add -q for JMESPath query on input data

Author: arraypad

Cargo.lock

@@ -21,6 +21,7 @@ exitcode = "1.1.2"
 failure = "0.1.6"
 handlebars = "2.0.2"
 indexmap = "1.3.0"
+jmespath = "0.2.0"
 log = "0.4.0"
 rayon = "1.2.0"
 serde = "1.0.102"

Cargo.toml

@@ -53,14 +53,14 @@ Homer Simpson <homer@example.com>
 
 ```
 
-#### Ping a list of EC2 instances returned from the AWS CLI.
+#### Query input data
 
-Since _each_ reads from stdin by default it plays nicely with other CLI tools including _jq_.
+You can use arbitrary [JMESPath](http://jmespath.org/) queries to extract rows from your input data:
 
 ```sh
-aws ec2 describe-instances | \
-	jq '.Instances' | \
-	each ping -c 1 {{PublicIpAddress}}
+curl https://cat-fact.herokuapp.com/facts |
+	each -f json -q 'all[?upvotes > `4`]' -- \
+	echo {{user.name.first}}: {{text}}
 ```
 
 #### Supply stdin to each command

README.md

@@ -126,7 +126,7 @@ impl Format for Csv {
 		Ok(false)
 	}
 
-	fn parse(&self, input: &mut dyn Read) -> Result<Vec<serde_json::Value>, Error> {
+	fn parse(&self, input: &mut dyn Read) -> Result<serde_json::Value, Error> {
 		let mut reader = self.reader_builder().from_reader(input);
 		let mut it = reader.records();
 
@@ -161,7 +161,7 @@ impl Format for Csv {
 			values.push(map.into());
 		}
 
-		Ok(values)
+		Ok(values.into())
 	}
 
 	fn write(&self, values: Vec<serde_json::Value>) -> Result<(), Error> {

src/formats/csv.rs

@@ -26,7 +26,7 @@ impl Format for Json {
 		Ok(header[0] as char == '[')
 	}
 
-	fn parse(&self, input: &mut dyn Read) -> Result<Vec<serde_json::Value>, Error> {
+	fn parse(&self, input: &mut dyn Read) -> Result<serde_json::Value, Error> {
 		// read to string first - see https://github.com/serde-rs/json/issues/160
 		let mut buffer = String::new();
 		input.read_to_string(&mut buffer)?;

src/formats/json.rs

@@ -16,7 +16,7 @@ pub trait Format {
 	fn set_arguments(&mut self, matches: &clap::ArgMatches) -> Result<(), Error>;
 	fn get_extensions(&self) -> &'static [&'static str];
 	fn is_valid_header(&self, header: &[u8]) -> Result<bool, Error>;
-	fn parse(&self, input: &mut dyn Read) -> Result<Vec<serde_json::Value>, Error>;
+	fn parse(&self, input: &mut dyn Read) -> Result<serde_json::Value, Error>;
 	fn write(&self, values: Vec<serde_json::Value>) -> Result<(), Error>;
 }
 

src/formats/mod.rs

@@ -45,6 +45,14 @@ fn main() {
 				.takes_value(true)
 				.possible_values(format_ids.as_slice()),
 		)
+		.arg(
+			Arg::with_name("query")
+				.short("q")
+				.long("query")
+				.value_name("QUERY")
+				.help("JMES query to apply to each input file")
+				.takes_value(true),
+		)
 		.arg(
 			Arg::with_name("output-format")
 				.short("F")
@@ -233,7 +241,7 @@ fn run(
 			},
 		};
 
-		let values = match format.parse(reader) {
+		let mut values = match format.parse(reader) {
 			Ok(values) => values,
 			Err(e) => {
 				return Err(EachError::Data {
@@ -242,9 +250,27 @@ fn run(
 			}
 		};
 
+		if let Some(query_str) = arg_matches.value_of("query") {
+			let query = jmespath::compile(&query_str).map_err(|e| EachError::Usage {
+				message: format!("Invalid JMES query: {}", e),
+			})?;
+
+			let query_result = query.search(values).map_err(|e| EachError::Data {
+				message: format!("Error evaluating JMES query: {}", e),
+			})?;
+
+			values = serde_json::to_value(query_result).map_err(|e| EachError::Data {
+				message: format!("Error converting query result to JSON value: {}", e),
+			})?;
+		}
+
+		let vec_values = values.as_array().ok_or_else(|| EachError::Data {
+			message: "Input values are not an array".to_string(),
+		})?;
+
 		match action {
-			Some(ref action) => process(&values, &action)?,
-			None => output_values.extend_from_slice(&values),
+			Some(ref action) => process(&vec_values, &action)?,
+			None => output_values.extend_from_slice(&vec_values),
 		}
 	}
 

src/main.rs

@@ -153,4 +153,24 @@ Homer Simpson <homer@example.com>
 			)
 			.unwrap();
 	}
+
+	#[test]
+	fn jmes_query() {
+		Assert::main_binary()
+			.with_args(&[
+				"-i",
+				PEOPLE_JSON_PATH,
+				"-q",
+				r#"[?starts_with(name, `"Bart"`)]"#,
+				"--",
+				"echo",
+				"-n",
+				"{{email}}",
+			])
+			.succeeds()
+			.and()
+			.stdout()
+			.is("bart@example.com")
+			.unwrap();
+	}
 }