pkh/src/logfmt.rs

//! Line classifiers rewriting raw subprocess output for the live UI
//!
//! Each classifier is a small stateful machine fed every captured line of a
//! build phase; it decides what to display (rewritten lines, warnings,
//! errors) and whether the line carries countable progress. Classifiers are
//! pure with respect to the UI: they only return [`Action`]s.

use std::sync::{LazyLock, OnceLock};

use crate::context::Stream;
use regex::Regex;

/// Maximum length of a rewritten line displayed in the rolling pane
pub(crate) const MAX_LINE_WIDTH: usize = 120;

/// What a classifier decided to do with a captured line
#[derive(Debug, Clone, PartialEq)]
pub enum Action {
    /// Drop the line (noise)
    Hidden,
    /// Display a rewritten line in the rolling pane
    Shown(String),
    /// Display a warning line (yellow)
    Warning(String),
    /// Display an error line (red, sticky)
    Error(String),
    /// Update the determinate progress bar
    Progress {
        /// Current position
        pos: u64,
        /// Total number of items (0 = unknown)
        total: u64,
    },
}

/// Stateful classifier turning raw subprocess lines into UI actions
pub trait Classifier: Send {
    /// Feed one captured line, returning the actions it produces
    fn feed(&mut self, stream: Stream, line: &str) -> Vec<Action>;
}

/// Truncate a line to [`MAX_LINE_WIDTH`], appending an ellipsis if cut
pub(crate) fn truncate(line: &str) -> String {
    if line.chars().count() <= MAX_LINE_WIDTH {
        line.to_string()
    } else {
        let cut: String = line.chars().take(MAX_LINE_WIDTH - 1).collect();
        format!("{}…", cut.trim_end())
    }
}

/// Classify apt-style severity prefixes (`E:` / `W:`)
fn apt_severity(line: &str) -> Option<Action> {
    if line.starts_with("E:") {
        Some(Action::Error(truncate(line)))
    } else if line.starts_with("W:") {
        Some(Action::Warning(truncate(line)))
    } else {
        None
    }
}

/// Classifier for `apt-get update` output
///
/// Collapses `Get:/Hit:/Ign:` lines into a running source counter and always
/// surfaces errors and warnings.
#[derive(Default)]
pub struct AptUpdateClassifier {
    sources: u64,
}

impl AptUpdateClassifier {
    /// Create a new classifier
    pub fn new() -> Self {
        Self::default()
    }
}

impl Classifier for AptUpdateClassifier {
    fn feed(&mut self, _stream: Stream, line: &str) -> Vec<Action> {
        if line.starts_with("Get:") || line.starts_with("Hit:") || line.starts_with("Ign:") {
            self.sources += 1;
            vec![Action::Shown(format!(
                "Updating package lists… ({} sources)",
                self.sources
            ))]
        } else if let Some(severity) = apt_severity(line) {
            vec![severity]
        } else {
            vec![Action::Hidden]
        }
    }
}

/// Classifier for `apt-get install` / `apt-get build-dep` output
///
/// Parses the upfront summary ("N upgraded, M newly installed, …") to derive
/// a total, then counts `Unpacking`/`Setting up` lines to drive a determinate
/// progress bar.
#[derive(Default)]
pub struct AptInstallClassifier {
    label: String,
    total: u64,
    done: u64,
}

impl AptInstallClassifier {
    /// Create a classifier for an install phase labeled `label`
    pub fn new(label: &str) -> Self {
        Self {
            label: label.to_string(),
            ..Default::default()
        }
    }

    fn progress(&self) -> Action {
        if self.total > 0 {
            Action::Progress {
                pos: self.done.min(self.total),
                total: self.total,
            }
        } else {
            Action::Hidden
        }
    }
}

impl Classifier for AptInstallClassifier {
    fn feed(&mut self, _stream: Stream, line: &str) -> Vec<Action> {
        static SUMMARY_RE: OnceLock<Regex> = OnceLock::new();
        let summary_re = SUMMARY_RE.get_or_init(|| {
            Regex::new(r"(\d+) (?:upgraded|newly installed|re-installed)").unwrap()
        });
        static UNPACK_RE: OnceLock<Regex> = OnceLock::new();
        let unpack_re =
            UNPACK_RE.get_or_init(|| Regex::new(r"^Unpacking ([^ ]+) \(([^)]+)\)").unwrap());
        static SETUP_RE: OnceLock<Regex> = OnceLock::new();
        let setup_re =
            SETUP_RE.get_or_init(|| Regex::new(r"^Setting up ([^ ]+) \(([^)]+)\)").unwrap());

        if summary_re.is_match(line) && !self.label.is_empty() {
            // Only accept the summary once: later lines may repeat counts
            if self.total == 0 {
                let total: u64 = summary_re
                    .captures_iter(line)
                    .filter_map(|c| c[1].parse::<u64>().ok())
                    .sum();
                self.total = total;
                vec![Action::Shown(format!("{}: {} packages", self.label, total))]
            } else {
                vec![Action::Hidden]
            }
        } else if let Some(caps) = unpack_re.captures(line) {
            self.done += 1;
            vec![
                Action::Shown(format!(
                    "{}: unpacking {} ({})",
                    self.label, &caps[1], &caps[2]
                )),
                self.progress(),
            ]
        } else if let Some(caps) = setup_re.captures(line) {
            self.done += 1;
            vec![
                Action::Shown(format!(
                    "{}: setting up {} ({})",
                    self.label, &caps[1], &caps[2]
                )),
                self.progress(),
            ]
        } else if let Some(severity) = apt_severity(line) {
            vec![severity]
        } else {
            vec![Action::Hidden]
        }
    }
}

/// Classifier for `quilt push -a` output
///
/// Driven by the number of patches listed in `debian/patches/series`, known
/// before the command runs.
pub struct QuiltClassifier {
    total: u64,
    applied: u64,
}

impl QuiltClassifier {
    /// Create a classifier expecting `total` patches
    pub fn new(total: usize) -> Self {
        Self {
            total: total as u64,
            applied: 0,
        }
    }
}

impl Classifier for QuiltClassifier {
    fn feed(&mut self, _stream: Stream, line: &str) -> Vec<Action> {
        static APPLYING_RE: OnceLock<Regex> = OnceLock::new();
        let applying_re =
            APPLYING_RE.get_or_init(|| Regex::new(r"^Applying patch ([^ ]+)").unwrap());

        if line.contains("failed") || line.contains("Failed") {
            // Check failures first: "Applying patch x failed" must not be
            // counted as a successful application
            vec![Action::Error(truncate(line))]
        } else if let Some(caps) = applying_re.captures(line) {
            self.applied += 1;
            let mut actions = vec![Action::Shown(format!("Applying patch {}", &caps[1]))];
            if self.total > 0 {
                actions.push(Action::Progress {
                    pos: self.applied.min(self.total),
                    total: self.total,
                });
            }
            actions
        } else if line.starts_with("Now at patch") {
            vec![Action::Shown(truncate(line))]
        } else {
            vec![Action::Hidden]
        }
    }
}

/// Classifier for make/cmake-based builds (`debian/rules build`, dh helpers)
///
/// Detects `[ 42%]`-style progress markers, hides directory enter/leave
/// noise, and shows compile/link/dh lines as well as kbuild status-tag
/// lines and warnings.
#[derive(Default)]
pub struct MakeClassifier {}

impl MakeClassifier {
    /// Create a new classifier
    pub fn new() -> Self {
        Self::default()
    }
}

impl Classifier for MakeClassifier {
    fn feed(&mut self, _stream: Stream, line: &str) -> Vec<Action> {
        static PERCENT_RE: OnceLock<Regex> = OnceLock::new();
        let percent_re = PERCENT_RE.get_or_init(|| Regex::new(r"\[\s*(\d+)%\]").unwrap());

        // kbuild abbreviates each invocation to an uppercase status tag
        // (`  CC [M] drivers/x.o`, `  HOSTCC scripts/foo`) instead of
        // printing the full command line, so the classic-make substring
        // rules below never fire for kernel builds; this tag list is the
        // kbuild counterpart of those substrings.
        static KBUILD_TAG_RE: LazyLock<Regex> = LazyLock::new(|| {
            Regex::new(
                r"^\s*(CC|HOSTCC|HOSTLD|HOSTAR|LD|AR|AS|OBJCOPY|STRIP|GEN|UPD|DESCEND|MODPOST|MODORDER|BTF|SHIPPED|INSTALL|INSTALLMOD|DEPMOD|SIGN|WRAP|SYSHDR|SYSTBL|KSYMS)\b",
            )
            .unwrap()
        });

        if let Some(caps) = percent_re.captures(line) {
            let pct: u64 = caps[1].parse().unwrap_or(0);
            return vec![
                Action::Shown(truncate(line)),
                Action::Progress {
                    pos: pct,
                    total: 100,
                },
            ];
        }

        if line.contains("error:")
            || line.contains("Error ")
            || line.contains("*** [")
            || line.contains("failed")
        {
            return vec![Action::Error(truncate(line))];
        }

        // gcc-style diagnostics (`file.c:12:3: warning: ...`) and kbuild's
        // uppercase banners (`WARNING: modpost: ...`) must not be drowned
        // out by the activity checks below.
        if line.contains("warning:") || line.trim_start().starts_with("WARNING:") {
            return vec![Action::Warning(truncate(line))];
        }

        if line.contains("make[")
            && (line.contains("Entering directory") || line.contains("Leaving directory"))
        {
            return vec![Action::Hidden];
        }

        if line.starts_with("dh_")
            || line.contains("gcc ")
            || line.contains("g++ ")
            || line.contains("cc ")
            || line.contains("clang")
            || line.contains("ld ")
            || line.contains("ar ")
            || KBUILD_TAG_RE.is_match(line)
        {
            return vec![Action::Shown(truncate(line))];
        }

        vec![Action::Hidden]
    }
}

/// Classifier for `dpkg-source` output (source-build phases)
///
/// The build pipeline pins `LC_ALL=C`, so dpkg-source emits stable English
/// messages prefixed with `info:` / `warning:` / `error:`; the prefix is
/// stripped and the severity drives the pane color. Raw `tar:` diagnostics
/// emitted while repacking tarballs are surfaced too.
#[derive(Default)]
pub struct DpkgSourceClassifier {}

impl DpkgSourceClassifier {
    /// Create a new classifier
    pub fn new() -> Self {
        Self::default()
    }
}

impl Classifier for DpkgSourceClassifier {
    fn feed(&mut self, _stream: Stream, line: &str) -> Vec<Action> {
        const PREFIX: &str = "dpkg-source: ";
        let rest = line.strip_prefix(PREFIX).unwrap_or(line);

        if let Some(rest) = rest.strip_prefix("info: ") {
            vec![Action::Shown(truncate(rest))]
        } else if let Some(rest) = rest.strip_prefix("warning: ") {
            vec![Action::Warning(truncate(rest))]
        } else if let Some(rest) = rest.strip_prefix("error: ") {
            vec![Action::Error(truncate(rest))]
        } else if let Some(tar) = rest.strip_prefix("tar: ") {
            // Diagnostics from the tarball repacking subprocess; warnings
            // about unknown header keywords are benign, real failures are not.
            let lower = tar.to_lowercase();
            if ["error", "cannot", "failed", "exited"]
                .iter()
                .any(|m| lower.contains(m))
            {
                vec![Action::Error(truncate(tar))]
            } else {
                vec![Action::Warning(truncate(tar))]
            }
        } else if line == PREFIX.trim_end() || rest.is_empty() {
            vec![Action::Hidden]
        } else {
            // Unprefixed output from a foreign subprocess: keep it visible
            vec![Action::Shown(truncate(line))]
        }
    }
}

/// Classifier for `mmdebstrap` output (chroot tarball creation)
///
/// mmdebstrap prefixes its own messages with `I:` / `W:` / `E:`; everything
/// else is chroot-internal apt/dpkg noise.
#[derive(Default)]
pub struct MmdebstrapClassifier {}

impl MmdebstrapClassifier {
    /// Create a new classifier
    pub fn new() -> Self {
        Self::default()
    }
}

impl Classifier for MmdebstrapClassifier {
    fn feed(&mut self, _stream: Stream, line: &str) -> Vec<Action> {
        if let Some(rest) = line.strip_prefix("I: ") {
            vec![Action::Shown(truncate(rest))]
        } else if let Some(rest) = line.strip_prefix("W: ") {
            vec![Action::Warning(truncate(rest))]
        } else if let Some(rest) = line.strip_prefix("E: ") {
            vec![Action::Error(truncate(rest))]
        } else if line.starts_with("Setting up ") {
            vec![Action::Shown(truncate(line))]
        } else {
            vec![Action::Hidden]
        }
    }
}

/// Generic fallback classifier: shows the last meaningful line and surfaces
/// obvious error/warning patterns.
#[derive(Default)]
pub struct GenericClassifier {}

impl GenericClassifier {
    /// Create a new classifier
    pub fn new() -> Self {
        Self::default()
    }
}

impl Classifier for GenericClassifier {
    fn feed(&mut self, _stream: Stream, line: &str) -> Vec<Action> {
        if line.starts_with("E:")
            || line.contains("error:")
            || line.contains("Error ")
            || line.contains("failed")
        {
            vec![Action::Error(truncate(line))]
        } else if line.starts_with("W:") {
            vec![Action::Warning(truncate(line))]
        } else {
            vec![Action::Shown(truncate(line))]
        }
    }
}

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

    fn feed_one(c: &mut dyn Classifier, line: &str) -> Vec<Action> {
        c.feed(Stream::Stdout, line)
    }

    #[test]
    fn test_apt_update_collapses_sources_and_surfaces_errors() {
        let mut c = AptUpdateClassifier::new();
        assert_eq!(
            feed_one(&mut c, "Hit:1 http://archive.ubuntu.com noble InRelease"),
            vec![Action::Shown(
                "Updating package lists… (1 sources)".to_string()
            )]
        );
        assert_eq!(
            feed_one(
                &mut c,
                "Get:2 http://security.ubuntu.com noble-security InRelease"
            ),
            vec![Action::Shown(
                "Updating package lists… (2 sources)".to_string()
            )]
        );
        assert_eq!(
            feed_one(&mut c, "E: Repository 'x' changed its 'suite' value"),
            vec![Action::Error(
                "E: Repository 'x' changed its 'suite' value".to_string()
            )]
        );
        assert_eq!(
            feed_one(&mut c, "Reading package lists..."),
            vec![Action::Hidden]
        );
    }

    #[test]
    fn test_apt_install_counts_packages() {
        let mut c = AptInstallClassifier::new("Installing build dependencies");
        // Summary line sets the total
        assert_eq!(
            feed_one(
                &mut c,
                "2 upgraded, 3 newly installed, 0 to remove and 0 not upgraded."
            ),
            vec![Action::Shown(
                "Installing build dependencies: 5 packages".to_string()
            )]
        );
        // Unpacking and setting up drive progress
        assert_eq!(
            feed_one(&mut c, "Unpacking libfoo (1.2-3)"),
            vec![
                Action::Shown(
                    "Installing build dependencies: unpacking libfoo (1.2-3)".to_string()
                ),
                Action::Progress { pos: 1, total: 5 }
            ]
        );
        assert_eq!(
            feed_one(&mut c, "Setting up libfoo (1.2-3)"),
            vec![
                Action::Shown(
                    "Installing build dependencies: setting up libfoo (1.2-3)".to_string()
                ),
                Action::Progress { pos: 2, total: 5 }
            ]
        );
    }

    #[test]
    fn test_quilt_counts_patches() {
        let mut c = QuiltClassifier::new(2);
        assert_eq!(
            feed_one(&mut c, "Applying patch debian/patches/foo.patch"),
            vec![
                Action::Shown("Applying patch debian/patches/foo.patch".to_string()),
                Action::Progress { pos: 1, total: 2 }
            ]
        );
        assert_eq!(
            feed_one(&mut c, "Applying patch debian/patches/bar.patch"),
            vec![
                Action::Shown("Applying patch debian/patches/bar.patch".to_string()),
                Action::Progress { pos: 2, total: 2 }
            ]
        );
        assert_eq!(
            feed_one(&mut c, "Applying patch x failed"),
            vec![Action::Error("Applying patch x failed".to_string())]
        );
    }

    #[test]
    fn test_make_detects_percent_and_hides_noise() {
        let mut c = MakeClassifier::new();
        assert_eq!(
            feed_one(
                &mut c,
                "[ 42%] Building CXX object CMakeFiles/hello.dir/hello.o"
            ),
            vec![
                Action::Shown(
                    "[ 42%] Building CXX object CMakeFiles/hello.dir/hello.o".to_string()
                ),
                Action::Progress {
                    pos: 42,
                    total: 100
                }
            ]
        );
        assert_eq!(
            feed_one(&mut c, "make[2]: Entering directory '/tmp/build'"),
            vec![Action::Hidden]
        );
        assert_eq!(
            feed_one(&mut c, "make[1]: *** [Makefile:531: hello.o] Error 1"),
            vec![Action::Error(
                "make[1]: *** [Makefile:531: hello.o] Error 1".to_string()
            )]
        );
        assert_eq!(
            feed_one(&mut c, "dh_auto_build"),
            vec![Action::Shown("dh_auto_build".to_string())]
        );
    }

    #[test]
    fn test_make_shows_kbuild_status_tags() {
        let mut c = MakeClassifier::new();
        assert_eq!(
            feed_one(&mut c, "  CC [M] drivers/net/foo.o"),
            vec![Action::Shown("  CC [M] drivers/net/foo.o".to_string())]
        );
        assert_eq!(
            feed_one(&mut c, "  LD [M] drivers/bar.ko"),
            vec![Action::Shown("  LD [M] drivers/bar.ko".to_string())]
        );
        assert_eq!(
            feed_one(&mut c, "  HOSTCC scripts/tool"),
            vec![Action::Shown("  HOSTCC scripts/tool".to_string())]
        );
        assert_eq!(
            feed_one(&mut c, "  MODPOST 1 modules"),
            vec![Action::Shown("  MODPOST 1 modules".to_string())]
        );
        // kbuild tag lines still roll the pane while make progress markers
        // update the progress bar
        assert_eq!(
            feed_one(&mut c, "[ 45%] Building C object x.o"),
            vec![
                Action::Shown("[ 45%] Building C object x.o".to_string()),
                Action::Progress {
                    pos: 45,
                    total: 100
                }
            ]
        );
        // plain noise stays hidden
        assert_eq!(
            feed_one(&mut c, "test -x debian/scripts/foo"),
            vec![Action::Hidden]
        );
    }

    #[test]
    fn test_make_shows_warnings_and_keeps_regressions() {
        let mut c = MakeClassifier::new();
        assert_eq!(
            feed_one(&mut c, "drivers/x.c:12:3: warning: unused variable 'y'"),
            vec![Action::Warning(
                "drivers/x.c:12:3: warning: unused variable 'y'".to_string()
            )]
        );
        assert_eq!(
            feed_one(&mut c, "WARNING: modpost: missing MODULE_LICENSE"),
            vec![Action::Warning(
                "WARNING: modpost: missing MODULE_LICENSE".to_string()
            )]
        );
        // configure-stage probe lines embed the toolchain name and must
        // stay visible
        assert_eq!(
            feed_one(
                &mut c,
                "checking whether riscv64-linux-gnu-gcc understands -c and -o together... (cached) yes"
            ),
            vec![Action::Shown(
                "checking whether riscv64-linux-gnu-gcc understands -c and -o together... (cached) yes"
                    .to_string()
            )]
        );
    }

    #[test]
    fn test_mmdebstrap_prefixes() {
        let mut c = MmdebstrapClassifier::new();
        assert_eq!(
            feed_one(&mut c, "I: chroot architecture is amd64"),
            vec![Action::Shown("chroot architecture is amd64".to_string())]
        );
        assert_eq!(
            feed_one(&mut c, "W: some warning"),
            vec![Action::Warning("some warning".to_string())]
        );
        assert_eq!(
            feed_one(&mut c, "Get:1 http://x InRelease"),
            vec![Action::Hidden]
        );
    }

    #[test]
    fn test_generic_shows_lines_and_errors() {
        let mut c = GenericClassifier::new();
        assert_eq!(
            feed_one(&mut c, "some random output"),
            vec![Action::Shown("some random output".to_string())]
        );
        assert_eq!(
            feed_one(&mut c, "something failed badly"),
            vec![Action::Error("something failed badly".to_string())]
        );
    }

    #[test]
    fn test_dpkg_source_severity_prefixes() {
        let mut c = DpkgSourceClassifier::new();
        assert_eq!(
            feed_one(
                &mut c,
                "dpkg-source: info: using patch list from debian/patches/series"
            ),
            vec![Action::Shown(
                "using patch list from debian/patches/series".to_string()
            )]
        );
        assert_eq!(
            feed_one(
                &mut c,
                "dpkg-source: info: applying patch debian/patches/reproducible.patch"
            ),
            vec![Action::Shown(
                "applying patch debian/patches/reproducible.patch".to_string()
            )]
        );
        assert_eq!(
            feed_one(
                &mut c,
                "dpkg-source: info: building hello in ../hello_2.10-5.dsc"
            ),
            vec![Action::Shown(
                "building hello in ../hello_2.10-5.dsc".to_string()
            )]
        );
        assert_eq!(
            feed_one(
                &mut c,
                "dpkg-source: warning: upstream signing key but no upstream signature"
            ),
            vec![Action::Warning(
                "upstream signing key but no upstream signature".to_string()
            )]
        );
        assert_eq!(
            feed_one(
                &mut c,
                "dpkg-source: error: unrepresentable changes to source"
            ),
            vec![Action::Error(
                "unrepresentable changes to source".to_string()
            )]
        );
    }

    #[test]
    fn test_dpkg_source_tar_and_unknown_lines() {
        let mut c = DpkgSourceClassifier::new();
        // Benign tar header-keyword warnings stay yellow
        assert_eq!(
            feed_one(
                &mut c,
                "tar: Ignoring unknown extended header keyword 'SCHILY.xattr.user.foo'"
            ),
            vec![Action::Warning(
                "Ignoring unknown extended header keyword 'SCHILY.xattr.user.foo'".to_string()
            )]
        );
        // Real tar failures are errors
        assert_eq!(
            feed_one(
                &mut c,
                "tar: ../hello_2.10.orig.tar.xz: Cannot open: No such file or directory"
            ),
            vec![Action::Error(
                "../hello_2.10.orig.tar.xz: Cannot open: No such file or directory".to_string()
            )]
        );
        // Unprefixed foreign output stays visible
        assert_eq!(
            feed_one(&mut c, "gpgv: Signature made Tue 01 Jan 2026"),
            vec![Action::Shown(
                "gpgv: Signature made Tue 01 Jan 2026".to_string()
            )]
        );
    }

    #[test]
    fn test_truncate_long_lines() {
        let long = "x".repeat(300);
        let truncated = truncate(&long);
        assert_eq!(truncated.chars().count(), MAX_LINE_WIDTH);
        assert!(truncated.ends_with('…'));
        assert_eq!(truncate("short"), "short");
    }
}