph(workdir: &Path, max: usize, file: Option<&Path>) -> Result<Vec<LogRow>, String> {
    log_graph_range(workdir, max, file, None)
}

/// `git log -L start,end:path` — the history of a line range (0014 wave
/// 4: selection archaeology). The graph flag is meaningless with -L;
/// rows come straight from the patch headers.
pub fn log_graph_range(
    workdir: &Path,
    max: usize,
    file: Option<&Path>,
    range: Option<(usize, usize)>,
) -> Result<Vec<LogRow>, String> {
    let mut cmd = std::process::Command::new("git");
    let (marker_fmt, ranged) = match range {
        Some(_) => ("%x01%h %an · %ar · %s%x00%H", true),
        None => ("%h %an · %ar · %s%x00%H", false),
    };
    cmd.args([
        "-C",
        &workdir.display().to_string(),
        "log",
        &format!("--format={marker_fmt}"),
        "-n",
        &max.to_string(),
    ]);
    match (file, range) {
        (Some(f), Some((a, b))) => {
            cmd.arg(format!("-L{a},{b}:{}", f.display()));
        }
        (Some(f), None) => {
            cmd.arg("--graph").arg("--").arg(f);
        }
        (None, None) => {
            cmd.arg("--graph");
        }
        (None, Some(_)) => return Err("-L needs a file".into()),
    }
    let out = cmd.output().map_err(|e| format!("spawn git log: {e}"))?;
    if !out.status.success() {
        return Err(String::from_utf8_lossy(&out.stderr).trim().to_string());
    }
    let text = String::from_utf8_lossy(&out.stdout);
    Ok(text
        .lines()
        // -L output carries patch text; only marked lines are commits
        .filter(|line| !ranged || line.starts_with('\x01'))
        .map(|line| {
            let line = line.strip_prefix('\x01').unwrap_or(line);
            // the format hides the full SHA after a NUL
            let (vis, sha) = match line.split_once('\0') {
                Some((v, s)) => (v.to_string(), Some(s.trim().to_string())),
                None => (line.to_string(), None),
            };
            LogRow { text: vis, sha }
        })
        .collect())
}

/// A blame card for one line (0001 pillar 3.3).
#[derive(Debug, Clone)]
pub struct BlameCard {
    pub sha: String,
    pub short_sha: String,
    pub author: String,
    pub age: String,
    pub summary: String,
    pub line: usize,
}

/// Blame one line of a file (1-based). Shells out; porcelain format.
pub fn blame_line(workdir: &Path, rel: &Path, line: usize) -> Result<BlameCard, String> {
    let out = std::process::Command::new("git")
        .args([
            "-C",
            &workdir.display().to_string(),
            "blame",
            "--line-porcelain",
            "-L",
            &format!("{line},{line}"),
            "--",
            &rel.display().to_string(),
        ])
        .output()
        .map_err(|e| format!("spawn git blame: {e}"))?;
    if !out.status.success() {
        return Err(String::from_utf8_lossy(&out.stderr).trim().to_string());
    }
    let text = String::from_utf8_lossy(&out.stdout);
    let mut sha = String::new();
    let mut author = String::new();
    let mut summary = String::new();
    let mut ts = 0i64;
    for l in text.lines() {
        if sha.is_empty()
            && !l.starts_with('\t')
            && l.chars().take(8).all(|c| c.is_ascii_hexdigit())
        {
            sha = l.split_whitespace().next().unwrap_or("").to_string();
        } else if let Some(a) = l.strip_prefix("author ") {
            author = a.to_string();
        } else if let Some(t) = l.strip_prefix("author-time ") {
            ts = t.parse().unwrap_or(0);
        } else if let Some(s) = l.strip_prefix("summary ") {
            summary = s.to_string();
        }
    }
    if sha.is_empty() {
        return Err("no blame for line".into());
    }
    Ok(BlameCard {
        short_sha: sha.chars().take(8).collect(),
        sha,
        author,
        age: rel_age(ts),
        summary,
        line,
    })
}

/// One line of a whole-file blame (0001 pillar 3.3, the toggleable
/// column). `age` is rendered at parse time; `ts` keeps "recent"
/// honest for the caller's coloring.
#[derive(Debug, Clone)]
pub struct BlameLine {
    pub sha: String,
    pub author: String,
    /// Human short form ("3h", "2d", "5mo"); "now" when uncommitted.
    pub age: String,
    /// Author time, unix seconds (0 = uncommitted).
    pub ts: i64,
}

impl BlameLine {
    /// Worktree lines git blame attributes to nobody (all-zero sha).
    pub fn is_uncommitted(&self) -> bool {
        !self.sha.is_empty() && self.sha.chars().all(|c| c == '0')
    }
}

/// Blame every line of a file (`--line-porcelain`; the gutter's data,
/// 0011 §3). Shells out on a job thread — never the input path.
pub fn blame_file(workdir: &Path, rel: &Path) -> Result<Vec<BlameLine>, String> {
    let out = std::process::Command::new("git")
        .args([
            "-C",
            &workdir.display().to_string(),
            "blame",
            "--line-porcelain",
            "--",
            &rel.display().to_string(),
        ])
        .output()
        .map_err(|e| format!("spawn git blame: {e}"))?;
    if !out.status.success() {
        return Err(String::from_utf8_lossy(&out.stderr).trim().to_string());
    }
    let mut lines = Vec::new();
    let mut sha = String::new();
    let mut author = String::new();
    let mut ts = 0i64;
    for l in String::from_utf8_lossy(&out.stdout).lines() {
        if let Some(content) = l.strip_prefix('\t') {
            // the record's content row closes it — porcelain repeats
            // the full header per line, so every tab row emits one
            let _ = content;
            if !sha.is_empty() {
                let uncommitted = sha.chars().all(|c| c == '0');
                lines.push(BlameLine {
                    sha: sha.clone(),
                    age: if uncommitted {
                        "now".into()
                    } else {
                        rel_age(ts)
                    },
                    author: if uncommitted {
                        "you".into()
                    } else {
                        author.clone()
                    },
                    ts: if uncommitted { 0 } else { ts },
                });
            }
            sha.clear();
            author.clear();
            ts = 0;
        } else if sha.is_empty()
            && !l.is_empty()
            && l.chars().take(40).all(|c| c.is_ascii_hexdigit())
        {
            sha = l.split_whitespace().next().unwrap_or("").to_string();
        } else if let Some(a) = l.strip_prefix("author ") {
            author = a.to_string();
        } else if let Some(t) = l.strip_prefix("author-time ") {
            ts = t.parse().unwrap_or(0);
        }
    }
    if lines.is_empty() {
        return Err("no blame for file".into());
    }
    Ok(lines)
}

/// Files changed by a commit: `path | +N -M` rows for the dive view.
#[derive(Debug, Clone)]
pub struct ChangedFile {
    pub path: PathBuf,
    pub added: usize,
    pub deleted: usize,
}

pub fn show_stat(workdir: &Path, sha: &str) -> Result<Vec<ChangedFile>, String> {
    let out = std::process::Command::new("git")
        .args([
            "-C",
            &workdir.display().to_string(),
            "show",
            "--numstat",
            "--format=",
            sha,
        ])
        .output()
        .map_err(|e| format!("spawn git show: {e}"))?;
    if !out.status.success() {
        return Err(String::from_utf8_lossy(&out.stderr).trim().to_string());
    }
    Ok(String::from_utf8_lossy(&out.stdout)
        .lines()
        .filter_map(|l| {
            let mut parts = l.split('\t');
            let added = parts.next()?.parse().ok()?;
            let deleted = parts.next()?.parse().ok()?;
            Some(ChangedFile {
                path: PathBuf::from(parts.next()?),
                added,
                deleted,
            })
        })
        .collect())
}

// ---- permalinks ----------------------------------------------------------

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Host {
    GitHub,
    GitLab,
    Bitbucket,
    Gitea,
    /// Unknown host: emit whatever HTTPS we can normalize to.
    Other,
}

pub struct Remote {
    pub host: Host,
    pub owner_repo: String, // "org/repo"
    pub base: String,       // "https://github.com"
}

/// Normalize a remote URL (SSH or HTTPS) to a web base. Priority
/// upstream > origin > rest is the caller's job (0001 pillar 3.3).
pub fn normalize_remote(url: &str) -> Option<Remote> {
    let url = url.trim().trim_end_matches(".git");
    let (base, path) = if let Some(rest) = url.strip_prefix("git@") {
        // git@host:org/repo
        let (host, path) = rest.split_once(':')?;
        (format!("https://{host}"), path.to_string())
    } else if let Some(rest) = url.strip_prefix("ssh://git@") {
        // ssh://git@host/org/repo
        let rest = rest.split('/').collect::<Vec<_>>();
        let host = rest.first()?;
        (format!("https://{host}"), rest[1..].join("/"))
    } else if url.starts_with("https://") || url.starts_with("http://") {
        let stripped = url
            .strip_prefix("https://")
            .or_else(|| url.strip_prefix("http://"))?;
        let (host, path) = stripped.split_once('/')?;
        (format!("https://{host}"), path.to_string())
    } else if let Some((host, path)) = url.split_once(':') {
        // scp syntax without user@: bare hostname or an ssh host alias
        // (`bbgithub:org/repo` — ~/.ssh/config supplies the real host)
        if host.contains('@') || host.contains('/') {
            return None;
        }
        let host = resolve_ssh_alias(host).unwrap_or_else(|| host.to_string());
        (format!("https://{host}"), path.to_string())
    } else {
        return None;
    };
    let host = match base.as_str() {
        "https://github.com" => Host::GitHub,
        "https://gitlab.com" => Host::GitLab,
        "https://bitbucket.org" => Host::Bitbucket,
        b if b.contains("gitea") => Host::Gitea,
        _ => Host::Other,
    };
    Some(Remote {
        host,
        owner_repo: path,
        base,
    })
}

/// Resolve an ssh host alias via `~/.ssh/config` Host blocks (exact
/// matches; wildcard blocks skipped). Enterprise GitHub setups live on
/// these — the alias exists so the hostname isn't repeated per clone.
fn resolve_ssh_alias(alias: &str) -> Option<String> {
    let home = std::env::var_os("HOME")?;
    let config = std::fs::read_to_string(PathBuf::from(home).join(".ssh").join("config")).ok()?;
    parse_ssh_alias(&config, alias)
}

fn parse_ssh_alias(config: &str, alias: &str) -> Option<String> {
    let mut in_block = false;
    for line in config.lines() {
        let line = line.trim();
        if line.is_empty() || line.starts_with('#') {
            continue;
        }
        let mut parts = line.split_whitespace();
        match parts.next().map(|k| k.to_ascii_lowercase()).as_deref() {
            Some("host") => in_block = parts.any(|h| h == alias),
            Some("hostname") if in_block => return parts.next().map(|h| h.to_string()),
            _ => {}
        }
    }
    None
}

/// Pick the permalink remote: upstream > origin > first remaining.
pub fn pick_remote(repo: &Repo) -> Option<Remote> {
    let remotes = repo.remotes();
    for name in ["upstream", "origin"] {
        if let Some(url) = remotes.iter().find(|(n, _)| n == name).map(|(_, u)| u) {
            if let Some(r) = normalize_remote(url) {
                return Some(r);
            }
        }
    }
    remotes.iter().find_map(|(_, u)| normalize_remote(u))
}

/// Build the immutable permalink for a file at 1-based lines. Branch is
/// always resolved to a commit SHA (0001 pillar 3.3).
/// The URL for a revisioned location (0014): pinned to the location's
/// revision — a commit surface links that commit, not HEAD.
pub fn permalink(repo: &Repo, loc: &crate::SourceLocation) -> Option<String> {
    let remote = pick_remote(repo)?;
    let (start_line, end_line) = loc.lines.unwrap_or((1, 1));
    let sha = match &loc.revision {
        crate::GitRevision::Head => repo.head_sha()?,
        crate::GitRevision::Commit(sha) => sha.clone(),
        // permalinks pin to an immutable commit — the index/worktree
        // resolve through HEAD (their content is already local state)
        crate::GitRevision::Index | crate::GitRevision::Worktree => repo.head_sha()?,
        crate::GitRevision::MergeBase(a, b) => repo.merge_base(a, b)?,
    };
    let frag = if start_line == end_line {
        format!("#L{start_line}")
    } else {
        format!("#L{start_line}-L{end_line}")
    };
    Some(format!(
        "{}/{}/blob/{}/{}{frag}",
        remote.base,
        remote.owner_repo,
        sha,
        loc.path.display()
    ))
}

/// Relative age, human short form ("3h", "2d", "5mo").
fn rel_age(ts: i64) -> String {
    let now = std::time::SystemTime::now()
        .duration_since(std::time::UNIX_EPOCH)
        .map(|d| d.as_secs() as i64)
        .unwrap_or(0);
    let age = (now - ts).max(0);
    match age {
        a if a < 3600 => format!("{}m", a / 60),
        a if a < 86400 => format!("{}h", a / 3600),
        a if a < 86400 * 30 => format!("{}d", a / 86400),
        a if a < 86400 * 365 => format!("{}mo", a / (86400 * 30)),
        a => format!("{}y", a / (86400 * 365)),
    }
}

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

    #[test]
    fn ssh_alias_resolves_via_config() {
        let config = "# comment\nHost bbgithub\n  HostName bbgithub.dev.bloomberg.com\n  User git\nHost *\n  ServerAliveInterval 30\n";
        assert_eq!(
            parse_ssh_alias(config, "bbgithub").as_deref(),
            Some("bbgithub.dev.bloomberg.com")
        );
        assert_eq!(parse_ssh_alias(config, "other"), None);
        // wildcard-only blocks don't claim aliases
        assert_eq!(parse_ssh_alias("Host *\n  HostName x", "bbgithub"), None);
    }

    #[test]
    fn scp_without_user_parses_as_bare_host() {
        // unresolved alias falls back to the bare name (matches what git
        // itself would attempt) — but with a config entry it resolves
        let r = normalize_remote("bbgithub:acme/demo.git");
        assert!(r.is_some(), "alias form parses");
    }

    #[test]
    fn reviewer_table() {
        // the first-week report's remote table, verbatim
        for url in [
            "https://github.com/acme/demo.git",
            "ssh://git@github.com/acme/demo.git",
            "git@github.com:acme/demo",
            "git@bbgithub.dev.bloomberg.com:acme/demo.git",
            "https://bbgithub.dev.bloomberg.com/acme/demo.git",
        ] {
            let r = normalize_remote(url);
            assert!(r.is_some(), "should parse: {url}");
        }
        // the ssh host-alias form parses (bare-host fallback; resolves
        // via ~/.ssh/config when an entry exists)
        assert!(normalize_remote("bbgithub:acme/demo.git").is_some());
    }

    #[test]
    fn normalizes_ssh_and_https() {
        let r = normalize_remote("git@github.com:stropdev/strop.git").unwrap();
        assert_eq!(
            (r.base.as_str(), r.owner_repo.as_str()),
            ("https://github.com", "stropdev/strop")
        );
        assert_eq!(r.host, Host::GitHub);
        let r = normalize_remote("https://gitlab.com/org/proj").unwrap();
        assert_eq!(r.host, Host::GitLab);
        assert_eq!(r.owner_repo, "org/proj");
        let r = normalize_remote("ssh://git@bitbucket.org/team/repo.git").unwrap();
        assert_eq!(r.host, Host::Bitbucket);
        assert!(normalize_remote("not a url").is_none());
    }

    /// Repo with two commits (f.rs grows a line), then a dirty edit —
    /// blame_file must attribute committed lines and flag dirty ones.
    #[test]
    fn blame_file_attributes_lines() {
        let dir = tempfile::tempdir().unwrap();
        let root = dir.path();
        let git = |args: &[&str]| {
            std::process::Command::new("git")
                .args(args)
                .current_dir(root)
                .output()
                .unwrap();
        };
        git(&["init", "-q"]);
        git(&["config", "user.email", "t@t.t"]);
        git(&["config", "user.name", "t"]);
        std::fs::write(root.join("f.rs"), "one\n").unwrap();
        git(&["add", "."]);
        git(&["commit", "-qm", "first"]);
        std::fs::write(root.join("f.rs"), "one\ntwo\n").unwrap();
        git(&["commit", "-qam", "second"]);

        let clean = blame_file(root, Path::new("f.rs")).unwrap();
        assert_eq!(clean.len(), 2, "one BlameLine per file line");
        assert_eq!(clean[0].author, "t");
        assert_eq!(clean[1].author, "t");
        assert_ne!(clean[0].sha, clean[1].sha, "two commits, two shas");
        assert!(!clean[0].is_uncommitted());

        // dirty worktree: the new line belongs to nobody
        std::fs::write(root.join("f.rs"), "one\ntwo\nthree\n").unwrap();
        let dirty = blame_file(root, Path::new("f.rs")).unwrap();
        assert_eq!(dirty.len(), 3);
        assert!(dirty[2].is_uncommitted(), "last line is uncommitted");
        assert_eq!(dirty[2].age, "now");
        assert_eq!(dirty[2].author, "you");
        assert_eq!(dirty[2].ts, 0);
    }

    #[test]
    fn blame_file_rejects_missing_file() {
        let dir = tempfile::tempdir().unwrap();
        assert!(blame_file(dir.path(), Path::new("nope.rs")).is_err());
    }
}