From 1d084981311133e14646b65eef2ee251cd6a9de1 Mon Sep 17 00:00:00 2001
From: Lucas Schwiderski <lucas@lschwiderski.de>
Date: Fri, 17 Feb 2023 11:10:56 +0100
Subject: [PATCH] feat(dtmt): Add command to print the dictionary

This is mostly helpful to check/debug whether the internal dictionary
actually contains the expected data. For manually looking through the
entire dictionary, opening the CSV file is still more convenient.
---
 Cargo.lock                        | 23 +++++++++++++++++++++
 crates/dtmt/Cargo.toml            |  1 +
 crates/dtmt/src/cmd/dictionary.rs | 34 +++++++++++++++++++++++++++++++
 lib/sdk/src/murmur/dictionary.rs  | 22 ++++++++++++++++++++
 lib/sdk/src/murmur/mod.rs         |  3 +--
 5 files changed, 81 insertions(+), 2 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index 4dddb00..2e3679a 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -291,6 +291,28 @@ dependencies = [
  "os_str_bytes",
 ]
 
+[[package]]
+name = "cli-table"
+version = "0.4.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "adfbb116d9e2c4be7011360d0c0bee565712c11e969c9609b25b619366dc379d"
+dependencies = [
+ "cli-table-derive",
+ "termcolor",
+ "unicode-width",
+]
+
+[[package]]
+name = "cli-table-derive"
+version = "0.4.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2af3bfb9da627b0a6c467624fb7963921433774ed435493b5c08a3053e829ad4"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
 [[package]]
 name = "clipboard-win"
 version = "4.4.2"
@@ -654,6 +676,7 @@ name = "dtmt"
 version = "0.2.0"
 dependencies = [
  "clap",
+ "cli-table",
  "color-eyre",
  "confy",
  "csv-async",
diff --git a/crates/dtmt/Cargo.toml b/crates/dtmt/Cargo.toml
index 73d2d72..cb8f646 100644
--- a/crates/dtmt/Cargo.toml
+++ b/crates/dtmt/Cargo.toml
@@ -26,6 +26,7 @@ confy = "0.5.1"
 zip = "0.6.3"
 string_template = "0.2.1"
 promptly = "0.3.1"
+cli-table = { version = "0.4.7", default-features = false, features = ["derive"] }
 
 [dev-dependencies]
 tempfile = "3.3.0"
diff --git a/crates/dtmt/src/cmd/dictionary.rs b/crates/dtmt/src/cmd/dictionary.rs
index 22a225b..e94e2ad 100644
--- a/crates/dtmt/src/cmd/dictionary.rs
+++ b/crates/dtmt/src/cmd/dictionary.rs
@@ -1,8 +1,10 @@
 use std::path::PathBuf;
 
 use clap::{value_parser, Arg, ArgAction, ArgMatches, Command, ValueEnum};
+use cli_table::{print_stdout, WithTitle};
 use color_eyre::eyre::{Context, Result};
 use color_eyre::{Help, SectionExt};
+use sdk::murmur::{IdString64, Murmur32, Murmur64};
 use tokio::fs::File;
 use tokio::io::{AsyncBufReadExt, BufReader};
 use tokio_stream::wrappers::LinesStream;
@@ -27,6 +29,29 @@ impl From<HashGroup> for sdk::murmur::HashGroup {
     }
 }
 
+#[derive(cli_table::Table)]
+struct TableRow {
+    #[table(title = "Value")]
+    value: String,
+    #[table(title = "Murmur64")]
+    long: Murmur64,
+    #[table(title = "Murmur32")]
+    short: Murmur32,
+    #[table(title = "Group")]
+    group: sdk::murmur::HashGroup,
+}
+
+impl From<&sdk::murmur::Entry> for TableRow {
+    fn from(entry: &sdk::murmur::Entry) -> Self {
+        Self {
+            value: entry.value().clone(),
+            long: entry.long(),
+            short: entry.short(),
+            group: entry.group(),
+        }
+    }
+}
+
 pub(crate) fn command_definition() -> Command {
     Command::new("dictionary")
         .about("Manipulate a hash dictionary file.")
@@ -67,6 +92,7 @@ pub(crate) fn command_definition() -> Command {
                         .value_parser(value_parser!(PathBuf)),
                 ),
         )
+        .subcommand(Command::new("show").about("Show the contents of the dictionary"))
         .subcommand(Command::new("save").about(
             "Save back the currently loaded dictionary, with hashes pre-computed. \
                 Pre-computing hashes speeds up loading large dictionaries, as they would \
@@ -176,6 +202,14 @@ pub(crate) async fn run(mut ctx: sdk::Context, matches: &ArgMatches) -> Result<(
                 .await
                 .wrap_err("Failed to write dictionary to disk")
         }
+        Some(("show", _)) => {
+            let lookup = &ctx.lookup;
+            let rows: Vec<_> = lookup.entries().iter().map(TableRow::from).collect();
+
+            print_stdout(rows.with_title())?;
+
+            Ok(())
+        }
         _ => unreachable!(
             "clap is configured to require a subcommand, and they're all handled above"
         ),
diff --git a/lib/sdk/src/murmur/dictionary.rs b/lib/sdk/src/murmur/dictionary.rs
index 322dded..2d51af1 100644
--- a/lib/sdk/src/murmur/dictionary.rs
+++ b/lib/sdk/src/murmur/dictionary.rs
@@ -55,6 +55,24 @@ pub struct Entry {
     group: HashGroup,
 }
 
+impl Entry {
+    pub fn value(&self) -> &String {
+        &self.value
+    }
+
+    pub fn long(&self) -> Murmur64 {
+        self.long
+    }
+
+    pub fn short(&self) -> Murmur32 {
+        self.short
+    }
+
+    pub fn group(&self) -> HashGroup {
+        self.group
+    }
+}
+
 pub struct Dictionary {
     entries: Vec<Entry>,
 }
@@ -172,4 +190,8 @@ impl Dictionary {
     pub fn is_empty(&self) -> bool {
         self.entries.is_empty()
     }
+
+    pub fn entries(&self) -> &Vec<Entry> {
+        &self.entries
+    }
 }
diff --git a/lib/sdk/src/murmur/mod.rs b/lib/sdk/src/murmur/mod.rs
index 95e66fa..784a6df 100644
--- a/lib/sdk/src/murmur/mod.rs
+++ b/lib/sdk/src/murmur/mod.rs
@@ -13,8 +13,7 @@ mod murmurhash64;
 
 pub const SEED: u32 = 0;
 
-pub use dictionary::Dictionary;
-pub use dictionary::HashGroup;
+pub use dictionary::{Dictionary, Entry, HashGroup};
 pub use murmurhash64::hash;
 pub use murmurhash64::hash32;
 pub use murmurhash64::hash_inverse as inverse;