Make issues filter label configurable per repo (#597)

Signed-off-by: Sergio Castaño Arteaga <[email protected]>

clotributor-registrar/src/registrar.rs

@@ -12,92 +12,6 @@ use tracing::{debug, error, info, instrument};
 /// Maximum time that can take processing a foundation data file.
 const FOUNDATION_TIMEOUT: u64 = 300;
 
-/// Represents a foundation registered in the database.
-#[derive(Debug, Clone, Serialize, Deserialize)]
-pub(crate) struct Foundation {
-    pub foundation_id: String,
-    pub data_url: String,
-}
-
-/// Represents a project to be registered or updated.
-#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
-pub(crate) struct Project {
-    pub name: String,
-
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub display_name: Option<String>,
-
-    pub description: String,
-
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub logo_url: Option<String>,
-
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub logo_dark_url: Option<String>,
-
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub devstats_url: Option<String>,
-
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub accepted_at: Option<String>,
-
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub maturity: Option<String>,
-
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub maintainers_wanted: Option<MaintainersWanted>,
-
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub digest: Option<String>,
-
-    pub repositories: Vec<Repository>,
-}
-
-impl Project {
-    /// Set the project's digest.
-    fn set_digest(&mut self) -> Result<()> {
-        let data = bincode::serialize(&self)?;
-        let digest = hex::encode(Sha256::digest(data));
-        self.digest = Some(digest);
-        Ok(())
-    }
-}
-
-/// Represents a project's repository.
-#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
-pub(crate) struct Repository {
-    pub name: String,
-    pub url: String,
-
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub exclude: Option<Vec<String>>,
-}
-
-/// Defines if the project is looking for maintainers, as well as some extra
-/// reference and contact information.
-#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
-#[serde(rename_all = "snake_case")]
-pub(crate) struct MaintainersWanted {
-    enabled: bool,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    links: Option<Vec<Link>>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    contacts: Option<Vec<Contact>>,
-}
-
-/// Represents some information about a link.
-#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
-pub(crate) struct Link {
-    title: Option<String>,
-    url: String,
-}
-
-/// Represents some information about a contact.
-#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
-pub(crate) struct Contact {
-    github_handle: String,
-}
-
 /// Process foundations registered in the database.
 #[instrument(skip_all, err)]
 pub(crate) async fn run(cfg: &Config, db: DynDB) -> Result<()> {
@@ -217,6 +131,95 @@ async fn process_foundation(
     Ok(())
 }
 
+/// Represents a foundation registered in the database.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub(crate) struct Foundation {
+    pub foundation_id: String,
+    pub data_url: String,
+}
+
+/// Represents a project to be registered or updated.
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub(crate) struct Project {
+    pub name: String,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub display_name: Option<String>,
+
+    pub description: String,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub logo_url: Option<String>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub logo_dark_url: Option<String>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub devstats_url: Option<String>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub accepted_at: Option<String>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub maturity: Option<String>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub maintainers_wanted: Option<MaintainersWanted>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub digest: Option<String>,
+
+    pub repositories: Vec<Repository>,
+}
+
+impl Project {
+    /// Set the project's digest.
+    fn set_digest(&mut self) -> Result<()> {
+        let data = bincode::serialize(&self)?;
+        let digest = hex::encode(Sha256::digest(data));
+        self.digest = Some(digest);
+        Ok(())
+    }
+}
+
+/// Represents a project's repository.
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub(crate) struct Repository {
+    pub name: String,
+    pub url: String,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub exclude: Option<Vec<String>>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub issues_filter_label: Option<String>,
+}
+
+/// Defines if the project is looking for maintainers, as well as some extra
+/// reference and contact information.
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
+pub(crate) struct MaintainersWanted {
+    enabled: bool,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    links: Option<Vec<Link>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    contacts: Option<Vec<Contact>>,
+}
+
+/// Represents some information about a link.
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub(crate) struct Link {
+    title: Option<String>,
+    url: String,
+}
+
+/// Represents some information about a contact.
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub(crate) struct Contact {
+    github_handle: String,
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -419,6 +422,7 @@ mod tests {
                         name: "artifact-hub".to_string(),
                         url: "https://github.com/artifacthub/hub".to_string(),
                         exclude: None,
+                        issues_filter_label: None,
                     }],
                     maintainers_wanted: None,
                 }),

clotributor-tracker/src/db.rs

@@ -62,6 +62,7 @@ impl DB for PgDB {
                     r.languages,
                     r.stars,
                     r.digest,
+                    r.issues_filter_label,
                     p.name as project_name,
                     p.foundation_id
                 from repository r
@@ -84,6 +85,7 @@ impl DB for PgDB {
                 languages: row.get("languages"),
                 stars: row.get("stars"),
                 digest: row.get("digest"),
+                issues_filter_label: row.get("issues_filter_label"),
                 project_name: row.get("project_name"),
                 foundation_id: row.get("foundation_id"),
             })

clotributor-tracker/src/github.rs

@@ -17,6 +17,9 @@ use time::{
 /// GitHub GraphQL API URL.
 const GITHUB_GRAPHQL_API_URL: &str = "https://api.github.com/graphql";
 
+/// Label used to filter the issues we want to track.
+const DEFAULT_ISSUES_FILTER_LABEL: &str = "help wanted";
+
 lazy_static! {
     static ref GITHUB_REPO_URL: Regex =
         Regex::new("^https://github.com/(?P<owner>[^/]+)/(?P<repo>[^/]+)/?$")
@@ -112,7 +115,12 @@ impl repo_view::RepoViewRepository {
 #[cfg_attr(test, automock)]
 pub(crate) trait GH {
     /// Get repository information from GitHub.
-    async fn repository(&self, token: &str, url: &str) -> Result<repo_view::RepoViewRepository>;
+    async fn repository(
+        &self,
+        token: &str,
+        url: &str,
+        issues_filter_label: &Option<String>,
+    ) -> Result<repo_view::RepoViewRepository>;
 }
 
 /// GH implementation backed by the GitHub GraphQL API.
@@ -127,16 +135,25 @@ impl GHGraphQL {
 
 #[async_trait]
 impl GH for GHGraphQL {
-    async fn repository(&self, token: &str, url: &str) -> Result<repo_view::RepoViewRepository> {
+    async fn repository(
+        &self,
+        token: &str,
+        url: &str,
+        issues_filter_label: &Option<String>,
+    ) -> Result<repo_view::RepoViewRepository> {
         // Do request to GraphQL API
         let http_client = setup_http_client(token)?;
         let (owner, repo) = get_owner_and_repo(url)?;
         let issues_since = OffsetDateTime::now_utc()
             .saturating_sub(365.days())
             .format(&Iso8601::DEFAULT)?;
+        let issues_label = issues_filter_label
+            .as_deref()
+            .unwrap_or(DEFAULT_ISSUES_FILTER_LABEL);
         let vars = repo_view::Variables {
             repo,
             owner,
+            issues_label: issues_label.to_string(),
             issues_since,
         };
         let req_body = &RepoView::build_query(vars);

clotributor-tracker/src/graphql/repo_view.graphql

@@ -1,12 +1,12 @@
-query RepoView($repo: String!, $owner: String!, $issues_since: DateTime!) {
+query RepoView($repo: String!, $owner: String!, $issues_label: String!, $issues_since: DateTime!) {
     repository(name: $repo, owner: $owner) {
         description
         homepageUrl
         issues (
             first: 50,
             filterBy: {
                 assignee: null,
-                labels: ["help wanted"],
+                labels: [$issues_label],
                 since: $issues_since,
                 states: [OPEN]
             },

clotributor-tracker/src/tracker.rs

@@ -110,7 +110,9 @@ async fn track_repository(
     debug!("started");
 
     // Fetch repository data from GitHub
-    let gh_repo = gh.repository(&gh_token, &repo.url).await?;
+    let gh_repo = gh
+        .repository(&gh_token, &repo.url, &repo.issues_filter_label)
+        .await?;
 
     // Update repository's GitHub data in db if needed
     let changed = repo.update_gh_data(&gh_repo)?;
@@ -169,6 +171,7 @@ pub(crate) struct Repository {
     pub languages: Option<Vec<String>>,
     pub stars: Option<i32>,
     pub digest: Option<String>,
+    pub issues_filter_label: Option<String>,
     pub project_name: String,
     pub foundation_id: String,
 }
@@ -647,9 +650,9 @@ mod tests {
                 }])))
             });
         gh.expect_repository()
-            .with(eq(TOKEN1), eq(REPOSITORY_URL))
+            .with(eq(TOKEN1), eq(REPOSITORY_URL), eq(None))
             .times(1)
-            .returning(|_, _| Box::pin(future::ready(Err(format_err!(FAKE_ERROR)))));
+            .returning(|_, _, _| Box::pin(future::ready(Err(format_err!(FAKE_ERROR)))));
 
         let result = run(&cfg, Arc::new(db), Arc::new(gh)).await;
         assert_eq!(result.unwrap_err().root_cause().to_string(), FAKE_ERROR);
@@ -672,9 +675,9 @@ mod tests {
                 }])))
             });
         gh.expect_repository()
-            .with(eq(TOKEN1), eq(REPOSITORY_URL))
+            .with(eq(TOKEN1), eq(REPOSITORY_URL), eq(None))
             .times(1)
-            .returning(|_, _| {
+            .returning(|_, _, _| {
                 Box::pin(future::ready(Ok(RepoViewRepository {
                     description: Some("description".to_string()),
                     homepage_url: None,

database/migrations/functions/projects/register_project.sql

@@ -50,15 +50,18 @@ begin
         insert into repository (
             name,
             url,
+            issues_filter_label,
             project_id
         ) values (
             v_repository->>'name',
             v_repository->>'url',
+            nullif(v_repository->>'issues_filter_label', ''),
             v_project_id
         )
         on conflict (project_id, url) do update
         set
             name = excluded.name,
+            issues_filter_label = excluded.issues_filter_label,
             updated_at = current_timestamp;
     end loop;
 

database/migrations/schema/003_issues_filter_label.sql

@@ -0,0 +1,5 @@
+alter table repository add column issues_filter_label text;
+
+---- create above / drop below ----
+
+alter table repository drop column issues_filter_label;