chore(ci): move changelog fragment validation to vdev/Rust

.github/workflows/changelog.yaml

@@ -9,6 +9,9 @@
 name: Changelog
 
 on:
+  pull_request:
+    types: [opened, synchronize, reopened, labeled, unlabeled]
+
   pull_request_target:
     types: [opened, synchronize, reopened, labeled, unlabeled]
 
@@ -47,8 +50,9 @@ jobs:
         with:
           ref: master
           sparse-checkout: |
-            scripts/check_changelog_fragments.sh
             changelog.d/
+            scripts/environment/release-flags.sh
+            scripts/environment/binstall.sh
           sparse-checkout-cone-mode: false
 
       # Checkout PR's changelog.d/ into tmp/
@@ -58,7 +62,13 @@ jobs:
           repository: ${{ github.event.pull_request.head.repo.full_name }}
           ref: ${{ github.event.pull_request.head.sha }}
           path: tmp
-          sparse-checkout: changelog.d/
+          sparse-checkout: |
+            changelog.d/
+            scripts/environment/prepare.sh # FIXME DO NOT MERGE
+
+      - name: Prepare vdev
+        if: env.SHOULD_RUN == 'true'
+        run: bash tmp/scripts/environment/prepare.sh --modules=vdev
 
       - name: Run changelog fragment checker
         if: env.SHOULD_RUN == 'true'
@@ -68,4 +78,4 @@ jobs:
 
           # Add files and then compare with HEAD instead of origin/master
           git add changelog.d/
-          MERGE_BASE=HEAD ./scripts/check_changelog_fragments.sh
+          vdev check changelog-fragments --merge-base HEAD

CONTRIBUTING.md

@@ -137,6 +137,7 @@ make check-licenses
 make check-fmt
 make check-clippy
 make check-component-docs
+make check-changelog-fragments
 
 # Some other checks that in our experience rarely fail on PRs.
 make check-deny
@@ -145,8 +146,6 @@ make check-version
 make check-examples
 make check-scripts
 
-./scripts/check_changelog_fragments.sh
-
 # The following check is very slow.
 # make check-component-features
 ```

Makefile

@@ -462,6 +462,11 @@ check-all: check-scripts check-deny check-component-docs check-licenses
 check-component-features: ## Check that all component features are setup properly
 	${MAYBE_ENVIRONMENT_EXEC} cargo vdev check component-features
 
+.PHONY: check-changelog-fragments
+check-changelog-fragments: ## Check that all component features are setup properly
+	${MAYBE_ENVIRONMENT_EXEC} cargo vdev check changelog-fragments
+
+
 .PHONY: check-clippy
 check-clippy: ## Check code with Clippy
 	${MAYBE_ENVIRONMENT_EXEC} cargo vdev check rust

changelog.d/README.md

@@ -23,7 +23,7 @@ This is enforced during CI.
 To mark a PR as not requiring user-facing changelog notes, add the label 'no-changelog'.
 
 To run the same check that is run in CI to validate that your changelog fragments have
-the correct syntax, commit the fragment additions and then run ./scripts/check_changelog_fragments.sh
+the correct syntax, commit the fragment additions and then run `make check-changelog-fragments`
 
 The format for fragments is: `<unique_name>.<fragment_type>.md`
 

scripts/environment/prepare.sh

@@ -49,6 +49,7 @@ Modules:
   wasm-pack
   markdownlint
   datadog-ci
+  vdev
 
 If a module requires rust then rustup will be automatically installed.
 By default, all modules are installed. To install only a subset:
@@ -77,7 +78,7 @@ contains_module() {
 # Always ensure git safe.directory is set
 git config --global --add safe.directory "$(pwd)"
 
-REQUIRES_RUSTUP=(dd-rust-license-tool cargo-deb cross cargo-nextest cargo-deny cargo-msrv wasm-pack)
+REQUIRES_RUSTUP=(dd-rust-license-tool cargo-deb cross cargo-nextest cargo-deny cargo-msrv wasm-pack vdev)
 
 REQUIRES_BINSTALL=("${REQUIRES_RUSTUP[@]}")
 unset -v 'REQUIRES_BINSTALL[0]' # remove dd-rust-license-tool
@@ -169,3 +170,9 @@ if contains_module datadog-ci; then
     sudo npm install -g @datadog/[email protected]
   fi
 fi
+
+if contains_module vdev; then
+  if [[ "$(vdev --version 2>/dev/null)" != "vdev 0.1.0" ]]; then
+    rustup run stable cargo "${install[@]}" vdev --version 0.1.0 --force --locked
+  fi
+fi

scripts/generate-release-cue.rb

@@ -199,7 +199,7 @@ def generate_changelog!(new_version)
     # these are handled in the upgrade guide separately.
 
     # NOTE: If the fragment types are altered, update both the 'changelog.d/README.md' and
-    #       'scripts/check_changelog_fragments.sh' accordingly.
+    #       'vdev/src/commands/check/changelog_fragments.rs' accordingly.
     type = ""
     if fragment_type == "breaking"
       type = "chore"

vdev/src/commands/check/changelog_fragments.rs

@@ -0,0 +1,148 @@
+use crate::git::added_files_against_merge_base;
+use crate::path_utils::get_changelog_dir;
+use anyhow::{anyhow, Context, Result};
+use std::ffi::OsStr;
+use std::path::{Path, PathBuf};
+
+const FRAGMENT_TYPES: &[&str] = &[
+    "breaking",
+    "security",
+    "deprecation",
+    "feature",
+    "enhancement",
+    "fix",
+];
+
+const DEFAULT_MAX_FRAGMENTS: usize = 1000;
+
+/// Validate changelog fragments added in this branch/PR.
+#[derive(clap::Args, Debug)]
+#[command()]
+pub struct Cli {
+    /// Changelog directory (defaults to `changelog.d`)
+    #[arg(long)]
+    changelog_dir: Option<PathBuf>,
+
+    /// Merge base to diff against (defaults to `origin/master`)
+    #[arg(long)]
+    merge_base: Option<String>,
+
+    /// Max fragments threshold (defaults to `1000`)
+    #[arg(long)]
+    max_fragments: Option<usize>,
+}
+
+impl Cli {
+    pub fn exec(&self) -> Result<()> {
+        let changelog_dir = self
+            .changelog_dir.clone().unwrap_or_else(get_changelog_dir);
+
+        if !changelog_dir.is_dir() {
+            error!(
+                "No ./{} found. This tool must be invoked from the root of the repo.",
+                changelog_dir.display()
+            );
+            std::process::exit(1);
+        }
+
+        let merge_base = self
+            .merge_base.clone()
+            .unwrap_or_else(|| "origin/master".to_string());
+
+        let max_fragments: usize = self
+            .max_fragments
+            .unwrap_or(DEFAULT_MAX_FRAGMENTS);
+
+        let fragments = added_files_against_merge_base(&merge_base, &changelog_dir)
+            .context("failed to collect added changelog fragments")?;
+
+        if fragments.is_empty() {
+            info!("No changelog fragments detected");
+            info!("If no changes necessitate user-facing explanations, add the GH label 'no-changelog'");
+            info!("Otherwise, add changelog fragments to {}/", changelog_dir.display());
+            info!("For details, see '{}/README.md'", changelog_dir.display());
+            std::process::exit(1);
+        }
+
+        if fragments.len() > max_fragments {
+            error!("Too many changelog fragments ({} > {max_fragments}).", fragments.len());
+            std::process::exit(1);
+        }
+
+        for path in fragments {
+            if let Some(name) = path.file_name().and_then(OsStr::to_str) {
+                if name == "README.md" {
+                    continue;
+                }
+                info!("Validating `{name}`");
+                validate_fragment_filename(name)?;
+                validate_fragment_contents(&changelog_dir.join(name), name)?;
+            } else {
+                return Err(anyhow!("unexpected path (no filename): {}", path.display()));
+            }
+        }
+
+        info!("changelog additions are valid.");
+        Ok(())
+    }
+}
+
+fn validate_fragment_filename(filename: &str) -> Result<()> {
+    // Expected: <unique_name>.<fragment_type>.md
+    let parts: Vec<&str> = filename.split('.').collect();
+    if parts.len() != 3 {
+        return Err(anyhow!(
+            "invalid fragment filename: {filename} - wrong number of period delimiters. \
+             expected '<unique_name>.<fragment_type>.md'",
+        ));
+    }
+
+    let fragment_type = parts[1];
+    if !FRAGMENT_TYPES.contains(&fragment_type) {
+        return Err(anyhow!(
+            "invalid fragment filename: {filename} - fragment type must be one of: {})",
+            FRAGMENT_TYPES.join("|"),
+        ));
+    }
+
+    if parts[2] != "md" {
+        return Err(anyhow!(
+            "invalid fragment filename: {filename} - extension must be markdown (.md)",
+        ));
+    }
+
+    Ok(())
+}
+
+fn validate_fragment_contents(path: &Path, filename: &str) -> Result<()> {
+    let content = std::fs::read_to_string(path).with_context(|| {
+        format!("failed to read fragment file: {}", path.to_string_lossy())
+    })?;
+
+    // Use last non-empty line to avoid false negatives due to trailing newline(s).
+    let last_line = content.lines().rev().find(|l| !l.trim().is_empty()).unwrap_or("");
+
+    if !last_line.starts_with("authors:") {
+        return Err(anyhow!("last line must start with 'authors: ' and include at least one name. ({filename})"));
+    }
+
+    // Split on the first colon, take the remainder as names
+    let names = last_line.split_once(':').map_or("", |(_, rest)| rest.trim());
+
+    if names.is_empty() {
+        return Err(anyhow!(
+            "author line is empty. ({})",
+            filename
+        ));
+    }
+
+    if names.contains('@') {
+        return Err(anyhow!("author should not be prefixed with @"));
+    }
+
+    if names.contains(',') {
+        return Err(anyhow!("authors should be space delimited, not comma delimited."));
+    }
+
+    Ok(())
+}

vdev/src/commands/check/mod.rs

@@ -1,10 +1,11 @@
 crate::cli_subcommands! {
     "Check parts of the Vector code base..."
+    docs,
+    events,
+    mod changelog_fragments,
     mod component_docs,
     mod component_features,
     mod deny,
-    docs,
-    events,
     mod examples,
     mod fmt,
     mod licenses,

vdev/src/commands/release/prepare.rs

@@ -2,6 +2,7 @@
 #![allow(clippy::print_stderr)]
 
 use crate::git;
+use crate::path_utils::get_repo_root;
 use crate::util::run_command;
 use anyhow::{anyhow, Result};
 use reqwest::blocking::Client;
@@ -389,10 +390,6 @@ impl Prepare {
 
 // FREE FUNCTIONS AFTER THIS LINE
 
-fn get_repo_root() -> PathBuf {
-    Path::new(env!("CARGO_MANIFEST_DIR")).parent().unwrap().to_path_buf()
-}
-
 fn get_latest_version_from_vector_tags() -> Result<Version> {
     let tags = run_command("git tag --list --sort=-v:refname");
     let latest_tag = tags