From 624843668aecaedc481bf3ac20842ddbec9ca25e Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Sun, 7 Jun 2020 18:57:26 -0400
Subject: [PATCH] netdoc: add an iterator that can parse annotated microdescs

---
 tor-netdoc/src/microdesc.rs        | 89 +++++++++++++++++++++++++++++-
 tor-netdoc/testdata/microdesc2.txt | 39 +++++++++++++
 2 files changed, 127 insertions(+), 1 deletion(-)
 create mode 100644 tor-netdoc/testdata/microdesc2.txt

diff --git a/tor-netdoc/src/microdesc.rs b/tor-netdoc/src/microdesc.rs
index 3b2742418..c26d9369a 100644
--- a/tor-netdoc/src/microdesc.rs
+++ b/tor-netdoc/src/microdesc.rs
@@ -48,6 +48,15 @@ pub struct Microdesc {
     // pr is obsolete and doesn't go here any more.
 }
 
+/// A microdescriptor annotated with additional data
+///
+/// TODO: rename this.
+#[allow(dead_code)]
+pub struct AnnotatedMicrodesc {
+    md: Microdesc,
+    ann: MicrodescAnnotation,
+}
+
 decl_keyword! {
     /// Keyword type for recognized objects in microdescriptors.
     MicrodescKW {
@@ -84,6 +93,12 @@ lazy_static! {
     };
 }
 
+impl Default for MicrodescAnnotation {
+    fn default() -> Self {
+        MicrodescAnnotation { last_listed: None }
+    }
+}
+
 impl MicrodescAnnotation {
     #[allow(dead_code)]
     fn parse_from_reader(
@@ -206,14 +221,86 @@ impl Microdesc {
         })
     }
 }
+
+/// An iterator that parses one or more (possible annnotated)
+/// microdescriptors from a string.
+pub struct MicrodescReader<'a> {
+    annotated: bool,
+    reader: NetDocReader<'a, MicrodescKW>,
+}
+
+/// Indicates whether we should parse an annotated list of votes or a
+/// non-annotated list.
+///
+/// TODO: Move this.
+#[derive(PartialEq, Debug)]
+pub enum AllowAnnotations {
+    /// Parsing a document where items might be annotated.
+    ///
+    /// Annotations are a list of zero or more items with keywords
+    /// beginning with @ that precede the items that are actually part
+    /// of the document.
+    AnnotationsAllowed,
+    /// Parsing a document where annotations are not allowed.
+    AnnotationsNotAllowed,
+}
+
+impl<'a> MicrodescReader<'a> {
+    /// Construct a MicrodescReader to take microdescriptors from a string
+    /// 's'.
+    pub fn new(s: &'a str, allow: AllowAnnotations) -> Self {
+        let reader = NetDocReader::new(s);
+        let annotated = allow == AllowAnnotations::AnnotationsAllowed;
+        MicrodescReader { annotated, reader }
+    }
+
+    fn take_annotation(&mut self) -> Result<MicrodescAnnotation> {
+        if self.annotated {
+            MicrodescAnnotation::parse_from_reader(&mut self.reader)
+        } else {
+            Ok(MicrodescAnnotation::default())
+        }
+    }
+
+    fn take_annotated_microdesc(&mut self) -> Result<AnnotatedMicrodesc> {
+        let ann = self.take_annotation()?;
+        let md = Microdesc::parse_from_reader(&mut self.reader)?;
+        Ok(AnnotatedMicrodesc { md, ann })
+        // TODO need to advance to next likely md.
+    }
+}
+
+impl<'a> Iterator for MicrodescReader<'a> {
+    type Item = Result<AnnotatedMicrodesc>;
+    fn next(&mut self) -> Option<Self::Item> {
+        // If there is no next token, we're at the end.
+        self.reader.iter().peek()?;
+
+        Some(self.take_annotated_microdesc())
+    }
+}
+
 #[cfg(test)]
 mod test {
     use super::*;
     const TESTDATA: &str = include_str!("../testdata/microdesc1.txt");
+    const TESTDATA2: &str = include_str!("../testdata/microdesc2.txt");
 
     #[test]
-    fn parse_arbitrary() -> Result<()> {
+    fn parse_single() -> Result<()> {
         let _md = Microdesc::parse(TESTDATA)?;
         Ok(())
     }
+
+    #[test]
+    fn parse_multi() -> Result<()> {
+        let mut n: u32 = 0;
+        for md in MicrodescReader::new(TESTDATA2, AllowAnnotations::AnnotationsAllowed) {
+            md?;
+            n += 1;
+        }
+        assert_eq!(n, 4);
+        Ok(())
+        // TODO: test actual contents.
+    }
 }
diff --git a/tor-netdoc/testdata/microdesc2.txt b/tor-netdoc/testdata/microdesc2.txt
new file mode 100644
index 000000000..c575dabfc
--- /dev/null
+++ b/tor-netdoc/testdata/microdesc2.txt
@@ -0,0 +1,39 @@
+@last-listed 2020-01-27 18:52:09
+onion-key
+-----BEGIN RSA PUBLIC KEY-----
+MIGJAoGBAMaCi6KwIpOaPFhyMJsJC+r516p0kZXnEuaJLgSR5uakJDW31qDlwD29
+dDLBwT4RmiD+OPNSjBpdEIJDbaDNGdgQLpdLgpl7v5ONoOQ3cRlbySOLmZ/7fYeW
+DE+4eDLemIqU2e3lEOnb/bpCdQ309FX5OiKDcNy8tqtWgoOdOE0/AgMBAAE=
+-----END RSA PUBLIC KEY-----
+ntor-onion-key XoldZTBKOhiUYWZgFD969XV/4IvBgEXHhV7o3ruebEc=
+family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
+p accept 43,80,443,853,873,993,995,1194,2086,3128,5222-5223,5269,8000,8008,8080,8232-8233,8443,8888,9418,11371
+p6 accept 43,80,443,853,873,993,995,1194,2086,3128,5222-5223,5269,8000,8008,8080,8232-8233,8443,8888,9418,11371
+id ed25519 LYX9yI5sG8+0aJf8odum0TVPkyYdaKeeC1vBcN2SMIQ
+@last-listed 2020-01-27 18:52:09
+onion-key
+-----BEGIN RSA PUBLIC KEY-----
+MIGJAoGBAMJi4/MtSJv4lYgLfq+z+2S90cGwIJJfp/mIxx0wX/q7jX6MWGG4+NQx
+gofp3zxAEgglTfdTsvFFzQN4PjVkuwyFmmcYukVmS86e35nmPT3Yq3DPSXBRCtX5
+mMXC68bQT+npnPnoafxMdQddWCATkSD8wBrUKix4nOOaOcoD73SfAgMBAAE=
+-----END RSA PUBLIC KEY-----
+ntor-onion-key siPT+/px5J0pgsTdAWxCNRHbghZ98E6QY2Ioxl7Y2x0=
+id ed25519 eQEthL1Cz57iO7GZ8RtS5lNiBVGCW9KH+VCBpOikywc
+@last-listed 2020-01-27 18:52:10
+onion-key
+-----BEGIN RSA PUBLIC KEY-----
+MIGJAoGBALWooFgPxb0v6TR6PtAlLZWQ5vsX8oTJCiD4yg1bPv2MA2KnquQj+lys
+H0yh/8MH814J/qFkjgTFt7An3U5VRnOyD9BAAJuT7FsqAFTlPzn2hK/PliE/K2/N
+XMeaV6brBOcYvhMR9EWV+zBRpvkkQWqSqNlVjZi78fD1s/y0TjzTAgMBAAE=
+-----END RSA PUBLIC KEY-----
+ntor-onion-key mp9h+hzj2D18B/dLvu4RlEaTktzqrSThY5hc0cSEoBg=
+id ed25519 3B+5CwspC+alleI5eA7jvxp6aGBC8QvF1D3wKBSwEDQ
+@last-listed 2020-01-27 18:52:09
+onion-key
+-----BEGIN RSA PUBLIC KEY-----
+MIGJAoGBAMESE/ElGU2kyaq0VJ7mwGvns6mhMm75ZyYDbR+Eaa3b18PVOslR4aEv
+3RJAt9N4Ph7N2VHZHVxwPKWViVDhZNMvOD5PJLb587KmTvD77Eo6MSh6umxy6xvo
+GfgJJlyrlBQdvxRZVfEmUvjdZixbJUCI9ydxQG7hBk1n89dgLnTJAgMBAAE=
+-----END RSA PUBLIC KEY-----
+ntor-onion-key wcEKLkC4GnqCN0LJxRecweZgYmTg4whSxzFmhAmJYGQ=
+id ed25519 1az7VeHFAgB0KwtxH+nQJSi3dw8rIl4iT5sMqeodTDg