Add url commands (#2274)

* scheme
* path
* query
* host

Cargo.lock

@@ -2584,6 +2584,7 @@ dependencies = [
  "typetag",
  "umask",
  "unicode-xid",
+ "url",
  "users",
  "uuid",
  "which",

crates/nu-cli/Cargo.toml

@@ -34,7 +34,7 @@ codespan-reporting = "0.9.5"
 csv = "1.1"
 ctrlc = {version = "3.1.4", optional = true}
 derive-new = "0.5.8"
-directories = {version = "2.0.2", optional = true}
+directories = {version = "2.0.2", optional = true} 
 dirs = {version = "2.0.2", optional = true}
 dunce = "1.0.1"
 eml-parser = "0.1.0"
@@ -94,6 +94,7 @@ rayon = "1.3.1"
 starship = {version = "0.43.0", optional = true}
 trash = {version = "1.0.1", optional = true}
 quick-xml = "0.18.1"
+url = {version = "2.1.1"}
 
 [target.'cfg(unix)'.dependencies]
 users = "0.10.0"

crates/nu-cli/src/cli.rs

@@ -420,6 +420,12 @@ pub fn create_default_context(
             whole_stream_command(PathExpand),
             whole_stream_command(PathExists),
             whole_stream_command(PathType),
+            // Url
+            whole_stream_command(UrlCommand),
+            whole_stream_command(UrlScheme),
+            whole_stream_command(UrlPath),
+            whole_stream_command(UrlHost),
+            whole_stream_command(UrlQuery),
         ]);
 
         #[cfg(feature = "clipboard")]

crates/nu-cli/src/commands.rs

@@ -115,6 +115,7 @@ pub(crate) mod to_yaml;
 pub(crate) mod trim;
 pub(crate) mod uniq;
 pub(crate) mod update;
+pub(crate) mod url_;
 pub(crate) mod version;
 pub(crate) mod what;
 pub(crate) mod where_;
@@ -247,6 +248,7 @@ pub(crate) use to_yaml::ToYAML;
 pub(crate) use touch::Touch;
 pub(crate) use trim::Trim;
 pub(crate) use uniq::Uniq;
+pub(crate) use url_::{UrlCommand, UrlHost, UrlPath, UrlQuery, UrlScheme};
 pub(crate) use version::Version;
 pub(crate) use what::What;
 pub(crate) use where_::Where;

crates/nu-cli/src/commands/url_/command.rs

@@ -0,0 +1,46 @@
+use crate::commands::WholeStreamCommand;
+use crate::prelude::*;
+use nu_errors::ShellError;
+use nu_protocol::{ReturnSuccess, Signature, UntaggedValue};
+
+pub struct Url;
+
+#[async_trait]
+impl WholeStreamCommand for Url {
+    fn name(&self) -> &str {
+        "url"
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::build("url")
+    }
+
+    fn usage(&self) -> &str {
+        "Apply url function"
+    }
+
+    async fn run(
+        &self,
+        _args: CommandArgs,
+        registry: &CommandRegistry,
+    ) -> Result<OutputStream, ShellError> {
+        let registry = registry.clone();
+
+        Ok(OutputStream::one(ReturnSuccess::value(
+            UntaggedValue::string(crate::commands::help::get_help(&Url, &registry))
+                .into_value(Tag::unknown()),
+        )))
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::Url;
+
+    #[test]
+    fn examples_work_as_expected() {
+        use crate::examples::test as test_examples;
+
+        test_examples(Url {})
+    }
+}

crates/nu-cli/src/commands/url_/host.rs

@@ -0,0 +1,58 @@
+use url::Url;
+
+use super::{operate, DefaultArguments};
+use crate::commands::WholeStreamCommand;
+use crate::prelude::*;
+use nu_errors::ShellError;
+use nu_protocol::{Signature, SyntaxShape, Value};
+
+pub struct UrlHost;
+
+#[async_trait]
+impl WholeStreamCommand for UrlHost {
+    fn name(&self) -> &str {
+        "url host"
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::build("url host")
+            .rest(SyntaxShape::ColumnPath, "optionally operate by column path")
+    }
+
+    fn usage(&self) -> &str {
+        "gets the host of a url"
+    }
+
+    async fn run(
+        &self,
+        args: CommandArgs,
+        registry: &CommandRegistry,
+    ) -> Result<OutputStream, ShellError> {
+        let (DefaultArguments { rest }, input) = args.process(&registry).await?;
+        operate(input, rest, &host).await
+    }
+
+    fn examples(&self) -> Vec<Example> {
+        vec![Example {
+            description: "Get host of a url",
+            example: "echo 'http://www.example.com/foo/bar' | url host",
+            result: Some(vec![Value::from("www.example.com")]),
+        }]
+    }
+}
+
+fn host(url: &Url) -> &str {
+    url.host_str().unwrap_or("")
+}
+
+#[cfg(test)]
+mod tests {
+    use super::UrlHost;
+
+    #[test]
+    fn examples_work_as_expected() {
+        use crate::examples::test as test_examples;
+
+        test_examples(UrlHost {})
+    }
+}

crates/nu-cli/src/commands/url_/mod.rs

@@ -0,0 +1,72 @@
+mod command;
+mod host;
+mod path;
+mod query;
+mod scheme;
+
+use crate::prelude::*;
+use nu_errors::ShellError;
+use nu_protocol::{ColumnPath, Primitive, ReturnSuccess, ShellTypeName, UntaggedValue, Value};
+use url::Url;
+
+pub use command::Url as UrlCommand;
+pub use host::UrlHost;
+pub use path::UrlPath;
+pub use query::UrlQuery;
+pub use scheme::UrlScheme;
+
+#[derive(Deserialize)]
+struct DefaultArguments {
+    rest: Vec<ColumnPath>,
+}
+
+fn handle_value<F>(action: &F, v: &Value) -> Result<Value, ShellError>
+where
+    F: Fn(&Url) -> &str + Send + 'static,
+{
+    let a = |url| UntaggedValue::string(action(url));
+    let v = match &v.value {
+        UntaggedValue::Primitive(Primitive::String(s))
+        | UntaggedValue::Primitive(Primitive::Line(s)) => match Url::parse(s) {
+            Ok(url) => a(&url).into_value(v.tag()),
+            Err(_) => UntaggedValue::string("").into_value(v.tag()),
+        },
+        other => {
+            let got = format!("got {}", other.type_name());
+            return Err(ShellError::labeled_error(
+                "value is not a string",
+                got,
+                v.tag().span,
+            ));
+        }
+    };
+    Ok(v)
+}
+
+async fn operate<F>(
+    input: crate::InputStream,
+    paths: Vec<ColumnPath>,
+    action: &'static F,
+) -> Result<OutputStream, ShellError>
+where
+    F: Fn(&Url) -> &str + Send + Sync + 'static,
+{
+    Ok(input
+        .map(move |v| {
+            if paths.is_empty() {
+                ReturnSuccess::value(handle_value(&action, &v)?)
+            } else {
+                let mut ret = v;
+
+                for path in &paths {
+                    ret = ret.swap_data_by_column_path(
+                        path,
+                        Box::new(move |old| handle_value(&action, &old)),
+                    )?;
+                }
+
+                ReturnSuccess::value(ret)
+            }
+        })
+        .to_output_stream())
+}

crates/nu-cli/src/commands/url_/path.rs

@@ -0,0 +1,61 @@
+use url::Url;
+
+use super::{operate, DefaultArguments};
+use crate::commands::WholeStreamCommand;
+use crate::prelude::*;
+use nu_errors::ShellError;
+use nu_protocol::{Signature, SyntaxShape, Value};
+
+pub struct UrlPath;
+
+#[async_trait]
+impl WholeStreamCommand for UrlPath {
+    fn name(&self) -> &str {
+        "url path"
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::build("url path")
+            .rest(SyntaxShape::ColumnPath, "optionally operate by column path")
+    }
+
+    fn usage(&self) -> &str {
+        "gets the path of a url"
+    }
+
+    async fn run(
+        &self,
+        args: CommandArgs,
+        registry: &CommandRegistry,
+    ) -> Result<OutputStream, ShellError> {
+        let (DefaultArguments { rest }, input) = args.process(&registry).await?;
+        operate(input, rest, &Url::path).await
+    }
+
+    fn examples(&self) -> Vec<Example> {
+        vec![
+            Example {
+                description: "Get path of a url",
+                example: "echo 'http://www.example.com/foo/bar' | url path",
+                result: Some(vec![Value::from("/foo/bar")]),
+            },
+            Example {
+                description: "A trailing slash will be reflected in the path",
+                example: "echo 'http://www.example.com' | url path",
+                result: Some(vec![Value::from("/")]),
+            },
+        ]
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::UrlPath;
+
+    #[test]
+    fn examples_work_as_expected() {
+        use crate::examples::test as test_examples;
+
+        test_examples(UrlPath {})
+    }
+}

crates/nu-cli/src/commands/url_/query.rs

@@ -0,0 +1,65 @@
+use url::Url;
+
+use super::{operate, DefaultArguments};
+use crate::commands::WholeStreamCommand;
+use crate::prelude::*;
+use nu_errors::ShellError;
+use nu_protocol::{Signature, SyntaxShape, Value};
+
+pub struct UrlQuery;
+
+#[async_trait]
+impl WholeStreamCommand for UrlQuery {
+    fn name(&self) -> &str {
+        "url query"
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::build("url query")
+            .rest(SyntaxShape::ColumnPath, "optionally operate by column path")
+    }
+
+    fn usage(&self) -> &str {
+        "gets the query of a url"
+    }
+
+    async fn run(
+        &self,
+        args: CommandArgs,
+        registry: &CommandRegistry,
+    ) -> Result<OutputStream, ShellError> {
+        let (DefaultArguments { rest }, input) = args.process(&registry).await?;
+        operate(input, rest, &query).await
+    }
+
+    fn examples(&self) -> Vec<Example> {
+        vec![
+            Example {
+                description: "Get query of a url",
+                example: "echo 'http://www.example.com/?foo=bar&baz=quux' | url query",
+                result: Some(vec![Value::from("foo=bar&baz=quux")]),
+            },
+            Example {
+                description: "No query gives the empty string",
+                example: "echo 'http://www.example.com/' | url query",
+                result: Some(vec![Value::from("")]),
+            },
+        ]
+    }
+}
+
+fn query(url: &Url) -> &str {
+    url.query().unwrap_or("")
+}
+
+#[cfg(test)]
+mod tests {
+    use super::UrlQuery;
+
+    #[test]
+    fn examples_work_as_expected() {
+        use crate::examples::test as test_examples;
+
+        test_examples(UrlQuery {})
+    }
+}

crates/nu-cli/src/commands/url_/scheme.rs

@@ -0,0 +1,60 @@
+use url::Url;
+
+use super::{operate, DefaultArguments};
+use crate::commands::WholeStreamCommand;
+use crate::prelude::*;
+use nu_errors::ShellError;
+use nu_protocol::{Signature, SyntaxShape, Value};
+
+pub struct UrlScheme;
+
+#[async_trait]
+impl WholeStreamCommand for UrlScheme {
+    fn name(&self) -> &str {
+        "url scheme"
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::build("url scheme").rest(SyntaxShape::ColumnPath, "optionally operate by path")
+    }
+
+    fn usage(&self) -> &str {
+        "gets the scheme (eg http, file) of a url"
+    }
+
+    async fn run(
+        &self,
+        args: CommandArgs,
+        registry: &CommandRegistry,
+    ) -> Result<OutputStream, ShellError> {
+        let (DefaultArguments { rest }, input) = args.process(&registry).await?;
+        operate(input, rest, &Url::scheme).await
+    }
+
+    fn examples(&self) -> Vec<Example> {
+        vec![
+            Example {
+                description: "Get scheme of a url",
+                example: "echo 'http://www.example.com' | url scheme",
+                result: Some(vec![Value::from("http")]),
+            },
+            Example {
+                description: "You get an empty string if there is no scheme",
+                example: "echo 'test' | url scheme",
+                result: Some(vec![Value::from("")]),
+            },
+        ]
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::UrlScheme;
+
+    #[test]
+    fn examples_work_as_expected() {
+        use crate::examples::test as test_examples;
+
+        test_examples(UrlScheme {})
+    }
+}