From 50fe17a6049c40d47835da3ec2db047d8132b2ef Mon Sep 17 00:00:00 2001
From: Yorhel <git@yorhel.nl>
Date: Tue, 15 Nov 2016 21:11:18 +0100
Subject: [PATCH] Indexer: Support .deb archives

---
 indexer/src/archive.rs | 22 ++++++++++++++++++++++
 indexer/src/open.rs    |  1 +
 indexer/src/pkg.rs     | 39 +++++++++++++++++++++++++++++++--------
 3 files changed, 54 insertions(+), 8 deletions(-)

diff --git a/indexer/src/archive.rs b/indexer/src/archive.rs
index c536dcb..a81755a 100644
--- a/indexer/src/archive.rs
+++ b/indexer/src/archive.rs
@@ -46,6 +46,14 @@ pub enum FileType {
     Other, // Also includes Link(<non-utf8-path>)
 }
 
+// Top-level formats, as in ARCHIVE_FORMAT_*
+#[derive(Debug,PartialEq,Eq)]
+pub enum Format {
+    Tar,
+    Ar,
+    Other, // Ultra lazyness
+}
+
 
 unsafe extern "C" fn archive_read_cb(_: *mut ffi::Struct_archive, data: *mut c_void, buf: *mut *const c_void) -> ssize_t {
     let arch: &mut Archive = &mut *(data as *mut Archive);
@@ -189,6 +197,19 @@ impl<'a> ArchiveEntry<'a> {
         unsafe { ffi::archive_entry_size(self.e) as usize }
     }
 
+    pub fn format(&self) -> Format {
+        // Interestingly, archive_format() is a property of the entry itself, not of the top-level
+        // archive. Hence it requires archive_read_next_header() and hence it's better placed as
+        // part of this ArchiveEntry object rather than the Archive object.
+        // ...that said, the top-level format isn't likely to change, it's the lower 16 bits that
+        // might be different.
+        match unsafe { ffi::archive_format(self.a.a) } >> 16 {
+            0x3 => Format::Tar,
+            0x7 => Format::Ar,
+            _   => Format::Other,
+        }
+    }
+
     fn symlink(&self) -> Option<String> {
         let c_str: &CStr = unsafe {
             let ptr = ffi::archive_entry_symlink(self.e);
@@ -295,6 +316,7 @@ mod tests {
         let mut ent = Archive::open_archive(&mut f).unwrap().unwrap();
 
         let t = |e:&mut ArchiveEntry, path, size, ft, cont| {
+            assert_eq!(e.format(), Format::Tar);
             assert_eq!(e.path(), path);
             assert_eq!(e.size(), size);
             assert_eq!(e.filetype(), ft);
diff --git a/indexer/src/open.rs b/indexer/src/open.rs
index ef7e8b2..6bbe82d 100644
--- a/indexer/src/open.rs
+++ b/indexer/src/open.rs
@@ -10,6 +10,7 @@ const CACHE_PATH: &'static str = "/var/tmp/manned-indexer";
 const CACHE_TIME: u64 = 23*3600;
 
 
+#[derive(Clone,Copy)]
 pub struct Path<'a> {
     pub path: &'a str,
     pub cache: bool,
diff --git a/indexer/src/pkg.rs b/indexer/src/pkg.rs
index 9d59898..11cfc9f 100644
--- a/indexer/src/pkg.rs
+++ b/indexer/src/pkg.rs
@@ -1,11 +1,11 @@
 use std;
-use std::io::Read;
+use std::io::{Error,ErrorKind,Read};
 use postgres;
 
 use open;
 use archread;
 use man;
-use archive::{Archive,ArchiveEntry};
+use archive::{Format,Archive,ArchiveEntry};
 
 pub struct PkgOpt<'a> {
     pub force: bool,
@@ -104,22 +104,45 @@ fn insert_link(tr: &postgres::GenericConnection, verid: i32, src: &str, dest: &s
 }
 
 
+fn with_pkg<F,T>(f: open::Path, cb: F) -> std::io::Result<T>
+    where F: FnOnce(Option<ArchiveEntry>) -> std::io::Result<T>
+{
+    let mut rd = f.open()?;
+    let ent = match Archive::open_archive(&mut rd)? {
+        None => return cb(None),
+        Some(x) => x,
+    };
+
+    // .deb ("2.0")
+    if ent.format() == Format::Ar && ent.path() == Some("debian-binary") {
+        let mut ent = ent.next()?;
+        while let Some(mut e) = ent {
+            if e.path().map(|p| p.starts_with("data.tar")) == Some(true) {
+                return cb(Archive::open_archive(&mut e)?);
+            }
+            ent = e.next()?
+        }
+        Err(Error::new(ErrorKind::Other, "Debian file without data.tar"))
+
+    // any other archive (Arch/FreeBSD .tar)
+    } else {
+        cb(Some(ent))
+    }
+}
+
+
 fn index_pkg(tr: &postgres::GenericConnection, opt: &PkgOpt, verid: i32) -> std::io::Result<()> {
     let indexfunc = |paths: &[&str], ent: &mut ArchiveEntry| {
         insert_man(tr, verid, paths, ent);
         Ok(()) /* Don't propagate errors, continue handling other man pages */
     };
 
-    let mut rd = try!(opt.file.open());
-    let missed = try!(archread::FileList::read(
-            try!(Archive::open_archive(&mut rd)),
-            man::ismanpath, &indexfunc))
+    let missed = with_pkg(opt.file, |e| { archread::FileList::read(e, man::ismanpath, &indexfunc) })?
         .links(|src, dest| { insert_link(tr, verid, src, dest) });
 
     if let Some(missed) = missed {
         warn!("Some links were missed, reading package again");
-        let mut rd = try!(opt.file.open());
-        try!(missed.read(try!(Archive::open_archive(&mut rd)), indexfunc));
+        with_pkg(opt.file, |e| { missed.read(e, indexfunc) })?
     }
     Ok(())
 }