fix(task): handle node_modules/.bin directory with byonm (#21386)

A bit hacky, but it works. Essentially, this will check for all the
scripts in the node_modules/.bin directory then force them to run with
Deno via deno_task_shell.

1 files changed, 149 insertions, 3 deletions

diff --git a/cli/tools/task.rs b/cli/tools/task.rs
index d929dc666..78d09f0c7 100644
--- a/cli/tools/task.rs
+++ b/cli/tools/task.rs
@@ -5,6 +5,8 @@ use crate::args::Flags;
 use crate::args::TaskFlags;
 use crate::colors;
 use crate::factory::CliFactory;
+use crate::npm::CliNpmResolver;
+use crate::npm::InnerCliNpmResolverRef;
 use crate::npm::ManagedCliNpmResolver;
 use crate::util::fs::canonicalize_path;
 use deno_core::anyhow::bail;
@@ -18,6 +20,8 @@ use deno_task_shell::ExecuteResult;
 use deno_task_shell::ShellCommand;
 use deno_task_shell::ShellCommandContext;
 use indexmap::IndexMap;
+use lazy_regex::Lazy;
+use regex::Regex;
 use std::collections::HashMap;
 use std::path::Path;
 use std::path::PathBuf;
@@ -115,11 +119,15 @@ pub async fn execute_script(
         output_task(&task_name, &script);
         let seq_list = deno_task_shell::parser::parse(&script)
           .with_context(|| format!("Error parsing script '{task_name}'."))?;
-        let npx_commands = match npm_resolver.as_managed() {
-          Some(npm_resolver) => {
+        let npx_commands = match npm_resolver.as_inner() {
+          InnerCliNpmResolverRef::Managed(npm_resolver) => {
             resolve_npm_commands(npm_resolver, node_resolver)?
           }
-          None => Default::default(),
+          InnerCliNpmResolverRef::Byonm(npm_resolver) => {
+            let node_modules_dir =
+              npm_resolver.root_node_modules_path().unwrap();
+            resolve_npm_commands_from_bin_dir(node_modules_dir)?
+          }
         };
         let env_vars = match npm_resolver.root_node_modules_path() {
           Some(dir_path) => collect_env_vars_with_node_modules_dir(dir_path),
@@ -294,6 +302,113 @@ impl ShellCommand for NpmPackageBinCommand {
   }
 }
 
+/// Runs a module in the node_modules folder.
+#[derive(Clone)]
+struct NodeModulesFileRunCommand {
+  command_name: String,
+  path: PathBuf,
+}
+
+impl ShellCommand for NodeModulesFileRunCommand {
+  fn execute(
+    &self,
+    mut context: ShellCommandContext,
+  ) -> LocalBoxFuture<'static, ExecuteResult> {
+    let mut args = vec![
+      "run".to_string(),
+      "--ext=js".to_string(),
+      "-A".to_string(),
+      self.path.to_string_lossy().to_string(),
+    ];
+    args.extend(context.args);
+    let executable_command =
+      deno_task_shell::ExecutableCommand::new("deno".to_string());
+    // set this environment variable so that the launched process knows the npm command name
+    context
+      .state
+      .apply_env_var("DENO_INTERNAL_NPM_CMD_NAME", &self.command_name);
+    executable_command.execute(ShellCommandContext { args, ..context })
+  }
+}
+
+fn resolve_npm_commands_from_bin_dir(
+  node_modules_dir: &Path,
+) -> Result<HashMap<String, Rc<dyn ShellCommand>>, AnyError> {
+  let mut result = HashMap::<String, Rc<dyn ShellCommand>>::new();
+  let bin_dir = node_modules_dir.join(".bin");
+  log::debug!("Resolving commands in '{}'.", bin_dir.display());
+  match std::fs::read_dir(&bin_dir) {
+    Ok(entries) => {
+      for entry in entries {
+        let Ok(entry) = entry else {
+          continue;
+        };
+        if let Some(command) = resolve_bin_dir_entry_command(entry) {
+          result.insert(command.command_name.clone(), Rc::new(command));
+        }
+      }
+    }
+    Err(err) => {
+      log::debug!("Failed read_dir for '{}': {:#}", bin_dir.display(), err);
+    }
+  }
+  Ok(result)
+}
+
+fn resolve_bin_dir_entry_command(
+  entry: std::fs::DirEntry,
+) -> Option<NodeModulesFileRunCommand> {
+  if entry.path().extension().is_some() {
+    return None; // only look at files without extensions (even on Windows)
+  }
+  let file_type = entry.file_type().ok()?;
+  let path = if file_type.is_file() {
+    entry.path()
+  } else if file_type.is_symlink() {
+    entry.path().canonicalize().ok()?
+  } else {
+    return None;
+  };
+  let text = std::fs::read_to_string(&path).ok()?;
+  let command_name = entry.file_name().to_string_lossy().to_string();
+  if let Some(path) = resolve_execution_path_from_npx_shim(path, &text) {
+    log::debug!(
+      "Resolved npx command '{}' to '{}'.",
+      command_name,
+      path.display()
+    );
+    Some(NodeModulesFileRunCommand { command_name, path })
+  } else {
+    log::debug!("Failed resolving npx command '{}'.", command_name);
+    None
+  }
+}
+
+/// This is not ideal, but it works ok because it allows us to bypass
+/// the shebang and execute the script directly with Deno.
+fn resolve_execution_path_from_npx_shim(
+  file_path: PathBuf,
+  text: &str,
+) -> Option<PathBuf> {
+  static SCRIPT_PATH_RE: Lazy<Regex> =
+    lazy_regex::lazy_regex!(r#""\$basedir\/([^"]+)" "\$@""#);
+
+  if text.starts_with("#!/usr/bin/env node") {
+    // launch this file itself because it's a JS file
+    Some(file_path)
+  } else {
+    // Search for...
+    // > "$basedir/../next/dist/bin/next" "$@"
+    // ...which is what it will look like on Windows
+    SCRIPT_PATH_RE
+      .captures(text)
+      .and_then(|c| c.get(1))
+      .map(|relative_path| {
+        file_path.parent().unwrap().join(relative_path.as_str())
+      })
+  }
+}
+
 fn resolve_npm_commands(
   npm_resolver: &ManagedCliNpmResolver,
   node_resolver: &NodeResolver,
@@ -351,4 +466,35 @@ mod test {
       HashMap::from([("PATH".to_string(), "/example".to_string())])
     );
   }
+
+  #[test]
+  fn test_resolve_execution_path_from_npx_shim() {
+    // example shim on unix
+    let unix_shim = r#"#!/usr/bin/env node
+"use strict";
+console.log('Hi!');
+"#;
+    let path = PathBuf::from("/node_modules/.bin/example");
+    assert_eq!(
+      resolve_execution_path_from_npx_shim(path.clone(), unix_shim).unwrap(),
+      path
+    );
+    // example shim on windows
+    let windows_shim = r#"#!/bin/sh
+basedir=$(dirname "$(echo "$0" | sed -e 's,\\,/,g')")
+
+case `uname` in
+    *CYGWIN*|*MINGW*|*MSYS*) basedir=`cygpath -w "$basedir"`;;
+esac
+
+if [ -x "$basedir/node" ]; then
+  exec "$basedir/node"  "$basedir/../example/bin/example" "$@"
+else 
+  exec node  "$basedir/../example/bin/example" "$@"
+fi"#;
+    assert_eq!(
+      resolve_execution_path_from_npx_shim(path.clone(), windows_shim).unwrap(),
+      path.parent().unwrap().join("../example/bin/example")
+    );
+  }
 }