workspace, licence, and the three record types

crates/corpus-core/src/main.rs

@@ -0,0 +1,56 @@
+//! Prints the JSON Schema for the record types, so a consumer in another
+//! language can generate its own bindings instead of reading Rust.
+
+use std::process::ExitCode;
+
+use corpus_core::{Chunk, Doc, Hit, exit};
+
+const USAGE: &str = "\
+usage: corpus-core schema [doc|chunk|hit|all]
+
+Prints the JSON Schema for one record type, or an object of all three.
+
+    cargo run -p corpus-core -- schema chunk > chunk.schema.json
+";
+
+fn main() -> ExitCode {
+    let mut args = std::env::args().skip(1);
+    match args.next().as_deref() {
+        Some("schema") => {
+            let which = args.next().unwrap_or_else(|| "all".into());
+            match schema(&which) {
+                Some(json) => {
+                    println!("{json}");
+                    ExitCode::from(exit::OK as u8)
+                }
+                None => {
+                    eprintln!("corpus-core: unknown record type {which:?}\n\n{USAGE}");
+                    ExitCode::from(exit::USAGE as u8)
+                }
+            }
+        }
+        Some("--help" | "-h") => {
+            println!("{USAGE}");
+            ExitCode::from(exit::OK as u8)
+        }
+        _ => {
+            eprintln!("{USAGE}");
+            ExitCode::from(exit::USAGE as u8)
+        }
+    }
+}
+
+fn schema(which: &str) -> Option<String> {
+    let json = match which {
+        "doc" => serde_json::to_string_pretty(&schemars::schema_for!(Doc)),
+        "chunk" => serde_json::to_string_pretty(&schemars::schema_for!(Chunk)),
+        "hit" => serde_json::to_string_pretty(&schemars::schema_for!(Hit)),
+        "all" => serde_json::to_string_pretty(&serde_json::json!({
+            "doc": schemars::schema_for!(Doc),
+            "chunk": schemars::schema_for!(Chunk),
+            "hit": schemars::schema_for!(Hit),
+        })),
+        _ => return None,
+    };
+    Some(json.expect("a schema is always serialisable"))
+}