Add subcommands to support data stream analysis.

Cargo.toml

@@ -12,6 +12,7 @@ keywords = ["format", "parse", "encode"]
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
+lowcharts = "*"
 anyhow = "1.0"
 clap = { version = "4.0.17", features = ["cargo"] }
 colored = "2.0.0"

src/bin/ion/commands/beta/analyze/depth.rs

@@ -0,0 +1,60 @@
+use std::fs::File;
+use crate::commands::{IonCliCommand, WithIonCliArgument};
+use anyhow::{bail, Context, Result};
+use clap::{ArgMatches, Command};
+use ion_rs::ElementReader;
+use ion_rs::Element;
+use ion_rs::{IonReader, IonType, Reader, ReaderBuilder};
+
+
+pub struct DepthCommand;
+
+impl IonCliCommand for DepthCommand {
+    fn name(&self) -> &'static str {
+        "depth"
+    }
+
+    fn about(&self) -> &'static str {
+        "Prints the maximum depth of the input ion stream."
+    }
+
+    fn configure_args(&self, command: Command) -> Command {
+        command.with_input()
+    }
+
+    fn run(&self, _command_path: &mut Vec<String>, args: &ArgMatches) -> Result<()> {
+        if let Some(input_file_iter) = args.get_many::<String>("input") {
+            for input_file in input_file_iter {
+                let file = File::open(input_file)
+                    .with_context(|| format!("Could not open file '{}'", input_file))?;
+                let mut reader = ReaderBuilder::new().build(file)?;
+                get_depth(&mut reader)?;
+            }
+        } else {
+            bail!("this command does not yet support reading from STDIN")
+        };
+        Ok(())
+    }
+}
+
+fn get_depth(reader: &mut Reader) -> Result<()> {
+    let mut max_depth = 0;
+    for element in reader.elements(){
+        let unwrap_element = element.unwrap();
+        max_depth = calculate_depth(&unwrap_element, 0);
+    }
+    println!("The maximum depth is {}",max_depth);
+    Ok(())
+}
+
+fn calculate_depth(element: &Element, depth: usize) -> usize {
+    return if element.ion_type().is_container() {
+        if element.ion_type() == IonType::Struct {
+            element.as_struct().unwrap().iter().map(|(_field_name,e)| calculate_depth(e, depth + 1)).max().unwrap_or(depth)
+        } else {
+            element.as_sequence().unwrap().into_iter().map(|e| calculate_depth(e, depth + 1)).max().unwrap_or(depth)
+        }
+    } else {
+        depth
+    }
+}

src/bin/ion/commands/beta/analyze/mod.rs

@@ -0,0 +1,25 @@
+pub mod count;
+pub mod size;
+pub mod depth;
+
+
+use crate::commands::IonCliCommand;
+use crate::commands::beta::analyze::count::CountCommand;
+use crate::commands::beta::analyze::size::SizeCommand;
+use crate::commands::beta::analyze::depth::DepthCommand;
+
+
+pub struct AnalyzeNamespace;
+
+impl IonCliCommand for AnalyzeNamespace {
+    fn name(&self) -> &'static str {
+        "analyze"
+    }
+
+    fn about(&self) -> &'static str {
+        "The 'analyze' command is a namespace for commands used for Ion stream statistical analysis."
+    }
+    fn subcommands(&self) -> Vec<Box<dyn IonCliCommand>> {
+        vec![Box::new(CountCommand), Box::new(SizeCommand), Box::new(DepthCommand)]
+    }
+}

src/bin/ion/commands/beta/analyze/size.rs

@@ -0,0 +1,71 @@
+use std::fs::File;
+use crate::commands::{IonCliCommand, WithIonCliArgument};
+use anyhow::{bail, Context, Result};
+use clap::{ArgMatches, Command};
+use ion_rs::{IonReader, RawBinaryReader, SystemReader, SystemStreamItem};
+use memmap::MmapOptions;
+use lowcharts::plot;
+
+
+pub struct SizeCommand;
+
+impl IonCliCommand for SizeCommand {
+    fn name(&self) -> &'static str {
+        "size"
+    }
+
+    fn about(&self) -> &'static str {
+        "Prints the overall min, max, mean size of top-level values in the input stream."
+    }
+
+    fn configure_args(&self, command: Command) -> Command {
+        command.with_input()
+    }
+
+    fn run(&self, _command_path: &mut Vec<String>, args: &ArgMatches) -> Result<()> {
+        if let Some(input_file_names) = args.get_many::<String>("input") {
+            for input_file in input_file_names {
+                let file = File::open(input_file.as_str())
+                    .with_context(|| format!("Could not open file '{}'", &input_file))?;
+                let mmap = unsafe {
+                    MmapOptions::new()
+                        .map(&file)
+                        .with_context(|| format!("Could not mmap '{}'", input_file))?
+                };
+                // Treat the mmap as a byte array.
+                let ion_data: &[u8] = &mmap[..];
+                let raw_reader = RawBinaryReader::new(ion_data);
+                let mut system_reader = SystemReader::new(raw_reader);
+                size_analyze(&mut system_reader);
+            }
+        } else {
+            bail!("this command does not yet support reading from STDIN")
+        }
+        Ok(())
+    }
+}
+
+fn size_analyze(reader: &mut SystemReader<RawBinaryReader<&[u8]>>) -> Result<()> {
+    let mut vec: Vec<f64> = Vec::new();
+    loop {
+        match reader.next()? {
+            SystemStreamItem::Value(_) => {
+                let mut size = 0;
+                if reader.annotations_length() != None {
+                    size = reader.annotations_length().unwrap() +  reader.header_length() + reader.value_length();
+                } else {
+                    size = reader.header_length() + reader.value_length();
+                }
+                vec.push(size as f64);
+            },
+            SystemStreamItem::Nothing => break,
+            _ => {}
+        }
+    }
+// Plot a histogram of the above vector, with 4 buckets and a precision
+// chosen by library
+    let options = plot::HistogramOptions { intervals: 4, ..Default::default() };
+    let histogram = plot::Histogram::new(&vec, options);
+    print!("{}", histogram);
+    Ok(())
+}

src/bin/ion/commands/beta/mod.rs

@@ -1,4 +1,3 @@
-pub mod count;
 pub mod from;
 
 #[cfg(feature = "beta-subcommands")]
@@ -9,15 +8,17 @@ pub mod primitive;
 pub mod schema;
 pub mod symtab;
 pub mod to;
+pub mod analyze;
+
 
-use crate::commands::beta::count::CountCommand;
 use crate::commands::beta::from::FromNamespace;
 #[cfg(feature = "beta-subcommands")]
 use crate::commands::beta::generate::GenerateCommand;
 use crate::commands::beta::head::HeadCommand;
 use crate::commands::beta::inspect::InspectCommand;
 use crate::commands::beta::primitive::PrimitiveCommand;
 use crate::commands::beta::schema::SchemaNamespace;
+use crate::commands::beta::analyze::AnalyzeNamespace;
 use crate::commands::beta::symtab::SymtabNamespace;
 use crate::commands::beta::to::ToNamespace;
 use crate::commands::IonCliCommand;
@@ -35,10 +36,10 @@ impl IonCliCommand for BetaNamespace {
 
     fn subcommands(&self) -> Vec<Box<dyn IonCliCommand>> {
         vec![
-            Box::new(CountCommand),
             Box::new(InspectCommand),
             Box::new(PrimitiveCommand),
             Box::new(SchemaNamespace),
+            Box::new(AnalyzeNamespace),
             Box::new(HeadCommand),
             Box::new(FromNamespace),
             Box::new(ToNamespace),

src/bin/ion/commands/beta/symtab/count.rs

@@ -0,0 +1,61 @@
+use crate::commands::{IonCliCommand, WithIonCliArgument};
+use anyhow::{bail, Context, Result};
+use clap::{ArgMatches, Command};
+use ion_rs::*;
+use std::fs::File;
+use ion_rs::RawBinaryReader;
+use memmap::MmapOptions;
+
+pub struct SymbolTableCommand;
+
+impl IonCliCommand for SymbolTableCommand {
+    fn name(&self) -> &'static str {
+        "count"
+    }
+
+    fn about(&self) -> &'static str {
+        "Prints the number of symbol tables."
+    }
+
+    fn configure_args(&self, command: Command) -> Command {
+        command.with_input()
+    }
+
+    fn run(&self, _command_path: &mut Vec<String>, args: &ArgMatches) -> Result<()> {
+        if let Some(input_file_names) = args.get_many::<String>("input") {
+            // Input files were specified, run the converter on each of them in turn
+            for input_file in input_file_names {
+                let file = File::open(input_file.as_str())
+                    .with_context(|| format!("Could not open file '{}'", &input_file))?;
+                let mmap = unsafe {
+                    MmapOptions::new()
+                        .map(&file)
+                        .with_context(|| format!("Could not mmap '{}'", input_file))?
+                };
+                // Treat the mmap as a byte array.
+                let ion_data: &[u8] = &mmap[..];
+                let raw_reader = RawBinaryReader::new(ion_data);
+                let mut system_reader = SystemReader::new(raw_reader);
+                symbol_tables(&mut system_reader);
+            }
+        } else {
+            bail!("this command does not yet support reading from STDIN")
+        }
+        Ok(())
+    }
+}
+
+fn symbol_tables(reader: &mut SystemReader<RawBinaryReader<&[u8]>>) -> Result<()> {
+    let mut  count = 0;
+    loop {
+        match reader.next()? {
+            SystemStreamItem::SymbolTableValue(IonType::Struct) => {
+            count += 1;
+        },
+            SystemStreamItem::Nothing => break,
+            _ => {}
+        }
+    }
+    println!("The number of symbol tables is {} ", count);
+    Ok(())
+}

src/bin/ion/commands/beta/symtab/mod.rs

@@ -1,7 +1,11 @@
+use crate::commands::beta::symtab::count::SymbolTableCommand;
 use crate::commands::beta::symtab::filter::SymtabFilterCommand;
+use crate::commands::beta::symtab::symbol_count::SymbolNumberCommand;
 use crate::commands::IonCliCommand;
 
 pub mod filter;
+pub mod count;
+pub mod symbol_count;
 
 pub struct SymtabNamespace;
 
@@ -15,6 +19,6 @@ impl IonCliCommand for SymtabNamespace {
     }
 
     fn subcommands(&self) -> Vec<Box<dyn IonCliCommand>> {
-        vec![Box::new(SymtabFilterCommand)]
+        vec![Box::new(SymtabFilterCommand), Box::new(SymbolTableCommand), Box::new(SymbolNumberCommand)]
     }
 }

src/bin/ion/commands/beta/symtab/symbol_count.rs

@@ -0,0 +1,63 @@
+use crate::commands::{IonCliCommand, WithIonCliArgument};
+use anyhow::{bail, Context, Result};
+use clap::{ArgMatches, Command};
+use ion_rs::*;
+use std::fs::{File};
+use memmap::MmapOptions;
+
+
+pub struct SymbolNumberCommand;
+
+impl IonCliCommand for SymbolNumberCommand {
+    fn name(&self) -> &'static str {
+        "symbol_count"
+    }
+
+    fn about(&self) -> &'static str {
+        "Prints the number of symbols."
+    }
+
+    fn configure_args(&self, command: Command) -> Command {
+        command.with_input()
+    }
+
+    fn run(&self, _command_path: &mut Vec<String>, args: &ArgMatches) -> Result<()> {
+        if let Some(input_file_names) = args.get_many::<String>("input") {
+            // Input files were specified, run the converter on each of them in turn
+            for input_file in input_file_names {
+                let file = File::open(input_file.as_str())
+                    .with_context(|| format!("Could not open file '{}'", &input_file))?;
+                let mmap = unsafe {
+                    MmapOptions::new()
+                        .map(&file)
+                        .with_context(|| format!("Could not mmap '{}'", input_file))?
+                };
+                // Treat the mmap as a byte array.
+                let ion_data: &[u8] = &mmap[..];
+                let raw_reader = RawBinaryReader::new(ion_data);
+                let mut system_reader = SystemReader::new(raw_reader);
+                symtab_number(&mut system_reader);
+            }
+        } else {
+            bail!("this command does not yet support reading from STDIN")
+        }
+        Ok(())
+    }
+}
+
+fn symtab_number(reader: &mut SystemReader<RawBinaryReader<&[u8]>>) -> Result<()> {
+    let mut count = 0;
+    loop {
+        match reader.next()? {
+            SystemStreamItem:: Value(_)=> {
+                let symbols_len = reader.symbol_table().symbols().iter().len();
+                // Reduce the number of system symbols.
+                count += symbols_len - 10;
+            }
+            SystemStreamItem::Nothing => break,
+            _ => {}
+        }
+    }
+    println!("The number of symbols is {}", count);
+    Ok(())
+}