add scaffolding for repl

- module
- clap

Cargo.toml

@@ -19,6 +19,7 @@ serde_json = "1.0"
 serde = { version = "1.0", features = ["derive", "rc"] }
 futures = "0.3"
 url = "2.5"
+clap = { version = "4.5.16", features = ["derive"] }
 
 [workspace]
 members = [".", "filter-proc-macro", "filter-types"]

src/frontend/mod.rs

@@ -1,5 +1,6 @@
 pub mod ast;
 mod parser;
+mod repl;
 mod scanner;
 
 pub use parser::{ParseError, Parser};

src/frontend/repl.rs

@@ -0,0 +1,2 @@
+#[derive(Debug, Clone)]
+pub struct Repl {}

src/main.rs

@@ -1,42 +1,69 @@
 #![forbid(unsafe_code)]
-use std::env;
+use std::path::PathBuf;
 
 use anyhow::Context;
+use clap::Parser as _;
 use scrapelect::{frontend::Parser, interpreter::Interpreter};
+use url::Url;
 
-#[tokio::main]
-async fn main() -> anyhow::Result<()> {
-    let mut args = env::args();
-    // skip name
-    let name = args.next().context("usage: scrapelect <filename> <url>")?;
+#[derive(Debug, clap::Parser)]
+#[command(version, args_conflicts_with_subcommands = true)]
+struct Interface {
+    #[command(subcommand)]
+    mode: Option<Mode>,
+    // There is some magic going on where `Option<RunArgs>` makes it required
+    // when the subcommand is not provided.
+    #[command(flatten)]
+    run: Option<RunArgs>,
+}
+
+#[derive(Debug, clap::Args)]
+struct RunArgs {
+    /// The `.scrp` file describing how to convert the web page into structured data
+    file: PathBuf,
+    /// The URL of the web page to start scraping at.
+    url: Url,
+}
 
-    let filename = args
-        .next()
-        .with_context(|| format!("usage: {name} <filename> <url>"))?;
+#[derive(Debug, clap::Subcommand)]
+enum Mode {
+    Run(RunArgs),
+    Repl,
+}
 
-    let url = args
-        .next()
-        .with_context(|| format!("usage: {name} <filename = {filename}> <url>"))?;
+#[tokio::main]
+async fn main() -> anyhow::Result<()> {
+    let args = Interface::parse();
 
-    let pgm = std::fs::read_to_string(&filename)
-        .with_context(|| format!("error reading file {filename}"))?;
+    match (args.mode, args.run) {
+        (Some(Mode::Run(run_args)), None) | (None, Some(run_args)) => {
+            let pgm = std::fs::read_to_string(&run_args.file)
+                .with_context(|| format!("error reading file {}", run_args.file.display()))?;
 
-    let parser = Parser::new(&pgm);
+            let parser = Parser::new(&pgm);
 
-    let ast = parser
-        .parse()
-        .with_context(|| format!("parse error in {filename}:"))?;
+            let ast = parser
+                .parse()
+                .with_context(|| format!("parse error in {}:", run_args.file.display()))?;
 
-    let interpreter = Interpreter::new(&ast);
+            let interpreter = Interpreter::new(&ast);
 
-    let results = interpreter
-        .interpret(
-            url.parse()
-                .with_context(|| format!("Couldn't parse `{url}` into a URL"))?,
-        )
-        .await?;
+            let results = interpreter.interpret(run_args.url).await?;
 
-    println!("{}", serde_json::to_string_pretty(&results)?);
+            println!("{}", serde_json::to_string_pretty(&results)?);
+        }
+        // TODO: investigate if the (None, None) branch is reachable (I think it isn't)
+        (Some(Mode::Repl), None) | (None, None) => {
+            todo!()
+        }
+        (Some(_), Some(_)) => {
+            unreachable!(
+                "This should be impossible to reach with clap's `args_conflicts_with_subcommands`.
+                If you see this error message, please file a GitHub issue with the arguments
+                you provided to `scrapelect`."
+            )
+        }
+    }
 
     Ok(())
 }