5 files changed, 331 insertions, 67 deletions

diff --git a/bin/Cargo.toml b/bin/Cargo.toml
index 9452c2b..0d6f970 100644
--- a/bin/Cargo.toml
+++ b/bin/Cargo.toml
@@ -8,5 +8,8 @@ edition = "2018"
 [dependencies]
 lib = { path = "../lib" }
 ariadne = "0.1.3"
-anyhow = "1.0"
 rnix = "0.9.0"
+clap = "3.0.0-beta.4"
+globset = "0.4.8"
+thiserror = "1.0.30"
+vfs = { path = "../vfs" }
diff --git a/bin/src/config.rs b/bin/src/config.rs
new file mode 100644
index 0000000..f2cf29d
--- /dev/null
+++ b/bin/src/config.rs
@@ -0,0 +1,170 @@
+use std::{default::Default, fs, path::PathBuf, str::FromStr};
+
+use clap::Clap;
+use globset::{GlobBuilder, GlobSetBuilder};
+use vfs::ReadOnlyVfs;
+
+use crate::err::ConfigErr;
+
+/// Static analysis and linting for the nix programming language
+#[derive(Clap, Debug)]
+#[clap(version = "0.1.0", author = "Akshay <[email protected]>")]
+pub struct Opts {
+    /// File or directory to run statix on
+    #[clap(default_value = ".")]
+    target: String,
+
+    // /// Path to statix config
+    // #[clap(short, long, default_value = ".statix.toml")]
+    // config: String,
+    /// Regex of file patterns to not lint
+    #[clap(short, long)]
+    ignore: Vec<String>,
+
+    /// Output format. Supported values: json, errfmt
+    #[clap(short = 'o', long)]
+    format: Option<OutFormat>,
+
+    #[clap(subcommand)]
+    pub subcmd: Option<SubCommand>,
+}
+
+#[derive(Clap, Debug)]
+#[clap(version = "0.1.0", author = "Akshay <[email protected]>")]
+pub enum SubCommand {
+    /// Find and fix issues raised by statix
+    Fix(Fix),
+}
+
+#[derive(Clap, Debug)]
+pub struct Fix {
+    /// Do not write to files, display a diff instead
+    #[clap(short = 'd', long = "dry-run")]
+    diff_only: bool,
+}
+
+#[derive(Debug, Copy, Clone)]
+pub enum OutFormat {
+    Json,
+    Errfmt,
+    StdErr,
+}
+
+impl Default for OutFormat {
+    fn default() -> Self {
+        OutFormat::StdErr
+    }
+}
+
+impl FromStr for OutFormat {
+    type Err = &'static str;
+
+    fn from_str(value: &str) -> Result<Self, Self::Err> {
+        match value.to_ascii_lowercase().as_str() {
+            "json" => Ok(Self::Json),
+            "errfmt" => Ok(Self::Errfmt),
+            "stderr" => Ok(Self::StdErr),
+            _ => Err("unknown output format, try: json, errfmt"),
+        }
+    }
+}
+
+#[derive(Debug)]
+pub struct LintConfig {
+    pub files: Vec<PathBuf>,
+    pub format: OutFormat,
+}
+
+impl LintConfig {
+    pub fn from_opts(opts: Opts) -> Result<Self, ConfigErr> {
+        let ignores = {
+            let mut set = GlobSetBuilder::new();
+            for pattern in opts.ignore {
+                let glob = GlobBuilder::new(&pattern).build().map_err(|err| {
+                    ConfigErr::InvalidGlob(err.glob().map(|i| i.to_owned()), err.kind().clone())
+                })?;
+                set.add(glob);
+            }
+            set.build().map_err(|err| {
+                ConfigErr::InvalidGlob(err.glob().map(|i| i.to_owned()), err.kind().clone())
+            })
+        }?;
+
+        let walker = dirs::Walker::new(opts.target).map_err(ConfigErr::InvalidPath)?;
+
+        let files = walker
+            .filter(|path| matches!(path.extension(), Some(e) if e == "nix"))
+            .filter(|path| !ignores.is_match(path))
+            .collect();
+        Ok(Self {
+            files,
+            format: opts.format.unwrap_or_default(),
+        })
+    }
+
+    pub fn vfs(&self) -> Result<ReadOnlyVfs, ConfigErr> {
+        let mut vfs = ReadOnlyVfs::default();
+        for file in self.files.iter() {
+            let _id = vfs.alloc_file_id(&file);
+            let data = fs::read_to_string(&file).map_err(ConfigErr::InvalidPath)?;
+            vfs.set_file_contents(&file, data.as_bytes());
+        }
+        Ok(vfs)
+    }
+}
+
+mod dirs {
+    use std::{
+        fs,
+        io::{self, Error, ErrorKind},
+        path::{Path, PathBuf},
+    };
+
+    #[derive(Default, Debug)]
+    pub struct Walker {
+        dirs: Vec<PathBuf>,
+        files: Vec<PathBuf>,
+    }
+
+    impl Walker {
+        pub fn new<P: AsRef<Path>>(target: P) -> io::Result<Self> {
+            let target = target.as_ref().to_path_buf();
+            if !target.exists() {
+                Err(Error::new(
+                    ErrorKind::NotFound,
+                    format!("file not found: {}", target.display()),
+                ))
+            } else if target.is_dir() {
+                Ok(Self {
+                    dirs: vec![target],
+                    ..Default::default()
+                })
+            } else {
+                Ok(Self {
+                    files: vec![target],
+                    ..Default::default()
+                })
+            }
+        }
+    }
+
+    impl Iterator for Walker {
+        type Item = PathBuf;
+        fn next(&mut self) -> Option<Self::Item> {
+            if let Some(dir) = self.dirs.pop() {
+                if dir.is_dir() {
+                    for entry in fs::read_dir(dir).ok()? {
+                        let entry = entry.ok()?;
+                        let path = entry.path();
+                        if path.is_dir() {
+                            self.dirs.push(path);
+                        } else if path.is_file() {
+                            self.files.push(path);
+                        }
+                    }
+                }
+            }
+            self.files.pop()
+        }
+    }
+}
diff --git a/bin/src/err.rs b/bin/src/err.rs
new file mode 100644
index 0000000..b3a79c2
--- /dev/null
+++ b/bin/src/err.rs
@@ -0,0 +1,28 @@
+use std::{io, path::PathBuf};
+
+use globset::ErrorKind;
+use rnix::parser::ParseError;
+use thiserror::Error;
+
+#[derive(Error, Debug)]
+pub enum ConfigErr {
+    #[error("error parsing glob `{0:?}`: {1}")]
+    InvalidGlob(Option<String>, ErrorKind),
+
+    #[error("path error: {0}")]
+    InvalidPath(#[from] io::Error),
+}
+
+#[derive(Error, Debug)]
+pub enum LintErr {
+    #[error("[{0}] syntax error: {1}")]
+    Parse(PathBuf, ParseError),
+}
+
+#[derive(Error, Debug)]
+pub enum StatixErr {
+    #[error("linter error: {0}")]
+    Lint(#[from] LintErr),
+    #[error("config error: {0}")]
+    Config(#[from] ConfigErr),
+}
diff --git a/bin/src/main.rs b/bin/src/main.rs
index ab99aee..b26151d 100644
--- a/bin/src/main.rs
+++ b/bin/src/main.rs
@@ -1,20 +1,28 @@
-use std::{
-    env, fs,
-    path::{Path, PathBuf},
-};
+#![feature(path_try_exists)]
+
+mod config;
+mod err;
+mod traits;
 
-use anyhow::{Context, Result};
-use ariadne::{
-    CharSet, Color, Config as CliConfig, Label, LabelAttach, Report as CliReport,
-    ReportKind as CliReportKind, Source,
+use std::io;
+
+use crate::{
+    err::{LintErr, StatixErr},
+    traits::{LintResult, WriteDiagnostic},
 };
-use lib::{Report, LINTS};
-use rnix::{TextRange, WalkEvent};
 
-fn analyze(source: &str) -> Result<Vec<Report>> {
-    let parsed = rnix::parse(source).as_result()?;
+use clap::Clap;
+use config::{LintConfig, Opts, SubCommand};
+use lib::LINTS;
+use rnix::WalkEvent;
+use vfs::VfsEntry;
 
-    Ok(parsed
+fn analyze<'ρ>(vfs_entry: VfsEntry<'ρ>) -> Result<LintResult, LintErr> {
+    let source = vfs_entry.contents;
+    let parsed = rnix::parse(source)
+        .as_result()
+        .map_err(|e| LintErr::Parse(vfs_entry.file_path.to_path_buf(), e))?;
+    let reports = parsed
         .node()
         .preorder_with_tokens()
         .filter_map(|event| match event {
@@ -27,61 +35,33 @@ fn analyze(source: &str) -> Result<Vec<Report>> {
             _ => None,
         })
         .flatten()
-        .collect())
+        .collect();
+    Ok(LintResult {
+        file_id: vfs_entry.file_id,
+        reports,
+    })
 }
 
-fn print_report(report: Report, file_src: &str, file_path: &Path) -> Result<()> {
-    let range = |at: TextRange| at.start().into()..at.end().into();
-    let src_id = file_path.to_str().unwrap_or("<unknown>");
-    let offset = report
-        .diagnostics
-        .iter()
-        .map(|d| d.at.start().into())
-        .min()
-        .unwrap_or(0usize);
-    report
-        .diagnostics
-        .iter()
-        .fold(
-            CliReport::build(CliReportKind::Warning, src_id, offset)
-                .with_config(
-                    CliConfig::default()
-                        .with_cross_gap(true)
-                        .with_multiline_arrows(false)
-                        .with_label_attach(LabelAttach::Middle)
-                        .with_char_set(CharSet::Unicode),
-                )
-                .with_message(report.note)
-                .with_code(report.code),
-            |cli_report, diagnostic| {
-                cli_report.with_label(
-                    Label::new((src_id, range(diagnostic.at)))
-                        .with_message(&diagnostic.message)
-                        .with_color(Color::Magenta),
-                )
-            },
-        )
-        .finish()
-        .eprint((src_id, Source::from(file_src)))
-        .context("failed to print report to stdout")
-}
-
-fn _main() -> Result<()> {
+fn _main() -> Result<(), StatixErr> {
     // TODO: accept cli args, construct a CLI config with a list of files to analyze
-    let args = env::args();
-    for (file_src, file_path, reports) in args
-        .skip(1)
-        .map(|s| PathBuf::from(&s))
-        .filter(|p| p.is_file())
-        .filter_map(|path| {
-            let s = fs::read_to_string(&path).ok()?;
-            analyze(&s)
-                .map(|analysis_result| (s, path, analysis_result))
-                .ok()
-        })
-    {
-        for r in reports {
-            print_report(r, &file_src, &file_path)?
+    let opts = Opts::parse();
+    match opts.subcmd {
+        Some(SubCommand::Fix(_)) => {}
+        None => {
+            let lint_config = LintConfig::from_opts(opts)?;
+            let vfs = lint_config.vfs()?;
+            let (reports, errors): (Vec<_>, Vec<_>) =
+                vfs.iter().map(analyze).partition(Result::is_ok);
+            let lint_results: Vec<_> = reports.into_iter().map(Result::unwrap).collect();
+            let errors: Vec<_> = errors.into_iter().map(Result::unwrap_err).collect();
+
+            let mut stderr = io::stderr();
+            lint_results.into_iter().for_each(|r| {
+                stderr.write(&r, &vfs).unwrap();
+            });
+            errors.into_iter().for_each(|e| {
+                eprintln!("{}", e);
+            });
         }
     }
     Ok(())
@@ -90,6 +70,6 @@ fn _main() -> Result<()> {
 fn main() {
     match _main() {
         Err(e) => eprintln!("{}", e),
-        _ => {}
+        _ => (),
     }
 }
diff --git a/bin/src/traits.rs b/bin/src/traits.rs
new file mode 100644
index 0000000..1807ad0
--- /dev/null
+++ b/bin/src/traits.rs
@@ -0,0 +1,83 @@
+use std::{
+    io::{self, Write},
+    str,
+};
+
+use ariadne::{
+    CharSet, Color, Config as CliConfig, Label, LabelAttach, Report as CliReport,
+    ReportKind as CliReportKind, Source, Fmt
+};
+use lib::Report;
+use rnix::TextRange;
+use vfs::{FileId, ReadOnlyVfs};
+
+#[derive(Debug)]
+pub struct LintResult {
+    pub file_id: FileId,
+    pub reports: Vec<Report>,
+}
+
+pub trait WriteDiagnostic {
+    fn write(&mut self, report: &LintResult, vfs: &ReadOnlyVfs) -> io::Result<()>;
+}
+
+impl<T> WriteDiagnostic for T
+where
+    T: Write,
+{
+    fn write(&mut self, lint_result: &LintResult, vfs: &ReadOnlyVfs) -> io::Result<()> {
+        let file_id = lint_result.file_id;
+        let src = str::from_utf8(vfs.get(file_id)).unwrap();
+        let path = vfs.file_path(file_id);
+        let range = |at: TextRange| at.start().into()..at.end().into();
+        let src_id = path.to_str().unwrap_or("<unknown>");
+        for report in lint_result.reports.iter() {
+            let offset = report
+                .diagnostics
+                .iter()
+                .map(|d| d.at.start().into())
+                .min()
+                .unwrap_or(0usize);
+            report
+                .diagnostics
+                .iter()
+                .fold(
+                    CliReport::build(CliReportKind::Warning, src_id, offset)
+                        .with_config(
+                            CliConfig::default()
+                                .with_cross_gap(true)
+                                .with_multiline_arrows(false)
+                                .with_label_attach(LabelAttach::Middle)
+                                .with_char_set(CharSet::Unicode),
+                        )
+                        .with_message(report.note)
+                        .with_code(report.code),
+                    |cli_report, diagnostic| {
+                        cli_report.with_label(
+                            Label::new((src_id, range(diagnostic.at)))
+                                .with_message(&colorize(&diagnostic.message))
+                                .with_color(Color::Magenta),
+                        )
+                    },
+                )
+                .finish()
+                .write((src_id, Source::from(src)), &mut *self)?;
+        }
+        Ok(())
+    }
+}
+
+// everything within backticks is colorized, backticks are removed
+fn colorize(message: &str) -> String {
+    message.split('`')
+        .enumerate()
+        .map(|(idx, part)| {
+            if idx % 2 == 1 {
+                part.fg(Color::Cyan).to_string()
+            } else {
+                part.to_string()
+            }
+        })
+        .collect::<Vec<_>>()
+        .join("")
+}