start-os/core/src/disk/util.rs

use std::collections::{BTreeMap, BTreeSet};
use std::path::{Path, PathBuf};

use chrono::{DateTime, Utc};
use color_eyre::eyre::{self, eyre};
use futures::TryStreamExt;
use nom::bytes::complete::{tag, take_till1};
use nom::character::complete::multispace1;
use nom::combinator::{opt, rest};
use nom::sequence::{pair, preceded, terminated};
use nom::{AsChar, IResult, Parser};
use regex::Regex;
use serde::{Deserialize, Serialize};
use tokio::process::Command;
use tracing::instrument;

use super::mount::filesystem::ReadOnly;
use super::mount::filesystem::block_dev::BlockDev;
use super::mount::guard::TmpMountGuard;
use crate::disk::OsPartitionInfo;
use crate::disk::mount::guard::GenericMountGuard;
use crate::hostname::Hostname;
use crate::prelude::*;
use crate::util::Invoke;
use crate::util::serde::IoFormat;

#[derive(Clone, Copy, Debug, Deserialize, Serialize)]
#[serde(rename_all = "camelCase")]
pub enum PartitionTable {
    Mbr,
    Gpt,
}

#[derive(Clone, Debug, Deserialize, Serialize)]
#[serde(rename_all = "camelCase")]
pub struct DiskInfo {
    pub logicalname: PathBuf,
    pub partition_table: Option<PartitionTable>,
    pub vendor: Option<String>,
    pub model: Option<String>,
    pub partitions: Vec<PartitionInfo>,
    pub capacity: u64,
    pub guid: Option<InternedString>,
}

#[derive(Clone, Debug, Deserialize, Serialize)]
#[serde(rename_all = "camelCase")]
pub struct PartitionInfo {
    pub logicalname: PathBuf,
    pub label: Option<String>,
    pub capacity: u64,
    pub used: Option<u64>,
    pub start_os: BTreeMap<String, StartOsRecoveryInfo>,
    pub guid: Option<InternedString>,
}

#[derive(Clone, Debug, Default, Deserialize, Serialize)]
#[serde(rename_all = "camelCase")]
pub struct StartOsRecoveryInfo {
    pub hostname: Hostname,
    pub version: exver::Version,
    pub timestamp: DateTime<Utc>,
    pub password_hash: Option<String>,
    pub wrapped_key: Option<String>,
}

const DISK_PATH: &str = "/dev/disk/by-path";
const SYS_BLOCK_PATH: &str = "/sys/block";

lazy_static::lazy_static! {
    static ref PARTITION_REGEX: Regex = Regex::new("-part[0-9]+$").unwrap();
}

#[instrument(skip_all)]
pub async fn get_partition_table<P: AsRef<Path>>(path: P) -> Result<Option<PartitionTable>, Error> {
    Ok(String::from_utf8(
        Command::new("fdisk")
            .arg("-l")
            .arg(path.as_ref())
            .invoke(crate::ErrorKind::BlockDevice)
            .await?,
    )?
    .lines()
    .find_map(|l| l.strip_prefix("Disklabel type:"))
    .and_then(|t| match t.trim() {
        "dos" => Some(PartitionTable::Mbr),
        "gpt" => Some(PartitionTable::Gpt),
        _ => None,
    }))
}

#[instrument(skip_all)]
pub async fn get_vendor<P: AsRef<Path>>(path: P) -> Result<Option<String>, Error> {
    let vendor = tokio::fs::read_to_string(
        Path::new(SYS_BLOCK_PATH)
            .join(path.as_ref().strip_prefix("/dev").map_err(|_| {
                Error::new(
                    eyre!("{}", t!("disk.util.not-canonical-block-device")),
                    crate::ErrorKind::BlockDevice,
                )
            })?)
            .join("device")
            .join("vendor"),
    )
    .await?
    .trim()
    .to_owned();
    Ok(if vendor.is_empty() {
        None
    } else {
        Some(vendor)
    })
}

#[instrument(skip_all)]
pub async fn get_model<P: AsRef<Path>>(path: P) -> Result<Option<String>, Error> {
    let model = tokio::fs::read_to_string(
        Path::new(SYS_BLOCK_PATH)
            .join(path.as_ref().strip_prefix("/dev").map_err(|_| {
                Error::new(
                    eyre!("{}", t!("disk.util.not-canonical-block-device")),
                    crate::ErrorKind::BlockDevice,
                )
            })?)
            .join("device")
            .join("model"),
    )
    .await?
    .trim()
    .to_owned();
    Ok(if model.is_empty() { None } else { Some(model) })
}

#[instrument(skip_all)]
pub async fn get_capacity<P: AsRef<Path>>(path: P) -> Result<u64, Error> {
    Ok(String::from_utf8(
        Command::new("blockdev")
            .arg("--getsize64")
            .arg(path.as_ref())
            .invoke(crate::ErrorKind::BlockDevice)
            .await?,
    )?
    .trim()
    .parse::<u64>()?)
}

#[instrument(skip_all)]
pub async fn get_label<P: AsRef<Path>>(path: P) -> Result<Option<String>, Error> {
    let label = String::from_utf8(
        Command::new("lsblk")
            .arg("-no")
            .arg("label")
            .arg(path.as_ref())
            .invoke(crate::ErrorKind::BlockDevice)
            .await?,
    )?
    .trim()
    .to_owned();
    Ok(if label.is_empty() { None } else { Some(label) })
}

#[instrument(skip_all)]
pub async fn get_used<P: AsRef<Path>>(path: P) -> Result<u64, Error> {
    Ok(String::from_utf8(
        Command::new("df")
            .arg("--output=used")
            .arg("--block-size=1")
            .arg(path.as_ref())
            .invoke(crate::ErrorKind::Filesystem)
            .await?,
    )?
    .lines()
    .skip(1)
    .next()
    .unwrap_or_default()
    .trim()
    .parse::<u64>()?)
}

#[instrument(skip_all)]
pub async fn get_available<P: AsRef<Path>>(path: P) -> Result<u64, Error> {
    Ok(String::from_utf8(
        Command::new("df")
            .arg("--output=avail")
            .arg("--block-size=1")
            .arg(path.as_ref())
            .invoke(crate::ErrorKind::Filesystem)
            .await?,
    )?
    .lines()
    .skip(1)
    .next()
    .unwrap_or_default()
    .trim()
    .parse::<u64>()?)
}

#[instrument(skip_all)]
pub async fn get_percentage<P: AsRef<Path>>(path: P) -> Result<u64, Error> {
    Ok(String::from_utf8(
        Command::new("df")
            .arg("--output=pcent")
            .arg(path.as_ref())
            .invoke(crate::ErrorKind::Filesystem)
            .await?,
    )?
    .lines()
    .skip(1)
    .next()
    .unwrap_or_default()
    .trim()
    .strip_suffix("%")
    .unwrap()
    .parse::<u64>()?)
}

#[instrument(skip_all)]
pub async fn pvscan() -> Result<BTreeMap<PathBuf, Option<InternedString>>, Error> {
    let pvscan_out = Command::new("pvscan")
        .invoke(crate::ErrorKind::DiskManagement)
        .await?;
    let pvscan_out_str = std::str::from_utf8(&pvscan_out)?;
    Ok(parse_pvscan_output(pvscan_out_str))
}

pub async fn recovery_info(
    mountpoint: impl AsRef<Path>,
) -> Result<BTreeMap<String, StartOsRecoveryInfo>, Error> {
    let backup_root = mountpoint.as_ref().join("StartOSBackups");
    let mut res = BTreeMap::new();
    if tokio::fs::metadata(&backup_root).await.is_ok() {
        let mut dir = tokio::fs::read_dir(&backup_root).await?;
        while let Some(entry) = dir.next_entry().await? {
            let server_id = entry.file_name().to_string_lossy().into_owned();
            let backup_unencrypted_metadata_path = backup_root
                .join(&server_id)
                .join("unencrypted-metadata.json");
            if tokio::fs::metadata(&backup_unencrypted_metadata_path)
                .await
                .is_ok()
            {
                res.insert(
                    server_id,
                    IoFormat::Json.from_slice(
                        &tokio::fs::read(&backup_unencrypted_metadata_path)
                            .await
                            .with_ctx(|_| {
                                (
                                    crate::ErrorKind::Filesystem,
                                    backup_unencrypted_metadata_path.display().to_string(),
                                )
                            })?,
                    )?,
                );
            }
        }
    }

    Ok(res)
}

/// Returns the canonical path of the source device for a given mount point,
/// or None if the mount point doesn't exist or isn't mounted.
#[instrument(skip_all)]
pub async fn get_mount_source(mountpoint: impl AsRef<Path>) -> Result<Option<PathBuf>, Error> {
    let mounts_content = tokio::fs::read_to_string("/proc/mounts")
        .await
        .with_ctx(|_| (crate::ErrorKind::Filesystem, "/proc/mounts"))?;

    let mountpoint = mountpoint.as_ref();
    for line in mounts_content.lines() {
        let mut parts = line.split_whitespace();
        let source = parts.next();
        let mount = parts.next();
        if let (Some(source), Some(mount)) = (source, mount) {
            if Path::new(mount) == mountpoint {
                // Try to canonicalize the source path
                if let Ok(canonical) = tokio::fs::canonicalize(source).await {
                    return Ok(Some(canonical));
                }
            }
        }
    }
    Ok(None)
}

#[instrument(skip_all)]
pub async fn list(os: &OsPartitionInfo) -> Result<Vec<DiskInfo>, Error> {
    struct DiskIndex {
        parts: BTreeSet<PathBuf>,
        internal: bool,
    }
    let disk_guids = pvscan().await?;
    let disks = tokio_stream::wrappers::ReadDirStream::new(
        tokio::fs::read_dir(DISK_PATH)
            .await
            .with_ctx(|_| (crate::ErrorKind::Filesystem, DISK_PATH))?,
    )
    .map_err(|e| {
        Error::new(
            eyre::Error::from(e).wrap_err(DISK_PATH),
            crate::ErrorKind::Filesystem,
        )
    })
    .try_fold(
        BTreeMap::<PathBuf, DiskIndex>::new(),
        |mut disks, dir_entry| async move {
            if dir_entry.file_type().await?.is_dir() {
                return Ok(disks);
            }
            if let Some(disk_path) = dir_entry.path().file_name().and_then(|s| s.to_str()) {
                let (disk_path, part_path) = if let Some(end) = PARTITION_REGEX.find(disk_path) {
                    (
                        disk_path.strip_suffix(end.as_str()).unwrap_or_default(),
                        Some(disk_path),
                    )
                } else {
                    (disk_path, None)
                };
                let disk_path = Path::new(DISK_PATH).join(disk_path);
                let disk = tokio::fs::canonicalize(&disk_path).await.with_ctx(|_| {
                    (
                        crate::ErrorKind::Filesystem,
                        disk_path.display().to_string(),
                    )
                })?;
                let part = if let Some(part_path) = part_path {
                    let part_path = Path::new(DISK_PATH).join(part_path);
                    let part = tokio::fs::canonicalize(&part_path).await.with_ctx(|_| {
                        (
                            crate::ErrorKind::Filesystem,
                            part_path.display().to_string(),
                        )
                    })?;
                    Some(part)
                } else {
                    None
                };
                if !disks.contains_key(&disk) {
                    disks.insert(
                        disk.clone(),
                        DiskIndex {
                            parts: BTreeSet::new(),
                            internal: false,
                        },
                    );
                }
                if let Some(part) = part {
                    if os.contains(&part) {
                        disks.get_mut(&disk).unwrap().internal = true;
                    } else {
                        disks.get_mut(&disk).unwrap().parts.insert(part);
                    }
                }
            }
            Ok(disks)
        },
    )
    .await?;

    let mut res = Vec::with_capacity(disks.len());
    for (disk, index) in disks {
        if index.internal {
            for part in index.parts {
                let mut disk_info = disk_info(disk.clone()).await;
                let part_info = part_info(part).await;
                disk_info.logicalname = part_info.logicalname.clone();
                disk_info.capacity = part_info.capacity;
                if let Some(g) = disk_guids.get(&disk_info.logicalname) {
                    disk_info.guid = g.clone();
                } else {
                    disk_info.partitions = vec![part_info];
                }
                res.push(disk_info);
            }
        } else {
            let mut disk_info = disk_info(disk).await;
            disk_info.partitions = Vec::with_capacity(index.parts.len());
            if let Some(g) = disk_guids.get(&disk_info.logicalname) {
                disk_info.guid = g.clone();
            } else {
                for part in index.parts {
                    let mut part_info = part_info(part).await;
                    if let Some(g) = disk_guids.get(&part_info.logicalname) {
                        part_info.guid = g.clone();
                    }
                    disk_info.partitions.push(part_info);
                }
            }
            res.push(disk_info);
        }
    }

    Ok(res)
}

async fn disk_info(disk: PathBuf) -> DiskInfo {
    let partition_table = get_partition_table(&disk)
        .await
        .map_err(|e| {
            tracing::warn!(
                "{}",
                t!(
                    "disk.util.could-not-get-partition-table",
                    disk = disk.display(),
                    error = e.source
                )
            )
        })
        .unwrap_or_default();
    let vendor = get_vendor(&disk)
        .await
        .map_err(|e| {
            tracing::warn!(
                "{}",
                t!(
                    "disk.util.could-not-get-vendor",
                    disk = disk.display(),
                    error = e.source
                )
            )
        })
        .unwrap_or_default();
    let model = get_model(&disk)
        .await
        .map_err(|e| {
            tracing::warn!(
                "{}",
                t!(
                    "disk.util.could-not-get-model",
                    disk = disk.display(),
                    error = e.source
                )
            )
        })
        .unwrap_or_default();
    let capacity = get_capacity(&disk)
        .await
        .map_err(|e| {
            tracing::warn!(
                "{}",
                t!(
                    "disk.util.could-not-get-capacity",
                    disk = disk.display(),
                    error = e.source
                )
            )
        })
        .unwrap_or_default();
    DiskInfo {
        logicalname: disk,
        partition_table,
        vendor,
        model,
        partitions: Vec::new(),
        capacity,
        guid: None,
    }
}

async fn part_info(part: PathBuf) -> PartitionInfo {
    let mut start_os = BTreeMap::new();
    let label = get_label(&part)
        .await
        .map_err(|e| {
            tracing::warn!(
                "{}",
                t!(
                    "disk.util.could-not-get-label",
                    part = part.display(),
                    error = e.source
                )
            )
        })
        .unwrap_or_default();
    let capacity = get_capacity(&part)
        .await
        .map_err(|e| {
            tracing::warn!(
                "{}",
                t!(
                    "disk.util.could-not-get-capacity-part",
                    part = part.display(),
                    error = e.source
                )
            )
        })
        .unwrap_or_default();
    let mut used = None;

    match TmpMountGuard::mount(&BlockDev::new(&part), ReadOnly).await {
        Err(e) => tracing::warn!(
            "{}",
            t!("disk.util.could-not-collect-usage-info", error = e.source)
        ),
        Ok(mount_guard) => {
            used = get_used(mount_guard.path())
                .await
                .map_err(|e| {
                    tracing::warn!(
                        "{}",
                        t!(
                            "disk.util.could-not-get-usage",
                            part = part.display(),
                            error = e.source
                        )
                    )
                })
                .ok();
            match recovery_info(mount_guard.path()).await {
                Ok(a) => {
                    start_os = a;
                }
                Err(e) => {
                    tracing::error!(
                        "{}",
                        t!("disk.util.error-fetching-backup-metadata", error = e)
                    );
                }
            }
            if let Err(e) = mount_guard.unmount().await {
                tracing::error!(
                    "{}",
                    t!(
                        "disk.util.error-unmounting-partition",
                        part = part.display(),
                        error = e
                    )
                );
            }
        }
    }

    PartitionInfo {
        logicalname: part,
        label,
        capacity,
        used,
        start_os,
        guid: None,
    }
}

fn parse_pvscan_output(pvscan_output: &str) -> BTreeMap<PathBuf, Option<InternedString>> {
    fn parse_line(line: &str) -> IResult<&str, (&str, Option<&str>)> {
        let pv_parse = preceded(
            tag("  PV "),
            terminated(take_till1(|c: char| c.is_space()), multispace1),
        );
        let vg_parse = preceded(
            opt(tag("is in exported ")),
            preceded(
                tag("VG "),
                terminated(take_till1(|c: char| c.is_space()), multispace1),
            ),
        );
        let mut parser = terminated(pair(pv_parse, opt(vg_parse)), rest);
        parser.parse(line)
    }
    let lines = pvscan_output.lines();
    let n = lines.clone().count();
    let entries = lines.take(n.saturating_sub(1));
    let mut ret = BTreeMap::new();
    for entry in entries {
        match parse_line(entry) {
            Ok((_, (pv, vg))) => {
                ret.insert(PathBuf::from(pv), vg.map(InternedString::intern));
            }
            Err(_) => {
                tracing::warn!("{}", t!("disk.util.failed-to-parse-pvscan", line = entry));
            }
        }
    }
    ret
}

#[test]
fn test_pvscan_parser() {
    let s1 = r#"  PV /dev/mapper/cryptdata   VG data            lvm2 [1.81 TiB / 0    free]
  PV /dev/sdb                                   lvm2 [931.51 GiB]
  Total: 2 [2.72 TiB] / in use: 1 [1.81 TiB] / in no VG: 1 [931.51 GiB]
"#;
    let s2 = r#"  PV /dev/sdb   VG EMBASSY_LZHJAENWGPCJJL6C6AXOD7OOOIJG7HFBV4GYRJH6HADXUCN4BRWQ   lvm2 [931.51 GiB / 0    free]
  Total: 1 [931.51 GiB] / in use: 1 [931.51 GiB] / in no VG: 0 [0   ]
"#;
    let s3 = r#"  PV /dev/mapper/cryptdata   VG data            lvm2 [1.81 TiB / 0    free]
  Total: 1 [1.81 TiB] / in use: 1 [1.81 TiB] / in no VG: 0 [0   ]
"#;
    let s4 = r#"  PV /dev/sda    is in exported VG EMBASSY_ZFHOCTYV3ZJMJW3OTFMG55LSQZLP667EDNZKDNUJKPJX5HE6S5HQ [931.51 GiB / 0    free]
  Total: 1 [931.51 GiB] / in use: 1 [931.51 GiB] / in no VG: 0 [0   ]
"#;
    println!("{:?}", parse_pvscan_output(s1));
    println!("{:?}", parse_pvscan_output(s2));
    println!("{:?}", parse_pvscan_output(s3));
    println!("{:?}", parse_pvscan_output(s4));
}