Add basic support for md5 hashing strings and binary data (#3197)

Cargo.lock

@@ -2796,6 +2796,12 @@ version = "0.6.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7e6bcd6433cff03a4bfc3d9834d504467db1f1cf6d0ea765d37d330249ed629d"
 
+[[package]]
+name = "md5"
+version = "0.7.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "490cc448043f947bae3cbee9c203358d62dbee0db12107a74be5c30ccfd09771"
+
 [[package]]
 name = "memchr"
 version = "2.3.4"
@@ -3265,6 +3271,7 @@ dependencies = [
  "itertools",
  "lazy_static 1.4.0",
  "log 0.4.14",
+ "md5 0.7.0",
  "meval",
  "minus",
  "nu-ansi-term",

crates/nu-command/Cargo.toml

@@ -62,6 +62,7 @@ indexmap = { version = "1.6.1", features = ["serde-1"] }
 itertools = "0.10.0"
 lazy_static = "1.*"
 log = "0.4.14"
+md5 = "0.7.0"
 meval = "0.2.0"
 minus = { version = "3.3.0", optional = true, features = ["async_std_lib", "search"] }
 num-bigint = { version = "0.3.1", features = ["serde"] }

crates/nu-command/src/commands.rs

@@ -202,7 +202,7 @@ pub(crate) use from_yaml::FromYML;
 pub(crate) use get::Command as Get;
 pub(crate) use group_by::Command as GroupBy;
 pub(crate) use group_by_date::GroupByDate;
-pub(crate) use hash_::{Hash, HashBase64};
+pub(crate) use hash_::{Hash, HashBase64, HashMd5};
 pub(crate) use headers::Headers;
 pub(crate) use help::Help;
 pub(crate) use histogram::Histogram;

crates/nu-command/src/commands/default_context.rs

@@ -74,6 +74,7 @@ pub fn create_default_context(interactive: bool) -> Result<EvaluationContext, Bo
             // Text manipulation
             whole_stream_command(Hash),
             whole_stream_command(HashBase64),
+            whole_stream_command(HashMd5),
             whole_stream_command(Split),
             whole_stream_command(SplitColumn),
             whole_stream_command(SplitRow),

crates/nu-command/src/commands/hash_/md5_.rs

@@ -0,0 +1,133 @@
+use crate::prelude::*;
+use nu_engine::WholeStreamCommand;
+use nu_errors::ShellError;
+use nu_protocol::ShellTypeName;
+use nu_protocol::{
+    ColumnPath, Primitive, ReturnSuccess, Signature, SyntaxShape, UntaggedValue, Value,
+};
+use nu_source::Tag;
+
+#[derive(Deserialize)]
+pub struct Arguments {
+    pub rest: Vec<ColumnPath>,
+}
+
+pub struct SubCommand;
+
+#[async_trait]
+impl WholeStreamCommand for SubCommand {
+    fn name(&self) -> &str {
+        "hash md5"
+    }
+
+    fn signature(&self) -> Signature {
+        Signature::build("hash md5").rest(
+            SyntaxShape::ColumnPath,
+            "optionally md5 encode data by column paths",
+        )
+    }
+
+    fn usage(&self) -> &str {
+        "md5 encode a value"
+    }
+
+    async fn run(&self, args: CommandArgs) -> Result<OutputStream, ShellError> {
+        operate(args).await
+    }
+
+    fn examples(&self) -> Vec<Example> {
+        vec![
+            Example {
+                description: "md5 encode a string",
+                example: "echo 'abcdefghijklmnopqrstuvwxyz' | hash md5",
+                result: Some(vec![UntaggedValue::string(
+                    "c3fcd3d76192e4007dfb496cca67e13b",
+                )
+                .into_untagged_value()]),
+            },
+            Example {
+                description: "md5 encode a file",
+                example: "open ./nu_0_24_1_windows.zip | hash md5",
+                result: Some(vec![UntaggedValue::string(
+                    "dcf30f2836a1a99fc55cf72e28272606",
+                )
+                .into_untagged_value()]),
+            },
+        ]
+    }
+}
+
+async fn operate(args: CommandArgs) -> Result<OutputStream, ShellError> {
+    let (Arguments { rest }, input) = args.process().await?;
+
+    let column_paths: Vec<_> = rest;
+
+    Ok(input
+        .map(move |v| {
+            if column_paths.is_empty() {
+                ReturnSuccess::value(action(&v, v.tag())?)
+            } else {
+                let mut ret = v;
+
+                for path in &column_paths {
+                    ret = ret.swap_data_by_column_path(
+                        path,
+                        Box::new(move |old| action(old, old.tag())),
+                    )?;
+                }
+
+                ReturnSuccess::value(ret)
+            }
+        })
+        .to_output_stream())
+}
+
+fn action(input: &Value, tag: impl Into<Tag>) -> Result<Value, ShellError> {
+    match &input.value {
+        UntaggedValue::Primitive(Primitive::String(s)) => {
+            let md5_digest = md5::compute(s.as_bytes());
+            Ok(UntaggedValue::string(&format!("{:x}", md5_digest)).into_value(tag))
+        }
+        UntaggedValue::Primitive(Primitive::Binary(bytes)) => {
+            let md5_digest = md5::compute(bytes);
+            Ok(UntaggedValue::string(&format!("{:x}", md5_digest)).into_value(tag))
+        }
+        other => {
+            let got = format!("got {}", other.type_name());
+            Err(ShellError::labeled_error(
+                "value is not supported for hashing as md5",
+                got,
+                tag.into().span,
+            ))
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::action;
+    use nu_protocol::{Primitive, UntaggedValue};
+    use nu_source::Tag;
+    use nu_test_support::value::string;
+
+    #[test]
+    fn md5_encode_string() {
+        let word = string("abcdefghijklmnopqrstuvwxyz");
+        let expected =
+            UntaggedValue::string("c3fcd3d76192e4007dfb496cca67e13b").into_untagged_value();
+
+        let actual = action(&word, Tag::unknown()).unwrap();
+        assert_eq!(actual, expected);
+    }
+
+    #[test]
+    fn md5_encode_bytes() {
+        let bytes = vec![0xC0, 0xFF, 0xEE];
+        let binary = UntaggedValue::Primitive(Primitive::Binary(bytes)).into_untagged_value();
+        let expected =
+            UntaggedValue::string("5f80e231382769b0102b1164cf722d83").into_untagged_value();
+
+        let actual = action(&binary, Tag::unknown()).unwrap();
+        assert_eq!(actual, expected);
+    }
+}

crates/nu-command/src/commands/hash_/mod.rs

@@ -1,5 +1,7 @@
 mod base64_;
 mod command;
+mod md5_;
 
 pub use base64_::SubCommand as HashBase64;
 pub use command::Command as Hash;
+pub use md5_::SubCommand as HashMd5;

crates/nu-command/tests/commands/hash_/mod.rs

@@ -83,3 +83,16 @@ fn error_use_both_flags() {
         .err
         .contains("only one of --decode and --encode flags can be used"));
 }
+
+#[test]
+fn md5_works_with_file() {
+    let actual = nu!(
+        cwd: "tests/fixtures/formats", pipeline(
+        r#"
+        open sample.db | hash md5 
+        "#
+        )
+    );
+
+    assert_eq!(actual.out, "4de97601d232c427977ef11db396c951");
+}