build: add differential tests against real dpkg-buildpackage

Add an automated differential test harness in build/mod.rs that builds
the same source tree twice - once with real 'dpkg-buildpackage -S
-I -i -nc -d --no-sign', once with the native pipeline - and compares
all produced artifacts:

- .dsc payload byte-for-byte,
- .changes field-by-field (checksum lines of the .buildinfo itself
  excluded, as its content legitimately differs on machine-dependent
  fields),
- .buildinfo structure (Installed-Build-Depends, Environment,
  Build-Date and Build-Tainted-By excluded).

src/build/changes.rs

@@ -34,6 +34,8 @@ pub struct ChangesInput {
     pub changed_by: Option<String>,
     /// Formatted per-package description lines (empty for source-only).
     pub descriptions: Vec<String>,
+    /// Bug numbers collected from the changelog (`Closes` field), if any.
+    pub closes: Option<String>,
     /// Rendered `Changes` field value from the changelog entry.
     pub changes_field: String,
     /// Computed artifact checksums (dsc, tarballs, debs, buildinfo).
@@ -111,6 +113,9 @@ pub fn render_changes(input: &ChangesInput) -> Paragraph {
         // Leading `\n`: pre-wrapped multiline field, dpkg-style.
         p.set("Description", &format!("\n{}", sorted.join("\n")));
     }
+    if let Some(closes) = &input.closes {
+        p.set("Closes", closes);
+    }
     p.set("Changes", &input.changes_field);
 
     if !input.checksums.is_empty() {
@@ -199,6 +204,7 @@ mod tests {
             maintainer: Some("A B <a@b.c>".to_string()),
             changed_by: Some("A B <a@b.c>".to_string()),
             descriptions: vec![],
+            closes: None,
             changes_field: "pkg (1.0) unstable; urgency=medium\n.\n * Something.".to_string(),
             checksums,
             files_list,

src/build/mod.rs

@@ -115,6 +115,28 @@ pub fn run_source_build(
     log::info!("source version {}", entry.version.full());
     log::info!("source distribution {}", entry.distribution);
 
+    // binNMU builds reference the *previous* (source) version in their
+    // artifact metadata, like dpkg-genchanges/genbuildinfo do.
+    let previous_version = if entry.binary_only {
+        crate::debian::changelog::parse_previous_version(&changelog_path)?
+    } else {
+        None
+    };
+    let source_display = if entry.binary_only {
+        match previous_version.as_deref() {
+            Some(prev) => format!("{} ({})", entry.source, prev),
+            None => entry.source.clone(),
+        }
+    } else {
+        entry.source.clone()
+    };
+    let binary_only_changes = entry.binary_only.then(|| {
+        format!(
+            "{}\n\n -- {} <{}>  {}",
+            entry.changes_field, entry.maintainer_name, entry.maintainer_email, entry.date_raw
+        )
+    });
+
     let sversion = entry.version.no_epoch();
     let dsc_name = format!("{}_{}.dsc", entry.source, sversion);
     let dsc_path = parent.join(&dsc_name);
@@ -186,11 +208,32 @@ pub fn run_source_build(
         .into());
     }
 
+    // Binary-only uploads redistribute the *previous* source: metadata
+    // references the previous version's .dsc (which must already exist in
+    // the output directory), exactly like dpkg-genchanges/genbuildinfo.
+    let ref_dsc_name = match previous_version.as_deref().filter(|_| entry.binary_only) {
+        Some(prev) => {
+            let prev_version = crate::debian::DebianVersion::parse(prev)?;
+            let name = format!("{}_{}.dsc", entry.source, prev_version.no_epoch());
+            if !parent.join(&name).exists() {
+                return Err(format!(
+                    "binary-only build requires the previous source '{} \
+                     {}' to exist next to the package",
+                    entry.source, prev
+                )
+                .into());
+            }
+            name
+        }
+        None => dsc_name.clone(),
+    };
+    let ref_dsc_path = parent.join(&ref_dsc_name);
+
     // ------------------------------------------------------------------
     // 6. .buildinfo generation (native dpkg-genbuildinfo equivalent)
     // ------------------------------------------------------------------
     let mut checksums = FileChecksums::new();
-    checksums.add_file_as(&dsc_path, &dsc_name)?;
+    checksums.add_file_as(&ref_dsc_path, &ref_dsc_name)?;
 
     let status_path = PathBuf::from("/var/lib/dpkg/status");
     let bd_fields = [ctrl.source.get("Build-Depends").unwrap_or("")];
@@ -199,11 +242,11 @@ pub fn run_source_build(
 
     let render_buildinfo_doc = |checksums: &FileChecksums| {
         buildinfo::render_buildinfo(&buildinfo::BuildInfoInput {
-            source: entry.source.clone(),
+            source: source_display.clone(),
             binaries: Vec::new(), // source-only build
             architecture: "source".to_string(),
             version: entry.version.full(),
-            binary_only_changes: None,
+            binary_only_changes: binary_only_changes.clone(),
             build_origin: vendor.clone(),
             build_architecture: arch_vars
                 .get("DEB_BUILD_ARCH")
@@ -234,15 +277,15 @@ pub fn run_source_build(
     // ------------------------------------------------------------------
     // 7. .changes generation (native dpkg-genchanges equivalent)
     // ------------------------------------------------------------------
-    // Pull the tarball checksums out of the generated .dsc so they are
+    // Pull the tarball checksums out of the referenced .dsc so they are
     // distributed through the .changes like dpkg-genchanges does, in the
     // order the .dsc itself lists them.
-    let dsc_content = std::fs::read_to_string(&dsc_path)
+    let dsc_content = std::fs::read_to_string(&ref_dsc_path)
-        .map_err(|e| format!("cannot read '{}': {}", dsc_path.display(), e))?;
+        .map_err(|e| format!("cannot read '{}': {}", ref_dsc_path.display(), e))?;
     let dsc_para = parse_paragraphs(&dsc_content)
         .into_iter()
         .next()
-        .ok_or_else(|| format!("'{}' is empty", dsc_path.display()))?;
+        .ok_or_else(|| format!("'{}' is empty", ref_dsc_path.display()))?;
 
     let mut tarball_paths = Vec::new();
     let mut dsc_file_names: Vec<String> = Vec::new();
@@ -277,7 +320,7 @@ pub fn run_source_build(
         }
     }
     for name in &dsc_file_names {
-        if name == &dsc_name {
+        if name == &ref_dsc_name {
             continue; // already computed directly above
         }
         let path = parent.join(name);
@@ -310,9 +353,13 @@ pub fn run_source_build(
     // dsc and tarballs use the source stanza defaults (not persisted into
     // debian/files, matching dpkg).
     let mut changes_files = files_list.clone();
-    changes_files.add(FilesEntry::new(&dsc_name, ctrl.section(), ctrl.priority()));
+    changes_files.add(FilesEntry::new(
+        &ref_dsc_name,
+        ctrl.section(),
+        ctrl.priority(),
+    ));
     for name in &dsc_file_names {
-        if name != &dsc_name {
+        if name != &ref_dsc_name {
             changes_files.add(FilesEntry::new(name, ctrl.section(), ctrl.priority()));
         }
     }
@@ -321,7 +368,7 @@ pub fn run_source_build(
     let render_changes_doc = |checksums: &FileChecksums| {
         changes::render_changes(&changes::ChangesInput {
             date: entry.date_raw.clone(),
-            source: entry.source.clone(),
+            source: source_display.clone(),
             binaries: Vec::new(), // source-only upload
             built_for_profiles: profiles.clone(),
             architecture: "source".to_string(),
@@ -331,6 +378,7 @@ pub fn run_source_build(
             maintainer: ctrl.source.get("Maintainer").map(str::to_string),
             changed_by: Some(changed_by.clone()),
             descriptions: Vec::new(),
+            closes: entry.closes.clone(),
             changes_field: entry.changes_field.clone(),
             checksums: checksums.clone(),
             files_list: changes_files.clone(),
@@ -357,8 +405,13 @@ pub fn run_source_build(
 
         println!("signfile {}", dsc_name);
         crate::utils::gpg::clearsign_file(&dsc_path, &keyid)?;
-        // The .dsc changed: refresh its checksums inside the .buildinfo.
+        // The freshly built .dsc changed: refresh its checksums inside the
+        // .buildinfo. For binary-only builds the metadata references the
+        // *previous* .dsc (untouched by this build), so there is nothing to
+        // refresh.
+        if !entry.binary_only {
             checksums.add_file_as(&dsc_path, &dsc_name)?;
+        }
         buildinfo::save_buildinfo(&buildinfo_path, &render_buildinfo_doc(&checksums))?;
 
         println!("signfile {}", buildinfo_name);
@@ -443,3 +496,602 @@ mod tests {
         assert_eq!(v.no_epoch(), "1.0-2");
     }
 }
+
+/// Differential tests: build synthetic (or real archive) source packages
+/// with both real `dpkg-buildpackage` and the native pipeline, then compare
+/// the produced `.dsc` / `.changes` / `.buildinfo`.
+///
+/// Fields that legitimately depend on machine state (`Installed-Build-Depends`,
+/// `Environment`, `Build-Date`, `Build-Tainted-By`) and checksum lines of the
+/// `.buildinfo` itself are excluded from the comparison; everything else must
+/// match byte-for-byte.
+#[cfg(test)]
+mod differential_tests {
+    use super::*;
+    use crate::debian::control::{Paragraph, parse_paragraphs};
+    use std::fs;
+
+    const MAINTAINER: &str = "Pkh Diff <pkh-diff@example.invalid>";
+    const DATE: &str = "Thu, 01 Jan 2026 00:00:00 +0000";
+
+    /// Specification of a synthetic source package.
+    struct FixtureSpec {
+        name: &'static str,
+        version: &'static str,
+        distribution: &'static str,
+        urgency: &'static str,
+        /// `debian/source/format` content ("3.0 (native)", "3.0 (quilt)", "1.0").
+        format: &'static str,
+        binaries: &'static [(&'static str, &'static str)],
+        section: &'static str,
+        priority: &'static str,
+        body: &'static [&'static str],
+        patches: &'static [&'static str],
+        extra_source_fields: &'static [(&'static str, &'static str)],
+        /// Two-entry changelog (previous entry) for binNMU cases.
+        with_previous_entry: bool,
+    }
+
+    impl FixtureSpec {
+        fn new(name: &'static str, version: &'static str, distribution: &'static str) -> Self {
+            FixtureSpec {
+                name,
+                version,
+                distribution,
+                urgency: "medium",
+                format: "3.0 (native)",
+                binaries: &[],
+                section: "utils",
+                priority: "optional",
+                body: &["* Something changed."],
+                patches: &[],
+                extra_source_fields: &[],
+                with_previous_entry: false,
+            }
+        }
+
+        fn sversion(&self) -> String {
+            // Version without epoch, as used in artifact file names.
+            self.version
+                .split_once(':')
+                .map(|(_, rest)| rest.to_string())
+                .unwrap_or_else(|| self.version.to_string())
+        }
+
+        fn changelog(&self) -> String {
+            let mut out = format!(
+                "{} ({}) {}; urgency={}\n\n",
+                self.name, self.version, self.distribution, self.urgency
+            );
+            for line in self.body {
+                out.push_str("  * ");
+                out.push_str(line);
+                out.push('\n');
+            }
+            out.push_str(&format!("\n -- {MAINTAINER}  {DATE}\n"));
+            if self.with_previous_entry {
+                let prev = match self.version.split_once(':') {
+                    Some((_, r)) => r.to_string(),
+                    None => self.version.to_string(),
+                };
+                // Turn "1.0-1+b1" into "1.0-1" for the previous entry.
+                let prev = prev.rsplit_once('+').map(|(p, _)| p).unwrap_or(&prev);
+                out.push_str(&format!(
+                    "\n{p} ({pv}) {d}; urgency={u}\n\n  * Initial release.\n\n -- {MAINTAINER}  {DATE}\n",
+                    p = self.name,
+                    pv = prev,
+                    d = self.distribution,
+                    u = self.urgency
+                ));
+            }
+            out
+        }
+
+        fn control(&self) -> String {
+            let mut out = format!(
+                "Source: {}\nSection: {}\nPriority: {}\nMaintainer: {MAINTAINER}\nBuild-Depends: build-essential\n",
+                self.name, self.section, self.priority
+            );
+            for (k, v) in self.extra_source_fields {
+                out.push_str(k);
+                out.push_str(": ");
+                out.push_str(v);
+                out.push('\n');
+            }
+            // dpkg-source refuses trees without any binary stanza.
+            let binaries: &[(&str, &str)] = if self.binaries.is_empty() {
+                &[(self.name, "all")]
+            } else {
+                self.binaries
+            };
+            for (bname, barch) in binaries {
+                out.push_str(&format!(
+                    "\nPackage: {bname}\nArchitecture: {barch}\nDescription: {bname} component\n A component of {}.\n",
+                    self.name
+                ));
+            }
+            out
+        }
+    }
+
+    /// Materialize a fixture tree under `base`, along with an orig tarball
+    /// when the source format requires one. Returns the tree path.
+    fn write_fixture(base: &Path, spec: &FixtureSpec) -> PathBuf {
+        let sversion = spec.sversion();
+        let dir_name = format!("{}-{}", spec.name, sversion);
+        let dir = base.join(&dir_name);
+        fs::create_dir_all(dir.join("debian/source")).expect("create debian/source");
+        if !spec.patches.is_empty() {
+            fs::create_dir_all(dir.join("debian/patches")).expect("create debian/patches");
+        }
+
+        fs::write(dir.join("hello.txt"), "upstream content v1\n").expect("write upstream file");
+        fs::write(dir.join("debian/changelog"), spec.changelog()).expect("write changelog");
+        fs::write(dir.join("debian/control"), spec.control()).expect("write control");
+        fs::write(
+            dir.join("debian/source/format"),
+            format!("{}\n", spec.format),
+        )
+        .expect("write format");
+
+        let rules = dir.join("debian/rules");
+        fs::write(&rules, "#!/usr/bin/make -f\n%:\n\tdh $@\n").expect("write rules");
+        #[cfg(unix)]
+        {
+            use std::os::unix::fs::PermissionsExt;
+            fs::set_permissions(&rules, fs::Permissions::from_mode(0o755)).expect("chmod rules");
+        }
+
+        for (i, patch) in spec.patches.iter().enumerate() {
+            let name = format!("{:02}-fix.patch", i + 1);
+            fs::write(dir.join("debian/patches").join(&name), format!("{patch}\n"))
+                .expect("write patch");
+        }
+        if !spec.patches.is_empty() {
+            let series = (1..=spec.patches.len())
+                .map(|i| format!("{:02}-fix.patch", i))
+                .collect::<Vec<_>>()
+                .join("\n");
+            fs::write(dir.join("debian/patches/series"), series + "\n").expect("write series");
+        }
+
+        // quilt and 1.0 formats need a pristine orig tarball next to the tree,
+        // named after the *upstream* version.
+        if spec.format != "3.0 (native)" {
+            let uversion = crate::debian::DebianVersion::parse(spec.version)
+                .expect("valid fixture version")
+                .upstream;
+            let upstream = base.join(".upstream");
+            fs::create_dir_all(upstream.join(&dir_name)).expect("create upstream dir");
+            fs::write(
+                upstream.join(&dir_name).join("hello.txt"),
+                "upstream content v1\n",
+            )
+            .expect("write pristine upstream file");
+
+            let ext = if spec.format == "1.0" { "gz" } else { "xz" };
+            let tarball = base.join(format!("{}_{}.orig.tar.{}", spec.name, uversion, ext));
+            let mut cmd = Command::new("tar");
+            cmd.current_dir(&upstream);
+            match ext {
+                "gz" => {
+                    cmd.arg("-cz");
+                }
+                _ => {
+                    cmd.arg("-cJ");
+                }
+            }
+            cmd.arg("-f").arg(&tarball).arg(&dir_name);
+            let status = cmd.status().expect("run tar");
+            assert!(status.success(), "tar failed for {}", tarball.display());
+        }
+
+        dir
+    }
+
+    fn copy_path(src: &Path, dst_root: &Path) {
+        let status = Command::new("cp")
+            .arg("-a")
+            .arg(src)
+            .arg(dst_root)
+            .status()
+            .expect("run cp -a");
+        assert!(
+            status.success(),
+            "cp -a {} {} failed",
+            src.display(),
+            dst_root.display()
+        );
+    }
+
+    fn run_dpkg(tree: &Path) {
+        let status = Command::new("dpkg-buildpackage")
+            .current_dir(tree)
+            .args(["-S", "-I", "-i", "-nc", "-d", "--no-sign"])
+            .status()
+            .expect("failed to run dpkg-buildpackage (is dpkg-dev installed?)");
+        assert!(status.success(), "dpkg-buildpackage failed");
+    }
+
+    fn strip_signature(text: &str) -> &str {
+        match text.find("-----BEGIN PGP SIGNATURE-----") {
+            Some(i) => &text[..i],
+            None => text,
+        }
+    }
+
+    fn first_paragraph(text: &str) -> Paragraph {
+        parse_paragraphs(text)
+            .into_iter()
+            .next()
+            .expect("document has a paragraph")
+    }
+
+    /// Compare a `.changes`: every field must be identical, except checksum
+    /// lines referring to the `.buildinfo` itself (whose content legitimately
+    /// differs on machine-dependent fields).
+    fn assert_changes_equivalent(golden: &Path, ours: &Path) {
+        let g = first_paragraph(strip_signature(
+            &fs::read_to_string(golden).expect("read golden changes"),
+        ));
+        let o = first_paragraph(strip_signature(
+            &fs::read_to_string(ours).expect("read our changes"),
+        ));
+
+        let gkeys: std::collections::BTreeSet<&str> = g.iter().map(|(k, _)| k).collect();
+        let okeys: std::collections::BTreeSet<&str> = o.iter().map(|(k, _)| k).collect();
+        assert_eq!(gkeys, okeys, ".changes field sets differ");
+
+        for (key, gvalue) in g.iter() {
+            let ovalue = o.get(key).unwrap();
+            if matches!(key, "Checksums-Sha1" | "Checksums-Sha256" | "Files") {
+                let without_buildinfo = |v: &str| -> String {
+                    v.lines()
+                        .filter(|l| !l.trim_end().ends_with("_source.buildinfo"))
+                        .collect::<Vec<_>>()
+                        .join("\n")
+                };
+                assert_eq!(
+                    without_buildinfo(gvalue),
+                    without_buildinfo(ovalue),
+                    ".changes field {} differs",
+                    key
+                );
+            } else {
+                assert_eq!(gvalue, ovalue, ".changes field {} differs", key);
+            }
+        }
+    }
+
+    /// Compare a `.buildinfo` structure, skipping machine-dependent fields.
+    fn assert_buildinfo_equivalent(golden: &Path, ours: &Path) {
+        const SKIP: &[&str] = &[
+            "Installed-Build-Depends",
+            "Environment",
+            "Build-Date",
+            "Build-Tainted-By",
+        ];
+        let g = first_paragraph(strip_signature(
+            &fs::read_to_string(golden).expect("read golden buildinfo"),
+        ));
+        let o = first_paragraph(strip_signature(
+            &fs::read_to_string(ours).expect("read our buildinfo"),
+        ));
+
+        for (key, gvalue) in g.iter() {
+            if SKIP.contains(&key) {
+                continue;
+            }
+            let ovalue = o
+                .get(key)
+                .unwrap_or_else(|| panic!(".buildinfo missing field {}", key));
+            assert_eq!(gvalue, ovalue, ".buildinfo field {} differs", key);
+        }
+        for (key, _) in o.iter() {
+            assert!(
+                SKIP.contains(&key) || g.get(key).is_some(),
+                ".buildinfo has unexpected extra field {}",
+                key
+            );
+        }
+    }
+
+    /// Build `src_tree` with both implementations and compare all artifacts.
+    fn differential_on_tree(src_tree: &Path) {
+        let src_parent = src_tree.parent().expect("tree has a parent directory");
+        let tree_name = src_tree.file_name().expect("tree has a name").to_owned();
+
+        let base = tempfile::tempdir().expect("tempdir");
+        let golden_root = base.path().join("golden");
+        let ours_root = base.path().join("ours");
+        fs::create_dir_all(&golden_root).expect("mkdir golden");
+        fs::create_dir_all(&ours_root).expect("mkdir ours");
+
+        copy_path(src_tree, &golden_root);
+        copy_path(src_tree, &ours_root);
+
+        // Sibling orig tarballs are required by quilt/1.0 formats; sibling
+        // .dsc files are required by binary-only (binNMU) builds.
+        for entry in fs::read_dir(src_parent).expect("list source parent") {
+            let entry = entry.expect("dir entry");
+            let name = entry.file_name().to_string_lossy().to_string();
+            if entry.path().is_file() && (name.contains(".orig.tar.") || name.ends_with(".dsc")) {
+                copy_path(&entry.path(), &golden_root);
+                copy_path(&entry.path(), &ours_root);
+            }
+        }
+
+        let golden_tree = golden_root.join(&tree_name);
+        let ours_tree = ours_root.join(&tree_name);
+
+        run_dpkg(&golden_tree);
+        run_source_build(&ours_tree, &SourceBuildOptions::default())
+            .expect("native source pipeline failed");
+
+        let entry =
+            crate::debian::parse_changelog_entry(&ours_tree.join("debian/changelog")).unwrap();
+        let sversion = entry.version.no_epoch();
+        let dsc = golden_root.join(format!("{}_{}.dsc", entry.source, sversion));
+        let changes = golden_root.join(format!("{}_{}_source.changes", entry.source, sversion));
+        let buildinfo = golden_root.join(format!("{}_{}_source.buildinfo", entry.source, sversion));
+
+        assert!(
+            dsc.exists() && changes.exists() && buildinfo.exists(),
+            "native pipeline did not produce all artifacts"
+        );
+
+        // The .dsc payload must be byte-identical (both unsigned here).
+        let g_dsc_text = fs::read_to_string(&dsc).expect("read golden dsc");
+        let o_dsc_text = fs::read_to_string(ours_root.join(&dsc)).expect("read our dsc");
+        assert_eq!(
+            strip_signature(&g_dsc_text),
+            strip_signature(&o_dsc_text),
+            ".dsc payload differs"
+        );
+
+        assert_changes_equivalent(&changes, &ours_root.join(&changes));
+        assert_buildinfo_equivalent(&buildinfo, &ours_root.join(&buildinfo));
+    }
+
+    fn differential_case(spec: &FixtureSpec) {
+        let base = tempfile::tempdir().expect("tempdir");
+        let tree = write_fixture(base.path(), spec);
+        differential_on_tree(&tree);
+    }
+
+    #[test]
+    fn diff_native_minimal() {
+        differential_case(&FixtureSpec::new("pkh-diff-a", "1.0-1", "unstable"));
+    }
+
+    #[test]
+    fn diff_native_epoch() {
+        differential_case(&FixtureSpec::new("pkh-diff-b", "3:2.4.1-1", "unstable"));
+    }
+
+    #[test]
+    fn diff_native_tilde() {
+        differential_case(&FixtureSpec::new("pkh-diff-c", "1.0~rc2-1", "unstable"));
+    }
+
+    #[test]
+    fn diff_quilt_ubuntu_focal_high_urgency() {
+        let mut spec = FixtureSpec::new("pkh-diff-d", "1.2-1", "focal");
+        spec.format = "3.0 (quilt)";
+        spec.patches = &[
+            "--- a/hello.txt\n+++ b/hello.txt\n@@ -1 +1 @@\n-upstream content v1\n+patched by debian\n",
+        ];
+        spec.urgency = "high";
+        differential_case(&spec);
+    }
+
+    #[test]
+    fn diff_quilt_noble_multiple_binaries() {
+        let mut spec = FixtureSpec::new("pkh-diff-e", "0.9-2", "noble");
+        spec.format = "3.0 (quilt)";
+        spec.patches = &[
+            "--- a/hello.txt\n+++ b/hello.txt\n@@ -1 +1 @@\n-upstream content v1\n+patched by debian\n",
+        ];
+        spec.binaries = &[("pkh-diff-e-bin", "any"), ("pkh-diff-e-common", "all")];
+        differential_case(&spec);
+    }
+
+    #[test]
+    fn diff_quilt_single_patch() {
+        let mut spec = FixtureSpec::new("pkh-diff-f", "2.10-3", "unstable");
+        spec.format = "3.0 (quilt)";
+        spec.patches = &[
+            "--- a/hello.txt\n+++ b/hello.txt\n@@ -1 +1 @@\n-upstream content v1\n+patched by debian\n",
+        ];
+        differential_case(&spec);
+    }
+
+    #[test]
+    fn diff_quilt_epoch_two_patches_jammy() {
+        let mut spec = FixtureSpec::new("pkh-diff-g", "9:1.5-2", "jammy");
+        spec.format = "3.0 (quilt)";
+        spec.patches = &[
+            "--- a/hello.txt\n+++ b/hello.txt\n@@ -1 +1 @@\n-upstream content v1\n+patched once\n",
+            "--- a/hello.txt\n+++ b/hello.txt\n@@ -1 +1 @@\n-patched once\n+patched twice\n",
+        ];
+        differential_case(&spec);
+    }
+
+    #[test]
+    fn diff_format_1_0() {
+        let mut spec = FixtureSpec::new("pkh-diff-h", "0.1-1", "unstable");
+        spec.format = "1.0";
+        differential_case(&spec);
+    }
+
+    #[test]
+    fn diff_binmu_binary_only() {
+        let mut spec = FixtureSpec::new("pkh-diff-i", "1.0-1+b1", "unstable");
+        spec.body = &["* Binary-only rebuild."];
+        spec.with_previous_entry = true;
+
+        // Binary-only metadata references the previous version's .dsc, which
+        // must already exist next to the package tree.
+        let base = tempfile::tempdir().expect("tempdir");
+        let tree = write_fixture(base.path(), &spec);
+        let prev_dsc = format!(
+            "Format: 3.0 (native)\nSource: {}\nBinary: {}\nArchitecture: all\nVersion: 1.0-1\nMaintainer: {MAINTAINER}\n",
+            spec.name, spec.name
+        );
+        fs::write(
+            base.path().join(format!("{}_1.0-1.dsc", spec.name)),
+            prev_dsc,
+        )
+        .expect("write previous dsc");
+        differential_on_tree(&tree);
+    }
+
+    #[test]
+    fn diff_closes_bugs() {
+        let mut spec = FixtureSpec::new("pkh-diff-j", "2.0-1", "unstable");
+        spec.body = &[
+            "* Fix crash (Closes: #123456)",
+            "* Another fix (Closes: #42)",
+        ];
+        differential_case(&spec);
+    }
+
+    #[test]
+    fn diff_unreleased_no_sign() {
+        differential_case(&FixtureSpec::new("pkh-diff-k", "1.1-1", "UNRELEASED"));
+    }
+
+    #[test]
+    fn diff_extra_control_fields() {
+        let mut spec = FixtureSpec::new("pkh-diff-l", "5.3-1", "trixie");
+        spec.extra_source_fields = &[
+            ("Homepage", "https://example.com/pkh-diff"),
+            ("Vcs-Git", "https://example.com/git/pkh-diff.git"),
+        ];
+        differential_case(&spec);
+    }
+
+    /// Differential check of a single real archive package: pull it with
+    /// pkh's own [`crate::pull`] (archive download mode) from `dist`
+    /// (optionally `series`), then compare artifacts produced by real
+    /// `dpkg-buildpackage` against the native pipeline. Requires network
+    /// access.
+    fn differential_real_archive_package(package: &str, dist: &str, series: Option<&str>) {
+        let fetch_dir = tempfile::tempdir().expect("tempdir");
+        let rt = tokio::runtime::Runtime::new().expect("tokio runtime");
+
+        let package_info = rt
+            .block_on(crate::package_info::lookup(
+                package,
+                None,
+                series,
+                "",
+                Some(dist),
+                None,
+                None,
+            ))
+            .unwrap_or_else(|e| panic!("package lookup failed for {}: {}", package, e));
+        rt.block_on(crate::pull::pull(
+            &package_info,
+            Some(fetch_dir.path()),
+            None,
+            true,
+        ))
+        .unwrap_or_else(|e| panic!("pull failed for {}: {}", package, e));
+
+        // pull extracts the source tree under '<fetch_dir>/<package>/<package>',
+        // with the orig tarball and .dsc alongside it.
+        let tree = fetch_dir.path().join(package).join(package);
+        assert!(
+            tree.join("debian/changelog").exists(),
+            "pulled tree for {} has no debian/changelog",
+            package
+        );
+        log::info!(
+            "differential test against real package: {} ({}/{})",
+            package,
+            dist,
+            series.unwrap_or("latest")
+        );
+        differential_on_tree(&tree);
+    }
+
+    #[test]
+    fn diff_real_hello_ubuntu_noble() {
+        differential_real_archive_package("hello", "ubuntu", Some("noble"));
+    }
+
+    #[test]
+    fn diff_real_dosfstools_debian_trixie() {
+        differential_real_archive_package("dosfstools", "debian", Some("trixie"));
+    }
+
+    #[test]
+    fn diff_real_sl_ubuntu_resolute() {
+        differential_real_archive_package("sl", "ubuntu", Some("resolute"));
+    }
+
+    #[test]
+    fn diff_real_linux_ubuntu_resolute() {
+        differential_real_archive_package("linux", "ubuntu", Some("resolute"));
+    }
+
+    #[test]
+    fn diff_real_linux_riscv_ubuntu_resolute() {
+        differential_real_archive_package("linux-riscv", "ubuntu", Some("resolute"));
+    }
+
+    #[test]
+    fn diff_real_2048_universe_ubuntu_end_to_end() {
+        differential_real_archive_package("2048", "ubuntu", Some("noble"));
+    }
+
+    #[test]
+    fn diff_real_1oom_contrib_debian_end_to_end() {
+        differential_real_archive_package("1oom", "debian", Some("trixie"));
+    }
+
+    #[test]
+    fn diff_real_agg_svn_fallback_ok() {
+        differential_real_archive_package("agg", "debian", Some("trixie"));
+    }
+
+    #[test]
+    fn diff_real_hello_debian_latest_end_to_end() {
+        differential_real_archive_package("hello", "debian", None);
+    }
+
+    #[test]
+    fn diff_real_hello_ubuntu_latest_end_to_end() {
+        differential_real_archive_package("hello", "ubuntu", None);
+    }
+
+    /// Arbitrary corpus via environment variables, for ad-hoc broad runs:
+    /// ```text
+    /// PKH_DIFF_PACKAGES="bash coreutils curl" \
+    /// PKH_DIFF_DIST=ubuntu \
+    /// PKH_DIFF_SERIES=noble \
+    ///     cargo test --lib differential_real_archive_packages -- --ignored
+    /// ```
+    #[test]
+    #[ignore = "requires network access; intended for ad-hoc broad runs"]
+    fn differential_real_archive_packages() {
+        let packages = std::env::var("PKH_DIFF_PACKAGES")
+            .unwrap_or_else(|_| "hello".to_string())
+            .split_whitespace()
+            .map(str::to_string)
+            .collect::<Vec<_>>();
+
+        assert!(
+            !packages.is_empty(),
+            "PKH_DIFF_PACKAGES contained no package names"
+        );
+
+        let dist = std::env::var("PKH_DIFF_DIST").unwrap_or_else(|_| "ubuntu".to_string());
+        let series = std::env::var("PKH_DIFF_SERIES").ok();
+
+        for name in &packages {
+            differential_real_archive_package(name, &dist, series.as_deref());
+        }
+    }
+}

src/debian/changelog.rs

@@ -3,6 +3,7 @@
 use std::path::Path;
 
 use chrono::DateTime;
+use regex::Regex;
 
 use super::version::DebianVersion;
 
@@ -30,6 +31,9 @@ pub struct ChangelogEntry {
     /// Value for the `.changes` `Changes` field: header line, blank lines
     /// converted to `.`, body lines verbatim; without the trailer line.
     pub changes_field: String,
+    /// Bug numbers collected from `(Closes: #NNN)` mentions in the body,
+    /// sorted numerically and de-duplicated (like dpkg's `find_closes`).
+    pub closes: Option<String>,
 }
 
 /// Parse the most recent entry of a Debian changelog file.
@@ -163,6 +167,8 @@ pub fn parse_changelog_entry(path: &Path) -> Result<ChangelogEntry, Box<dyn std:
         changes_field.push_str(&body_lines.join("\n"));
     }
 
+    let closes = find_closes(&body_lines);
+
     Ok(ChangelogEntry {
         source,
         version,
@@ -174,9 +180,67 @@ pub fn parse_changelog_entry(path: &Path) -> Result<ChangelogEntry, Box<dyn std:
         date_raw,
         timestamp,
         changes_field,
+        closes,
     })
 }
 
+/// Extract bug numbers from `(Closes: #NNN)` mentions in changelog body
+/// lines, sorted numerically and de-duplicated (a lenient port of dpkg's
+/// `find_closes`).
+fn find_closes(body_lines: &[String]) -> Option<String> {
+    let re = Regex::new(r"(?i)\(closes:\s*([^)]*)\)").ok()?;
+    let mut numbers: Vec<u64> = Vec::new();
+    for line in body_lines {
+        for capture in re.captures_iter(line) {
+            if let Some(inner) = capture.get(1) {
+                for token in inner.as_str().split(|c: char| !c.is_ascii_digit()) {
+                    if let Ok(n) = token.parse::<u64>() {
+                        numbers.push(n);
+                    }
+                }
+            }
+        }
+    }
+    if numbers.is_empty() {
+        return None;
+    }
+    numbers.sort_unstable();
+    numbers.dedup();
+    Some(
+        numbers
+            .iter()
+            .map(u64::to_string)
+            .collect::<Vec<_>>()
+            .join(" "),
+    )
+}
+
+/// Return the version of the *previous* changelog entry (the second header
+/// in the file), or `None` when only one entry exists.
+pub fn parse_previous_version(path: &Path) -> Result<Option<String>, Box<dyn std::error::Error>> {
+    let content = std::fs::read_to_string(path)
+        .map_err(|e| format!("failed to read changelog '{}': {}", path.display(), e))?;
+
+    let mut seen_first = false;
+    for line in content.lines() {
+        let line = line.trim_end();
+        if looks_like_header(line) {
+            if !seen_first {
+                seen_first = true;
+                continue;
+            }
+            let open = line.find('(').ok_or_else(|| {
+                format!("invalid changelog header in '{}': {}", path.display(), line)
+            })?;
+            let close = line[open..]
+                .find(')')
+                .ok_or_else(|| format!("unbalanced parenthesis in changelog header '{}'", line))?;
+            return Ok(Some(line[open + 1..open + close].to_string()));
+        }
+    }
+    Ok(None)
+}
+
 /// Heuristic check for a changelog entry header line
 /// (`name (version) dist; urgency=...`).
 fn looks_like_header(line: &str) -> bool {