activates optional trim in 'from csv' and 'from tsv' (#5326)

crates/nu-command/src/formats/from/csv.rs

@@ -1,4 +1,4 @@
-use super::delimited::from_delimited_data;
+use super::delimited::{from_delimited_data, trim_from_str};
 
 use nu_engine::CallExt;
 use nu_protocol::ast::Call;
@@ -26,6 +26,12 @@ impl Command for FromCsv {
                 "don't treat the first row as column names",
                 Some('n'),
             )
+            .named(
+                "trim",
+                SyntaxShape::String,
+                "drop leading and trailing whitespaces around headers names and/or field values",
+                Some('t'),
+            )
             .category(Category::Formats)
     }
 
@@ -65,6 +71,21 @@ impl Command for FromCsv {
                 example: "open data.txt | from csv --separator ';'",
                 result: None,
             },
+            Example {
+                description: "Convert semicolon-separated data to a table, dropping all possible whitespaces around header names and field values",
+                example: "open data.txt | from csv --trim all",
+                result: None,
+            },
+            Example {
+                description: "Convert semicolon-separated data to a table, dropping all possible whitespaces around header names",
+                example: "open data.txt | from csv --trim headers",
+                result: None,
+            },
+            Example {
+                description: "Convert semicolon-separated data to a table, dropping all possible whitespaces around field values",
+                example: "open data.txt | from csv --trim fields",
+                result: None,
+            },
         ]
     }
 }
@@ -79,6 +100,7 @@ fn from_csv(
 
     let noheaders = call.has_flag("noheaders");
     let separator: Option<Value> = call.get_flag(engine_state, stack, "separator")?;
+    let trim: Option<Value> = call.get_flag(engine_state, stack, "trim")?;
     let config = engine_state.get_config();
 
     let sep = match separator {
@@ -99,7 +121,9 @@ fn from_csv(
         _ => ',',
     };
 
-    from_delimited_data(noheaders, sep, input, name, config)
+    let trim = trim_from_str(trim)?;
+
+    from_delimited_data(noheaders, sep, trim, input, name, config)
 }
 
 #[cfg(test)]

crates/nu-command/src/formats/from/delimited.rs

@@ -1,15 +1,17 @@
-use csv::ReaderBuilder;
+use csv::{ReaderBuilder, Trim};
 use nu_protocol::{Config, IntoPipelineData, PipelineData, ShellError, Span, Value};
 
 fn from_delimited_string_to_value(
     s: String,
     noheaders: bool,
     separator: char,
+    trim: Trim,
     span: Span,
 ) -> Result<Value, csv::Error> {
     let mut reader = ReaderBuilder::new()
         .has_headers(!noheaders)
         .delimiter(separator as u8)
+        .trim(trim)
         .from_reader(s.as_bytes());
 
     let headers = if noheaders {
@@ -48,6 +50,7 @@ fn from_delimited_string_to_value(
 pub fn from_delimited_data(
     noheaders: bool,
     sep: char,
+    trim: Trim,
     input: PipelineData,
     name: Span,
     config: &Config,
@@ -55,8 +58,25 @@ pub fn from_delimited_data(
     let concat_string = input.collect_string("", config)?;
 
     Ok(
-        from_delimited_string_to_value(concat_string, noheaders, sep, name)
+        from_delimited_string_to_value(concat_string, noheaders, sep, trim, name)
             .map_err(|x| ShellError::DelimiterError(x.to_string(), name))?
             .into_pipeline_data(),
     )
 }
+
+pub fn trim_from_str(trim: Option<Value>) -> Result<Trim, ShellError> {
+    match trim {
+        Some(Value::String { val: item, span }) => match item.as_str() {
+            "all" => Ok(Trim::All),
+            "headers" => Ok(Trim::Headers),
+            "fields" => Ok(Trim::Fields),
+            "none" => Ok(Trim::None),
+            _ => Err(ShellError::UnsupportedInput(
+                "the only possible values for trim are 'all', 'headers', 'fields' and 'none'"
+                    .into(),
+                span,
+            )),
+        },
+        _ => Ok(Trim::None),
+    }
+}

crates/nu-command/src/formats/from/tsv.rs

@@ -1,8 +1,9 @@
-use super::delimited::from_delimited_data;
+use super::delimited::{from_delimited_data, trim_from_str};
 
+use nu_engine::CallExt;
 use nu_protocol::ast::Call;
 use nu_protocol::engine::{Command, EngineState, Stack};
-use nu_protocol::{Category, Config, Example, PipelineData, ShellError, Signature};
+use nu_protocol::{Category, Example, PipelineData, ShellError, Signature, SyntaxShape, Value};
 
 #[derive(Clone)]
 pub struct FromTsv;
@@ -19,6 +20,12 @@ impl Command for FromTsv {
                 "don't treat the first row as column names",
                 Some('n'),
             )
+            .named(
+                "trim",
+                SyntaxShape::String,
+                "drop leading and trailing whitespaces around headers names and/or field values",
+                Some('t'),
+            )
             .category(Category::Formats)
     }
 
@@ -29,12 +36,11 @@ impl Command for FromTsv {
     fn run(
         &self,
         engine_state: &EngineState,
-        _stack: &mut Stack,
+        stack: &mut Stack,
         call: &Call,
         input: PipelineData,
     ) -> Result<nu_protocol::PipelineData, ShellError> {
-        let config = engine_state.get_config();
-        from_tsv(call, input, config)
+        from_tsv(engine_state, stack, call, input)
     }
 
     fn examples(&self) -> Vec<Example> {
@@ -49,16 +55,45 @@ impl Command for FromTsv {
                 example: r#"echo $'a1(char tab)b1(char tab)c1(char nl)a2(char tab)b2(char tab)c2' | save tsv-data | open tsv-data | from tsv -n"#,
                 result: None,
             },
+            Example {
+                description: "Create a tsv file without header columns and open it, removing all unnecessary whitespaces",
+                example: r#"echo $'a1(char tab)b1(char tab)c1(char nl)a2(char tab)b2(char tab)c2' | save tsv-data | open tsv-data | from tsv --trim all"#,
+                result: None,
+            },
+            Example {
+                description: "Create a tsv file without header columns and open it, removing all unnecessary whitespaces in the header names",
+                example: r#"echo $'a1(char tab)b1(char tab)c1(char nl)a2(char tab)b2(char tab)c2' | save tsv-data | open tsv-data | from tsv --trim headers"#,
+                result: None,
+            },
+            Example {
+                description: "Create a tsv file without header columns and open it, removing all unnecessary whitespaces in the field values",
+                example: r#"echo $'a1(char tab)b1(char tab)c1(char nl)a2(char tab)b2(char tab)c2' | save tsv-data | open tsv-data | from tsv --trim fields"#,
+                result: None,
+            },
         ]
     }
 }
 
-fn from_tsv(call: &Call, input: PipelineData, config: &Config) -> Result<PipelineData, ShellError> {
+fn from_tsv(
+    engine_state: &EngineState,
+    stack: &mut Stack,
+    call: &Call,
+    input: PipelineData,
+) -> Result<PipelineData, ShellError> {
     let name = call.head;
 
     let noheaders = call.has_flag("noheaders");
+    let trim: Option<Value> = call.get_flag(engine_state, stack, "trim")?;
+    let trim = trim_from_str(trim)?;
 
-    from_delimited_data(noheaders, '\t', input, name, config)
+    from_delimited_data(
+        noheaders,
+        '\t',
+        trim,
+        input,
+        name,
+        engine_state.get_config(),
+    )
 }
 
 #[cfg(test)]