chore(tools): add cargo package metadata; fix formatting and clippy unused-result fixes

Author: mikkihugo

.github/workflows/publish-snapshot.yml

@@ -73,10 +73,11 @@ jobs:
             jq . "$OUT" || true
             exit 1
           fi
-          # ensure each object has id and name fields
-          if ! jq -e '.[0] | has("id") and has("name")' "$OUT" >/dev/null; then
-            echo "Snapshot entries appear to be missing required fields (id/name)"
-            jq '.[0]' "$OUT" || true
+          # ensure every object has non-empty string id and name fields
+          if ! jq -e 'all(.[]; (has("id") and has("name") and (.id|type=="string") and (.name|type=="string") and (.id != "") and (.name != "")))' "$OUT" >/dev/null; then
+            echo "Snapshot validation failed: one or more entries are missing required fields 'id' or 'name', or they are empty/non-string"
+            # Show up to first 5 offending entries
+            jq 'map(select( (has("id")|not) or (has("name")|not) or (.id|type!="string") or (.name|type!="string") or (.id=="") or (.name=="") )) | .[0:5]' "$OUT" || true
             exit 1
           fi
 

Cargo.lock

@@ -127,6 +127,12 @@ default = []
 validate-metadata = []
 sync-tool = ["reqwest", "tokio", "regex"]
 
+[workspace]
+members = [
+    "tools/linguist_to_snapshot",
+    "tools/generate_snapshot_job",
+]
+
 [build-dependencies]
 
 # ═══════════════════════════════════════════════════════════════════════════

Cargo.toml

@@ -2,6 +2,12 @@
 name = "generate_snapshot_job"
 version = "0.1.0"
 edition = "2021"
+description = "Wrapper CLI used in CI to generate the canonical snapshot by invoking the linguist_to_snapshot converter."
+license = "MIT OR Apache-2.0"
+repository = "https://github.com/Singularity-ng/singularity-language-registry"
+readme = "../../README.md"
+keywords = ["snapshot", "ci", "linguist"]
+categories = ["command-line-utilities", "development-tools"]
 
 [dependencies]
 anyhow = "1.0"

tools/generate_snapshot_job/Cargo.toml

@@ -1,7 +1,7 @@
 use anyhow::Result;
 use clap::Parser;
-use std::process::Command;
 use std::path::PathBuf;
+use std::process::Command;
 
 #[derive(Parser)]
 struct Args {

tools/generate_snapshot_job/src/main.rs

@@ -62,7 +62,7 @@ fn parse_vendor_yml(content: &str) -> HashSet<String> {
                 .to_string();
 
             if !cleaned.is_empty() && !cleaned.contains('(') && cleaned.len() < 100 {
-                patterns.insert(cleaned);
+                let _ = patterns.insert(cleaned);
             }
         }
     }
@@ -88,7 +88,7 @@ fn parse_generated_rb(content: &str) -> HashSet<String> {
             if let Some(matched) = cap.get(1) {
                 let pattern = matched.as_str().to_string();
                 if pattern.len() < 100 && !pattern.contains("//") {
-                    patterns.insert(pattern);
+                    let _ = patterns.insert(pattern);
                 }
             }
         }
@@ -118,7 +118,7 @@ fn parse_heuristics_yml(content: &str) -> HashMap<String, Vec<String>> {
             // Save previous extension's patterns
             if let Some(ext) = current_extension.take() {
                 if !patterns.is_empty() {
-                    heuristics.insert(ext, patterns.clone());
+                    let _ = heuristics.insert(ext, patterns.clone());
                     patterns.clear();
                 }
             }
@@ -142,7 +142,7 @@ fn parse_heuristics_yml(content: &str) -> HashMap<String, Vec<String>> {
     // Save last extension
     if let Some(ext) = current_extension {
         if !patterns.is_empty() {
-            heuristics.insert(ext, patterns);
+            let _ = heuristics.insert(ext, patterns);
         }
     }
 
@@ -166,16 +166,16 @@ fn categorize_patterns(
         if pattern.starts_with('.') && pattern.len() < 10 {
             // It's an extension
             if binary_exts.contains(&pattern.as_str()) {
-                binary.insert(pattern.clone());
+                let _ = binary.insert(pattern.clone());
             } else if pattern.contains("pb")
                 || pattern.contains("generated")
                 || pattern.contains("proto")
             {
-                generated.insert(pattern.clone());
+                let _ = generated.insert(pattern.clone());
             }
         } else {
             // It's a path
-            vendored.insert(pattern.clone());
+            let _ = vendored.insert(pattern.clone());
         }
     }
 

tools/linguist_sync.rs

@@ -2,6 +2,12 @@
 name = "linguist_to_snapshot"
 version = "0.1.0"
 edition = "2021"
+description = "Convert GitHub Linguist languages.yml/JSON into the registry snapshot JSON format."
+license = "MIT OR Apache-2.0"
+repository = "https://github.com/Singularity-ng/singularity-language-registry"
+readme = "../../README.md"
+keywords = ["linguist", "snapshot", "converter", "languages"]
+categories = ["command-line-utilities", "development-tools"]
 
 [dependencies]
 clap = { version = "4.2", features = ["derive"] }

tools/linguist_to_snapshot/Cargo.toml

@@ -74,19 +74,40 @@ fn main() -> Result<()> {
             let id = k.as_str().unwrap_or_default().to_string();
             let name = id.clone();
             // Map some fields
-            let extensions = v.get(&serde_yaml::Value::from("extensions")).and_then(|x| x.as_sequence()).map(|seq| {
-                seq.iter().filter_map(|e| e.as_str().map(|s| s.to_string())).collect()
-            }).unwrap_or_default();
+            let extensions = v
+                .get(&serde_yaml::Value::from("extensions"))
+                .and_then(|x| x.as_sequence())
+                .map(|seq| {
+                    seq.iter()
+                        .filter_map(|e| e.as_str().map(|s| s.to_string()))
+                        .collect()
+                })
+                .unwrap_or_default();
 
-            let aliases = v.get(&serde_yaml::Value::from("aliases")).and_then(|x| x.as_sequence()).map(|seq| {
-                seq.iter().filter_map(|e| e.as_str().map(|s| s.to_string())).collect()
-            }).unwrap_or_default();
+            let aliases = v
+                .get(&serde_yaml::Value::from("aliases"))
+                .and_then(|x| x.as_sequence())
+                .map(|seq| {
+                    seq.iter()
+                        .filter_map(|e| e.as_str().map(|s| s.to_string()))
+                        .collect()
+                })
+                .unwrap_or_default();
 
-            let mime_types = v.get(&serde_yaml::Value::from("mime_types")).and_then(|x| x.as_sequence()).map(|seq| {
-                seq.iter().filter_map(|e| e.as_str().map(|s| s.to_string())).collect()
-            }).unwrap_or_default();
+            let mime_types = v
+                .get(&serde_yaml::Value::from("mime_types"))
+                .and_then(|x| x.as_sequence())
+                .map(|seq| {
+                    seq.iter()
+                        .filter_map(|e| e.as_str().map(|s| s.to_string()))
+                        .collect()
+                })
+                .unwrap_or_default();
 
-            let tree_sitter_language = v.get(&serde_yaml::Value::from("tree_sitter_language")).and_then(|x| x.as_str()).map(|s| s.to_string());
+            let tree_sitter_language = v
+                .get(&serde_yaml::Value::from("tree_sitter_language"))
+                .and_then(|x| x.as_str())
+                .map(|s| s.to_string());
 
             snapshots.push(SnapshotEntry {
                 id,