Added row groups data

pq2json/Cargo.toml

@@ -8,7 +8,7 @@ edition = "2018"
 clap = "2"
 parquet = { git = "https://github.com/rzheka/arrow.git", branch = "dev" }
 itertools = "0.8"
-serde = "1"
+serde = { version = "1", features = ["derive"] }
 serde_json = "1"
 num-bigint = "0.2"
 chrono = "0.4"

pq2json/src/main.rs

@@ -101,6 +101,13 @@ fn main() {
                 .help("Print Kusto schema")
                 .takes_value(false)
                 .required(false),
+        )
+		.arg(
+            Arg::with_name("rowgroups")
+                .long("rowgroups")
+                .help("Print Row Groups Metadata")
+                .takes_value(false)
+                .required(false),
         )
         .arg(
             Arg::with_name("INPUT")
@@ -142,6 +149,8 @@ fn main() {
         schema::print_schema(input)
     } else if matches.is_present("cslschema") {
         schema::print_csl_schema(input)
+    } else if matches.is_present("rowgroups") {
+        schema::print_row_groups_metadata(input)
     } else {
         converter::convert(&settings, input, output)
     };

pq2json/src/schema.rs

@@ -8,6 +8,7 @@ use parquet::file::reader::{FileReader, SerializedFileReader};
 use parquet::schema::printer::{print_file_metadata, print_parquet_metadata};
 use parquet::schema::types::Type;
 use serde_json::Value;
+use serde::Serialize;
 
 /// Prints Parquet file schema information
 ///
@@ -107,3 +108,48 @@ fn field_csl_schema(field_type: &Type) -> (&str, &str) {
         Type::GroupType { ref basic_info, .. } => (basic_info.name(), "dynamic"),
     }
 }
+
+/// Prints limited row groups metadata of a specified Parquet file as JSON,
+/// for each row group its size in bytes and the number of rows.
+///
+/// Arguments:
+///
+/// * `input_file` - Parquet file path
+///
+pub fn print_row_groups_metadata(input_file: &str) -> Result<(), Box<dyn Error>> {
+    let file = File::open(&Path::new(input_file))?;
+    let reader = SerializedFileReader::new(file)?;
+    let row_groups_count = reader.metadata().num_row_groups();
+    let row_groups = Value::Array(
+        reader
+            .metadata()
+            .row_groups()
+            .iter()
+            .map(|row_group_metadata| {
+                let mut map = serde_json::Map::with_capacity(2);
+                map.insert(
+                    String::from("numberOfRows"),
+                    Value::String(row_group_metadata.num_rows().to_string()),
+                );
+                map.insert(
+                    String::from("totalByteSize"),
+                    Value::String(row_group_metadata.total_byte_size().to_string()),
+                );
+                Value::Object(map)
+            })
+            .collect_vec(),
+    );
+
+    let row_groups_object = RowGroups {
+        count: row_groups_count,
+        row_groups,
+    };
+    println!("{}", serde_json::to_string(&row_groups_object)?);
+    Ok(())
+}
+
+#[derive(Serialize)]
+pub struct RowGroups {
+    pub count: usize,
+    pub row_groups: Value,
+}