indexer: Add support for indexing RPM repositories

This code hasn't been thoroughly tested, I'll see how things go when
indexing a live repo.

And XML parsing sucks in every language.

4 files changed, 200 insertions, 0 deletions

diff --git a/indexer/Cargo.lock b/indexer/Cargo.lock
index 04d00cd..eb506b6 100644
--- a/indexer/Cargo.lock
+++ b/indexer/Cargo.lock
@@ -12,6 +12,7 @@ dependencies = [
  "libc 0.2.18 (registry+https://github.com/rust-lang/crates.io-index)",
  "log 0.3.6 (registry+https://github.com/rust-lang/crates.io-index)",
  "postgres 0.13.5 (registry+https://github.com/rust-lang/crates.io-index)",
+ "quick-xml 0.5.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "regex 0.2.1 (registry+https://github.com/rust-lang/crates.io-index)",
  "ring 0.6.2 (registry+https://github.com/rust-lang/crates.io-index)",
  "url 1.2.4 (registry+https://github.com/rust-lang/crates.io-index)",
@@ -351,6 +352,14 @@ dependencies = [
 ]
 
 [[package]]
+name = "quick-xml"
+version = "0.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+dependencies = [
+ "log 0.3.6 (registry+https://github.com/rust-lang/crates.io-index)",
+]
+
+[[package]]
 name = "redox_syscall"
 version = "0.1.16"
 source = "registry+https://github.com/rust-lang/crates.io-index"
@@ -575,6 +584,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 "checksum postgres 0.13.5 (registry+https://github.com/rust-lang/crates.io-index)" = "585ca978431cddac0aa926246f18fe30a47401eabbe9bbda573dc60389c10ea1"
 "checksum postgres-protocol 0.2.0 (registry+https://github.com/rust-lang/crates.io-index)" = "283e27d237a5772ef00c9e3f97e632f9a565ff514761af3e88e129576af7077c"
 "checksum postgres-shared 0.1.0 (registry+https://github.com/rust-lang/crates.io-index)" = "6f09b8819c2586032ed23bfbe95f6edfbebdc18bf9d0fe02c1f785f659958fbb"
+"checksum quick-xml 0.5.0 (registry+https://github.com/rust-lang/crates.io-index)" = "e685d9ea689e56229debf59cb6d24e28021a9c950bbd988af24e43da3ea2bd79"
 "checksum redox_syscall 0.1.16 (registry+https://github.com/rust-lang/crates.io-index)" = "8dd35cc9a8bdec562c757e3d43c1526b5c6d2653e23e2315065bc25556550753"
 "checksum regex 0.2.1 (registry+https://github.com/rust-lang/crates.io-index)" = "4278c17d0f6d62dfef0ab00028feb45bd7d2102843f80763474eeb1be8a10c01"
 "checksum regex-syntax 0.4.0 (registry+https://github.com/rust-lang/crates.io-index)" = "2f9191b1f57603095f105d317e375d19b1c9c5c3185ea9633a99a6dcbed04457"
diff --git a/indexer/Cargo.toml b/indexer/Cargo.toml
index 4113ded..bf62eb8 100644
--- a/indexer/Cargo.toml
+++ b/indexer/Cargo.toml
@@ -17,3 +17,4 @@ clap = "2.20.0"
 hyper = { version = "0.10.0", default-features = false }
 url = "1.2.3"
 chrono = "0.2.25"
+quick-xml = "0.5.0"
diff --git a/indexer/src/main.rs b/indexer/src/main.rs
index 5399e09..2941951 100644
--- a/indexer/src/main.rs
+++ b/indexer/src/main.rs
@@ -11,6 +11,7 @@ extern crate postgres;
 extern crate hyper;
 extern crate url;
 extern crate chrono;
+extern crate quick_xml;
 
 mod archive;
 mod archread;
@@ -22,6 +23,7 @@ mod sys_deb;
 mod sys_freebsd1;
 mod sys_freebsd2;
 mod sys_rpmdir;
+mod sys_rpm;
 
 
 // Convenience function to get a system id by short-name. Panics if the system doesn't exist.
@@ -80,6 +82,12 @@ fn main() {
             (@arg cat: --cat +required +takes_value "Category to set for all packages")
             (@arg mirror: --mirror +required +takes_value "Mirror URL")
         )
+        (@subcommand rpm =>
+            (about: "Index an RPM repository")
+            (@arg sys: --sys +required +takes_value "System short-name")
+            (@arg cat: --cat +required +takes_value "Category to set for all packages")
+            (@arg mirror: --mirror +required +takes_value "Mirror URL")
+        )
     ).get_matches();
 
     unsafe { pkg::DRY_RUN = arg.is_present("dry") };
@@ -169,5 +177,13 @@ fn main() {
         ).unwrap_or_else(|e| error!("{}", e));
     }
 
+    if let Some(matches) = arg.subcommand_matches("rpm") {
+        sys_rpm::sync(&db,
+            sysbyshort(&db, matches.value_of("sys").unwrap()),
+            matches.value_of("cat").unwrap(),
+            matches.value_of("mirror").unwrap()
+        ).unwrap_or_else(|e| error!("{}", e));
+    }
+
     trace!("Exiting");
 }
diff --git a/indexer/src/sys_rpm.rs b/indexer/src/sys_rpm.rs
new file mode 100644
index 0000000..a30d93d
--- /dev/null
+++ b/indexer/src/sys_rpm.rs
@@ -0,0 +1,173 @@
+use std::collections::HashSet;
+use std::io::BufReader;
+use std::str::FromStr;
+use std::error::Error;
+use chrono::NaiveDateTime;
+use postgres;
+use quick_xml as xml;
+
+use archive;
+use open;
+use pkg;
+use man;
+
+
+fn xml_getattr(e: &xml::Element, attr: &str) -> Result<String,Box<Error>> {
+    for kv in e.unescaped_attributes() {
+        let (key, val) = kv.map_err(|(e,_)| e)?;
+        if key == attr.as_bytes() {
+            return Ok(String::from_utf8(val.into_owned())?);
+        }
+    }
+    Err(Box::new(xml::error::Error::EOL))
+}
+
+
+#[derive(Default)]
+struct PkgInfo {
+    name: Option<String>,
+    arch: Option<String>,
+    ver: Option<String>,
+    date: Option<i64>,
+    path: Option<String>,
+    hasman: bool,
+}
+
+
+// Shared function to read primary.xml.gz and filelists.xml.gz. Runs the callback for each package
+// with the info that was found.
+fn readpkgs<F>(url: String, mut cb: F) -> Result<(),Box<Error>>
+    where F: FnMut(PkgInfo)
+{
+    debug!("Reading {}", url);
+    let mut fd = open::Path{path: &url, cache: true, canbelocal: false}.open()?;
+    let xml = xml::XmlReader::from_reader(
+        BufReader::new(
+            archive::Archive::open_raw(&mut fd)?
+        )
+    ).trim_text(true);
+
+    let mut savestr = false;
+    let mut saved = None;
+    let mut pkg = PkgInfo::default();
+
+    let arch_src = Some("src".to_string());
+
+    for event in xml {
+        let event = event.map_err(|(e,_)| e)?;
+        match event {
+
+            xml::Event::Start(ref e) =>
+                match e.name() {
+                    b"name" |
+                    b"file" |
+                    b"arch"     => savestr  = true,
+                    b"version"  => pkg.ver  = Some(format!("{}-{}", xml_getattr(e, "ver")?, xml_getattr(e, "rel")?)),
+                    b"location" => pkg.path = Some(xml_getattr(e, "href")?),
+                    b"time"     => pkg.date = Some(i64::from_str(&xml_getattr(e, "build")?)?),
+                    b"package"  => {
+                        pkg.name = xml_getattr(e, "name").ok();
+                        pkg.arch = xml_getattr(e, "arch").ok();
+                    },
+                    _ => (),
+                },
+
+            xml::Event::Text(e) =>
+                if savestr {
+                    saved = Some(e.into_unescaped_string()?);
+                    savestr = false
+                },
+
+            xml::Event::End(ref e) => {
+                savestr = false;
+                match e.name() {
+                    b"name" => pkg.name = Some(saved.take().unwrap()),
+                    b"arch" => pkg.arch = Some(saved.take().unwrap()),
+                    b"file" => pkg.hasman = pkg.hasman || man::ismanpath(&saved.take().unwrap()),
+                    b"package" => {
+                        if pkg.arch != arch_src {
+                            cb(pkg);
+                        }
+                        pkg = PkgInfo::default();
+                    },
+                    _ => (),
+                };
+            },
+
+            _ => (),
+        }
+    }
+    Ok(())
+}
+
+
+// Reads repomd.xml and returns the path to the primary.xml.gz and filelists.xml.gz
+fn repomd(url: String) -> Result<(String,String),Box<Error>> {
+    debug!("Reading {}", url);
+    let mut fd = open::Path{path: &url, cache: true, canbelocal: false}.open()?;
+    let xml = xml::XmlReader::from_reader(
+        BufReader::new(
+            archive::Archive::open_raw(&mut fd)?
+        )
+    ).trim_text(true);
+
+    let mut primary = String::new();
+    let mut filelists = String::new();
+    let mut datatype = 0;
+
+    for event in xml {
+        if let xml::Event::Start(ref e) = event.map_err(|(e,_)| e)? {
+            match e.name() {
+                b"data" =>
+                    datatype = match &xml_getattr(e, "type")? as &str {
+                        "primary"   => 1,
+                        "filelists" => 2,
+                        _           => 0,
+                    },
+
+                b"location" =>
+                    match datatype {
+                        1 => primary   = xml_getattr(e, "href")?,
+                        2 => filelists = xml_getattr(e, "href")?,
+                        _ => (),
+                    },
+
+                _ => (),
+            }
+        }
+    }
+    Ok((primary, filelists))
+}
+
+
+pub fn sync(pg: &postgres::GenericConnection, sys: i32, cat: &str, mirror: &str) -> Result<(),Box<Error>> {
+    let(primary, filelists) = repomd(format!("{}repodata/repomd.xml", mirror))?;
+
+    let mut pkgswithman = HashSet::new();
+    readpkgs(format!("{}{}", mirror, filelists), |pkg| {
+        if pkg.hasman { pkgswithman.insert(pkg.name.unwrap()); () }
+    })?;
+
+    readpkgs(format!("{}{}", mirror, primary), |pkg| {
+        let name = pkg.name.unwrap();
+        if pkgswithman.contains(&name) {
+            let uri = format!("{}{}", mirror, pkg.path.unwrap());
+            let date = NaiveDateTime::from_timestamp(pkg.date.unwrap(), 0).format("%Y-%m-%d").to_string();
+            pkg::pkg(pg, pkg::PkgOpt{
+                force: false,
+                sys: sys,
+                cat: cat,
+                pkg: &name,
+                ver: &pkg.ver.unwrap(),
+                date: pkg::Date::Known(&date),
+                arch: Some(&pkg.arch.unwrap()),
+                file: open::Path{
+                    path: &uri,
+                    cache: false,
+                    canbelocal: false,
+                },
+            });
+        }
+    })?;
+    Ok(())
+}