Move all gitignore matching to separate crate.

This PR introduces a new sub-crate, `ignore`, which primarily provides a
fast recursive directory iterator that respects ignore files like
gitignore and other configurable filtering rules based on globs or even
file types.

This results in a substantial source of complexity moved out of ripgrep's
core and into a reusable component that others can now (hopefully)
benefit from.

While much of the ignore code carried over from ripgrep's core, a
substantial portion of it was rewritten with the following goals in
mind:

1. Reuse matchers built from gitignore files across directory iteration.
2. Design the matcher data structure to be amenable for parallelizing
   directory iteration. (Indeed, writing the parallel iterator is the
   next step.)

Fixes #9, #44, #45

ignore/src/lib.rs

@@ -0,0 +1,300 @@
+/*!
+The ignore crate provides a fast recursive directory iterator that respects
+various filters such as globs, file types and `.gitignore` files. The precise
+matching rules and precedence is explained in the documentation for
+`WalkBuilder`.
+
+Secondarily, this crate exposes gitignore and file type matchers for use cases
+that demand more fine-grained control.
+
+# Example
+
+This example shows the most basic usage of this crate. This code will
+recursively traverse the current directory while automatically filtering out
+files and directories according to ignore globs found in files like
+`.ignore` and `.gitignore`:
+
+
+```rust,no_run
+use ignore::Walk;
+
+for result in Walk::new("./") {
+    // Each item yielded by the iterator is either a directory entry or an
+    // error, so either print the path or the error.
+    match result {
+        Ok(entry) => println!("{}", entry.path().display()),
+        Err(err) => println!("ERROR: {}", err),
+    }
+}
+```
+
+# Example: advanced
+
+By default, the recursive directory iterator will ignore hidden files and
+directories. This can be disabled by building the iterator with `WalkBuilder`:
+
+```rust,no_run
+use ignore::WalkBuilder;
+
+for result in WalkBuilder::new("./").hidden(false).build() {
+    println!("{:?}", result);
+}
+```
+
+See the documentation for `WalkBuilder` for many other options.
+*/
+
+extern crate globset;
+#[macro_use]
+extern crate lazy_static;
+#[macro_use]
+extern crate log;
+extern crate memchr;
+extern crate regex;
+#[cfg(test)]
+extern crate tempdir;
+extern crate thread_local;
+extern crate walkdir;
+
+use std::error;
+use std::fmt;
+use std::io;
+use std::path::{Path, PathBuf};
+
+pub use walk::{DirEntry, Walk, WalkBuilder};
+
+mod dir;
+pub mod gitignore;
+mod pathutil;
+pub mod overrides;
+pub mod types;
+mod walk;
+
+/// Represents an error that can occur when parsing a gitignore file.
+#[derive(Debug)]
+pub enum Error {
+    /// A collection of "soft" errors. These occur when adding an ignore
+    /// file partially succeeded.
+    Partial(Vec<Error>),
+    /// An error associated with a specific line number.
+    WithLineNumber { line: u64, err: Box<Error> },
+    /// An error associated with a particular file path.
+    WithPath { path: PathBuf, err: Box<Error> },
+    /// An error that occurs when doing I/O, such as reading an ignore file.
+    Io(io::Error),
+    /// An error that occurs when trying to parse a glob.
+    Glob(String),
+    /// A type selection for a file type that is not defined.
+    UnrecognizedFileType(String),
+    /// A user specified file type definition could not be parsed.
+    InvalidDefinition,
+}
+
+impl Error {
+    /// Returns true if this is a partial error.
+    ///
+    /// A partial error occurs when only some operations failed while others
+    /// may have succeeded. For example, an ignore file may contain an invalid
+    /// glob among otherwise valid globs.
+    pub fn is_partial(&self) -> bool {
+        match *self {
+            Error::Partial(_) => true,
+            Error::WithLineNumber { ref err, .. } => err.is_partial(),
+            Error::WithPath { ref err, .. } => err.is_partial(),
+            _ => false,
+        }
+    }
+
+    /// Returns true if this error is exclusively an I/O error.
+    pub fn is_io(&self) -> bool {
+        match *self {
+            Error::Partial(ref errs) => errs.len() == 1 && errs[0].is_io(),
+            Error::WithLineNumber { ref err, .. } => err.is_io(),
+            Error::WithPath { ref err, .. } => err.is_io(),
+            Error::Io(_) => true,
+            Error::Glob(_) => false,
+            Error::UnrecognizedFileType(_) => false,
+            Error::InvalidDefinition => false,
+        }
+    }
+
+    /// Turn an error into a tagged error with the given file path.
+    fn with_path<P: AsRef<Path>>(self, path: P) -> Error {
+        Error::WithPath {
+            path: path.as_ref().to_path_buf(),
+            err: Box::new(self),
+        }
+    }
+
+    /// Turn an error into a tagged error with the given file path and line
+    /// number. If path is empty, then it is omitted from the error.
+    fn tagged<P: AsRef<Path>>(self, path: P, lineno: u64) -> Error {
+        let errline = Error::WithLineNumber {
+            line: lineno,
+            err: Box::new(self),
+        };
+        if path.as_ref().as_os_str().is_empty() {
+            return errline;
+        }
+        errline.with_path(path)
+    }
+}
+
+impl error::Error for Error {
+    fn description(&self) -> &str {
+        match *self {
+            Error::Partial(_) => "partial error",
+            Error::WithLineNumber { ref err, .. } => err.description(),
+            Error::WithPath { ref err, .. } => err.description(),
+            Error::Io(ref err) => err.description(),
+            Error::Glob(ref msg) => msg,
+            Error::UnrecognizedFileType(_) => "unrecognized file type",
+            Error::InvalidDefinition => "invalid definition",
+        }
+    }
+}
+
+impl fmt::Display for Error {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        match *self {
+            Error::Partial(ref errs) => {
+                let msgs: Vec<String> =
+                    errs.iter().map(|err| err.to_string()).collect();
+                write!(f, "{}", msgs.join("\n"))
+            }
+            Error::WithLineNumber { line, ref err } => {
+                write!(f, "line {}: {}", line, err)
+            }
+            Error::WithPath { ref path, ref err } => {
+                write!(f, "{}: {}", path.display(), err)
+            }
+            Error::Io(ref err) => err.fmt(f),
+            Error::Glob(ref msg) => write!(f, "{}", msg),
+            Error::UnrecognizedFileType(ref ty) => {
+                write!(f, "unrecognized file type: {}", ty)
+            }
+            Error::InvalidDefinition => {
+                write!(f, "invalid definition (format is type:glob, e.g., \
+                           html:*.html)")
+            }
+        }
+    }
+}
+
+impl From<io::Error> for Error {
+    fn from(err: io::Error) -> Error {
+        Error::Io(err)
+    }
+}
+
+#[derive(Debug, Default)]
+struct PartialErrorBuilder(Vec<Error>);
+
+impl PartialErrorBuilder {
+    fn push(&mut self, err: Error) {
+        self.0.push(err);
+    }
+
+    fn push_ignore_io(&mut self, err: Error) {
+        if !err.is_io() {
+            self.push(err);
+        }
+    }
+
+    fn maybe_push(&mut self, err: Option<Error>) {
+        if let Some(err) = err {
+            self.push(err);
+        }
+    }
+
+    fn maybe_push_ignore_io(&mut self, err: Option<Error>) {
+        if let Some(err) = err {
+            self.push_ignore_io(err);
+        }
+    }
+
+    fn into_error_option(mut self) -> Option<Error> {
+        if self.0.is_empty() {
+            None
+        } else if self.0.len() == 1 {
+            Some(self.0.pop().unwrap())
+        } else {
+            Some(Error::Partial(self.0))
+        }
+    }
+}
+
+/// The result of a glob match.
+///
+/// The type parameter `T` typically refers to a type that provides more
+/// information about a particular match. For example, it might identify
+/// the specific gitignore file and the specific glob pattern that caused
+/// the match.
+#[derive(Clone, Debug)]
+pub enum Match<T> {
+    /// The path didn't match any glob.
+    None,
+    /// The highest precedent glob matched indicates the path should be
+    /// ignored.
+    Ignore(T),
+    /// The highest precedent glob matched indicates the path should be
+    /// whitelisted.
+    Whitelist(T),
+}
+
+impl<T> Match<T> {
+    /// Returns true if the match result didn't match any globs.
+    pub fn is_none(&self) -> bool {
+        match *self {
+            Match::None => true,
+            Match::Ignore(_) | Match::Whitelist(_) => false,
+        }
+    }
+
+    /// Returns true if the match result implies the path should be ignored.
+    pub fn is_ignore(&self) -> bool {
+        match *self {
+            Match::Ignore(_) => true,
+            Match::None | Match::Whitelist(_) => false,
+        }
+    }
+
+    /// Returns true if the match result implies the path should be
+    /// whitelisted.
+    pub fn is_whitelist(&self) -> bool {
+        match *self {
+            Match::Whitelist(_) => true,
+            Match::None | Match::Ignore(_) => false,
+        }
+    }
+
+    /// Inverts the match so that `Ignore` becomes `Whitelist` and
+    /// `Whitelist` becomes `Ignore`. A non-match remains the same.
+    pub fn invert(self) -> Match<T> {
+        match self {
+            Match::None => Match::None,
+            Match::Ignore(t) => Match::Whitelist(t),
+            Match::Whitelist(t) => Match::Ignore(t),
+        }
+    }
+
+    /// Return the value inside this match if it exists.
+    pub fn inner(&self) -> Option<&T> {
+        match *self {
+            Match::None => None,
+            Match::Ignore(ref t) => Some(t),
+            Match::Whitelist(ref t) => Some(t),
+        }
+    }
+
+    /// Apply the given function to the value inside this match.
+    ///
+    /// If the match has no value, then return the match unchanged.
+    pub fn map<U, F: FnOnce(T) -> U>(self, f: F) -> Match<U> {
+        match self {
+            Match::None => Match::None,
+            Match::Ignore(t) => Match::Ignore(f(t)),
+            Match::Whitelist(t) => Match::Whitelist(f(t)),
+        }
+    }
+}