add tracking for PRs landing in channels

7 files changed, 213 insertions, 33 deletions

diff --git a/Cargo.lock b/Cargo.lock
index e2a562d..c90ea47 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -681,6 +681,7 @@ dependencies = [
  "graphql_client",
  "log",
  "pretty_env_logger",
+ "regex",
  "reqwest",
  "rss",
  "serde",
@@ -970,9 +971,9 @@ dependencies = [
 
 [[package]]
 name = "regex"
-version = "1.5.6"
+version = "1.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d83f127d94bdbcda4c8cc2e50f6f84f4b611f69c902699ca385a39c3a75f9ff1"
+checksum = "4c4eb3267174b8c6c2f654116623910a0fef09c4753f8dd83db29c48a0df988b"
 dependencies = [
  "aho-corasick",
  "memchr",
@@ -981,9 +982,9 @@ dependencies = [
 
 [[package]]
 name = "regex-syntax"
-version = "0.6.26"
+version = "0.6.27"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "49b3de9ec5dc0a3417da371aab17d729997c15010e7fd24ff707773a33bddb64"
+checksum = "a3f87b73ce11b1619a3c6332f45341e0047173771e8b8b73f87bfeefb7b56244"
 
 [[package]]
 name = "remove_dir_all"
diff --git a/Cargo.toml b/Cargo.toml
index f96e157..517561e 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -12,6 +12,7 @@ clap = { version = "3.1.18", features = [ "derive" ] }
 graphql_client = { version = "0.10", features = [ "reqwest-blocking" ] }
 log = "0.4"
 pretty_env_logger = "0.4"
+regex = "1.6"
 reqwest = { version = "0.11.10", features = [ "json", "blocking" ] }
 rss = "2.0.1"
 serde = "1.0"
diff --git a/module.nix b/module.nix
index e11048d..a418c29 100644
--- a/module.nix
+++ b/module.nix
@@ -51,6 +51,18 @@ in {
               type = types.str;
               description = "Name of the label.";
             };
+
+            channels = mkOption {
+              type = types.attrsOf (types.listOf types.str);
+              description = ''
+                Branches to check for PR landing events (ie. changes of a
+                PR showing up in a given branch). Useful for channels.
+
+                Branch names my regular expressions. Targets can refer to
+                captures in the base ref regex with $1, $2 etc.
+              '';
+              default = {};
+            };
           };
         });
         default = [];
@@ -78,7 +90,7 @@ in {
     systemd.services.label-tracker = {
       startAt = cfg.startAt;
 
-      path = [ self.packages.${config.nixpkgs.system}.label-tracker ];
+      path = [ pkgs.git self.packages.${config.nixpkgs.system}.label-tracker ];
       environment.RUST_LOG = "info";
       script = ''
         set -euo pipefail
@@ -92,6 +104,11 @@ in {
                 owner = escapeShellArg args.owner;
                 repo = escapeShellArg args.repo;
                 label = escapeShellArg args.label;
+                patterns = escapeShellArg
+                  (concatStringsSep ","
+                    (mapAttrsToList
+                      (base: targets: "${base}:${concatStringsSep " " targets}")
+                      args.channels));
               in ''
                 (
                   umask 0077
@@ -100,7 +117,9 @@ in {
                     label-tracker init states/${name} ${owner} ${repo} ${label}
                   fi
                   label-tracker sync-issues states/${name}
-                  label-tracker sync-prs states/${name}
+                  label-tracker sync-prs states/${name} \
+                    -l states/${name}.git \
+                    -p ${patterns}
                 )
                 (
                   umask 0027
diff --git a/src/github.rs b/src/github.rs
index 6e499be..1cb90be 100644
--- a/src/github.rs
+++ b/src/github.rs
@@ -9,6 +9,7 @@ use crate::types::{DateTime, Issue, PullRequest, HTML, URI};
 const API_URL: &str = "https://api.github.com/graphql";
 
 type Cursor = String;
+type GitObjectID = String;
 
 pub struct Github {
     client: reqwest::blocking::Client,
@@ -119,6 +120,8 @@ impl ChunkedQuery for PullsQuery {
                 last_update: n.updated_at,
                 url: n.url,
                 base_ref: n.base_ref_name,
+                merge_commit: n.merge_commit.map(|c| c.oid),
+                landed_in: Default::default(),
             })
             .collect();
         let cursor = match (self.since, infos.last()) {
diff --git a/src/main.rs b/src/main.rs
index 7dba807..1d09d86 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -7,22 +7,61 @@ mod github;
 mod types;
 
 use std::{
-    collections::{btree_map::Entry, BTreeMap},
+    collections::{btree_map::Entry, BTreeMap, BTreeSet},
     env,
     fs::File,
     io::{BufReader, BufWriter},
-    path::{Path, PathBuf},
+    path::{Path, PathBuf}, process, str::FromStr,
 };
 
 use anyhow::{Context, Result};
 use chrono::{Duration, Utc};
 use clap::{Args, Parser, Subcommand};
 use github::Github;
+use regex::Regex;
 use rss::{Channel, ChannelBuilder, Guid, Item, ItemBuilder};
 use serde_json::{from_reader, to_writer};
 use tempfile::NamedTempFile;
 use types::{DateTime, IssueAction, PullAction, State, STATE_VERSION};
 
+#[derive(Debug)]
+struct ChannelPatterns {
+    patterns: Vec<(Regex, Vec<String>)>,
+}
+
+impl ChannelPatterns {
+    fn find_channels(&self, base: &str) -> BTreeSet<String> {
+        self.patterns
+            .iter()
+            .flat_map(|(b, c)| match b.find_at(base, 0) {
+                Some(m) if m.end() == base.len() => Some((b, c)),
+                _ => None
+            })
+            .flat_map(|(b, c)| c.iter().map(|chan| b.replace(base, chan).to_string()))
+            .collect()
+    }
+}
+
+impl FromStr for ChannelPatterns {
+    type Err = anyhow::Error;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        let patterns = s
+            .split(",")
+            .map(|s| {
+                match s.trim().split_once(":") {
+                    Some((base, channels)) => Ok((
+                        Regex::new(base)?,
+                        channels.split_whitespace().map(|s| s.to_owned()).collect::<Vec<_>>()
+                    )),
+                    None => bail!("invalid channel pattern `{s}`"),
+                }
+            })
+            .collect::<Result<_>>()?;
+        Ok(ChannelPatterns { patterns })
+    }
+}
+
 #[derive(Parser)]
 #[clap(version)]
 /// Poll github issues and PRs by label and generate RSS feeds.
@@ -48,9 +87,9 @@ enum Command {
         label: String,
     },
     /// Sync issues on a state.
-    SyncIssues(SyncArgs),
+    SyncIssues(SyncIssuesArgs),
     /// Sync pull requests on a state.
-    SyncPrs(SyncArgs),
+    SyncPrs(SyncPrsArgs),
     /// Emit an RSS feed for issue changes.
     EmitIssues(EmitArgs),
     /// Emit an RSS feed for PR changes.
@@ -58,9 +97,23 @@ enum Command {
 }
 
 #[derive(Args)]
-struct SyncArgs {
+struct SyncIssuesArgs {
+    /// State to sync.
+    state_file: PathBuf,
+}
+
+#[derive(Args)]
+struct SyncPrsArgs {
     /// State to sync.
     state_file: PathBuf,
+
+    /// Path to git repo used for landing detection.
+    #[clap(short = 'l', long)]
+    local_repo: PathBuf,
+
+    /// PR landing patterns.
+    #[clap(short = 'p', long)]
+    patterns: ChannelPatterns,
 }
 
 #[derive(Args)]
@@ -77,11 +130,12 @@ struct EmitArgs {
     out: Option<PathBuf>,
 }
 
-fn with_state<F>(state_file: PathBuf, f: F) -> Result<()>
+fn with_state<F>(state_file: impl AsRef<Path>, f: F) -> Result<()>
 where
     F: FnOnce(State) -> Result<Option<State>>,
 {
-    let old_state: State = from_reader(BufReader::new(File::open(&state_file)?))?;
+    let state_file = state_file.as_ref();
+    let old_state: State = from_reader(BufReader::new(File::open(state_file)?))?;
     if old_state.version != STATE_VERSION {
         bail!(
             "expected state version {}, got {}",
@@ -107,7 +161,7 @@ where
     Ok(())
 }
 
-fn with_state_and_github<F>(state_file: PathBuf, f: F) -> Result<()>
+fn with_state_and_github<F>(state_file: impl AsRef<Path>, f: F) -> Result<()>
 where
     F: FnOnce(State, &Github) -> Result<Option<State>>,
 {
@@ -126,7 +180,10 @@ where
     })
 }
 
-fn sync_issues(mut state: State, github: &github::Github) -> Result<Option<State>> {
+fn sync_issues(
+    mut state: State,
+    github: &github::Github,
+) -> Result<Option<State>> {
     let issues = github.query_issues(state.issues_updated)?;
 
     let mut new_history = vec![];
@@ -161,7 +218,13 @@ fn sync_issues(mut state: State, github: &github::Github) -> Result<Option<State
     Ok(Some(state))
 }
 
-fn sync_prs(mut state: State, github: &github::Github) -> Result<Option<State>> {
+fn sync_prs(
+    mut state: State,
+    github: &github::Github,
+    local_repo: impl AsRef<Path>,
+    channel_patterns: &ChannelPatterns,
+) -> Result<Option<State>> {
+    let local_repo = local_repo.as_ref();
     let prs = github.query_pulls(state.pull_requests_updated)?;
 
     let mut new_history = vec![];
@@ -180,7 +243,7 @@ fn sync_prs(mut state: State, github: &github::Github) -> Result<Option<State>>
                 if (stored.is_open, stored.is_merged) != (updated.is_open, updated.is_merged) {
                     new_history.push((updated.last_update, updated.id.clone(), pr_state(false)));
                 }
-                *stored = updated;
+                stored.update(updated);
             }
             Entry::Vacant(e) => {
                 new_history.push((updated.last_update, updated.id.clone(), pr_state(true)));
@@ -189,7 +252,67 @@ fn sync_prs(mut state: State, github: &github::Github) -> Result<Option<State>>
         }
     }
 
-    new_history.sort_by(|a, b| (a.0, &a.1).cmp(&(b.0, &b.1)));
+    let mut git_cmd = process::Command::new("git");
+    let kind = if !local_repo.exists() {
+        let url = format!("https://github.com/{}/{}", &state.owner, &state.repo);
+        git_cmd.arg("clone").args([&url, "--filter", "tree:0", "--bare"]).arg(local_repo);
+        "clone"
+    } else {
+        git_cmd.arg("-C")
+               .arg(local_repo)
+               .args(["fetch", "--force", "--prune", "origin", "refs/heads/*:refs/heads/*"]);
+        "fetch"
+    };
+
+    let git_status = git_cmd.spawn()?.wait()?;
+    if !git_status.success() {
+        bail!("{kind} failed: {git_status}");
+    }
+
+    let branches = state.pull_requests
+                        .values()
+                        .map(|pr| pr.base_ref.clone())
+                        .collect::<BTreeSet<_>>();
+    let patterns = branches.iter()
+                           .map(|b| (b.as_str(), channel_patterns.find_channels(b)))
+                           .filter(|(_, cs)| !cs.is_empty())
+                           .collect::<BTreeMap<_, _>>();
+
+    for pr in state.pull_requests.values_mut() {
+        let merge = match pr.merge_commit.as_ref() {
+            Some(m) => m,
+            None => continue,
+        };
+        let chans = match patterns.get(pr.base_ref.as_str()) {
+            Some(chans) if chans != &pr.landed_in => chans,
+            _ => continue,
+        };
+        let landed = process::Command::new("git")
+            .arg("-C").arg(local_repo)
+            .args(["branch", "--contains", &merge, "--list"])
+            .args(chans)
+            .output()?;
+        let landed = if landed.status.success() {
+            std::str::from_utf8(&landed.stdout)?
+                .split_whitespace()
+                .filter(|&b| !pr.landed_in.contains(b))
+                .map(|s| s.to_owned())
+                .collect::<Vec<_>>()
+        } else {
+            bail!(
+                "failed to check landing status of {}: {}, {}",
+                pr.id,
+                landed.status,
+                String::from_utf8_lossy(&landed.stderr));
+        };
+        if landed.is_empty() {
+            continue;
+        }
+        pr.landed_in.extend(landed.iter().cloned());
+        new_history.push((Utc::now(), pr.id.clone(), PullAction::Landed(landed)));
+    }
+
+    new_history.sort_by(|a, b| (a.0, &a.1, &a.2).cmp(&(b.0, &b.1, &b.2)));
     if let Some(&(at, _, _)) = new_history.last() {
         state.pull_requests_updated = Some(at);
     }
@@ -198,11 +321,15 @@ fn sync_prs(mut state: State, github: &github::Github) -> Result<Option<State>>
     Ok(Some(state))
 }
 
-fn format_history<V, A: Copy, F: Fn(&V, DateTime, A) -> Item>(
+fn format_history<V, A: Clone, F: Fn(&V, DateTime, &A) -> Item>(
     items: &BTreeMap<String, V>,
     history: &[(DateTime, String, A)],
     age_hours: u32,
     format_entry: F,
+    // backwards compat of GUIDs requires this. we need either a different ID format
+    // or an id suffix to give landing events unique ids in all cases, and the suffix
+    // is easier for now
+    id_suffix: impl Fn(&A) -> String,
 ) -> Result<Vec<Item>> {
     let since = Utc::now() - Duration::hours(age_hours as i64);
 
@@ -217,10 +344,10 @@ fn format_history<V, A: Copy, F: Fn(&V, DateTime, A) -> Item>(
             };
             Ok(Item {
                 guid: Some(Guid {
-                    value: format!("{}/{}", changed.to_rfc3339(), id),
+                    value: format!("{}/{}{}", changed.to_rfc3339(), id, id_suffix(how)),
                     permalink: false,
                 }),
-                ..format_entry(entry, *changed, *how)
+                ..format_entry(entry, *changed, how)
             })
         })
         .collect::<Result<Vec<_>, _>>()
@@ -248,6 +375,7 @@ fn emit_issues(state: &State, age_hours: u32) -> Result<Channel> {
             };
             new_rss_item(tag, &issue.title, &issue.url, changed, &issue.body)
         },
+        |_| String::default(),
     )?;
 
     let channel = ChannelBuilder::default()
@@ -267,16 +395,21 @@ fn emit_prs(state: &State, age_hours: u32) -> Result<Channel> {
         &state.pull_history,
         age_hours,
         |pr, changed, how| {
-            let tag = match how {
-                PullAction::New => "[NEW]",
-                PullAction::NewMerged => "[NEW][MERGED]",
-                PullAction::Closed => "[CLOSED]",
-                PullAction::NewClosed => "[NEW][CLOSED]",
-                PullAction::Merged => "[MERGED]",
+            let (tag, refs) = match how {
+                PullAction::New => ("[NEW]", None),
+                PullAction::NewMerged => ("[NEW][MERGED]", None),
+                PullAction::Closed => ("[CLOSED]", None),
+                PullAction::NewClosed => ("[NEW][CLOSED]", None),
+                PullAction::Merged => ("[MERGED]", None),
+                PullAction::Landed(l) => ("[LANDED]", Some(l.join(" "))),
             };
-            let info = format!("{}({})", tag, pr.base_ref);
+            let info = format!("{}({})", tag, refs.as_ref().unwrap_or_else(|| &pr.base_ref));
             new_rss_item(&info, &pr.title, &pr.url, changed, &pr.body)
         },
+        |how| match how {
+            PullAction::Landed(chans) => format!("/landed/{}", chans.join("/")),
+            _ => String::default(),
+        },
     )?;
 
     let channel = ChannelBuilder::default()
@@ -329,10 +462,15 @@ fn main() -> Result<()> {
             to_writer(file, &state)?;
         }
         Command::SyncIssues(cmd) => {
-            with_state_and_github(cmd.state_file, sync_issues)?;
+            with_state_and_github(&cmd.state_file, sync_issues)?;
         }
         Command::SyncPrs(cmd) => {
-            with_state_and_github(cmd.state_file, sync_prs)?;
+            with_state_and_github(
+                &cmd.state_file,
+                |s, g| {
+                    sync_prs(s, g, cmd.local_repo, &cmd.patterns)
+                },
+            )?;
         }
         Command::EmitIssues(cmd) => {
             with_state(cmd.state_file, |s| {
diff --git a/src/pulls.graphql b/src/pulls.graphql
index a25f247..840e665 100644
--- a/src/pulls.graphql
+++ b/src/pulls.graphql
@@ -21,8 +21,11 @@ query PullsQuery($owner: String!, $name: String!, $label: String!, $after: Strin
           title
           updatedAt
           url
+          mergeCommit {
+            oid
+          }
         }
       }
     }
   }
-}
\ No newline at end of file
+}
diff --git a/src/types.rs b/src/types.rs
index a960647..a18331c 100644
--- a/src/types.rs
+++ b/src/types.rs
@@ -1,6 +1,6 @@
 #![allow(non_camel_case_types)]
 
-use std::collections::BTreeMap;
+use std::collections::{BTreeMap, BTreeSet};
 
 use serde::{Deserialize, Serialize};
 
@@ -59,9 +59,23 @@ pub struct PullRequest {
     pub last_update: DateTime,
     pub url: String,
     pub base_ref: String,
+    pub merge_commit: Option<String>,
+
+    // non-github fields
+    #[serde(default)]
+    pub landed_in: BTreeSet<String>,
 }
 
-#[derive(Debug, PartialEq, Eq, Clone, Copy, Serialize, Deserialize)]
+impl PullRequest {
+    pub fn update(&mut self, from: PullRequest) {
+        *self = PullRequest {
+            landed_in: std::mem::replace(&mut self.landed_in, BTreeSet::new()),
+            ..from
+        }
+    }
+}
+
+#[derive(Debug, PartialOrd, Ord, PartialEq, Eq, Clone, Serialize, Deserialize)]
 #[serde(deny_unknown_fields)]
 pub enum PullAction {
     New,
@@ -69,4 +83,5 @@ pub enum PullAction {
     NewClosed,
     Merged,
     NewMerged,
+    Landed(Vec<String>)
 }