rust/ripgrep
1
0
Fork 0
mirror of https://github.com/BurntSushi/ripgrep.git synced 2025-06-25 14:22:54 +02:00
Code Issues Releases Activity

libripgrep: initial commit introducing libripgrep

Browse Source 
libripgrep is not any one library, but rather, a collection of libraries
that roughly separate the following key distinct phases in a grep
implementation:

  1. Pattern matching (e.g., by a regex engine).
  2. Searching a file using a pattern matcher.
  3. Printing results.

Ultimately, both (1) and (3) are defined by de-coupled interfaces, of
which there may be multiple implementations. Namely, (1) is satisfied by
the `Matcher` trait in the `grep-matcher` crate and (3) is satisfied by
the `Sink` trait in the `grep2` crate. The searcher (2) ties everything
together and finds results using a matcher and reports those results
using a `Sink` implementation.

Closes #162
This commit is contained in:
Andrew Gallant
2018-04-29 09:29:52 -04:00
parent 0958837ee1
commit d9ca529356
68 changed files with 18010 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

366
grep-printer/src/color.rs Normal file
View File

@ -0,0 +1,366 @@
use std::error;
use std::fmt;
use std::str::FromStr;
use termcolor::{Color, ColorSpec, ParseColorError};
/// An error that can occur when parsing color specifications.
#[derive(Clone, Debug, Eq, PartialEq)]
pub enum ColorError {
/// This occurs when an unrecognized output type is used.
UnrecognizedOutType(String),
/// This occurs when an unrecognized spec type is used.
UnrecognizedSpecType(String),
/// This occurs when an unrecognized color name is used.
UnrecognizedColor(String, String),
/// This occurs when an unrecognized style attribute is used.
UnrecognizedStyle(String),
/// This occurs when the format of a color specification is invalid.
InvalidFormat(String),
}
impl error::Error for ColorError {
fn description(&self) -> &str {
match *self {
ColorError::UnrecognizedOutType(_) => "unrecognized output type",
ColorError::UnrecognizedSpecType(_) => "unrecognized spec type",
ColorError::UnrecognizedColor(_, _) => "unrecognized color name",
ColorError::UnrecognizedStyle(_) => "unrecognized style attribute",
ColorError::InvalidFormat(_) => "invalid color spec",
}
}
}
impl ColorError {
fn from_parse_error(err: ParseColorError) -> ColorError {
ColorError::UnrecognizedColor(
err.invalid().to_string(),
err.to_string(),
)
}
}
impl fmt::Display for ColorError {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match *self {
ColorError::UnrecognizedOutType(ref name) => {
write!(
f,
"unrecognized output type '{}'. Choose from: \
path, line, column, match.",
name,
)
}
ColorError::UnrecognizedSpecType(ref name) => {
write!(
f,
"unrecognized spec type '{}'. Choose from: \
fg, bg, style, none.",
name,
)
}
ColorError::UnrecognizedColor(_, ref msg) => {
write!(f, "{}", msg)
}
ColorError::UnrecognizedStyle(ref name) => {
write!(
f,
"unrecognized style attribute '{}'. Choose from: \
nobold, bold, nointense, intense, nounderline, \
underline.",
name,
)
}
ColorError::InvalidFormat(ref original) => {
write!(
f,
"invalid color spec format: '{}'. Valid format \
is '(path|line|column|match):(fg|bg|style):(value)'.",
original,
)
}
}
}
}
/// A merged set of color specifications.
///
/// This set of color specifications represents the various color types that
/// are supported by the printers in this crate. A set of color specifications
/// can be created from a sequence of
/// [`UserColorSpec`s](struct.UserColorSpec.html).
#[derive(Clone, Debug, Default, Eq, PartialEq)]
pub struct ColorSpecs {
path: ColorSpec,
line: ColorSpec,
column: ColorSpec,
matched: ColorSpec,
}
/// A single color specification provided by the user.
///
/// ## Format
///
/// The format of a `Spec` is a triple: `{type}:{attribute}:{value}`. Each
/// component is defined as follows:
///
/// * `{type}` can be one of `path`, `line`, `column` or `match`.
/// * `{attribute}` can be one of `fg`, `bg` or `style`. `{attribute}` may also
/// be the special value `none`, in which case, `{value}` can be omitted.
/// * `{value}` is either a color name (for `fg`/`bg`) or a style instruction.
///
/// `{type}` controls which part of the output should be styled.
///
/// When `{attribute}` is `none`, then this should cause any existing style
/// settings to be cleared for the specified `type`.
///
/// `{value}` should be a color when `{attribute}` is `fg` or `bg`, or it
/// should be a style instruction when `{attribute}` is `style`. When
/// `{attribute}` is `none`, `{value}` must be omitted.
///
/// Valid colors are `black`, `blue`, `green`, `red`, `cyan`, `magenta`,
/// `yellow`, `white`. Extended colors can also be specified, and are formatted
/// as `x` (for 256-bit colors) or `x,x,x` (for 24-bit true color), where
/// `x` is a number between 0 and 255 inclusive. `x` may be given as a normal
/// decimal number of a hexadecimal number, where the latter is prefixed by
/// `0x`.
///
/// Valid style instructions are `nobold`, `bold`, `intense`, `nointense`,
/// `underline`, `nounderline`.
///
/// ## Example
///
/// The standard way to build a `UserColorSpec` is to parse it from a string.
/// Once multiple `UserColorSpec`s have been constructed, they can be provided
/// to the standard printer where they will automatically be applied to the
/// output.
///
/// A `UserColorSpec` can also be converted to a `termcolor::ColorSpec`:
///
/// ```rust
/// extern crate grep_printer;
/// extern crate termcolor;
///
/// # fn main() {
/// use termcolor::{Color, ColorSpec};
/// use grep_printer::UserColorSpec;
///
/// let user_spec1: UserColorSpec = "path:fg:blue".parse().unwrap();
/// let user_spec2: UserColorSpec = "match:bg:0xff,0x7f,0x00".parse().unwrap();
///
/// let spec1 = user_spec1.to_color_spec();
/// let spec2 = user_spec2.to_color_spec();
///
/// assert_eq!(spec1.fg(), Some(&Color::Blue));
/// assert_eq!(spec2.bg(), Some(&Color::Rgb(0xFF, 0x7F, 0x00)));
/// # }
/// ```
#[derive(Clone, Debug, Eq, PartialEq)]
pub struct UserColorSpec {
ty: OutType,
value: SpecValue,
}
impl UserColorSpec {
/// Convert this user provided color specification to a specification that
/// can be used with `termcolor`. This drops the type of this specification
/// (where the type indicates where the color is applied in the standard
/// printer, e.g., to the file path or the line numbers, etc.).
pub fn to_color_spec(&self) -> ColorSpec {
let mut spec = ColorSpec::default();
self.value.merge_into(&mut spec);
spec
}
}
/// The actual value given by the specification.
#[derive(Clone, Debug, Eq, PartialEq)]
enum SpecValue {
None,
Fg(Color),
Bg(Color),
Style(Style),
}
/// The set of configurable portions of ripgrep's output.
#[derive(Clone, Debug, Eq, PartialEq)]
enum OutType {
Path,
Line,
Column,
Match,
}
/// The specification type.
#[derive(Clone, Debug, Eq, PartialEq)]
enum SpecType {
Fg,
Bg,
Style,
None,
}
/// The set of available styles for use in the terminal.
#[derive(Clone, Debug, Eq, PartialEq)]
enum Style {
Bold,
NoBold,
Intense,
NoIntense,
Underline,
NoUnderline
}
impl ColorSpecs {
/// Create color specifications from a list of user supplied
/// specifications.
pub fn new(specs: &[UserColorSpec]) -> ColorSpecs {
let mut merged = ColorSpecs::default();
for spec in specs {
match spec.ty {
OutType::Path => spec.merge_into(&mut merged.path),
OutType::Line => spec.merge_into(&mut merged.line),
OutType::Column => spec.merge_into(&mut merged.column),
OutType::Match => spec.merge_into(&mut merged.matched),
}
}
merged
}
/// Return the color specification for coloring file paths.
pub fn path(&self) -> &ColorSpec {
&self.path
}
/// Return the color specification for coloring line numbers.
pub fn line(&self) -> &ColorSpec {
&self.line
}
/// Return the color specification for coloring column numbers.
pub fn column(&self) -> &ColorSpec {
&self.column
}
/// Return the color specification for coloring matched text.
pub fn matched(&self) -> &ColorSpec {
&self.matched
}
}
impl UserColorSpec {
/// Merge this spec into the given color specification.
fn merge_into(&self, cspec: &mut ColorSpec) {
self.value.merge_into(cspec);
}
}
impl SpecValue {
/// Merge this spec value into the given color specification.
fn merge_into(&self, cspec: &mut ColorSpec) {
match *self {
SpecValue::None => cspec.clear(),
SpecValue::Fg(ref color) => { cspec.set_fg(Some(color.clone())); }
SpecValue::Bg(ref color) => { cspec.set_bg(Some(color.clone())); }
SpecValue::Style(ref style) => {
match *style {
Style::Bold => { cspec.set_bold(true); }
Style::NoBold => { cspec.set_bold(false); }
Style::Intense => { cspec.set_intense(true); }
Style::NoIntense => { cspec.set_intense(false); }
Style::Underline => { cspec.set_underline(true); }
Style::NoUnderline => { cspec.set_underline(false); }
}
}
}
}
}
impl FromStr for UserColorSpec {
type Err = ColorError;
fn from_str(s: &str) -> Result<UserColorSpec, ColorError> {
let pieces: Vec<&str> = s.split(':').collect();
if pieces.len() <= 1 || pieces.len() > 3 {
return Err(ColorError::InvalidFormat(s.to_string()));
}
let otype: OutType = pieces[0].parse()?;
match pieces[1].parse()? {
SpecType::None => {
Ok(UserColorSpec {
ty: otype,
value: SpecValue::None,
})
}
SpecType::Style => {
if pieces.len() < 3 {
return Err(ColorError::InvalidFormat(s.to_string()));
}
let style: Style = pieces[2].parse()?;
Ok(UserColorSpec { ty: otype, value: SpecValue::Style(style) })
}
SpecType::Fg => {
if pieces.len() < 3 {
return Err(ColorError::InvalidFormat(s.to_string()));
}
let color: Color = pieces[2]
.parse()
.map_err(ColorError::from_parse_error)?;
Ok(UserColorSpec { ty: otype, value: SpecValue::Fg(color) })
}
SpecType::Bg => {
if pieces.len() < 3 {
return Err(ColorError::InvalidFormat(s.to_string()));
}
let color: Color = pieces[2]
.parse()
.map_err(ColorError::from_parse_error)?;
Ok(UserColorSpec { ty: otype, value: SpecValue::Bg(color) })
}
}
}
}
impl FromStr for OutType {
type Err = ColorError;
fn from_str(s: &str) -> Result<OutType, ColorError> {
match &*s.to_lowercase() {
"path" => Ok(OutType::Path),
"line" => Ok(OutType::Line),
"column" => Ok(OutType::Column),
"match" => Ok(OutType::Match),
_ => Err(ColorError::UnrecognizedOutType(s.to_string())),
}
}
}
impl FromStr for SpecType {
type Err = ColorError;
fn from_str(s: &str) -> Result<SpecType, ColorError> {
match &*s.to_lowercase() {
"fg" => Ok(SpecType::Fg),
"bg" => Ok(SpecType::Bg),
"style" => Ok(SpecType::Style),
"none" => Ok(SpecType::None),
_ => Err(ColorError::UnrecognizedSpecType(s.to_string())),
}
}
}
impl FromStr for Style {
type Err = ColorError;
fn from_str(s: &str) -> Result<Style, ColorError> {
match &*s.to_lowercase() {
"bold" => Ok(Style::Bold),
"nobold" => Ok(Style::NoBold),
"intense" => Ok(Style::Intense),
"nointense" => Ok(Style::NoIntense),
"underline" => Ok(Style::Underline),
"nounderline" => Ok(Style::NoUnderline),
_ => Err(ColorError::UnrecognizedStyle(s.to_string())),
}
}
}

90
grep-printer/src/counter.rs Normal file
View File

@ -0,0 +1,90 @@
use std::io::{self, Write};
use termcolor::{ColorSpec, WriteColor};
/// A writer that counts the number of bytes that have been successfully
/// written.
#[derive(Clone, Debug)]
pub struct CounterWriter<W> {
wtr: W,
count: u64,
total_count: u64,
}
impl<W: Write> CounterWriter<W> {
pub fn new(wtr: W) -> CounterWriter<W> {
CounterWriter { wtr: wtr, count: 0, total_count: 0 }
}
}
impl<W> CounterWriter<W> {
/// Returns the total number of bytes written since construction or the
/// last time `reset` was called.
pub fn count(&self) -> u64 {
self.count
}
/// Returns the total number of bytes written since construction.
pub fn total_count(&self) -> u64 {
self.total_count + self.count
}
/// Resets the number of bytes written to `0`.
pub fn reset_count(&mut self) {
self.total_count += self.count;
self.count = 0;
}
/// Clear resets all counting related state for this writer.
///
/// After this call, the total count of bytes written to the underlying
/// writer is erased and reset.
#[allow(dead_code)]
pub fn clear(&mut self) {
self.count = 0;
self.total_count = 0;
}
#[allow(dead_code)]
pub fn get_ref(&self) -> &W {
&self.wtr
}
pub fn get_mut(&mut self) -> &mut W {
&mut self.wtr
}
pub fn into_inner(self) -> W {
self.wtr
}
}
impl<W: Write> Write for CounterWriter<W> {
fn write(&mut self, buf: &[u8]) -> Result<usize, io::Error> {
let n = self.wtr.write(buf)?;
self.count += n as u64;
Ok(n)
}
fn flush(&mut self) -> Result<(), io::Error> {
self.wtr.flush()
}
}
impl<W: WriteColor> WriteColor for CounterWriter<W> {
fn supports_color(&self) -> bool {
self.wtr.supports_color()
}
fn set_color(&mut self, spec: &ColorSpec) -> io::Result<()> {
self.wtr.set_color(spec)
}
fn reset(&mut self) -> io::Result<()> {
self.wtr.reset()
}
fn is_synchronous(&self) -> bool {
self.wtr.is_synchronous()
}
}

921
grep-printer/src/json.rs Normal file
View File

@ -0,0 +1,921 @@
use std::io::{self, Write};
use std::path::Path;
use std::time::Instant;
use grep_matcher::{Match, Matcher};
use grep_searcher::{
Searcher,
Sink, SinkError, SinkContext, SinkContextKind, SinkFinish, SinkMatch,
};
use serde_json as json;
use counter::CounterWriter;
use jsont;
use stats::Stats;
/// The configuration for the JSON printer.
///
/// This is manipulated by the JSONBuilder and then referenced by the actual
/// implementation. Once a printer is build, the configuration is frozen and
/// cannot changed.
#[derive(Debug, Clone)]
struct Config {
pretty: bool,
max_matches: Option<u64>,
always_begin_end: bool,
}
impl Default for Config {
fn default() -> Config {
Config {
pretty: false,
max_matches: None,
always_begin_end: false,
}
}
}
/// A builder for a JSON lines printer.
///
/// The builder permits configuring how the printer behaves. The JSON printer
/// has fewer configuration options than the standard printer because it is
/// a structured format, and the printer always attempts to find the most
/// information possible.
///
/// Some configuration options, such as whether line numbers are included or
/// whether contextual lines are shown, are drawn directly from the
/// `grep_searcher::Searcher`'s configuration.
///
/// Once a `JSON` printer is built, its configuration cannot be changed.
#[derive(Clone, Debug)]
pub struct JSONBuilder {
config: Config,
}
impl JSONBuilder {
/// Return a new builder for configuring the JSON printer.
pub fn new() -> JSONBuilder {
JSONBuilder { config: Config::default() }
}
/// Create a JSON printer that writes results to the given writer.
pub fn build<W: io::Write>(&self, wtr: W) -> JSON<W> {
JSON {
config: self.config.clone(),
wtr: CounterWriter::new(wtr),
matches: vec![],
}
}
/// Print JSON in a pretty printed format.
///
/// Enabling this will no longer produce a "JSON lines" format, in that
/// each JSON object printed may span multiple lines.
///
/// This is disabled by default.
pub fn pretty(&mut self, yes: bool) -> &mut JSONBuilder {
self.config.pretty = yes;
self
}
/// Set the maximum amount of matches that are printed.
///
/// If multi line search is enabled and a match spans multiple lines, then
/// that match is counted exactly once for the purposes of enforcing this
/// limit, regardless of how many lines it spans.
pub fn max_matches(&mut self, limit: Option<u64>) -> &mut JSONBuilder {
self.config.max_matches = limit;
self
}
/// When enabled, the `begin` and `end` messages are always emitted, even
/// when no match is found.
///
/// When disabled, the `begin` and `end` messages are only shown if there
/// is at least one `match` or `context` message.
///
/// This is disabled by default.
pub fn always_begin_end(&mut self, yes: bool) -> &mut JSONBuilder {
self.config.always_begin_end = yes;
self
}
}
/// The JSON printer, which emits results in a JSON lines format.
///
/// This type is generic over `W`, which represents any implementation of
/// the standard library `io::Write` trait.
///
/// # Format
///
/// This section describes the JSON format used by this printer.
///
/// To skip the rigamarole, take a look at the
/// [example](#example)
/// at the end.
///
/// ## Overview
///
/// The format of this printer is the [JSON Lines](http://jsonlines.org/)
/// format. Specifically, this printer emits a sequence of messages, where
/// each message is encoded as a single JSON value on a single line. There are
/// four different types of messages (and this number may expand over time):
///
/// * **begin** - A message that indicates a file is being searched.
/// * **end** - A message the indicates a file is done being searched. This
/// message also include summary statistics about the search.
/// * **match** - A message that indicates a match was found. This includes
/// the text and offsets of the match.
/// * **context** - A message that indicates a contextual line was found.
/// This includes the text of the line, along with any match information if
/// the search was inverted.
///
/// Every message is encoded in the same envelope format, which includes a tag
/// indicating the message type along with an object for the payload:
///
/// ```json
/// {
/// "type": "{begin|end|match|context}",
/// "data": { ... }
/// }
/// ```
///
/// The message itself is encoded in the envelope's `data` key.
///
/// ## Text encoding
///
/// Before describing each message format, we first must briefly discuss text
/// encoding, since it factors into every type of message. In particular, JSON
/// may only be encoded in UTF-8, UTF-16 or UTF-32. For the purposes of this
/// printer, we need only worry about UTF-8. The problem here is that searching
/// is not limited to UTF-8 exclusively, which in turn implies that matches
/// may be reported that contain invalid UTF-8. Moreover, this printer may
/// also print file paths, and the encoding of file paths is itself not
/// guarnateed to be valid UTF-8. Therefore, this printer must deal with the
/// presence of invalid UTF-8 somehow. The printer could silently ignore such
/// things completely, or even lossily transcode invalid UTF-8 to valid UTF-8
/// by replacing all invalid sequences with the Unicode replacement character.
/// However, this would prevent consumers of this format from accessing the
/// original data in a non-lossy way.
///
/// Therefore, this printer will emit valid UTF-8 encoded bytes as normal
/// JSON strings and otherwise base64 encode data that isn't valid UTF-8. To
/// communicate whether this process occurs or not, strings are keyed by the
/// name `text` where as arbitrary bytes are keyed by `bytes`.
///
/// For example, when a path is included in a message, it is formatted like so,
/// if and only if the path is valid UTF-8:
///
/// ```json
/// {
/// "path": {
/// "text": "/home/ubuntu/lib.rs"
/// }
/// }
/// ```
///
/// If instead our path was `/home/ubuntu/lib\xFF.rs`, where the `\xFF` byte
/// makes it invalid UTF-8, the path would instead be encoded like so:
///
/// ```json
/// {
/// "path": {
/// "bytes": "L2hvbWUvdWJ1bnR1L2xpYv8ucnM="
/// }
/// }
/// ```
///
/// This same representation is used for reporting matches as well.
///
/// The printer guarantees that the `text` field is used whenever the
/// underlying bytes are valid UTF-8.
///
/// ## Wire format
///
/// This section documents the wire format emitted by this printer, starting
/// with the four types of messages.
///
/// Each message has its own format, and is contained inside an envelope that
/// indicates the type of message. The envelope has these fields:
///
/// * **type** - A string indicating the type of this message. It may be one
/// of four possible strings: `begin`, `end`, `match` or `context`. This
/// list may expand over time.
/// * **data** - The actual message data. The format of this field depends on
/// the value of `type`. The possible message formats are
/// [`begin`](#message-begin),
/// [`end`](#message-end),
/// [`match`](#message-match),
/// [`context`](#message-context).
///
/// #### Message: **begin**
///
/// This message indicates that a search has begun. It has these fields:
///
/// * **path** - An
/// [arbitrary data object](#object-arbitrary-data)
/// representing the file path corresponding to the search, if one is
/// present. If no file path is available, then this field is `null`.
///
/// #### Message: **end**
///
/// This message indicates that a search has finished. It has these fields:
///
/// * **path** - An
/// [arbitrary data object](#object-arbitrary-data)
/// representing the file path corresponding to the search, if one is
/// present. If no file path is available, then this field is `null`.
/// * **binary_offset** - The absolute offset in the data searched
/// corresponding to the place at which binary data was detected. If no
/// binary data was detected (or if binary detection was disabled), then this
/// field is `null`.
/// * **stats** - A [`stats` object](#object-stats) that contains summary
/// statistics for the previous search.
///
/// #### Message: **match**
///
/// This message indicates that a match has been found. A match generally
/// corresponds to a single line of text, although it may correspond to
/// multiple lines if the search can emit matches over multiple lines. It
/// has these fields:
///
/// * **path** - An
/// [arbitrary data object](#object-arbitrary-data)
/// representing the file path corresponding to the search, if one is
/// present. If no file path is available, then this field is `null`.
/// * **lines** - An
/// [arbitrary data object](#object-arbitrary-data)
/// representing one or more lines contained in this match.
/// * **line_number** - If the searcher has been configured to report line
/// numbers, then this corresponds to the line number of the first line
/// in `lines`. If no line numbers are available, then this is `null`.
/// * **absolute_offset** - The absolute byte offset corresponding to the start
/// of `lines` in the data being searched.
/// * **submatches** - An array of [`submatch` objects](#object-submatch)
/// corresponding to matches in `lines`. The offsets included in each
/// `submatch` correspond to byte offsets into `lines`. (If `lines` is base64
/// encoded, then the byte offsets correspond to the data after base64
/// decoding.) The `submatch` objects are guaranteed to be sorted by their
/// starting offsets. Note that it is possible for this array to be empty,
/// for example, when searching reports inverted matches.
///
/// #### Message: **context**
///
/// This message indicates that a contextual line has been found. A contextual
/// line is a line that doesn't contain a match, but is generally adjacent to
/// a line that does contain a match. The precise way in which contextual lines
/// are reported is determined by the searcher. It has these fields, which are
/// exactly the same fields found in a [`match`](#message-match):
///
/// * **path** - An
/// [arbitrary data object](#object-arbitrary-data)
/// representing the file path corresponding to the search, if one is
/// present. If no file path is available, then this field is `null`.
/// * **lines** - An
/// [arbitrary data object](#object-arbitrary-data)
/// representing one or more lines contained in this context. This includes
/// line terminators, if they're present.
/// * **line_number** - If the searcher has been configured to report line
/// numbers, then this corresponds to the line number of the first line
/// in `lines`. If no line numbers are available, then this is `null`.
/// * **absolute_offset** - The absolute byte offset corresponding to the start
/// of `lines` in the data being searched.
/// * **submatches** - An array of [`submatch` objects](#object-submatch)
/// corresponding to matches in `lines`. The offsets included in each
/// `submatch` correspond to byte offsets into `lines`. (If `lines` is base64
/// encoded, then the byte offsets correspond to the data after base64
/// decoding.) The `submatch` objects are guaranteed to be sorted by
/// their starting offsets. Note that it is possible for this array to be
/// non-empty, for example, when searching reports inverted matches such that
/// the original matcher could match things in the contextual lines.
///
/// #### Object: **submatch**
///
/// This object describes submatches found within `match` or `context`
/// messages. The `start` and `end` fields indicate the half-open interval on
/// which the match occurs (`start` is included, but `end` is not). It is
/// guaranteed that `start <= end`. It has these fields:
///
/// * **match** - An
/// [arbitrary data object](#object-arbitrary-data)
/// corresponding to the text in this submatch.
/// * **start** - A byte offset indicating the start of this match. This offset
/// is generally reported in terms of the parent object's data. For example,
/// the `lines` field in the
/// [`match`](#message-match) or [`context`](#message-context)
/// messages.
/// * **end** - A byte offset indicating the end of this match. This offset
/// is generally reported in terms of the parent object's data. For example,
/// the `lines` field in the
/// [`match`](#message-match) or [`context`](#message-context)
/// messages.
///
/// #### Object: **stats**
///
/// This object is included in messages and contains summary statistics about
/// a search. It has these fields:
///
/// * **elapsed** - A [`duration` object](#object-duration) describing the
/// length of time that elapsed while performing the search.
/// * **searches** - The number of searches that have run. For this printer,
/// this value is always `1`. (Implementations may emit additional message
/// types that use this same `stats` object that represents summary
/// statistics over multiple searches.)
/// * **searches_with_match** - The number of searches that have run that have
/// found at least one match. This is never more than `searches`.
/// * **bytes_searched** - The total number of bytes that have been searched.
/// * **bytes_printed** - The total number of bytes that have been printed.
/// This includes everything emitted by this printer.
/// * **matched_lines** - The total number of lines that participated in a
/// match. When matches may contain multiple lines, then this includes every
/// line that is part of every match.
/// * **matches** - The total number of matches. There may be multiple matches
/// per line. When matches may contain multiple lines, each match is counted
/// only once, regardless of how many lines it spans.
///
/// #### Object: **duration**
///
/// This object includes a few fields for describing a duration. Two of its
/// fields, `secs` and `nanos`, can be combined to give nanosecond precision
/// on systems that support it. It has these fields:
///
/// * **secs** - A whole number of seconds indicating the length of this
/// duration.
/// * **nanos** - A fractional part of this duration represent by nanoseconds.
/// If nanosecond precision isn't supported, then this is typically rounded
/// up to the nearest number of nanoseconds.
/// * **human** - A human readable string describing the length of the
/// duration. The format of the string is itself unspecified.
///
/// #### Object: **arbitrary data**
///
/// This object is used whenever arbitrary data needs to be represented as a
/// JSON value. This object contains two fields, where generally only one of
/// the fields is present:
///
/// * **text** - A normal JSON string that is UTF-8 encoded. This field is
/// populated if and only if the underlying data is valid UTF-8.
/// * **bytes** - A normal JSON string that is a base64 encoding of the
/// underlying bytes.
///
/// More information on the motivation for this representation can be seen in
/// the section [text encoding](#text-encoding) above.
///
/// ## Example
///
/// This section shows a small example that includes all message types.
///
/// Here's the file we want to search, located at `/home/andrew/sherlock`:
///
/// ```text
/// For the Doctor Watsons of this world, as opposed to the Sherlock
/// Holmeses, success in the province of detective work must always
/// be, to a very large extent, the result of luck. Sherlock Holmes
/// can extract a clew from a wisp of straw or a flake of cigar ash;
/// but Doctor Watson has to have it taken out for him and dusted,
/// and exhibited clearly, with a label attached.
/// ```
///
/// Searching for `Watson` with a `before_context` of `1` with line numbers
/// enabled shows something like this using the standard printer:
///
/// ```text
/// sherlock:1:For the Doctor Watsons of this world, as opposed to the Sherlock
/// --
/// sherlock-4-can extract a clew from a wisp of straw or a flake of cigar ash;
/// sherlock:5:but Doctor Watson has to have it taken out for him and dusted,
/// ```
///
/// Here's what the same search looks like using the JSON wire format described
/// above, where in we show semi-prettified JSON (instead of a strict JSON
/// Lines format), for illustrative purposes:
///
/// ```json
/// {
/// "type": "begin",
/// "data": {
/// "path": {"text": "/home/andrew/sherlock"}}
/// }
/// }
/// {
/// "type": "match",
/// "data": {
/// "path": {"text": "/home/andrew/sherlock"},
/// "lines": {"text": "For the Doctor Watsons of this world, as opposed to the Sherlock\n"},
/// "line_number": 1,
/// "absolute_offset": 0,
/// "submatches": [
/// {"match": {"text": "Watson"}, "start": 15, "end": 21}
/// ]
/// }
/// }
/// {
/// "type": "context",
/// "data": {
/// "path": {"text": "/home/andrew/sherlock"},
/// "lines": {"text": "can extract a clew from a wisp of straw or a flake of cigar ash;\n"},
/// "line_number": 4,
/// "absolute_offset": 193,
/// "submatches": []
/// }
/// }
/// {
/// "type": "match",
/// "data": {
/// "path": {"text": "/home/andrew/sherlock"},
/// "lines": {"text": "but Doctor Watson has to have it taken out for him and dusted,\n"},
/// "line_number": 5,
/// "absolute_offset": 258,
/// "submatches": [
/// {"match": {"text": "Watson"}, "start": 11, "end": 17}
/// ]
/// }
/// }
/// {
/// "type": "end",
/// "data": {
/// "path": {"text": "/home/andrew/sherlock"},
/// "binary_offset": null,
/// "stats": {
/// "elapsed": {"secs": 0, "nanos": 36296, "human": "0.0000s"},
/// "searches": 1,
/// "searches_with_match": 1,
/// "bytes_searched": 367,
/// "bytes_printed": 1151,
/// "matched_lines": 2,
/// "matches": 2
/// }
/// }
/// }
/// ```
#[derive(Debug)]
pub struct JSON<W> {
config: Config,
wtr: CounterWriter<W>,
matches: Vec<Match>,
}
impl<W: io::Write> JSON<W> {
/// Return a JSON lines printer with a default configuration that writes
/// matches to the given writer.
pub fn new(wtr: W) -> JSON<W> {
JSONBuilder::new().build(wtr)
}
/// Return an implementation of `Sink` for the JSON printer.
///
/// This does not associate the printer with a file path, which means this
/// implementation will never print a file path along with the matches.
pub fn sink<'s, M: Matcher>(
&'s mut self,
matcher: M,
) -> JSONSink<'static, 's, M, W> {
JSONSink {
matcher: matcher,
json: self,
path: None,
start_time: Instant::now(),
match_count: 0,
after_context_remaining: 0,
binary_byte_offset: None,
begin_printed: false,
stats: Stats::new(),
}
}
/// Return an implementation of `Sink` associated with a file path.
///
/// When the printer is associated with a path, then it may, depending on
/// its configuration, print the path along with the matches found.
pub fn sink_with_path<'p, 's, M, P>(
&'s mut self,
matcher: M,
path: &'p P,
) -> JSONSink<'p, 's, M, W>
where M: Matcher,
P: ?Sized + AsRef<Path>,
{
JSONSink {
matcher: matcher,
json: self,
path: Some(path.as_ref()),
start_time: Instant::now(),
match_count: 0,
after_context_remaining: 0,
binary_byte_offset: None,
begin_printed: false,
stats: Stats::new(),
}
}
/// Write the given message followed by a new line. The new line is
/// determined from the configuration of the given searcher.
fn write_message(&mut self, message: &jsont::Message) -> io::Result<()> {
if self.config.pretty {
json::to_writer_pretty(&mut self.wtr, message)?;
} else {
json::to_writer(&mut self.wtr, message)?;
}
self.wtr.write(&[b'\n'])?;
Ok(())
}
}
impl<W> JSON<W> {
/// Returns true if and only if this printer has written at least one byte
/// to the underlying writer during any of the previous searches.
pub fn has_written(&self) -> bool {
self.wtr.total_count() > 0
}
/// Return a mutable reference to the underlying writer.
pub fn get_mut(&mut self) -> &mut W {
self.wtr.get_mut()
}
/// Consume this printer and return back ownership of the underlying
/// writer.
pub fn into_inner(self) -> W {
self.wtr.into_inner()
}
}
/// An implementation of `Sink` associated with a matcher and an optional file
/// path for the JSON printer.
///
/// This type is generic over a few type parameters:
///
/// * `'p` refers to the lifetime of the file path, if one is provided. When
/// no file path is given, then this is `'static`.
/// * `'s` refers to the lifetime of the
/// [`JSON`](struct.JSON.html)
/// printer that this type borrows.
/// * `M` refers to the type of matcher used by
/// `grep_searcher::Searcher` that is reporting results to this sink.
/// * `W` refers to the underlying writer that this printer is writing its
/// output to.
#[derive(Debug)]
pub struct JSONSink<'p, 's, M: Matcher, W: 's> {
matcher: M,
json: &'s mut JSON<W>,
path: Option<&'p Path>,
start_time: Instant,
match_count: u64,
after_context_remaining: u64,
binary_byte_offset: Option<u64>,
begin_printed: bool,
stats: Stats,
}
impl<'p, 's, M: Matcher, W: io::Write> JSONSink<'p, 's, M, W> {
/// Returns true if and only if this printer received a match in the
/// previous search.
///
/// This is unaffected by the result of searches before the previous
/// search.
pub fn has_match(&self) -> bool {
self.match_count > 0
}
/// Return the total number of matches reported to this sink.
///
/// This corresponds to the number of times `Sink::matched` is called.
pub fn match_count(&self) -> u64 {
self.match_count
}
/// If binary data was found in the previous search, this returns the
/// offset at which the binary data was first detected.
///
/// The offset returned is an absolute offset relative to the entire
/// set of bytes searched.
///
/// This is unaffected by the result of searches before the previous
/// search. e.g., If the search prior to the previous search found binary
/// data but the previous search found no binary data, then this will
/// return `None`.
pub fn binary_byte_offset(&self) -> Option<u64> {
self.binary_byte_offset
}
/// Return a reference to the stats produced by the printer for all
/// searches executed on this sink.
pub fn stats(&self) -> &Stats {
&self.stats
}
/// Execute the matcher over the given bytes and record the match
/// locations if the current configuration demands match granularity.
fn record_matches(&mut self, bytes: &[u8]) -> io::Result<()> {
self.json.matches.clear();
// If printing requires knowing the location of each individual match,
// then compute and stored those right now for use later. While this
// adds an extra copy for storing the matches, we do amortize the
// allocation for it and this greatly simplifies the printing logic to
// the extent that it's easy to ensure that we never do more than
// one search to find the matches.
let matches = &mut self.json.matches;
self.matcher.find_iter(bytes, |m| {
matches.push(m);
true
}).map_err(io::Error::error_message)?;
// Don't report empty matches appearing at the end of the bytes.
if !matches.is_empty()
&& matches.last().unwrap().is_empty()
&& matches.last().unwrap().start() >= bytes.len()
{
matches.pop().unwrap();
}
Ok(())
}
/// Returns true if this printer should quit.
///
/// This implements the logic for handling quitting after seeing a certain
/// amount of matches. In most cases, the logic is simple, but we must
/// permit all "after" contextual lines to print after reaching the limit.
fn should_quit(&self) -> bool {
let limit = match self.json.config.max_matches {
None => return false,
Some(limit) => limit,
};
if self.match_count < limit {
return false;
}
self.after_context_remaining == 0
}
/// Write the "begin" message.
fn write_begin_message(&mut self) -> io::Result<()> {
if self.begin_printed {
return Ok(());
}
let msg = jsont::Message::Begin(jsont::Begin {
path: self.path,
});
self.json.write_message(&msg)?;
self.begin_printed = true;
Ok(())
}
}
impl<'p, 's, M: Matcher, W: io::Write> Sink for JSONSink<'p, 's, M, W> {
type Error = io::Error;
fn matched(
&mut self,
searcher: &Searcher,
mat: &SinkMatch,
) -> Result<bool, io::Error> {
self.write_begin_message()?;
self.match_count += 1;
self.after_context_remaining = searcher.after_context() as u64;
self.record_matches(mat.bytes())?;
self.stats.add_matches(self.json.matches.len() as u64);
self.stats.add_matched_lines(mat.lines().count() as u64);
let submatches = SubMatches::new(mat.bytes(), &self.json.matches);
let msg = jsont::Message::Match(jsont::Match {
path: self.path,
lines: mat.bytes(),
line_number: mat.line_number(),
absolute_offset: mat.absolute_byte_offset(),
submatches: submatches.as_slice(),
});
self.json.write_message(&msg)?;
Ok(!self.should_quit())
}
fn context(
&mut self,
searcher: &Searcher,
ctx: &SinkContext,
) -> Result<bool, io::Error> {
self.write_begin_message()?;
self.json.matches.clear();
if ctx.kind() == &SinkContextKind::After {
self.after_context_remaining =
self.after_context_remaining.saturating_sub(1);
}
let submatches =
if searcher.invert_match() {
self.record_matches(ctx.bytes())?;
SubMatches::new(ctx.bytes(), &self.json.matches)
} else {
SubMatches::empty()
};
let msg = jsont::Message::Context(jsont::Context {
path: self.path,
lines: ctx.bytes(),
line_number: ctx.line_number(),
absolute_offset: ctx.absolute_byte_offset(),
submatches: submatches.as_slice(),
});
self.json.write_message(&msg)?;
Ok(!self.should_quit())
}
fn begin(
&mut self,
_searcher: &Searcher,
) -> Result<bool, io::Error> {
self.json.wtr.reset_count();
self.start_time = Instant::now();
self.match_count = 0;
self.after_context_remaining = 0;
self.binary_byte_offset = None;
if self.json.config.max_matches == Some(0) {
return Ok(false);
}
if !self.json.config.always_begin_end {
return Ok(true);
}
self.write_begin_message()?;
Ok(true)
}
fn finish(
&mut self,
_searcher: &Searcher,
finish: &SinkFinish,
) -> Result<(), io::Error> {
if !self.begin_printed {
return Ok(());
}
self.binary_byte_offset = finish.binary_byte_offset();
self.stats.add_elapsed(self.start_time.elapsed());
self.stats.add_searches(1);
if self.match_count > 0 {
self.stats.add_searches_with_match(1);
}
self.stats.add_bytes_searched(finish.byte_count());
self.stats.add_bytes_printed(self.json.wtr.count());
let msg = jsont::Message::End(jsont::End {
path: self.path,
binary_offset: finish.binary_byte_offset(),
stats: self.stats.clone(),
});
self.json.write_message(&msg)?;
Ok(())
}
}
/// SubMatches represents a set of matches in a contiguous range of bytes.
///
/// A simpler representation for this would just simply be `Vec<SubMatch>`,
/// but the common case is exactly one match per range of bytes, which we
/// specialize here using a fixed size array without any allocation.
enum SubMatches<'a> {
Empty,
Small([jsont::SubMatch<'a>; 1]),
Big(Vec<jsont::SubMatch<'a>>),
}
impl<'a> SubMatches<'a> {
/// Create a new set of match ranges from a set of matches and the
/// corresponding bytes that those matches apply to.
fn new(bytes: &'a[u8], matches: &[Match]) -> SubMatches<'a> {
if matches.len() == 1 {
let mat = matches[0];
SubMatches::Small([jsont::SubMatch {
m: &bytes[mat],
start: mat.start(),
end: mat.end(),
}])
} else {
let mut match_ranges = vec![];
for &mat in matches {
match_ranges.push(jsont::SubMatch {
m: &bytes[mat],
start: mat.start(),
end: mat.end(),
});
}
SubMatches::Big(match_ranges)
}
}
/// Create an empty set of match ranges.
fn empty() -> SubMatches<'static> {
SubMatches::Empty
}
/// Return this set of match ranges as a slice.
fn as_slice(&self) -> &[jsont::SubMatch] {
match *self {
SubMatches::Empty => &[],
SubMatches::Small(ref x) => x,
SubMatches::Big(ref x) => x,
}
}
}
#[cfg(test)]
mod tests {
use grep_regex::RegexMatcher;
use grep_searcher::SearcherBuilder;
use super::{JSON, JSONBuilder};
const SHERLOCK: &'static [u8] = b"\
For the Doctor Watsons of this world, as opposed to the Sherlock
Holmeses, success in the province of detective work must always
be, to a very large extent, the result of luck. Sherlock Holmes
can extract a clew from a wisp of straw or a flake of cigar ash;
but Doctor Watson has to have it taken out for him and dusted,
and exhibited clearly, with a label attached.
";
fn printer_contents(
printer: &mut JSON<Vec<u8>>,
) -> String {
String::from_utf8(printer.get_mut().to_owned()).unwrap()
}
#[test]
fn binary_detection() {
use grep_searcher::BinaryDetection;
const BINARY: &'static [u8] = b"\
For the Doctor Watsons of this world, as opposed to the Sherlock
Holmeses, success in the province of detective work must always
be, to a very large extent, the result of luck. Sherlock Holmes
can extract a clew \x00 from a wisp of straw or a flake of cigar ash;
but Doctor Watson has to have it taken out for him and dusted,
and exhibited clearly, with a label attached.\
";
let matcher = RegexMatcher::new(
r"Watson"
).unwrap();
let mut printer = JSONBuilder::new()
.build(vec![]);
SearcherBuilder::new()
.binary_detection(BinaryDetection::quit(b'\x00'))
.heap_limit(Some(80))
.build()
.search_reader(&matcher, BINARY, printer.sink(&matcher))
.unwrap();
let got = printer_contents(&mut printer);
assert_eq!(got.lines().count(), 3);
let last = got.lines().last().unwrap();
assert!(last.contains(r#""binary_offset":212,"#));
}
#[test]
fn max_matches() {
let matcher = RegexMatcher::new(
r"Watson"
).unwrap();
let mut printer = JSONBuilder::new()
.max_matches(Some(1))
.build(vec![]);
SearcherBuilder::new()
.build()
.search_reader(&matcher, SHERLOCK, printer.sink(&matcher))
.unwrap();
let got = printer_contents(&mut printer);
assert_eq!(got.lines().count(), 3);
}
#[test]
fn no_match() {
let matcher = RegexMatcher::new(
r"DOES NOT MATCH"
).unwrap();
let mut printer = JSONBuilder::new()
.build(vec![]);
SearcherBuilder::new()
.build()
.search_reader(&matcher, SHERLOCK, printer.sink(&matcher))
.unwrap();
let got = printer_contents(&mut printer);
assert!(got.is_empty());
}
#[test]
fn always_begin_end_no_match() {
let matcher = RegexMatcher::new(
r"DOES NOT MATCH"
).unwrap();
let mut printer = JSONBuilder::new()
.always_begin_end(true)
.build(vec![]);
SearcherBuilder::new()
.build()
.search_reader(&matcher, SHERLOCK, printer.sink(&matcher))
.unwrap();
let got = printer_contents(&mut printer);
assert_eq!(got.lines().count(), 2);
assert!(got.contains("begin") && got.contains("end"));
}
}

213
grep-printer/src/jsont.rs Normal file
View File

@ -0,0 +1,213 @@
// This module defines the types we use for JSON serialization. We specifically
// omit deserialization, partially because there isn't a clear use case for
// them at this time, but also because deserialization will complicate things.
// Namely, the types below are designed in a way that permits JSON
// serialization with little or no allocation. Allocation is often quite
// convenient for deserialization however, so these types would become a bit
// more complex.
use std::borrow::Cow;
use std::path::Path;
use std::str;
use base64;
use serde::{Serialize, Serializer};
use stats::Stats;
#[derive(Serialize)]
#[serde(tag = "type", content = "data")]
#[serde(rename_all = "snake_case")]
pub enum Message<'a> {
Begin(Begin<'a>),
End(End<'a>),
Match(Match<'a>),
Context(Context<'a>),
}
#[derive(Serialize)]
pub struct Begin<'a> {
#[serde(serialize_with = "ser_path")]
pub path: Option<&'a Path>,
}
#[derive(Serialize)]
pub struct End<'a> {
#[serde(serialize_with = "ser_path")]
pub path: Option<&'a Path>,
pub binary_offset: Option<u64>,
pub stats: Stats,
}
#[derive(Serialize)]
pub struct Match<'a> {
#[serde(serialize_with = "ser_path")]
pub path: Option<&'a Path>,
#[serde(serialize_with = "ser_bytes")]
pub lines: &'a [u8],
pub line_number: Option<u64>,
pub absolute_offset: u64,
pub submatches: &'a [SubMatch<'a>],
}
#[derive(Serialize)]
pub struct Context<'a> {
#[serde(serialize_with = "ser_path")]
pub path: Option<&'a Path>,
#[serde(serialize_with = "ser_bytes")]
pub lines: &'a [u8],
pub line_number: Option<u64>,
pub absolute_offset: u64,
pub submatches: &'a [SubMatch<'a>],
}
#[derive(Serialize)]
pub struct SubMatch<'a> {
#[serde(rename = "match")]
#[serde(serialize_with = "ser_bytes")]
pub m: &'a [u8],
pub start: usize,
pub end: usize,
}
/// Data represents things that look like strings, but may actually not be
/// valid UTF-8. To handle this, `Data` is serialized as an object with one
/// of two keys: `text` (for valid UTF-8) or `bytes` (for invalid UTF-8).
///
/// The happy path is valid UTF-8, which streams right through as-is, since
/// it is natively supported by JSON. When invalid UTF-8 is found, then it is
/// represented as arbitrary bytes and base64 encoded.
#[derive(Clone, Debug, Hash, PartialEq, Eq, Serialize)]
#[serde(untagged)]
enum Data<'a> {
Text { text: Cow<'a, str> },
Bytes {
#[serde(serialize_with = "to_base64")]
bytes: &'a [u8],
},
}
impl<'a> Data<'a> {
fn from_bytes(bytes: &[u8]) -> Data {
match str::from_utf8(bytes) {
Ok(text) => Data::Text { text: Cow::Borrowed(text) },
Err(_) => Data::Bytes { bytes },
}
}
#[cfg(unix)]
fn from_path(path: &Path) -> Data {
use std::os::unix::ffi::OsStrExt;
match path.to_str() {
Some(text) => Data::Text { text: Cow::Borrowed(text) },
None => Data::Bytes { bytes: path.as_os_str().as_bytes() },
}
}
#[cfg(not(unix))]
fn from_path(path: &Path) -> Data {
// Using lossy conversion means some paths won't round trip precisely,
// but it's not clear what we should actually do. Serde rejects
// non-UTF-8 paths, and OsStr's are serialized as a sequence of UTF-16
// code units on Windows. Neither seem appropriate for this use case,
// so we do the easy thing for now.
Data::Text { text: path.to_string_lossy() }
}
// Unused deserialization routines.
/*
fn into_bytes(self) -> Vec<u8> {
match self {
Data::Text { text } => text.into_bytes(),
Data::Bytes { bytes } => bytes,
}
}
#[cfg(unix)]
fn into_path_buf(&self) -> PathBuf {
use std::os::unix::ffi::OsStrExt;
match self {
Data::Text { text } => PathBuf::from(text),
Data::Bytes { bytes } => {
PathBuf::from(OsStr::from_bytes(bytes))
}
}
}
#[cfg(not(unix))]
fn into_path_buf(&self) -> PathBuf {
match self {
Data::Text { text } => PathBuf::from(text),
Data::Bytes { bytes } => {
PathBuf::from(String::from_utf8_lossy(&bytes).into_owned())
}
}
}
*/
}
fn to_base64<T, S>(
bytes: T,
ser: S,
) -> Result<S::Ok, S::Error>
where T: AsRef<[u8]>,
S: Serializer
{
ser.serialize_str(&base64::encode(&bytes))
}
fn ser_bytes<T, S>(
bytes: T,
ser: S,
) -> Result<S::Ok, S::Error>
where T: AsRef<[u8]>,
S: Serializer
{
Data::from_bytes(bytes.as_ref()).serialize(ser)
}
fn ser_path<P, S>(
path: &Option<P>,
ser: S,
) -> Result<S::Ok, S::Error>
where P: AsRef<Path>,
S: Serializer
{
path.as_ref().map(|p| Data::from_path(p.as_ref())).serialize(ser)
}
// The following are some deserialization helpers, in case we decide to support
// deserialization of the above types.
/*
fn from_base64<'de, D>(
de: D,
) -> Result<Vec<u8>, D::Error>
where D: Deserializer<'de>
{
let encoded = String::deserialize(de)?;
let decoded = base64::decode(encoded.as_bytes())
.map_err(D::Error::custom)?;
Ok(decoded)
}
fn deser_bytes<'de, D>(
de: D,
) -> Result<Vec<u8>, D::Error>
where D: Deserializer<'de>
{
Data::deserialize(de).map(|datum| datum.into_bytes())
}
fn deser_path<'de, D>(
de: D,
) -> Result<Option<PathBuf>, D::Error>
where D: Deserializer<'de>
{
Option::<Data>::deserialize(de)
.map(|opt| opt.map(|datum| datum.into_path_buf()))
}
*/

106
grep-printer/src/lib.rs Normal file
View File

@ -0,0 +1,106 @@
/*!
This crate provides featureful and fast printers that interoperate with the
[`grep-searcher`](https://docs.rs/grep-searcher)
crate.
# Brief overview
The [`Standard`](struct.Standard.html) printer shows results in a human
readable format, and is modeled after the formats used by standard grep-like
tools. Features include, but are not limited to, cross platform terminal
coloring, search & replace, multi-line result handling and reporting summary
statistics.
The [`JSON`](struct.JSON.html) printer shows results in a machine readable
format. To facilitate a stream of search results, the format uses
[JSON Lines](http://jsonlines.org/)
by emitting a series of messages as search results are found.
The [`Summary`](struct.Summary.html) printer shows *aggregate* results for a
single search in a human readable format, and is modeled after similar formats
found in standard grep-like tools. This printer is useful for showing the total
number of matches and/or printing file paths that either contain or don't
contain matches.
# Example
This example shows how to create a "standard" printer and execute a search.
```
extern crate grep_regex;
extern crate grep_printer;
extern crate grep_searcher;
use std::error::Error;
use grep_regex::RegexMatcher;
use grep_printer::Standard;
use grep_searcher::Searcher;
const SHERLOCK: &'static [u8] = b"\
For the Doctor Watsons of this world, as opposed to the Sherlock
Holmeses, success in the province of detective work must always
be, to a very large extent, the result of luck. Sherlock Holmes
can extract a clew from a wisp of straw or a flake of cigar ash;
but Doctor Watson has to have it taken out for him and dusted,
and exhibited clearly, with a label attached.
";
# fn main() { example().unwrap(); }
fn example() -> Result<(), Box<Error>> {
let matcher = RegexMatcher::new(r"Sherlock")?;
let mut printer = Standard::new_no_color(vec![]);
Searcher::new().search_slice(&matcher, SHERLOCK, printer.sink(&matcher))?;
// into_inner gives us back the underlying writer we provided to
// new_no_color, which is wrapped in a termcolor::NoColor. Thus, a second
// into_inner gives us back the actual buffer.
let output = String::from_utf8(printer.into_inner().into_inner())?;
let expected = "\
1:For the Doctor Watsons of this world, as opposed to the Sherlock
3:be, to a very large extent, the result of luck. Sherlock Holmes
";
assert_eq!(output, expected);
Ok(())
}
```
*/
#![deny(missing_docs)]
#[cfg(feature = "serde1")]
extern crate base64;
extern crate grep_matcher;
#[cfg(test)]
extern crate grep_regex;
extern crate grep_searcher;
#[cfg(feature = "serde1")]
extern crate serde;
#[cfg(feature = "serde1")]
#[macro_use]
extern crate serde_derive;
#[cfg(feature = "serde1")]
extern crate serde_json;
extern crate termcolor;
pub use color::{ColorError, ColorSpecs, UserColorSpec};
#[cfg(feature = "serde1")]
pub use json::{JSON, JSONBuilder, JSONSink};
pub use standard::{Standard, StandardBuilder, StandardSink};
pub use stats::Stats;
pub use summary::{Summary, SummaryBuilder, SummaryKind, SummarySink};
pub use util::PrinterPath;
#[macro_use]
mod macros;
mod color;
mod counter;
#[cfg(feature = "serde1")]
mod json;
#[cfg(feature = "serde1")]
mod jsont;
mod standard;
mod stats;
mod summary;
mod util;

23
grep-printer/src/macros.rs Normal file
View File

@ -0,0 +1,23 @@
#[cfg(test)]
#[macro_export]
macro_rules! assert_eq_printed {
($expected:expr, $got:expr) => {
let expected = &*$expected;
let got = &*$got;
if expected != got {
panic!("
printed outputs differ!
expected:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
{}
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
got:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
{}
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
", expected, got);
}
}
}

3049
grep-printer/src/standard.rs Normal file
View File

File diff suppressed because it is too large Load Diff

147
grep-printer/src/stats.rs Normal file
View File

@ -0,0 +1,147 @@
use std::ops::{Add, AddAssign};
use std::time::Duration;
use util::NiceDuration;
/// Summary statistics produced at the end of a search.
///
/// When statistics are reported by a printer, they correspond to all searches
/// executed with that printer.
#[derive(Clone, Debug, Default, PartialEq, Eq)]
#[cfg_attr(feature = "serde1", derive(Serialize))]
pub struct Stats {
elapsed: NiceDuration,
searches: u64,
searches_with_match: u64,
bytes_searched: u64,
bytes_printed: u64,
matched_lines: u64,
matches: u64,
}
impl Add for Stats {
type Output = Stats;
fn add(self, rhs: Stats) -> Stats {
self + &rhs
}
}
impl<'a> Add<&'a Stats> for Stats {
type Output = Stats;
fn add(self, rhs: &'a Stats) -> Stats {
Stats {
elapsed: NiceDuration(self.elapsed.0 + rhs.elapsed.0),
searches: self.searches + rhs.searches,
searches_with_match:
self.searches_with_match + rhs.searches_with_match,
bytes_searched: self.bytes_searched + rhs.bytes_searched,
bytes_printed: self.bytes_printed + rhs.bytes_printed,
matched_lines: self.matched_lines + rhs.matched_lines,
matches: self.matches + rhs.matches,
}
}
}
impl AddAssign for Stats {
fn add_assign(&mut self, rhs: Stats) {
*self += &rhs;
}
}
impl<'a> AddAssign<&'a Stats> for Stats {
fn add_assign(&mut self, rhs: &'a Stats) {
self.elapsed.0 += rhs.elapsed.0;
self.searches += rhs.searches;
self.searches_with_match += rhs.searches_with_match;
self.bytes_searched += rhs.bytes_searched;
self.bytes_printed += rhs.bytes_printed;
self.matched_lines += rhs.matched_lines;
self.matches += rhs.matches;
}
}
impl Stats {
/// Return a new value for tracking aggregate statistics across searches.
///
/// All statistics are set to `0`.
pub fn new() -> Stats {
Stats::default()
}
/// Return the total amount of time elapsed.
pub fn elapsed(&self) -> Duration {
self.elapsed.0
}
/// Return the total number of searches executed.
pub fn searches(&self) -> u64 {
self.searches
}
/// Return the total number of searches that found at least one match.
pub fn searches_with_match(&self) -> u64 {
self.searches_with_match
}
/// Return the total number of bytes searched.
pub fn bytes_searched(&self) -> u64 {
self.bytes_searched
}
/// Return the total number of bytes printed.
pub fn bytes_printed(&self) -> u64 {
self.bytes_printed
}
/// Return the total number of lines that participated in a match.
///
/// When matches may contain multiple lines then this includes every line
/// that is part of every match.
pub fn matched_lines(&self) -> u64 {
self.matched_lines
}
/// Return the total number of matches.
///
/// There may be multiple matches per line.
pub fn matches(&self) -> u64 {
self.matches
}
/// Add to the elapsed time.
pub fn add_elapsed(&mut self, duration: Duration) {
self.elapsed.0 += duration;
}
/// Add to the number of searches executed.
pub fn add_searches(&mut self, n: u64) {
self.searches += n;
}
/// Add to the number of searches that found at least one match.
pub fn add_searches_with_match(&mut self, n: u64) {
self.searches_with_match += n;
}
/// Add to the total number of bytes searched.
pub fn add_bytes_searched(&mut self, n: u64) {
self.bytes_searched += n;
}
/// Add to the total number of bytes printed.
pub fn add_bytes_printed(&mut self, n: u64) {
self.bytes_printed += n;
}
/// Add to the total number of lines that participated in a match.
pub fn add_matched_lines(&mut self, n: u64) {
self.matched_lines += n;
}
/// Add to the total number of matches.
pub fn add_matches(&mut self, n: u64) {
self.matches += n;
}
}

1068
grep-printer/src/summary.rs Normal file
View File

File diff suppressed because it is too large Load Diff

392
grep-printer/src/util.rs Normal file
View File

@ -0,0 +1,392 @@
use std::borrow::Cow;
use std::fmt;
use std::io;
use std::path::Path;
use std::time;
use grep_matcher::{Captures, LineTerminator, Match, Matcher};
use grep_searcher::{
LineIter,
SinkError, SinkContext, SinkContextKind, SinkMatch,
};
#[cfg(feature = "serde1")]
use serde::{Serialize, Serializer};
/// A type for handling replacements while amortizing allocation.
pub struct Replacer<M: Matcher> {
space: Option<Space<M>>,
}
struct Space<M: Matcher> {
/// The place to store capture locations.
caps: M::Captures,
/// The place to write a replacement to.
dst: Vec<u8>,
/// The place to store match offsets in terms of `dst`.
matches: Vec<Match>,
}
impl<M: Matcher> fmt::Debug for Replacer<M> {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
let (dst, matches) = self.replacement().unwrap_or((&[], &[]));
f.debug_struct("Replacer")
.field("dst", &dst)
.field("matches", &matches)
.finish()
}
}
impl<M: Matcher> Replacer<M> {
/// Create a new replacer for use with a particular matcher.
///
/// This constructor does not allocate. Instead, space for dealing with
/// replacements is allocated lazily only when needed.
pub fn new() -> Replacer<M> {
Replacer { space: None }
}
/// Executes a replacement on the given subject string by replacing all
/// matches with the given replacement. To access the result of the
/// replacement, use the `replacement` method.
///
/// This can fail if the underlying matcher reports an error.
pub fn replace_all<'a>(
&'a mut self,
matcher: &M,
subject: &[u8],
replacement: &[u8],
) -> io::Result<()> {
{
let &mut Space {
ref mut dst,
ref mut caps,
ref mut matches,
} = self.allocate(matcher)?;
dst.clear();
matches.clear();
matcher.replace_with_captures(
subject,
caps,
dst,
|caps, dst| {
let start = dst.len();
caps.interpolate(
|name| matcher.capture_index(name),
subject,
replacement,
dst,
);
let end = dst.len();
matches.push(Match::new(start, end));
true
},
).map_err(io::Error::error_message)?;
}
Ok(())
}
/// Return the result of the prior replacement and the match offsets for
/// all replacement occurrences within the returned replacement buffer.
///
/// If no replacement has occurred then `None` is returned.
pub fn replacement<'a>(&'a self) -> Option<(&'a [u8], &'a [Match])> {
match self.space {
None => None,
Some(ref space) => {
if space.matches.is_empty() {
None
} else {
Some((&space.dst, &space.matches))
}
}
}
}
/// Clear space used for performing a replacement.
///
/// Subsequent calls to `replacement` after calling `clear` (but before
/// executing another replacement) will always return `None`.
pub fn clear(&mut self) {
if let Some(ref mut space) = self.space {
space.dst.clear();
space.matches.clear();
}
}
/// Allocate space for replacements when used with the given matcher and
/// return a mutable reference to that space.
///
/// This can fail if allocating space for capture locations from the given
/// matcher fails.
fn allocate(&mut self, matcher: &M) -> io::Result<&mut Space<M>> {
if self.space.is_none() {
let caps = matcher
.new_captures()
.map_err(io::Error::error_message)?;
self.space = Some(Space {
caps: caps,
dst: vec![],
matches: vec![],
});
}
Ok(self.space.as_mut().unwrap())
}
}
/// A simple layer of abstraction over either a match or a contextual line
/// reported by the searcher.
///
/// In particular, this provides an API that unions the `SinkMatch` and
/// `SinkContext` types while also exposing a list of all individual match
/// locations.
///
/// While this serves as a convenient mechanism to abstract over `SinkMatch`
/// and `SinkContext`, this also provides a way to abstract over replacements.
/// Namely, after a replacement, a `Sunk` value can be constructed using the
/// results of the replacement instead of the bytes reported directly by the
/// searcher.
#[derive(Debug)]
pub struct Sunk<'a> {
bytes: &'a [u8],
absolute_byte_offset: u64,
line_number: Option<u64>,
context_kind: Option<&'a SinkContextKind>,
matches: &'a [Match],
original_matches: &'a [Match],
}
impl<'a> Sunk<'a> {
#[inline]
pub fn empty() -> Sunk<'static> {
Sunk {
bytes: &[],
absolute_byte_offset: 0,
line_number: None,
context_kind: None,
matches: &[],
original_matches: &[],
}
}
#[inline]
pub fn from_sink_match(
sunk: &'a SinkMatch<'a>,
original_matches: &'a [Match],
replacement: Option<(&'a [u8], &'a [Match])>,
) -> Sunk<'a> {
let (bytes, matches) = replacement.unwrap_or_else(|| {
(sunk.bytes(), original_matches)
});
Sunk {
bytes: bytes,
absolute_byte_offset: sunk.absolute_byte_offset(),
line_number: sunk.line_number(),
context_kind: None,
matches: matches,
original_matches: original_matches,
}
}
#[inline]
pub fn from_sink_context(
sunk: &'a SinkContext<'a>,
original_matches: &'a [Match],
replacement: Option<(&'a [u8], &'a [Match])>,
) -> Sunk<'a> {
let (bytes, matches) = replacement.unwrap_or_else(|| {
(sunk.bytes(), original_matches)
});
Sunk {
bytes: bytes,
absolute_byte_offset: sunk.absolute_byte_offset(),
line_number: sunk.line_number(),
context_kind: Some(sunk.kind()),
matches: matches,
original_matches: original_matches,
}
}
#[inline]
pub fn context_kind(&self) -> Option<&'a SinkContextKind> {
self.context_kind
}
#[inline]
pub fn bytes(&self) -> &'a [u8] {
self.bytes
}
#[inline]
pub fn matches(&self) -> &'a [Match] {
self.matches
}
#[inline]
pub fn original_matches(&self) -> &'a [Match] {
self.original_matches
}
#[inline]
pub fn lines(&self, line_term: u8) -> LineIter<'a> {
LineIter::new(line_term, self.bytes())
}
#[inline]
pub fn absolute_byte_offset(&self) -> u64 {
self.absolute_byte_offset
}
#[inline]
pub fn line_number(&self) -> Option<u64> {
self.line_number
}
}
/// A simple encapsulation of a file path used by a printer.
///
/// This represents any transforms that we might want to perform on the path,
/// such as converting it to valid UTF-8 and/or replacing its separator with
/// something else. This allows us to amortize work if we are printing the
/// file path for every match.
///
/// In the common case, no transformation is needed, which lets us avoid the
/// allocation. Typically, only Windows requires a transform, since we can't
/// access the raw bytes of a path directly and first need to lossily convert
/// to UTF-8. Windows is also typically where the path separator replacement
/// is used, e.g., in cygwin environments to use `/` instead of `\`.
///
/// Users of this type are expected to construct it from a normal `Path`
/// found in the standard library. It can then be written to any `io::Write`
/// implementation using the `as_bytes` method. This achieves platform
/// portability with a small cost: on Windows, paths that are not valid UTF-16
/// will not roundtrip correctly.
#[derive(Clone, Debug)]
pub struct PrinterPath<'a>(Cow<'a, [u8]>);
impl<'a> PrinterPath<'a> {
/// Create a new path suitable for printing.
pub fn new(path: &'a Path) -> PrinterPath<'a> {
PrinterPath::new_impl(path)
}
#[cfg(unix)]
fn new_impl(path: &'a Path) -> PrinterPath<'a> {
use std::os::unix::ffi::OsStrExt;
PrinterPath(Cow::Borrowed(path.as_os_str().as_bytes()))
}
#[cfg(not(unix))]
fn new_impl(path: &'a Path) -> PrinterPath<'a> {
PrinterPath(match path.to_string_lossy() {
Cow::Owned(path) => Cow::Owned(path.into_bytes()),
Cow::Borrowed(path) => Cow::Borrowed(path.as_bytes()),
})
}
/// Create a new printer path from the given path which can be efficiently
/// written to a writer without allocation.
///
/// If the given separator is present, then any separators in `path` are
/// replaced with it.
pub fn with_separator(path: &'a Path, sep: Option<u8>) -> PrinterPath<'a> {
let mut ppath = PrinterPath::new(path);
if let Some(sep) = sep {
ppath.replace_separator(sep);
}
ppath
}
/// Replace the path separator in this path with the given separator
/// and do it in place. On Windows, both `/` and `\` are treated as
/// path separators that are both replaced by `new_sep`. In all other
/// environments, only `/` is treated as a path separator.
fn replace_separator(&mut self, new_sep: u8) {
let transformed_path: Vec<_> = self.as_bytes().iter().map(|&b| {
if b == b'/' || (cfg!(windows) && b == b'\\') {
new_sep
} else {
b
}
}).collect();
self.0 = Cow::Owned(transformed_path);
}
/// Return the raw bytes for this path.
pub fn as_bytes(&self) -> &[u8] {
&*self.0
}
}
/// A type that provides "nicer" Display and Serialize impls for
/// std::time::Duration. The serialization format should actually be compatible
/// with the Deserialize impl for std::time::Duration, since this type only
/// adds new fields.
#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)]
pub struct NiceDuration(pub time::Duration);
impl fmt::Display for NiceDuration {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
write!(f, "{:0.6}s", self.fractional_seconds())
}
}
impl NiceDuration {
/// Returns the number of seconds in this duration in fraction form.
/// The number to the left of the decimal point is the number of seconds,
/// and the number to the right is the number of milliseconds.
fn fractional_seconds(&self) -> f64 {
let fractional = (self.0.subsec_nanos() as f64) / 1_000_000_000.0;
self.0.as_secs() as f64 + fractional
}
}
#[cfg(feature = "serde1")]
impl Serialize for NiceDuration {
fn serialize<S: Serializer>(&self, ser: S) -> Result<S::Ok, S::Error> {
use serde::ser::SerializeStruct;
let mut state = ser.serialize_struct("Duration", 2)?;
state.serialize_field("secs", &self.0.as_secs())?;
state.serialize_field("nanos", &self.0.subsec_nanos())?;
state.serialize_field("human", &format!("{}", self))?;
state.end()
}
}
/// Trim prefix ASCII spaces from the given slice and return the corresponding
/// range.
///
/// This stops trimming a prefix as soon as it sees non-whitespace or a line
/// terminator.
pub fn trim_ascii_prefix_range(
line_term: LineTerminator,
slice: &[u8],
range: Match,
) -> Match {
fn is_space(b: u8) -> bool {
match b {
b'\t' | b'\n' | b'\x0B' | b'\x0C' | b'\r' | b' ' => true,
_ => false,
}
}
let count = slice[range]
.iter()
.take_while(|&&b| -> bool {
is_space(b) && !line_term.as_bytes().contains(&b)
})
.count();
range.with_start(range.start() + count)
}
/// Trim prefix ASCII spaces from the given slice and return the corresponding
/// sub-slice.
pub fn trim_ascii_prefix(line_term: LineTerminator, slice: &[u8]) -> &[u8] {
let range = trim_ascii_prefix_range(
line_term,
slice,
Match::new(0, slice.len()),
);
&slice[range]
}