Add JSON output option to CLI (#294)

Cargo.toml

@@ -4,7 +4,7 @@ resolver = "2"
 
 [workspace.package]
 description = "Get the language distribution stats of your repository"
-version = "0.10.0"
+version = "0.10.1"
 edition = "2021"
 repository = "https://github.com/spenserblack/gengo"
 readme = "README.md"

gengo-bin/Cargo.toml

@@ -24,6 +24,8 @@ clap = { version = "4", features = ["derive", "wrap_help"] }
 gengo = { path = "../gengo", version = "0.10", default-features = false }
 indexmap = "2"
 owo-colors = { version = ">=3, <=4", optional = true }
+serde_json = "1"
 
 [dev-dependencies]
-insta = "1"
+insta = { version = "1", features = ["json"] }
+serde_json = "1"

gengo-bin/src/cli.rs

@@ -1,5 +1,5 @@
 use clap::Error as ClapError;
-use clap::{Parser, Subcommand};
+use clap::{Parser, Subcommand, ValueEnum};
 use gengo::{analysis::SummaryOpts, Analysis, Builder, Directory, Git};
 use indexmap::IndexMap;
 use std::error::Error as BaseError;
@@ -28,15 +28,24 @@ pub struct CLI {
     #[arg(short = 'l', long, default_value = "1048576")]
     read_limit: usize,
     /// Report on all files, even if they are not detectable.
+    ///
+    /// This only applies to the pretty format, as machine-readable
+    /// formats always include all files.
     #[arg(short = 'a', long)]
     all: bool,
     /// Include detailed statistics for each language.
+    ///
+    /// This only applies to the pretty format, as machine-readable
+    /// formats always include detailed statistics.
     #[arg(short = 'b', long)]
     breakdown: bool,
     /// Force the output to not have colors.
     #[cfg(feature = "color")]
     #[arg(long)]
     no_color: bool,
+    /// The format to use for output.
+    #[arg(short = 'F', long, default_value = "pretty")]
+    format: Format,
 }
 
 #[derive(Subcommand)]
@@ -58,6 +67,14 @@ enum Commands {
     },
 }
 
+#[derive(ValueEnum, Debug, Clone)]
+enum Format {
+    /// Output for humans.
+    Pretty,
+    /// JSON output.
+    Json,
+}
+
 impl CLI {
     pub fn run(&self, mut out: impl Write, mut err: impl Write) -> Result<(), io::Error> {
         let results = self.command.analyze(self.read_limit);
@@ -69,6 +86,11 @@ impl CLI {
             }
         };
 
+        match self.format {
+            Format::Pretty => (),
+            Format::Json => return self.run_json(results, out, err),
+        }
+
         let mut summary_opts: SummaryOpts = Default::default();
         summary_opts.all = self.all;
         let summary = results.summary_with(summary_opts);
@@ -102,6 +124,22 @@ impl CLI {
         Ok(())
     }
 
+    fn run_json(
+        &self,
+        analysis: Analysis,
+        mut out: impl Write,
+        mut _err: impl Write,
+    ) -> Result<(), io::Error> {
+        match serde_json::to_string(&analysis) {
+            Ok(s) => writeln!(out, "{s}")?,
+            Err(e) => {
+                writeln!(out, "failed to serialize to JSON: {e}")?;
+                return Ok(());
+            }
+        };
+        Ok(())
+    }
+
     fn run_breakdown(
         &self,
         mut out: impl Write,

gengo-bin/tests/snapshots/test_cli__json_output_on_javascript_repo.snap

@@ -0,0 +1,66 @@
+---
+source: gengo-bin/tests/test_cli.rs
+expression: json
+---
+{
+  "bin.js": {
+    "detectable": true,
+    "documentation": false,
+    "generated": false,
+    "language": {
+      "category": "programming",
+      "color": "#F0DC4E",
+      "name": "JavaScript"
+    },
+    "size": 28,
+    "vendored": false
+  },
+  "dist/bin.js": {
+    "detectable": true,
+    "documentation": false,
+    "generated": true,
+    "language": {
+      "category": "prose",
+      "color": "#000000",
+      "name": "Plain Text"
+    },
+    "size": 62,
+    "vendored": false
+  },
+  "docs/index.html": {
+    "detectable": false,
+    "documentation": true,
+    "generated": false,
+    "language": {
+      "category": "markup",
+      "color": "#E96228",
+      "name": "HTML"
+    },
+    "size": 26,
+    "vendored": false
+  },
+  "node_modules/my-dependency/index.js": {
+    "detectable": false,
+    "documentation": false,
+    "generated": false,
+    "language": {
+      "category": "programming",
+      "color": "#F0DC4E",
+      "name": "JavaScript"
+    },
+    "size": 29,
+    "vendored": true
+  },
+  "src/bin.ts": {
+    "detectable": true,
+    "documentation": false,
+    "generated": false,
+    "language": {
+      "category": "programming",
+      "color": "#2F74C0",
+      "name": "TypeScript"
+    },
+    "size": 62,
+    "vendored": false
+  }
+}

gengo-bin/tests/test_cli.rs

@@ -1,4 +1,4 @@
-use insta::assert_snapshot;
+use insta::{assert_json_snapshot, assert_snapshot};
 use std::io::{self, Write};
 
 const ROOT: &str = env!("CARGO_MANIFEST_DIR");
@@ -25,6 +25,19 @@ macro_rules! assert_stdout_snapshot {
     }};
 }
 
+// TODO This feels like a bit of a hack.
+macro_rules! assert_stdout_json_snapshot {
+    ($cli_args:expr $(,)?) => {{
+        let cli = gengo_bin::cli::try_new_from($cli_args).unwrap();
+        let mut stdout = Vec::new();
+        let mut stderr = NullWriter;
+        cli.run(&mut stdout, &mut stderr).unwrap();
+        let stdout = String::from_utf8(stdout).unwrap();
+        let json: serde_json::Value = serde_json::from_str(&stdout).unwrap();
+        assert_json_snapshot!(json);
+    }};
+}
+
 #[test]
 #[cfg(not(feature = "color"))]
 fn test_javascript_repo() {
@@ -67,5 +80,19 @@ fn test_color_breakdown_javascript_repo() {
     ]);
 }
 
-//
 // TODO Add test_javascript_repo_windows
+
+#[test]
+#[cfg_attr(windows, ignore)]
+fn test_json_output_on_javascript_repo() {
+    assert_stdout_json_snapshot!(&[
+        "gengo",
+        "--format",
+        "json",
+        "git",
+        "-r",
+        "test/javascript",
+        "-R",
+        ROOT,
+    ]);
+}

gengo/Cargo.toml

@@ -25,7 +25,7 @@ gix = { version = ">= 0.56, <= 0.57", default-features = false, features = [
 ] }
 glob = "0.3"
 ignore = "0.4"
-indexmap = { version = "2", features = ["serde"] }
+indexmap = { version = "2", features = ["rayon", "serde"] }
 once_cell = "1"
 owo-colors = { version = ">=3, <=4", optional = true }
 rayon = "1"

gengo/src/analysis/mod.rs

@@ -1,5 +1,6 @@
 use super::Entry;
 use indexmap::IndexMap;
+use serde::Serialize;
 
 use std::path::PathBuf;
 
@@ -10,11 +11,11 @@ pub use summary::Summary;
 mod summary;
 
 /// The result of analyzing a repository along with all of its submodules.
-#[derive(Debug)]
-pub struct Analysis(pub(super) Vec<(PathBuf, Entry)>);
+#[derive(Debug, Serialize)]
+pub struct Analysis(pub(super) IndexMap<PathBuf, Entry>);
 
 impl Analysis {
-    pub fn iter(&self) -> impl Iterator<Item = &(PathBuf, Entry)> {
+    pub fn iter(&self) -> impl Iterator<Item = (&PathBuf, &Entry)> {
         let results = &self.0;
         results.iter()
     }

gengo/src/lib.rs

@@ -26,6 +26,7 @@ use generated::Generated;
 
 pub use file_source::{Directory, FileSource, Git};
 use glob::MatchOptions;
+use indexmap::IndexMap;
 pub use languages::analyzer::Analyzers;
 use languages::Category;
 pub use languages::Language;
@@ -35,7 +36,8 @@ use std::path::Path;
 
 use vendored::Vendored;
 
-use rayon::prelude::{ParallelBridge, ParallelIterator};
+use rayon::prelude::{FromParallelIterator, ParallelBridge, ParallelIterator};
+use serde::Serialize;
 
 pub mod analysis;
 mod builder;
@@ -70,7 +72,7 @@ impl<FS: for<'fs> FileSource<'fs>> Gengo<FS> {
     /// Analyzes each file in the repository at the given revision.
     pub fn analyze(&self) -> Result<Analysis> {
         let state = self.file_source.state()?;
-        let entries: Vec<(_, _)> = self
+        let entries = self
             .file_source
             .entries()?
             .par_bridge()
@@ -80,8 +82,8 @@ impl<FS: for<'fs> FileSource<'fs>> Gengo<FS> {
                 let entry = self.analyze_blob(&filepath, contents, state)?;
                 Some((filepath.as_ref().to_owned(), entry))
             })
-            .filter_map(|entry| entry)
-            .collect();
+            .filter_map(|entry| entry);
+        let entries = IndexMap::from_par_iter(entries);
 
         Ok(Analysis(entries))
     }
@@ -148,7 +150,7 @@ impl<FS: for<'fs> FileSource<'fs>> Gengo<FS> {
 }
 
 /// A single entry in the language statistics.
-#[derive(Debug)]
+#[derive(Debug, Serialize)]
 pub struct Entry {
     /// The detected language.
     language: Language,

gengo/tests/gengo_test.rs

@@ -14,7 +14,7 @@ fn test_git_javascript() {
     let gengo = Builder::new(git).analyzers(analyzers).build().unwrap();
     let analysis = gengo.analyze().unwrap();
     let mut results: Vec<_> = analysis.iter().collect();
-    results.sort_by_key(|(path, _)| path);
+    results.sort_by_key(|(path, _)| path.to_owned());
     insta::assert_debug_snapshot!(results);
 }