Attempt to guess the content type of a file when opening with --raw (#13521)

# Description
Attempt to guess the content type of a file when opening with --raw and
set it in the pipeline metadata.

<img width="644" alt="Screenshot 2024-08-02 at 11 30 10"
src="https://github.com/user-attachments/assets/071f0967-c4dd-405a-b8c8-f7aa073efa98">


# User-Facing Changes
- Content of files can be directly piped into commands like `http post`
with the content type set appropriately when using `--raw`.

crates/nu-command/src/filesystem/open.rs

@@ -146,11 +146,19 @@ impl Command for Open {
                         }
                     };
 
+                    let content_type = if raw {
+                        path.extension()
+                            .map(|ext| ext.to_string_lossy().to_string())
+                            .and_then(|ref s| detect_content_type(s))
+                    } else {
+                        None
+                    };
+
                     let stream = PipelineData::ByteStream(
                         ByteStream::file(file, call_span, engine_state.signals().clone()),
                         Some(PipelineMetadata {
                             data_source: DataSource::FilePath(path.to_path_buf()),
-                            content_type: None,
+                            content_type,
                         }),
                     );
 
@@ -268,3 +276,22 @@ fn extract_extensions(filename: &str) -> Vec<String> {
 
     extensions
 }
+
+fn detect_content_type(extension: &str) -> Option<String> {
+    // This will allow the overriding of metadata to be consistent with
+    // the content type
+    match extension {
+        // Per RFC-9512, application/yaml should be used
+        "yaml" | "yml" => Some("application/yaml".to_string()),
+        _ => mime_guess::from_ext(extension)
+            .first()
+            .map(|mime| mime.to_string()),
+    }
+}
+
+#[cfg(test)]
+mod test {
+
+    #[test]
+    fn test_content_type() {}
+}

crates/nu-command/src/formats/to/toml.rs

@@ -1,6 +1,6 @@
 use chrono::{DateTime, Datelike, FixedOffset, Timelike};
 use nu_engine::command_prelude::*;
-use nu_protocol::ast::PathMember;
+use nu_protocol::{ast::PathMember, PipelineMetadata};
 
 #[derive(Clone)]
 pub struct ToToml;
@@ -100,9 +100,18 @@ fn toml_into_pipeline_data(
     toml_value: &toml::Value,
     value_type: Type,
     span: Span,
+    metadata: Option<PipelineMetadata>,
 ) -> Result<PipelineData, ShellError> {
+    let new_md = Some(
+        metadata
+            .unwrap_or_default()
+            .with_content_type(Some("text/x-toml".into())),
+    );
+
     match toml::to_string_pretty(&toml_value) {
-        Ok(serde_toml_string) => Ok(Value::string(serde_toml_string, span).into_pipeline_data()),
+        Ok(serde_toml_string) => {
+            Ok(Value::string(serde_toml_string, span).into_pipeline_data_with_metadata(new_md))
+        }
         _ => Ok(Value::error(
             ShellError::CantConvert {
                 to_type: "TOML".into(),
@@ -112,7 +121,7 @@ fn toml_into_pipeline_data(
             },
             span,
         )
-        .into_pipeline_data()),
+        .into_pipeline_data_with_metadata(new_md)),
     }
 }
 
@@ -139,6 +148,7 @@ fn to_toml(
     input: PipelineData,
     span: Span,
 ) -> Result<PipelineData, ShellError> {
+    let metadata = input.metadata();
     let value = input.into_value(span)?;
 
     let toml_value = value_to_toml_value(engine_state, &value, span)?;
@@ -148,10 +158,11 @@ fn to_toml(
                 vec.iter().next().expect("this should never trigger"),
                 value.get_type(),
                 span,
+                metadata,
             ),
-            _ => toml_into_pipeline_data(&toml_value, value.get_type(), span),
+            _ => toml_into_pipeline_data(&toml_value, value.get_type(), span, metadata),
         },
-        _ => toml_into_pipeline_data(&toml_value, value.get_type(), span),
+        _ => toml_into_pipeline_data(&toml_value, value.get_type(), span, metadata),
     }
 }
 

crates/nu-command/src/formats/to/yaml.rs

@@ -98,6 +98,7 @@ fn to_yaml(input: PipelineData, head: Span) -> Result<PipelineData, ShellError>
     let metadata = input
         .metadata()
         .unwrap_or_default()
+        // Per RFC-9512, application/yaml should be used
         .with_content_type(Some("application/yaml".into()));
     let value = input.into_value(head)?;
 

crates/nu-command/tests/commands/open.rs

@@ -379,3 +379,27 @@ fn open_files_inside_glob_metachars_dir() {
         assert!(actual.out.contains("hello"));
     });
 }
+
+#[test]
+fn test_content_types_with_open_raw() {
+    Playground::setup("open_files_content_type_test", |dirs, _| {
+        let result = nu!(cwd: dirs.formats(), "open --raw random_numbers.csv | metadata");
+        assert!(result.out.contains("text/csv"));
+        let result = nu!(cwd: dirs.formats(), "open --raw caco3_plastics.tsv | metadata");
+        assert!(result.out.contains("text/tab-separated-values"));
+        let result = nu!(cwd: dirs.formats(), "open --raw sample-simple.json | metadata");
+        assert!(result.out.contains("application/json"));
+        let result = nu!(cwd: dirs.formats(), "open --raw sample.ini | metadata");
+        assert!(result.out.contains("text/plain"));
+        let result = nu!(cwd: dirs.formats(), "open --raw sample_data.xlsx | metadata");
+        assert!(result.out.contains("vnd.openxmlformats-officedocument"));
+        let result = nu!(cwd: dirs.formats(), "open --raw sample_def.nu | metadata");
+        assert!(!result.out.contains("content_type"));
+        let result = nu!(cwd: dirs.formats(), "open --raw sample.eml | metadata");
+        assert!(result.out.contains("message/rfc822"));
+        let result = nu!(cwd: dirs.formats(), "open --raw cargo_sample.toml | metadata");
+        assert!(result.out.contains("text/x-toml"));
+        let result = nu!(cwd: dirs.formats(), "open --raw appveyor.yml | metadata");
+        assert!(result.out.contains("application/yaml"));
+    })
+}

crates/nu-test-support/src/lib.rs

@@ -9,6 +9,7 @@ use std::process::ExitStatus;
 // Needs to be reexported for `nu!` macro
 pub use nu_path;
 
+#[derive(Debug)]
 pub struct Outcome {
     pub out: String,
     pub err: String,