<!DOCTYPE HTML>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>Error Handling - The Rust Programming Language</title>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
<meta name="description" content="">
<meta name="viewport" content="width=device-width, initial-scale=1">
<base href="">
<link rel="stylesheet" href="book.css">
<link href="https://fonts.googleapis.com/css?family=Open+Sans:300italic,400italic,600italic,700italic,800italic,400,300,600,700,800" rel="stylesheet" type="text/css">
<link href="https://fonts.googleapis.com/css?family=Source+Code+Pro:500" rel="stylesheet" type="text/css">
<link rel="shortcut icon" href="favicon.png">
<!-- Font Awesome -->
<link rel="stylesheet" href="https://maxcdn.bootstrapcdn.com/font-awesome/4.3.0/css/font-awesome.min.css">
<link rel="stylesheet" href="highlight.css">
<link rel="stylesheet" href="tomorrow-night.css">
<!-- MathJax -->
<script type="text/javascript" src="https://cdn.mathjax.org/mathjax/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>
<!-- Fetch JQuery from CDN but have a local fallback -->
<script src="https://code.jquery.com/jquery-2.1.4.min.js"></script>
<script>
if (typeof jQuery == 'undefined') {
document.write(unescape("%3Cscript src='jquery.js'%3E%3C/script%3E"));
}
</script>
</head>
<body class="light">
<!-- Set the theme before any content is loaded, prevents flash -->
<script type="text/javascript">
var theme = localStorage.getItem('theme');
if (theme == null) { theme = 'light'; }
$('body').removeClass().addClass(theme);
</script>
<!-- Hide / unhide sidebar before it is displayed -->
<script type="text/javascript">
var sidebar = localStorage.getItem('sidebar');
if (sidebar === "hidden") { $("html").addClass("sidebar-hidden") }
else if (sidebar === "visible") { $("html").addClass("sidebar-visible") }
</script>
<div id="sidebar" class="sidebar">
<ul class="chapter"><li class="affix"><a href="README.html">Introduction</a></li><li><a href="getting-started.html"><strong>1.</strong> Getting Started</a></li><li><a href="guessing-game.html"><strong>2.</strong> Tutorial: Guessing Game</a></li><li><a href="syntax-and-semantics.html"><strong>3.</strong> Syntax and Semantics</a></li><li><ul class="section"><li><a href="variable-bindings.html"><strong>3.1.</strong> Variable Bindings</a></li><li><a href="functions.html"><strong>3.2.</strong> Functions</a></li><li><a href="primitive-types.html"><strong>3.3.</strong> Primitive Types</a></li><li><a href="comments.html"><strong>3.4.</strong> Comments</a></li><li><a href="if.html"><strong>3.5.</strong> if</a></li><li><a href="loops.html"><strong>3.6.</strong> Loops</a></li><li><a href="vectors.html"><strong>3.7.</strong> Vectors</a></li><li><a href="ownership.html"><strong>3.8.</strong> Ownership</a></li><li><a href="references-and-borrowing.html"><strong>3.9.</strong> References and Borrowing</a></li><li><a href="lifetimes.html"><strong>3.10.</strong> Lifetimes</a></li><li><a href="mutability.html"><strong>3.11.</strong> Mutability</a></li><li><a href="structs.html"><strong>3.12.</strong> Structs</a></li><li><a href="enums.html"><strong>3.13.</strong> Enums</a></li><li><a href="match.html"><strong>3.14.</strong> Match</a></li><li><a href="patterns.html"><strong>3.15.</strong> Patterns</a></li><li><a href="method-syntax.html"><strong>3.16.</strong> Method Syntax</a></li><li><a href="strings.html"><strong>3.17.</strong> Strings</a></li><li><a href="generics.html"><strong>3.18.</strong> Generics</a></li><li><a href="traits.html"><strong>3.19.</strong> Traits</a></li><li><a href="drop.html"><strong>3.20.</strong> Drop</a></li><li><a href="if-let.html"><strong>3.21.</strong> if let</a></li><li><a href="trait-objects.html"><strong>3.22.</strong> Trait Objects</a></li><li><a href="closures.html"><strong>3.23.</strong> Closures</a></li><li><a href="ufcs.html"><strong>3.24.</strong> Universal Function Call Syntax</a></li><li><a href="crates-and-modules.html"><strong>3.25.</strong> Crates and Modules</a></li><li><a href="const-and-static.html"><strong>3.26.</strong> <code>const</code> and <code>static</code></a></li><li><a href="attributes.html"><strong>3.27.</strong> Attributes</a></li><li><a href="type-aliases.html"><strong>3.28.</strong> <code>type</code> aliases</a></li><li><a href="casting-between-types.html"><strong>3.29.</strong> Casting between types</a></li><li><a href="associated-types.html"><strong>3.30.</strong> Associated Types</a></li><li><a href="unsized-types.html"><strong>3.31.</strong> Unsized Types</a></li><li><a href="operators-and-overloading.html"><strong>3.32.</strong> Operators and Overloading</a></li><li><a href="deref-coercions.html"><strong>3.33.</strong> Deref coercions</a></li><li><a href="macros.html"><strong>3.34.</strong> Macros</a></li><li><a href="raw-pointers.html"><strong>3.35.</strong> Raw Pointers</a></li><li><a href="unsafe.html"><strong>3.36.</strong> <code>unsafe</code></a></li></ul></li><li><a href="effective-rust.html"><strong>4.</strong> Effective Rust</a></li><li><ul class="section"><li><a href="the-stack-and-the-heap.html"><strong>4.1.</strong> The Stack and the Heap</a></li><li><a href="testing.html"><strong>4.2.</strong> Testing</a></li><li><a href="conditional-compilation.html"><strong>4.3.</strong> Conditional Compilation</a></li><li><a href="documentation.html"><strong>4.4.</strong> Documentation</a></li><li><a href="iterators.html"><strong>4.5.</strong> Iterators</a></li><li><a href="concurrency.html"><strong>4.6.</strong> Concurrency</a></li><li><a href="error-handling.html" class="active"><strong>4.7.</strong> Error Handling</a></li><li><a href="choosing-your-guarantees.html"><strong>4.8.</strong> Choosing your Guarantees</a></li><li><a href="ffi.html"><strong>4.9.</strong> FFI</a></li><li><a href="borrow-and-asref.html"><strong>4.10.</strong> Borrow and AsRef</a></li><li><a href="release-channels.html"><strong>4.11.</strong> Release Channels</a></li><li><a href="using-rust-without-the-standard-library.html"><strong>4.12.</strong> Using Rust without the standard library</a></li><li><a href="procedural-macros.html"><strong>4.13.</strong> Procedural Macros (and custom derive)</a></li></ul></li><li><a href="glossary.html"><strong>5.</strong> Glossary</a></li><li><a href="syntax-index.html"><strong>6.</strong> Syntax Index</a></li><li><a href="bibliography.html"><strong>7.</strong> Bibliography</a></li></ul>
</div>
<div id="page-wrapper" class="page-wrapper">
<div class="page">
<div id="menu-bar" class="menu-bar">
<div class="left-buttons">
<i id="sidebar-toggle" class="fa fa-bars"></i>
<i id="theme-toggle" class="fa fa-paint-brush"></i>
</div>
<h1 class="menu-title">The Rust Programming Language</h1>
<div class="right-buttons">
<i id="print-button" class="fa fa-print" title="Print this book"></i>
</div>
</div>
<div id="content" class="content">
<a class="header" href="error-handling.html#error-handling" id="error-handling"><h1>Error Handling</h1></a>
<p>Like most programming languages, Rust encourages the programmer to handle
errors in a particular way. Generally speaking, error handling is divided into
two broad categories: exceptions and return values. Rust opts for return
values.</p>
<p>In this section, we intend to provide a comprehensive treatment of how to deal
with errors in Rust. More than that, we will attempt to introduce error handling
one piece at a time so that you'll come away with a solid working knowledge of
how everything fits together.</p>
<p>When done naïvely, error handling in Rust can be verbose and annoying. This
section will explore those stumbling blocks and demonstrate how to use the
standard library to make error handling concise and ergonomic.</p>
<a class="header" href="error-handling.html#table-of-contents" id="table-of-contents"><h1>Table of Contents</h1></a>
<p>This section is very long, mostly because we start at the very beginning with
sum types and combinators, and try to motivate the way Rust does error handling
incrementally. As such, programmers with experience in other expressive type
systems may want to jump around.</p>
<ul>
<li><a href="error-handling.html#the-basics">The Basics</a>
<ul>
<li><a href="error-handling.html#unwrapping-explained">Unwrapping explained</a></li>
<li><a href="error-handling.html#the-option-type">The <code>Option</code> type</a>
<ul>
<li><a href="error-handling.html#composing-optiont-values">Composing <code>Option<T></code> values</a></li>
</ul>
</li>
<li><a href="error-handling.html#the-result-type">The <code>Result</code> type</a>
<ul>
<li><a href="error-handling.html#parsing-integers">Parsing integers</a></li>
<li><a href="error-handling.html#the-result-type-alias-idiom">The <code>Result</code> type alias idiom</a></li>
</ul>
</li>
<li><a href="error-handling.html#a-brief-interlude-unwrapping-isnt-evil">A brief interlude: unwrapping isn't evil</a></li>
</ul>
</li>
<li><a href="error-handling.html#working-with-multiple-error-types">Working with multiple error types</a>
<ul>
<li><a href="error-handling.html#composing-option-and-result">Composing <code>Option</code> and <code>Result</code></a></li>
<li><a href="error-handling.html#the-limits-of-combinators">The limits of combinators</a></li>
<li><a href="error-handling.html#early-returns">Early returns</a></li>
<li><a href="error-handling.html#the-try-macro">The <code>try!</code> macro</a></li>
<li><a href="error-handling.html#defining-your-own-error-type">Defining your own error type</a></li>
</ul>
</li>
<li><a href="error-handling.html#standard-library-traits-used-for-error-handling">Standard library traits used for error handling</a>
<ul>
<li><a href="error-handling.html#the-error-trait">The <code>Error</code> trait</a></li>
<li><a href="error-handling.html#the-from-trait">The <code>From</code> trait</a></li>
<li><a href="error-handling.html#the-real-try-macro">The real <code>try!</code> macro</a></li>
<li><a href="error-handling.html#composing-custom-error-types">Composing custom error types</a></li>
<li><a href="error-handling.html#advice-for-library-writers">Advice for library writers</a></li>
</ul>
</li>
<li><a href="error-handling.html#case-study-a-program-to-read-population-data">Case study: A program to read population data</a>
<ul>
<li><a href="error-handling.html#initial-setup">Initial setup</a></li>
<li><a href="error-handling.html#argument-parsing">Argument parsing</a></li>
<li><a href="error-handling.html#writing-the-logic">Writing the logic</a></li>
<li><a href="error-handling.html#error-handling-with-boxerror">Error handling with <code>Box<Error></code></a></li>
<li><a href="error-handling.html#reading-from-stdin">Reading from stdin</a></li>
<li><a href="error-handling.html#error-handling-with-a-custom-type">Error handling with a custom type</a></li>
<li><a href="error-handling.html#adding-functionality">Adding functionality</a></li>
</ul>
</li>
<li><a href="error-handling.html#the-short-story">The short story</a></li>
</ul>
<a class="header" href="error-handling.html#the-basics" id="the-basics"><h1>The Basics</h1></a>
<p>You can think of error handling as using <em>case analysis</em> to determine whether
a computation was successful or not. As you will see, the key to ergonomic error
handling is reducing the amount of explicit case analysis the programmer has to
do while keeping code composable.</p>
<p>Keeping code composable is important, because without that requirement, we
could <a href="../../std/macro.panic.html"><code>panic</code></a> whenever we
come across something unexpected. (<code>panic</code> causes the current task to unwind,
and in most cases, the entire program aborts.) Here's an example:</p>
<pre><pre class="playpen"><code class="language-rust should_panic">// Guess a number between 1 and 10.
// If it matches the number we had in mind, return `true`. Else, return `false`.
fn guess(n: i32) -> bool {
if n < 1 || n > 10 {
panic!("Invalid number: {}", n);
}
n == 5
}
fn main() {
guess(11);
}
</code></pre></pre>
<p>If you try running this code, the program will crash with a message like this:</p>
<pre><code class="language-text">thread 'main' panicked at 'Invalid number: 11', src/bin/panic-simple.rs:5
</code></pre>
<p>Here's another example that is slightly less contrived. A program that accepts
an integer as an argument, doubles it and prints it.</p>
<p><span id="code-unwrap-double"></span></p>
<pre><pre class="playpen"><code class="language-rust should_panic">use std::env;
fn main() {
let mut argv = env::args();
let arg: String = argv.nth(1).unwrap(); // error 1
let n: i32 = arg.parse().unwrap(); // error 2
println!("{}", 2 * n);
}
</code></pre></pre>
<p>If you give this program zero arguments (error 1) or if the first argument
isn't an integer (error 2), the program will panic just like in the first
example.</p>
<p>You can think of this style of error handling as similar to a bull running
through a china shop. The bull will get to where it wants to go, but it will
trample everything in the process.</p>
<a class="header" href="error-handling.html#unwrapping-explained" id="unwrapping-explained"><h2>Unwrapping explained</h2></a>
<p>In the previous example, we claimed
that the program would simply panic if it reached one of the two error
conditions, yet, the program does not include an explicit call to <code>panic</code> like
the first example. This is because the
panic is embedded in the calls to <code>unwrap</code>.</p>
<p>To “unwrap” something in Rust is to say, “Give me the result of the
computation, and if there was an error, panic and stop the program.”
It would be better if we showed the code for unwrapping because it is so
simple, but to do that, we will first need to explore the <code>Option</code> and <code>Result</code>
types. Both of these types have a method called <code>unwrap</code> defined on them.</p>
<a class="header" href="error-handling.html#the-option-type" id="the-option-type"><h3>The <code>Option</code> type</h3></a>
<p>The <code>Option</code> type is <a href="../../std/option/enum.Option.html">defined in the standard library</a>:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
enum Option<T> {
None,
Some(T),
}
#}</code></pre></pre>
<p>The <code>Option</code> type is a way to use Rust's type system to express the
<em>possibility of absence</em>. Encoding the possibility of absence into the type
system is an important concept because it will cause the compiler to force the
programmer to handle that absence. Let's take a look at an example that tries
to find a character in a string:</p>
<p><span id="code-option-ex-string-find"></span></p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
// Searches `haystack` for the Unicode character `needle`. If one is found, the
// byte offset of the character is returned. Otherwise, `None` is returned.
fn find(haystack: &str, needle: char) -> Option<usize> {
for (offset, c) in haystack.char_indices() {
if c == needle {
return Some(offset);
}
}
None
}
#}</code></pre></pre>
<p>Notice that when this function finds a matching character, it doesn't only
return the <code>offset</code>. Instead, it returns <code>Some(offset)</code>. <code>Some</code> is a variant or
a <em>value constructor</em> for the <code>Option</code> type. You can think of it as a function
with the type <code>fn<T>(value: T) -> Option<T></code>. Correspondingly, <code>None</code> is also a
value constructor, except it has no arguments. You can think of <code>None</code> as a
function with the type <code>fn<T>() -> Option<T></code>.</p>
<p>This might seem like much ado about nothing, but this is only half of the
story. The other half is <em>using</em> the <code>find</code> function we've written. Let's try
to use it to find the extension in a file name.</p>
<pre><pre class="playpen"><code class="language-rust"># fn find(haystack: &str, needle: char) -> Option<usize> { haystack.find(needle) }
fn main() {
let file_name = "foobar.rs";
match find(file_name, '.') {
None => println!("No file extension found."),
Some(i) => println!("File extension: {}", &file_name[i+1..]),
}
}
</code></pre></pre>
<p>This code uses <a href="patterns.html">pattern matching</a> to do <em>case
analysis</em> on the <code>Option<usize></code> returned by the <code>find</code> function. In fact, case
analysis is the only way to get at the value stored inside an <code>Option<T></code>. This
means that you, as the programmer, must handle the case when an <code>Option<T></code> is
<code>None</code> instead of <code>Some(t)</code>.</p>
<p>But wait, what about <code>unwrap</code>, which we used <a href="error-handling.html#code-unwrap-double">previously</a>?
There was no case analysis there! Instead, the case analysis was put inside the
<code>unwrap</code> method for you. You could define it yourself if you want:</p>
<p><span id="code-option-def-unwrap"></span></p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
enum Option<T> {
None,
Some(T),
}
impl<T> Option<T> {
fn unwrap(self) -> T {
match self {
Option::Some(val) => val,
Option::None =>
panic!("called `Option::unwrap()` on a `None` value"),
}
}
}
#}</code></pre></pre>
<p>The <code>unwrap</code> method <em>abstracts away the case analysis</em>. This is precisely the thing
that makes <code>unwrap</code> ergonomic to use. Unfortunately, that <code>panic!</code> means that
<code>unwrap</code> is not composable: it is the bull in the china shop.</p>
<a class="header" href="error-handling.html#composing-optiont-values" id="composing-optiont-values"><h3>Composing <code>Option<T></code> values</h3></a>
<p>In an <a href="error-handling.html#code-option-ex-string-find">example from before</a>,
we saw how to use <code>find</code> to discover the extension in a file name. Of course,
not all file names have a <code>.</code> in them, so it's possible that the file name has
no extension. This <em>possibility of absence</em> is encoded into the types using
<code>Option<T></code>. In other words, the compiler will force us to address the
possibility that an extension does not exist. In our case, we only print out a
message saying as such.</p>
<p>Getting the extension of a file name is a pretty common operation, so it makes
sense to put it into a function:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
# fn find(haystack: &str, needle: char) -> Option<usize> { haystack.find(needle) }
// Returns the extension of the given file name, where the extension is defined
// as all characters following the first `.`.
// If `file_name` has no `.`, then `None` is returned.
fn extension_explicit(file_name: &str) -> Option<&str> {
match find(file_name, '.') {
None => None,
Some(i) => Some(&file_name[i+1..]),
}
}
#}</code></pre></pre>
<p>(Pro-tip: don't use this code. Use the
<a href="../../std/path/struct.Path.html#method.extension"><code>extension</code></a>
method in the standard library instead.)</p>
<p>The code stays simple, but the important thing to notice is that the type of
<code>find</code> forces us to consider the possibility of absence. This is a good thing
because it means the compiler won't let us accidentally forget about the case
where a file name doesn't have an extension. On the other hand, doing explicit
case analysis like we've done in <code>extension_explicit</code> every time can get a bit
tiresome.</p>
<p>In fact, the case analysis in <code>extension_explicit</code> follows a very common
pattern: <em>map</em> a function on to the value inside of an <code>Option<T></code>, unless the
option is <code>None</code>, in which case, return <code>None</code>.</p>
<p>Rust has parametric polymorphism, so it is very easy to define a combinator
that abstracts this pattern:</p>
<p><span id="code-option-map"></span></p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
fn map<F, T, A>(option: Option<T>, f: F) -> Option<A> where F: FnOnce(T) -> A {
match option {
None => None,
Some(value) => Some(f(value)),
}
}
#}</code></pre></pre>
<p>Indeed, <code>map</code> is <a href="../../std/option/enum.Option.html#method.map">defined as a method</a> on <code>Option<T></code> in the standard library.
As a method, it has a slightly different signature: methods take <code>self</code>, <code>&self</code>,
or <code>&mut self</code> as their first argument.</p>
<p>Armed with our new combinator, we can rewrite our <code>extension_explicit</code> method
to get rid of the case analysis:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
# fn find(haystack: &str, needle: char) -> Option<usize> { haystack.find(needle) }
// Returns the extension of the given file name, where the extension is defined
// as all characters following the first `.`.
// If `file_name` has no `.`, then `None` is returned.
fn extension(file_name: &str) -> Option<&str> {
find(file_name, '.').map(|i| &file_name[i+1..])
}
#}</code></pre></pre>
<p>One other pattern we commonly find is assigning a default value to the case
when an <code>Option</code> value is <code>None</code>. For example, maybe your program assumes that
the extension of a file is <code>rs</code> even if none is present. As you might imagine,
the case analysis for this is not specific to file extensions - it can work
with any <code>Option<T></code>:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
fn unwrap_or<T>(option: Option<T>, default: T) -> T {
match option {
None => default,
Some(value) => value,
}
}
#}</code></pre></pre>
<p>Like with <code>map</code> above, the standard library implementation is a method instead
of a free function.</p>
<p>The trick here is that the default value must have the same type as the value
that might be inside the <code>Option<T></code>. Using it is dead simple in our case:</p>
<pre><pre class="playpen"><code class="language-rust"># fn find(haystack: &str, needle: char) -> Option<usize> {
# for (offset, c) in haystack.char_indices() {
# if c == needle {
# return Some(offset);
# }
# }
# None
# }
#
# fn extension(file_name: &str) -> Option<&str> {
# find(file_name, '.').map(|i| &file_name[i+1..])
# }
fn main() {
assert_eq!(extension("foobar.csv").unwrap_or("rs"), "csv");
assert_eq!(extension("foobar").unwrap_or("rs"), "rs");
}
</code></pre></pre>
<p>(Note that <code>unwrap_or</code> is <a href="../../std/option/enum.Option.html#method.unwrap_or">defined as a method</a> on <code>Option<T></code> in the
standard library, so we use that here instead of the free-standing function we
defined above. Don't forget to check out the more general <a href="../../std/option/enum.Option.html#method.unwrap_or_else"><code>unwrap_or_else</code></a>
method.)</p>
<p>There is one more combinator that we think is worth paying special attention to:
<code>and_then</code>. It makes it easy to compose distinct computations that admit the
<em>possibility of absence</em>. For example, much of the code in this section is
about finding an extension given a file name. In order to do this, you first
need the file name which is typically extracted from a file <em>path</em>. While most
file paths have a file name, not <em>all</em> of them do. For example, <code>.</code>, <code>..</code> or
<code>/</code>.</p>
<p>So, we are tasked with the challenge of finding an extension given a file
<em>path</em>. Let's start with explicit case analysis:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
# fn extension(file_name: &str) -> Option<&str> { None }
fn file_path_ext_explicit(file_path: &str) -> Option<&str> {
match file_name(file_path) {
None => None,
Some(name) => match extension(name) {
None => None,
Some(ext) => Some(ext),
}
}
}
fn file_name(file_path: &str) -> Option<&str> {
// Implementation elided.
unimplemented!()
}
#}</code></pre></pre>
<p>You might think that we could use the <code>map</code> combinator to reduce the case
analysis, but its type doesn't quite fit...</p>
<pre><code class="language-rust ignore">fn file_path_ext(file_path: &str) -> Option<&str> {
file_name(file_path).map(|x| extension(x)) // This causes a compilation error.
}
</code></pre>
<p>The <code>map</code> function here wraps the value returned by the <code>extension</code> function
inside an <code>Option<_></code> and since the <code>extension</code> function itself returns an
<code>Option<&str></code> the expression <code>file_name(file_path).map(|x| extension(x))</code>
actually returns an <code>Option<Option<&str>></code>.</p>
<p>But since <code>file_path_ext</code> just returns <code>Option<&str></code> (and not
<code>Option<Option<&str>></code>) we get a compilation error.</p>
<p>The result of the function taken by map as input is <em>always</em> <a href="error-handling.html#code-option-map">rewrapped with
<code>Some</code></a>. Instead, we need something like <code>map</code>, but which
allows the caller to return a <code>Option<_></code> directly without wrapping it in
another <code>Option<_></code>.</p>
<p>Its generic implementation is even simpler than <code>map</code>:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
fn and_then<F, T, A>(option: Option<T>, f: F) -> Option<A>
where F: FnOnce(T) -> Option<A> {
match option {
None => None,
Some(value) => f(value),
}
}
#}</code></pre></pre>
<p>Now we can rewrite our <code>file_path_ext</code> function without explicit case analysis:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
# fn extension(file_name: &str) -> Option<&str> { None }
# fn file_name(file_path: &str) -> Option<&str> { None }
fn file_path_ext(file_path: &str) -> Option<&str> {
file_name(file_path).and_then(extension)
}
#}</code></pre></pre>
<p>Side note: Since <code>and_then</code> essentially works like <code>map</code> but returns an
<code>Option<_></code> instead of an <code>Option<Option<_>></code> it is known as <code>flatmap</code> in some
other languages.</p>
<p>The <code>Option</code> type has many other combinators <a href="../../std/option/enum.Option.html">defined in the standard
library</a>. It is a good idea to skim this list and familiarize
yourself with what's available—they can often reduce case analysis
for you. Familiarizing yourself with these combinators will pay
dividends because many of them are also defined (with similar
semantics) for <code>Result</code>, which we will talk about next.</p>
<p>Combinators make using types like <code>Option</code> ergonomic because they reduce
explicit case analysis. They are also composable because they permit the caller
to handle the possibility of absence in their own way. Methods like <code>unwrap</code>
remove choices because they will panic if <code>Option<T></code> is <code>None</code>.</p>
<a class="header" href="error-handling.html#the-result-type" id="the-result-type"><h2>The <code>Result</code> type</h2></a>
<p>The <code>Result</code> type is also
<a href="../../std/result/index.html">defined in the standard library</a>:</p>
<p><span id="code-result-def"></span></p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
enum Result<T, E> {
Ok(T),
Err(E),
}
#}</code></pre></pre>
<p>The <code>Result</code> type is a richer version of <code>Option</code>. Instead of expressing the
possibility of <em>absence</em> like <code>Option</code> does, <code>Result</code> expresses the possibility
of <em>error</em>. Usually, the <em>error</em> is used to explain why the execution of some
computation failed. This is a strictly more general form of <code>Option</code>. Consider
the following type alias, which is semantically equivalent to the real
<code>Option<T></code> in every way:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
type Option<T> = Result<T, ()>;
#}</code></pre></pre>
<p>This fixes the second type parameter of <code>Result</code> to always be <code>()</code> (pronounced
“unit” or “empty tuple”). Exactly one value inhabits the <code>()</code> type: <code>()</code>. (Yup,
the type and value level terms have the same notation!)</p>
<p>The <code>Result</code> type is a way of representing one of two possible outcomes in a
computation. By convention, one outcome is meant to be expected or “<code>Ok</code>” while
the other outcome is meant to be unexpected or “<code>Err</code>”.</p>
<p>Just like <code>Option</code>, the <code>Result</code> type also has an
<a href="../../std/result/enum.Result.html#method.unwrap"><code>unwrap</code> method
defined</a>
in the standard library. Let's define it:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
# enum Result<T, E> { Ok(T), Err(E) }
impl<T, E: ::std::fmt::Debug> Result<T, E> {
fn unwrap(self) -> T {
match self {
Result::Ok(val) => val,
Result::Err(err) =>
panic!("called `Result::unwrap()` on an `Err` value: {:?}", err),
}
}
}
#}</code></pre></pre>
<p>This is effectively the same as our <a href="error-handling.html#code-option-def-unwrap">definition for
<code>Option::unwrap</code></a>, except it includes the
error value in the <code>panic!</code> message. This makes debugging easier, but
it also requires us to add a <a href="../../std/fmt/trait.Debug.html"><code>Debug</code></a> constraint on the <code>E</code> type
parameter (which represents our error type). Since the vast majority
of types should satisfy the <code>Debug</code> constraint, this tends to work out
in practice. (<code>Debug</code> on a type simply means that there's a reasonable
way to print a human readable description of values with that type.)</p>
<p>OK, let's move on to an example.</p>
<a class="header" href="error-handling.html#parsing-integers" id="parsing-integers"><h3>Parsing integers</h3></a>
<p>The Rust standard library makes converting strings to integers dead simple.
It's so easy in fact, that it is very tempting to write something like the
following:</p>
<pre><pre class="playpen"><code class="language-rust">fn double_number(number_str: &str) -> i32 {
2 * number_str.parse::<i32>().unwrap()
}
fn main() {
let n: i32 = double_number("10");
assert_eq!(n, 20);
}
</code></pre></pre>
<p>At this point, you should be skeptical of calling <code>unwrap</code>. For example, if
the string doesn't parse as a number, you'll get a panic:</p>
<pre><code class="language-text">thread 'main' panicked at 'called `Result::unwrap()` on an `Err` value: ParseIntError { kind: InvalidDigit }', /home/rustbuild/src/rust-buildbot/slave/beta-dist-rustc-linux/build/src/libcore/result.rs:729
</code></pre>
<p>This is rather unsightly, and if this happened inside a library you're
using, you might be understandably annoyed. Instead, we should try to
handle the error in our function and let the caller decide what to
do. This means changing the return type of <code>double_number</code>. But to
what? Well, that requires looking at the signature of the <a href="../../std/primitive.str.html#method.parse"><code>parse</code>
method</a> in the standard library:</p>
<pre><code class="language-rust ignore">impl str {
fn parse<F: FromStr>(&self) -> Result<F, F::Err>;
}
</code></pre>
<p>Hmm. So we at least know that we need to use a <code>Result</code>. Certainly, it's
possible that this could have returned an <code>Option</code>. After all, a string either
parses as a number or it doesn't, right? That's certainly a reasonable way to
go, but the implementation internally distinguishes <em>why</em> the string didn't
parse as an integer. (Whether it's an empty string, an invalid digit, too big
or too small.) Therefore, using a <code>Result</code> makes sense because we want to
provide more information than simply “absence.” We want to say <em>why</em> the
parsing failed. You should try to emulate this line of reasoning when faced
with a choice between <code>Option</code> and <code>Result</code>. If you can provide detailed error
information, then you probably should. (We'll see more on this later.)</p>
<p>OK, but how do we write our return type? The <code>parse</code> method as defined
above is generic over all the different number types defined in the
standard library. We could (and probably should) also make our
function generic, but let's favor explicitness for the moment. We only
care about <code>i32</code>, so we need to <a href="../../std/primitive.i32.html">find its implementation of
<code>FromStr</code></a> (do a <code>CTRL-F</code> in your browser
for “FromStr”) and look at its <a href="associated-types.html">associated type</a> <code>Err</code>. We did
this so we can find the concrete error type. In this case, it's
<a href="../../std/num/struct.ParseIntError.html"><code>std::num::ParseIntError</code></a>.
Finally, we can rewrite our function:</p>
<pre><pre class="playpen"><code class="language-rust">use std::num::ParseIntError;
fn double_number(number_str: &str) -> Result<i32, ParseIntError> {
match number_str.parse::<i32>() {
Ok(n) => Ok(2 * n),
Err(err) => Err(err),
}
}
fn main() {
match double_number("10") {
Ok(n) => assert_eq!(n, 20),
Err(err) => println!("Error: {:?}", err),
}
}
</code></pre></pre>
<p>This is a little better, but now we've written a lot more code! The case
analysis has once again bitten us.</p>
<p>Combinators to the rescue! Just like <code>Option</code>, <code>Result</code> has lots of combinators
defined as methods. There is a large intersection of common combinators between
<code>Result</code> and <code>Option</code>. In particular, <code>map</code> is part of that intersection:</p>
<pre><pre class="playpen"><code class="language-rust">use std::num::ParseIntError;
fn double_number(number_str: &str) -> Result<i32, ParseIntError> {
number_str.parse::<i32>().map(|n| 2 * n)
}
fn main() {
match double_number("10") {
Ok(n) => assert_eq!(n, 20),
Err(err) => println!("Error: {:?}", err),
}
}
</code></pre></pre>
<p>The usual suspects are all there for <code>Result</code>, including
<a href="../../std/result/enum.Result.html#method.unwrap_or"><code>unwrap_or</code></a> and
<a href="../../std/result/enum.Result.html#method.and_then"><code>and_then</code></a>.
Additionally, since <code>Result</code> has a second type parameter, there are
combinators that affect only the error type, such as
<a href="../../std/result/enum.Result.html#method.map_err"><code>map_err</code></a> (instead of
<code>map</code>) and <a href="../../std/result/enum.Result.html#method.or_else"><code>or_else</code></a>
(instead of <code>and_then</code>).</p>
<a class="header" href="error-handling.html#the-result-type-alias-idiom" id="the-result-type-alias-idiom"><h3>The <code>Result</code> type alias idiom</h3></a>
<p>In the standard library, you may frequently see types like
<code>Result<i32></code>. But wait, <a href="error-handling.html#code-result-def">we defined <code>Result</code></a> to
have two type parameters. How can we get away with only specifying
one? The key is to define a <code>Result</code> type alias that <em>fixes</em> one of
the type parameters to a particular type. Usually the fixed type is
the error type. For example, our previous example parsing integers
could be rewritten like this:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
use std::num::ParseIntError;
use std::result;
type Result<T> = result::Result<T, ParseIntError>;
fn double_number(number_str: &str) -> Result<i32> {
unimplemented!();
}
#}</code></pre></pre>
<p>Why would we do this? Well, if we have a lot of functions that could return
<code>ParseIntError</code>, then it's much more convenient to define an alias that always
uses <code>ParseIntError</code> so that we don't have to write it out all the time.</p>
<p>The most prominent place this idiom is used in the standard library is
with <a href="../../std/io/type.Result.html"><code>io::Result</code></a>. Typically, one writes
<code>io::Result<T></code>, which makes it clear that you're using the <code>io</code>
module's type alias instead of the plain definition from
<code>std::result</code>. (This idiom is also used for
<a href="../../std/fmt/type.Result.html"><code>fmt::Result</code></a>.)</p>
<a class="header" href="error-handling.html#a-brief-interlude-unwrapping-isnt-evil" id="a-brief-interlude-unwrapping-isnt-evil"><h2>A brief interlude: unwrapping isn't evil</h2></a>
<p>If you've been following along, you might have noticed that I've taken a pretty
hard line against calling methods like <code>unwrap</code> that could <code>panic</code> and abort
your program. <em>Generally speaking</em>, this is good advice.</p>
<p>However, <code>unwrap</code> can still be used judiciously. What exactly justifies use of
<code>unwrap</code> is somewhat of a grey area and reasonable people can disagree. I'll
summarize some of my <em>opinions</em> on the matter.</p>
<ul>
<li><strong>In examples and quick 'n' dirty code.</strong> Sometimes you're writing examples
or a quick program, and error handling simply isn't important. Beating the
convenience of <code>unwrap</code> can be hard in such scenarios, so it is very
appealing.</li>
<li><strong>When panicking indicates a bug in the program.</strong> When the invariants of
your code should prevent a certain case from happening (like, say, popping
from an empty stack), then panicking can be permissible. This is because it
exposes a bug in your program. This can be explicit, like from an <code>assert!</code>
failing, or it could be because your index into an array was out of bounds.</li>
</ul>
<p>This is probably not an exhaustive list. Moreover, when using an
<code>Option</code>, it is often better to use its
<a href="../../std/option/enum.Option.html#method.expect"><code>expect</code></a>
method. <code>expect</code> does exactly the same thing as <code>unwrap</code>, except it
prints a message you give to <code>expect</code>. This makes the resulting panic
a bit nicer to deal with, since it will show your message instead of
“called unwrap on a <code>None</code> value.”</p>
<p>My advice boils down to this: use good judgment. There's a reason why the words
“never do X” or “Y is considered harmful” don't appear in my writing. There are
trade offs to all things, and it is up to you as the programmer to determine
what is acceptable for your use cases. My goal is only to help you evaluate
trade offs as accurately as possible.</p>
<p>Now that we've covered the basics of error handling in Rust, and
explained unwrapping, let's start exploring more of the standard
library.</p>
<a class="header" href="error-handling.html#working-with-multiple-error-types" id="working-with-multiple-error-types"><h1>Working with multiple error types</h1></a>
<p>Thus far, we've looked at error handling where everything was either an
<code>Option<T></code> or a <code>Result<T, SomeError></code>. But what happens when you have both an
<code>Option</code> and a <code>Result</code>? Or what if you have a <code>Result<T, Error1></code> and a
<code>Result<T, Error2></code>? Handling <em>composition of distinct error types</em> is the next
challenge in front of us, and it will be the major theme throughout the rest of
this section.</p>
<a class="header" href="error-handling.html#composing-option-and-result" id="composing-option-and-result"><h2>Composing <code>Option</code> and <code>Result</code></h2></a>
<p>So far, I've talked about combinators defined for <code>Option</code> and combinators
defined for <code>Result</code>. We can use these combinators to compose results of
different computations without doing explicit case analysis.</p>
<p>Of course, in real code, things aren't always as clean. Sometimes you have a
mix of <code>Option</code> and <code>Result</code> types. Must we resort to explicit case analysis,
or can we continue using combinators?</p>
<p>For now, let's revisit one of the first examples in this section:</p>
<pre><pre class="playpen"><code class="language-rust should_panic">use std::env;
fn main() {
let mut argv = env::args();
let arg: String = argv.nth(1).unwrap(); // error 1
let n: i32 = arg.parse().unwrap(); // error 2
println!("{}", 2 * n);
}
</code></pre></pre>
<p>Given our new found knowledge of <code>Option</code>, <code>Result</code> and their various
combinators, we should try to rewrite this so that errors are handled properly
and the program doesn't panic if there's an error.</p>
<p>The tricky aspect here is that <code>argv.nth(1)</code> produces an <code>Option</code> while
<code>arg.parse()</code> produces a <code>Result</code>. These aren't directly composable. When faced
with both an <code>Option</code> and a <code>Result</code>, the solution is <em>usually</em> to convert the
<code>Option</code> to a <code>Result</code>. In our case, the absence of a command line parameter
(from <code>env::args()</code>) means the user didn't invoke the program correctly. We
could use a <code>String</code> to describe the error. Let's try:</p>
<p><span id="code-error-double-string"></span></p>
<pre><pre class="playpen"><code class="language-rust">use std::env;
fn double_arg(mut argv: env::Args) -> Result<i32, String> {
argv.nth(1)
.ok_or("Please give at least one argument".to_owned())
.and_then(|arg| arg.parse::<i32>().map_err(|err| err.to_string()))
.map(|n| 2 * n)
}
fn main() {
match double_arg(env::args()) {
Ok(n) => println!("{}", n),
Err(err) => println!("Error: {}", err),
}
}
</code></pre></pre>
<p>There are a couple new things in this example. The first is the use of the
<a href="../../std/option/enum.Option.html#method.ok_or"><code>Option::ok_or</code></a>
combinator. This is one way to convert an <code>Option</code> into a <code>Result</code>. The
conversion requires you to specify what error to use if <code>Option</code> is <code>None</code>.
Like the other combinators we've seen, its definition is very simple:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
fn ok_or<T, E>(option: Option<T>, err: E) -> Result<T, E> {
match option {
Some(val) => Ok(val),
None => Err(err),
}
}
#}</code></pre></pre>
<p>The other new combinator used here is
<a href="../../std/result/enum.Result.html#method.map_err"><code>Result::map_err</code></a>.
This is like <code>Result::map</code>, except it maps a function on to the <em>error</em>
portion of a <code>Result</code> value. If the <code>Result</code> is an <code>Ok(...)</code> value, then it is
returned unmodified.</p>
<p>We use <code>map_err</code> here because it is necessary for the error types to remain
the same (because of our use of <code>and_then</code>). Since we chose to convert the
<code>Option<String></code> (from <code>argv.nth(1)</code>) to a <code>Result<String, String></code>, we must
also convert the <code>ParseIntError</code> from <code>arg.parse()</code> to a <code>String</code>.</p>
<a class="header" href="error-handling.html#the-limits-of-combinators" id="the-limits-of-combinators"><h2>The limits of combinators</h2></a>
<p>Doing IO and parsing input is a very common task, and it's one that I
personally have done a lot of in Rust. Therefore, we will use (and continue to
use) IO and various parsing routines to exemplify error handling.</p>
<p>Let's start simple. We are tasked with opening a file, reading all of its
contents and converting its contents to a number. Then we multiply it by <code>2</code>
and print the output.</p>
<p>Although I've tried to convince you not to use <code>unwrap</code>, it can be useful
to first write your code using <code>unwrap</code>. It allows you to focus on your problem
instead of the error handling, and it exposes the points where proper error
handling need to occur. Let's start there so we can get a handle on the code,
and then refactor it to use better error handling.</p>
<pre><pre class="playpen"><code class="language-rust should_panic">use std::fs::File;
use std::io::Read;
use std::path::Path;
fn file_double<P: AsRef<Path>>(file_path: P) -> i32 {
let mut file = File::open(file_path).unwrap(); // error 1
let mut contents = String::new();
file.read_to_string(&mut contents).unwrap(); // error 2
let n: i32 = contents.trim().parse().unwrap(); // error 3
2 * n
}
fn main() {
let doubled = file_double("foobar");
println!("{}", doubled);
}
</code></pre></pre>
<p>(N.B. The <code>AsRef<Path></code> is used because those are the
<a href="../../std/fs/struct.File.html#method.open">same bounds used on
<code>std::fs::File::open</code></a>.
This makes it ergonomic to use any kind of string as a file path.)</p>
<p>There are three different errors that can occur here:</p>
<ol>
<li>A problem opening the file.</li>
<li>A problem reading data from the file.</li>
<li>A problem parsing the data as a number.</li>
</ol>
<p>The first two problems are described via the
<a href="../../std/io/struct.Error.html"><code>std::io::Error</code></a> type. We know this
because of the return types of
<a href="../../std/fs/struct.File.html#method.open"><code>std::fs::File::open</code></a> and
<a href="../../std/io/trait.Read.html#method.read_to_string"><code>std::io::Read::read_to_string</code></a>.
(Note that they both use the <a href="error-handling.html#the-result-type-alias-idiom"><code>Result</code> type alias
idiom</a> described previously. If you
click on the <code>Result</code> type, you'll <a href="../../std/io/type.Result.html">see the type
alias</a>, and consequently, the underlying
<code>io::Error</code> type.) The third problem is described by the
<a href="../../std/num/struct.ParseIntError.html"><code>std::num::ParseIntError</code></a>
type. The <code>io::Error</code> type in particular is <em>pervasive</em> throughout the
standard library. You will see it again and again.</p>
<p>Let's start the process of refactoring the <code>file_double</code> function. To make this
function composable with other components of the program, it should <em>not</em> panic
if any of the above error conditions are met. Effectively, this means that the
function should <em>return an error</em> if any of its operations fail. Our problem is
that the return type of <code>file_double</code> is <code>i32</code>, which does not give us any
useful way of reporting an error. Thus, we must start by changing the return
type from <code>i32</code> to something else.</p>
<p>The first thing we need to decide: should we use <code>Option</code> or <code>Result</code>? We
certainly could use <code>Option</code> very easily. If any of the three errors occur, we
could simply return <code>None</code>. This will work <em>and it is better than panicking</em>,
but we can do a lot better. Instead, we should pass some detail about the error
that occurred. Since we want to express the <em>possibility of error</em>, we should
use <code>Result<i32, E></code>. But what should <code>E</code> be? Since two <em>different</em> types of
errors can occur, we need to convert them to a common type. One such type is
<code>String</code>. Let's see how that impacts our code:</p>
<pre><pre class="playpen"><code class="language-rust">use std::fs::File;
use std::io::Read;
use std::path::Path;
fn file_double<P: AsRef<Path>>(file_path: P) -> Result<i32, String> {
File::open(file_path)
.map_err(|err| err.to_string())
.and_then(|mut file| {
let mut contents = String::new();
file.read_to_string(&mut contents)
.map_err(|err| err.to_string())
.map(|_| contents)
})
.and_then(|contents| {
contents.trim().parse::<i32>()
.map_err(|err| err.to_string())
})
.map(|n| 2 * n)
}
fn main() {
match file_double("foobar") {
Ok(n) => println!("{}", n),
Err(err) => println!("Error: {}", err),
}
}
</code></pre></pre>
<p>This code looks a bit hairy. It can take quite a bit of practice before code
like this becomes easy to write. The way we write it is by <em>following the
types</em>. As soon as we changed the return type of <code>file_double</code> to
<code>Result<i32, String></code>, we had to start looking for the right combinators. In
this case, we only used three different combinators: <code>and_then</code>, <code>map</code> and
<code>map_err</code>.</p>
<p><code>and_then</code> is used to chain multiple computations where each computation could
return an error. After opening the file, there are two more computations that
could fail: reading from the file and parsing the contents as a number.
Correspondingly, there are two calls to <code>and_then</code>.</p>
<p><code>map</code> is used to apply a function to the <code>Ok(...)</code> value of a <code>Result</code>. For
example, the very last call to <code>map</code> multiplies the <code>Ok(...)</code> value (which is
an <code>i32</code>) by <code>2</code>. If an error had occurred before that point, this operation
would have been skipped because of how <code>map</code> is defined.</p>
<p><code>map_err</code> is the trick that makes all of this work. <code>map_err</code> is like
<code>map</code>, except it applies a function to the <code>Err(...)</code> value of a <code>Result</code>. In
this case, we want to convert all of our errors to one type: <code>String</code>. Since
both <code>io::Error</code> and <code>num::ParseIntError</code> implement <code>ToString</code>, we can call the
<code>to_string()</code> method to convert them.</p>
<p>With all of that said, the code is still hairy. Mastering use of combinators is
important, but they have their limits. Let's try a different approach: early
returns.</p>
<a class="header" href="error-handling.html#early-returns" id="early-returns"><h2>Early returns</h2></a>
<p>I'd like to take the code from the previous section and rewrite it using <em>early
returns</em>. Early returns let you exit the function early. We can't return early
in <code>file_double</code> from inside another closure, so we'll need to revert back to
explicit case analysis.</p>
<pre><pre class="playpen"><code class="language-rust">use std::fs::File;
use std::io::Read;
use std::path::Path;
fn file_double<P: AsRef<Path>>(file_path: P) -> Result<i32, String> {
let mut file = match File::open(file_path) {
Ok(file) => file,
Err(err) => return Err(err.to_string()),
};
let mut contents = String::new();
if let Err(err) = file.read_to_string(&mut contents) {
return Err(err.to_string());
}
let n: i32 = match contents.trim().parse() {
Ok(n) => n,
Err(err) => return Err(err.to_string()),
};
Ok(2 * n)
}
fn main() {
match file_double("foobar") {
Ok(n) => println!("{}", n),
Err(err) => println!("Error: {}", err),
}
}
</code></pre></pre>
<p>Reasonable people can disagree over whether this code is better than the code
that uses combinators, but if you aren't familiar with the combinator approach,
this code looks simpler to read to me. It uses explicit case analysis with
<code>match</code> and <code>if let</code>. If an error occurs, it simply stops executing the
function and returns the error (by converting it to a string).</p>
<p>Isn't this a step backwards though? Previously, we said that the key to
ergonomic error handling is reducing explicit case analysis, yet we've reverted
back to explicit case analysis here. It turns out, there are <em>multiple</em> ways to
reduce explicit case analysis. Combinators aren't the only way.</p>
<a class="header" href="error-handling.html#the-try-macro" id="the-try-macro"><h2>The <code>try!</code> macro</h2></a>
<p>A cornerstone of error handling in Rust is the <code>try!</code> macro. The <code>try!</code> macro
abstracts case analysis like combinators, but unlike combinators, it also
abstracts <em>control flow</em>. Namely, it can abstract the <em>early return</em> pattern
seen above.</p>
<p>Here is a simplified definition of a <code>try!</code> macro:</p>
<p><span id="code-try-def-simple"></span></p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
macro_rules! try {
($e:expr) => (match $e {
Ok(val) => val,
Err(err) => return Err(err),
});
}
#}</code></pre></pre>
<p>(The <a href="../../std/macro.try.html">real definition</a> is a bit more
sophisticated. We will address that later.)</p>
<p>Using the <code>try!</code> macro makes it very easy to simplify our last example. Since
it does the case analysis and the early return for us, we get tighter code that
is easier to read:</p>
<pre><pre class="playpen"><code class="language-rust">use std::fs::File;
use std::io::Read;
use std::path::Path;
fn file_double<P: AsRef<Path>>(file_path: P) -> Result<i32, String> {
let mut file = try!(File::open(file_path).map_err(|e| e.to_string()));
let mut contents = String::new();
try!(file.read_to_string(&mut contents).map_err(|e| e.to_string()));
let n = try!(contents.trim().parse::<i32>().map_err(|e| e.to_string()));
Ok(2 * n)
}
fn main() {
match file_double("foobar") {
Ok(n) => println!("{}", n),
Err(err) => println!("Error: {}", err),
}
}
</code></pre></pre>
<p>The <code>map_err</code> calls are still necessary given
<a href="error-handling.html#code-try-def-simple">our definition of <code>try!</code></a>.
This is because the error types still need to be converted to <code>String</code>.
The good news is that we will soon learn how to remove those <code>map_err</code> calls!
The bad news is that we will need to learn a bit more about a couple important
traits in the standard library before we can remove the <code>map_err</code> calls.</p>
<a class="header" href="error-handling.html#defining-your-own-error-type" id="defining-your-own-error-type"><h2>Defining your own error type</h2></a>
<p>Before we dive into some of the standard library error traits, I'd like to wrap
up this section by removing the use of <code>String</code> as our error type in the
previous examples.</p>
<p>Using <code>String</code> as we did in our previous examples is convenient because it's
easy to convert errors to strings, or even make up your own errors as strings
on the spot. However, using <code>String</code> for your errors has some downsides.</p>
<p>The first downside is that the error messages tend to clutter your
code. It's possible to define the error messages elsewhere, but unless
you're unusually disciplined, it is very tempting to embed the error
message into your code. Indeed, we did exactly this in a <a href="error-handling.html#code-error-double-string">previous
example</a>.</p>
<p>The second and more important downside is that <code>String</code>s are <em>lossy</em>. That is,
if all errors are converted to strings, then the errors we pass to the caller
become completely opaque. The only reasonable thing the caller can do with a
<code>String</code> error is show it to the user. Certainly, inspecting the string to
determine the type of error is not robust. (Admittedly, this downside is far
more important inside of a library as opposed to, say, an application.)</p>
<p>For example, the <code>io::Error</code> type embeds an
<a href="../../std/io/enum.ErrorKind.html"><code>io::ErrorKind</code></a>,
which is <em>structured data</em> that represents what went wrong during an IO
operation. This is important because you might want to react differently
depending on the error. (e.g., A <code>BrokenPipe</code> error might mean quitting your
program gracefully while a <code>NotFound</code> error might mean exiting with an error
code and showing an error to the user.) With <code>io::ErrorKind</code>, the caller can
examine the type of an error with case analysis, which is strictly superior
to trying to tease out the details of an error inside of a <code>String</code>.</p>
<p>Instead of using a <code>String</code> as an error type in our previous example of reading
an integer from a file, we can define our own error type that represents errors
with <em>structured data</em>. We endeavor to not drop information from underlying
errors in case the caller wants to inspect the details.</p>
<p>The ideal way to represent <em>one of many possibilities</em> is to define our own
sum type using <code>enum</code>. In our case, an error is either an <code>io::Error</code> or a
<code>num::ParseIntError</code>, so a natural definition arises:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
use std::io;
use std::num;
// We derive `Debug` because all types should probably derive `Debug`.
// This gives us a reasonable human readable description of `CliError` values.
#[derive(Debug)]
enum CliError {
Io(io::Error),
Parse(num::ParseIntError),
}
#}</code></pre></pre>
<p>Tweaking our code is very easy. Instead of converting errors to strings, we
simply convert them to our <code>CliError</code> type using the corresponding value
constructor:</p>
<pre><pre class="playpen"><code class="language-rust"># #[derive(Debug)]
# enum CliError { Io(::std::io::Error), Parse(::std::num::ParseIntError) }
use std::fs::File;
use std::io::Read;
use std::path::Path;
fn file_double<P: AsRef<Path>>(file_path: P) -> Result<i32, CliError> {
let mut file = try!(File::open(file_path).map_err(CliError::Io));
let mut contents = String::new();
try!(file.read_to_string(&mut contents).map_err(CliError::Io));
let n: i32 = try!(contents.trim().parse().map_err(CliError::Parse));
Ok(2 * n)
}
fn main() {
match file_double("foobar") {
Ok(n) => println!("{}", n),
Err(err) => println!("Error: {:?}", err),
}
}
</code></pre></pre>
<p>The only change here is switching <code>map_err(|e| e.to_string())</code> (which converts
errors to strings) to <code>map_err(CliError::Io)</code> or <code>map_err(CliError::Parse)</code>.
The <em>caller</em> gets to decide the level of detail to report to the user. In
effect, using a <code>String</code> as an error type removes choices from the caller while
using a custom <code>enum</code> error type like <code>CliError</code> gives the caller all of the
conveniences as before in addition to <em>structured data</em> describing the error.</p>
<p>A rule of thumb is to define your own error type, but a <code>String</code> error type
will do in a pinch, particularly if you're writing an application. If you're
writing a library, defining your own error type should be strongly preferred so
that you don't remove choices from the caller unnecessarily.</p>
<a class="header" href="error-handling.html#standard-library-traits-used-for-error-handling" id="standard-library-traits-used-for-error-handling"><h1>Standard library traits used for error handling</h1></a>
<p>The standard library defines two integral traits for error handling:
<a href="../../std/error/trait.Error.html"><code>std::error::Error</code></a> and
<a href="../../std/convert/trait.From.html"><code>std::convert::From</code></a>. While <code>Error</code>
is designed specifically for generically describing errors, the <code>From</code>
trait serves a more general role for converting values between two
distinct types.</p>
<a class="header" href="error-handling.html#the-error-trait" id="the-error-trait"><h2>The <code>Error</code> trait</h2></a>
<p>The <code>Error</code> trait is <a href="../../std/error/trait.Error.html">defined in the standard
library</a>:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
use std::fmt::{Debug, Display};
trait Error: Debug + Display {
/// A short description of the error.
fn description(&self) -> &str;
/// The lower level cause of this error, if any.
fn cause(&self) -> Option<&Error> { None }
}
#}</code></pre></pre>
<p>This trait is super generic because it is meant to be implemented for <em>all</em>
types that represent errors. This will prove useful for writing composable code
as we'll see later. Otherwise, the trait allows you to do at least the
following things:</p>
<ul>
<li>Obtain a <code>Debug</code> representation of the error.</li>
<li>Obtain a user-facing <code>Display</code> representation of the error.</li>
<li>Obtain a short description of the error (via the <code>description</code> method).</li>
<li>Inspect the causal chain of an error, if one exists (via the <code>cause</code> method).</li>
</ul>
<p>The first two are a result of <code>Error</code> requiring impls for both <code>Debug</code> and
<code>Display</code>. The latter two are from the two methods defined on <code>Error</code>. The
power of <code>Error</code> comes from the fact that all error types impl <code>Error</code>, which
means errors can be existentially quantified as a
<a href="trait-objects.html">trait object</a>.
This manifests as either <code>Box<Error></code> or <code>&Error</code>. Indeed, the <code>cause</code> method
returns an <code>&Error</code>, which is itself a trait object. We'll revisit the
<code>Error</code> trait's utility as a trait object later.</p>
<p>For now, it suffices to show an example implementing the <code>Error</code> trait. Let's
use the error type we defined in the
<a href="error-handling.html#defining-your-own-error-type">previous section</a>:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
use std::io;
use std::num;
// We derive `Debug` because all types should probably derive `Debug`.
// This gives us a reasonable human readable description of `CliError` values.
#[derive(Debug)]
enum CliError {
Io(io::Error),
Parse(num::ParseIntError),
}
#}</code></pre></pre>
<p>This particular error type represents the possibility of two types of errors
occurring: an error dealing with I/O or an error converting a string to a
number. The error could represent as many error types as you want by adding new
variants to the <code>enum</code> definition.</p>
<p>Implementing <code>Error</code> is pretty straight-forward. It's mostly going to be a lot
explicit case analysis.</p>
<pre><code class="language-rust ignore">use std::error;
use std::fmt;
impl fmt::Display for CliError {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match *self {
// Both underlying errors already impl `Display`, so we defer to
// their implementations.
CliError::Io(ref err) => write!(f, "IO error: {}", err),
CliError::Parse(ref err) => write!(f, "Parse error: {}", err),
}
}
}
impl error::Error for CliError {
fn description(&self) -> &str {
// Both underlying errors already impl `Error`, so we defer to their
// implementations.
match *self {
CliError::Io(ref err) => err.description(),
CliError::Parse(ref err) => err.description(),
}
}
fn cause(&self) -> Option<&error::Error> {
match *self {
// N.B. Both of these implicitly cast `err` from their concrete
// types (either `&io::Error` or `&num::ParseIntError`)
// to a trait object `&Error`. This works because both error types
// implement `Error`.
CliError::Io(ref err) => Some(err),
CliError::Parse(ref err) => Some(err),
}
}
}
</code></pre>
<p>We note that this is a very typical implementation of <code>Error</code>: match on your
different error types and satisfy the contracts defined for <code>description</code> and
<code>cause</code>.</p>
<a class="header" href="error-handling.html#the-from-trait" id="the-from-trait"><h2>The <code>From</code> trait</h2></a>
<p>The <code>std::convert::From</code> trait is
<a href="../../std/convert/trait.From.html">defined in the standard
library</a>:</p>
<p><span id="code-from-def"></span></p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
trait From<T> {
fn from(T) -> Self;
}
#}</code></pre></pre>
<p>Deliciously simple, yes? <code>From</code> is very useful because it gives us a generic
way to talk about conversion <em>from</em> a particular type <code>T</code> to some other type
(in this case, “some other type” is the subject of the impl, or <code>Self</code>).
The crux of <code>From</code> is the
<a href="../../std/convert/trait.From.html">set of implementations provided by the standard
library</a>.</p>
<p>Here are a few simple examples demonstrating how <code>From</code> works:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
let string: String = From::from("foo");
let bytes: Vec<u8> = From::from("foo");
let cow: ::std::borrow::Cow<str> = From::from("foo");
#}</code></pre></pre>
<p>OK, so <code>From</code> is useful for converting between strings. But what about errors?
It turns out, there is one critical impl:</p>
<pre><code class="language-rust ignore">impl<'a, E: Error + 'a> From<E> for Box<Error + 'a>
</code></pre>
<p>This impl says that for <em>any</em> type that impls <code>Error</code>, we can convert it to a
trait object <code>Box<Error></code>. This may not seem terribly surprising, but it is
useful in a generic context.</p>
<p>Remember the two errors we were dealing with previously? Specifically,
<code>io::Error</code> and <code>num::ParseIntError</code>. Since both impl <code>Error</code>, they work with
<code>From</code>:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
use std::error::Error;
use std::fs;
use std::io;
use std::num;
// We have to jump through some hoops to actually get error values:
let io_err: io::Error = io::Error::last_os_error();
let parse_err: num::ParseIntError = "not a number".parse::<i32>().unwrap_err();
// OK, here are the conversions:
let err1: Box<Error> = From::from(io_err);
let err2: Box<Error> = From::from(parse_err);
#}</code></pre></pre>
<p>There is a really important pattern to recognize here. Both <code>err1</code> and <code>err2</code>
have the <em>same type</em>. This is because they are existentially quantified types,
or trait objects. In particular, their underlying type is <em>erased</em> from the
compiler's knowledge, so it truly sees <code>err1</code> and <code>err2</code> as exactly the same.
Additionally, we constructed <code>err1</code> and <code>err2</code> using precisely the same
function call: <code>From::from</code>. This is because <code>From::from</code> is overloaded on both
its argument and its return type.</p>
<p>This pattern is important because it solves a problem we had earlier: it gives
us a way to reliably convert errors to the same type using the same function.</p>
<p>Time to revisit an old friend; the <code>try!</code> macro.</p>
<a class="header" href="error-handling.html#the-real-try-macro" id="the-real-try-macro"><h2>The real <code>try!</code> macro</h2></a>
<p>Previously, we presented this definition of <code>try!</code>:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
macro_rules! try {
($e:expr) => (match $e {
Ok(val) => val,
Err(err) => return Err(err),
});
}
#}</code></pre></pre>
<p>This is not its real definition. Its real definition is
<a href="../../std/macro.try.html">in the standard library</a>:</p>
<p><span id="code-try-def"></span></p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
macro_rules! try {
($e:expr) => (match $e {
Ok(val) => val,
Err(err) => return Err(::std::convert::From::from(err)),
});
}
#}</code></pre></pre>
<p>There's one tiny but powerful change: the error value is passed through
<code>From::from</code>. This makes the <code>try!</code> macro a lot more powerful because it gives
you automatic type conversion for free.</p>
<p>Armed with our more powerful <code>try!</code> macro, let's take a look at code we wrote
previously to read a file and convert its contents to an integer:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
use std::fs::File;
use std::io::Read;
use std::path::Path;
fn file_double<P: AsRef<Path>>(file_path: P) -> Result<i32, String> {
let mut file = try!(File::open(file_path).map_err(|e| e.to_string()));
let mut contents = String::new();
try!(file.read_to_string(&mut contents).map_err(|e| e.to_string()));
let n = try!(contents.trim().parse::<i32>().map_err(|e| e.to_string()));
Ok(2 * n)
}
#}</code></pre></pre>
<p>Earlier, we promised that we could get rid of the <code>map_err</code> calls. Indeed, all
we have to do is pick a type that <code>From</code> works with. As we saw in the previous
section, <code>From</code> has an impl that lets it convert any error type into a
<code>Box<Error></code>:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
use std::error::Error;
use std::fs::File;
use std::io::Read;
use std::path::Path;
fn file_double<P: AsRef<Path>>(file_path: P) -> Result<i32, Box<Error>> {
let mut file = try!(File::open(file_path));
let mut contents = String::new();
try!(file.read_to_string(&mut contents));
let n = try!(contents.trim().parse::<i32>());
Ok(2 * n)
}
#}</code></pre></pre>
<p>We are getting very close to ideal error handling. Our code has very little
overhead as a result from error handling because the <code>try!</code> macro encapsulates
three things simultaneously:</p>
<ol>
<li>Case analysis.</li>
<li>Control flow.</li>
<li>Error type conversion.</li>
</ol>
<p>When all three things are combined, we get code that is unencumbered by
combinators, calls to <code>unwrap</code> or case analysis.</p>
<p>There's one little nit left: the <code>Box<Error></code> type is <em>opaque</em>. If we
return a <code>Box<Error></code> to the caller, the caller can't (easily) inspect
underlying error type. The situation is certainly better than <code>String</code>
because the caller can call methods like
<a href="../../std/error/trait.Error.html#tymethod.description"><code>description</code></a>
and <a href="../../std/error/trait.Error.html#method.cause"><code>cause</code></a>, but the
limitation remains: <code>Box<Error></code> is opaque. (N.B. This isn't entirely
true because Rust does have runtime reflection, which is useful in
some scenarios that are <a href="https://crates.io/crates/error">beyond the scope of this
section</a>.)</p>
<p>It's time to revisit our custom <code>CliError</code> type and tie everything together.</p>
<a class="header" href="error-handling.html#composing-custom-error-types" id="composing-custom-error-types"><h2>Composing custom error types</h2></a>
<p>In the last section, we looked at the real <code>try!</code> macro and how it does
automatic type conversion for us by calling <code>From::from</code> on the error value.
In particular, we converted errors to <code>Box<Error></code>, which works, but the type
is opaque to callers.</p>
<p>To fix this, we use the same remedy that we're already familiar with: a custom
error type. Once again, here is the code that reads the contents of a file and
converts it to an integer:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
use std::fs::File;
use std::io::{self, Read};
use std::num;
use std::path::Path;
// We derive `Debug` because all types should probably derive `Debug`.
// This gives us a reasonable human readable description of `CliError` values.
#[derive(Debug)]
enum CliError {
Io(io::Error),
Parse(num::ParseIntError),
}
fn file_double_verbose<P: AsRef<Path>>(file_path: P) -> Result<i32, CliError> {
let mut file = try!(File::open(file_path).map_err(CliError::Io));
let mut contents = String::new();
try!(file.read_to_string(&mut contents).map_err(CliError::Io));
let n: i32 = try!(contents.trim().parse().map_err(CliError::Parse));
Ok(2 * n)
}
#}</code></pre></pre>
<p>Notice that we still have the calls to <code>map_err</code>. Why? Well, recall the
definitions of <a href="error-handling.html#code-try-def"><code>try!</code></a> and <a href="error-handling.html#code-from-def"><code>From</code></a>. The
problem is that there is no <code>From</code> impl that allows us to convert from error
types like <code>io::Error</code> and <code>num::ParseIntError</code> to our own custom <code>CliError</code>.
Of course, it is easy to fix this! Since we defined <code>CliError</code>, we can impl
<code>From</code> with it:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
# #[derive(Debug)]
# enum CliError { Io(io::Error), Parse(num::ParseIntError) }
use std::io;
use std::num;
impl From<io::Error> for CliError {
fn from(err: io::Error) -> CliError {
CliError::Io(err)
}
}
impl From<num::ParseIntError> for CliError {
fn from(err: num::ParseIntError) -> CliError {
CliError::Parse(err)
}
}
#}</code></pre></pre>
<p>All these impls are doing is teaching <code>From</code> how to create a <code>CliError</code> from
other error types. In our case, construction is as simple as invoking the
corresponding value constructor. Indeed, it is <em>typically</em> this easy.</p>
<p>We can finally rewrite <code>file_double</code>:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
# use std::io;
# use std::num;
# enum CliError { Io(::std::io::Error), Parse(::std::num::ParseIntError) }
# impl From<io::Error> for CliError {
# fn from(err: io::Error) -> CliError { CliError::Io(err) }
# }
# impl From<num::ParseIntError> for CliError {
# fn from(err: num::ParseIntError) -> CliError { CliError::Parse(err) }
# }
use std::fs::File;
use std::io::Read;
use std::path::Path;
fn file_double<P: AsRef<Path>>(file_path: P) -> Result<i32, CliError> {
let mut file = try!(File::open(file_path));
let mut contents = String::new();
try!(file.read_to_string(&mut contents));
let n: i32 = try!(contents.trim().parse());
Ok(2 * n)
}
#}</code></pre></pre>
<p>The only thing we did here was remove the calls to <code>map_err</code>. They are no
longer needed because the <code>try!</code> macro invokes <code>From::from</code> on the error value.
This works because we've provided <code>From</code> impls for all the error types that
could appear.</p>
<p>If we modified our <code>file_double</code> function to perform some other operation, say,
convert a string to a float, then we'd need to add a new variant to our error
type:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
use std::io;
use std::num;
enum CliError {
Io(io::Error),
ParseInt(num::ParseIntError),
ParseFloat(num::ParseFloatError),
}
#}</code></pre></pre>
<p>And add a new <code>From</code> impl:</p>
<pre><pre class="playpen"><code class="language-rust"># #![allow(unused_variables)]
#fn main() {
# enum CliError {
# Io(::std::io::Error),
# ParseInt(num::ParseIntError),
# ParseFloat(num::ParseFloatError),
# }
use std::num;
impl From<num::ParseFloatError> for CliError {
fn from(err: num::ParseFloatError) -> CliError {
CliError::ParseFloat(err)
}
}
#}</code></pre></pre>
<p>And that's it!</p>
<a class="header" href="error-handling.html#advice-for-library-writers" id="advice-for-library-writers"><h2>Advice for library writers</h2></a>
<p>If your library needs to report custom errors, then you should
probably define your own error type. It's up to you whether or not to
expose its representation (like
<a href="../../std/io/enum.ErrorKind.html"><code>ErrorKind</code></a>) or keep it hidden (like
<a href="../../std/num/struct.ParseIntError.html"><code>ParseIntError</code></a>). Regardless
of how you do it, it's usually good practice to at least provide some
information about the error beyond its <code>String</code>
representation. But certainly, this will vary depending on use cases.</p>
<p>At a minimum, you should probably implement the
<a href="../../std/error/trait.Error.html"><code>Error</code></a>
trait. This will give users of your library some minimum flexibility for
<a href="error-handling.html#the-real-try-macro">composing errors</a>. Implementing the <code>Error</code> trait also
means that users are guaranteed the ability to obtain a string representation
of an error (because it requires impls for both <code>fmt::Debug</code> and
<code>fmt::Display</code>).</p>
<p>Beyond that, it can also be useful to provide implementations of <code>From</code> on your
error types. This allows you (the library author) and your users to
<a href="error-handling.html#composing-custom-error-types">compose more detailed errors</a>. For example,
<a href="http://burntsushi.net/rustdoc/csv/enum.Error.html"><code>csv::Error</code></a>
provides <code>From</code> impls for both <code>io::Error</code> and <code>byteorder::Error</code>.</p>
<p>Finally, depending on your tastes, you may also want to define a
<a href="error-handling.html#the-result-type-alias-idiom"><code>Result</code> type alias</a>, particularly if your
library defines a single error type. This is used in the standard library
for <a href="../../std/io/type.Result.html"><code>io::Result</code></a>
and <a href="../../std/fmt/type.Result.html"><code>fmt::Result</code></a>.</p>
<a class="header" href="error-handling.html#case-study-a-program-to-read-population-data" id="case-study-a-program-to-read-population-data"><h1>Case study: A program to read population data</h1></a>
<p>This section was long, and depending on your background, it might be
rather dense. While there is plenty of example code to go along with
the prose, most of it was specifically designed to be pedagogical. So,
we're going to do something new: a case study.</p>
<p>For this, we're going to build up a command line program that lets you
query world population data. The objective is simple: you give it a location
and it will tell you the population. Despite the simplicity, there is a lot
that can go wrong!</p>
<p>The data we'll be using comes from the <a href="https://github.com/petewarden/dstkdata">Data Science
Toolkit</a>. I've prepared some data from it for this exercise. You
can either grab the <a href="http://burntsushi.net/stuff/worldcitiespop.csv.gz">world population data</a> (41MB gzip compressed,
145MB uncompressed) or only the <a href="http://burntsushi.net/stuff/uscitiespop.csv.gz">US population data</a> (2.2MB gzip
compressed, 7.2MB uncompressed).</p>
<p>Up until now, we've kept the code limited to Rust's standard library. For a real
task like this though, we'll want to at least use something to parse CSV data,
parse the program arguments and decode that stuff into Rust types automatically. For that, we'll use the
<a href="https://crates.io/crates/csv"><code>csv</code></a>,
and <a href="https://crates.io/crates/rustc-serialize"><code>rustc-serialize</code></a> crates.</p>
<a class="header" href="error-handling.html#initial-setup" id="initial-setup"><h2>Initial setup</h2></a>
<p>We're not going to spend a lot of time on setting up a project with
Cargo because it is already covered well in <a href="getting-started.html#hello-cargo">the Cargo
section</a> and <a href="http://doc.crates.io/guide.html">Cargo's documentation</a>.</p>
<p>To get started from scratch, run <code>cargo new --bin city-pop</code> and make sure your
<code>Cargo.toml</code> looks something like this:</p>
<pre><code class="language-text">[package]
name = "city-pop"
version = "0.1.0"
authors = ["Andrew Gallant <jamslam@gmail.com>"]
[[bin]]
name = "city-pop"
[dependencies]
csv = "0.*"
rustc-serialize = "0.*"
getopts = "0.*"
</code></pre>
<p>You should already be able to run:</p>
<pre><code class="language-text">cargo build --release
./target/release/city-pop
# Outputs: Hello, world!
</code></pre>
<a class="header" href="error-handling.html#argument-parsing" id="argument-parsing"><h2>Argument parsing</h2></a>
<p>Let's get argument parsing out of the way. We won't go into too much
detail on Getopts, but there is <a href="http://doc.rust-lang.org/getopts/getopts/index.html">some good documentation</a>
describing it. The short story is that Getopts generates an argument
parser and a help message from a vector of options (The fact that it
is a vector is hidden behind a struct and a set of methods). Once the
parsing is done, the parser returns a struct that records matches
for defined options, and remaining "free" arguments.
From there, we can get information about the flags, for
instance, whether they were passed in, and what arguments they
had. Here's our program with the appropriate <code>extern crate</code>
statements, and the basic argument setup for Getopts:</p>
<pre><code class="language-rust ignore">extern crate getopts;
extern crate rustc_serialize;
use getopts::Options;
use std::env;
fn print_usage(program: &str, opts: Options) {
println!("{}", opts.usage(&format!("Usage: {} [options] <data-path> <city>", program)));
}
fn main() {
let args: Vec<String> = env::args().collect();
let program = &args[0];
let mut opts = Options::new();
opts.optflag("h", "help", "Show this usage message.");
let matches = match opts.parse(&args[1..]) {
Ok(m) => { m }
Err(e) => { panic!(e.to_string()) }
};
if matches.opt_present("h") {
print_usage(&program, opts);
return;
}
let data_path = &matches.free[0];
let city: &str = &matches.free[1];
// Do stuff with information.
}
</code></pre>
<p>First, we get a vector of the arguments passed into our program. We
then store the first one, knowing that it is our program's name. Once
that's done, we set up our argument flags, in this case a simplistic
help message flag. Once we have the argument flags set up, we use
<code>Options.parse</code> to parse the argument vector (starting from index one,
because index 0 is the program name). If this was successful, we
assign matches to the parsed object, if not, we panic. Once past that,
we test if the user passed in the help flag, and if so print the usage
message. The option help messages are constructed by Getopts, so all
we have to do to print the usage message is tell it what we want it to
print for the program name and template. If the user has not passed in
the help flag, we assign the proper variables to their corresponding
arguments.</p>
<a class="header" href="error-handling.html#writing-the-logic" id="writing-the-logic"><h2>Writing the logic</h2></a>
<p>We all write code differently, but error handling is usually the last thing we
want to think about. This isn't great for the overall design of a program, but
it can be useful for rapid prototyping. Because Rust forces us to be explicit
about error handling (by making us call <code>unwrap</code>), it is easy to see which
parts of our program can cause errors.</p>
<p>In this case study, the logic is really simple. All we need to do is parse the
CSV data given to us and print out a field in matching rows. Let's do it. (Make
sure to add <code>extern crate csv;</code> to the top of your file.)</p>
<pre><code class="language-rust ignore">use std::fs::File;
// This struct represents the data in each row of the CSV file.
// Type based decoding absolves us of a lot of the nitty gritty error
// handling, like parsing strings as integers or floats.
#[derive(Debug, RustcDecodable)]
struct Row {
country: String,
city: String,
accent_city: String,
region: String,
// Not every row has data for the population, latitude or longitude!
// So we express them as `Option` types, which admits the possibility of
// absence. The CSV parser will fill in the correct value for us.
population: Option<u64>,
latitude: Option<f64>,
longitude: Option<f64>,
}
fn print_usage(program: &str, opts: Options) {
println!("{}", opts.usage(&format!("Usage: {} [options] <data-path> <city>", program)));
}
fn main() {
let args: Vec<String> = env::args().collect();
let program = &args[0];
let mut opts = Options::new();
opts.optflag("h", "help", "Show this usage message.");
let matches = match opts.parse(&args[1..]) {
Ok(m) => { m }
Err(e) => { panic!(e.to_string()) }
};
if matches.opt_present("h") {
print_usage(&program, opts);
return;
}
let data_path = &matches.free[0];
let city: &str = &matches.free[1];
let file = File::open(data_path).unwrap();
let mut rdr = csv::Reader::from_reader(file);
for row in rdr.decode::<Row>() {
let row = row.unwrap();
if row.city == city {
println!("{}, {}: {:?}",
row.city, row.country,
row.population.expect("population count"));
}
}
}
</code></pre>
<p>Let's outline the errors. We can start with the obvious: the three places that
<code>unwrap</code> is called:</p>
<ol>
<li><a href="../../std/fs/struct.File.html#method.open"><code>File::open</code></a>
can return an
<a href="../../std/io/struct.Error.html"><code>io::Error</code></a>.</li>
<li><a href="http://burntsushi.net/rustdoc/csv/struct.Reader.html#method.decode"><code>csv::Reader::decode</code></a>
decodes one record at a time, and
<a href="http://burntsushi.net/rustdoc/csv/struct.DecodedRecords.html">decoding a
record</a>
(look at the <code>Item</code> associated type on the <code>Iterator</code> impl)
can produce a
<a href="http://burntsushi.net/rustdoc/csv/enum.Error.html"><code>csv::Error</code></a>.</li>
<li>If <code>row.population</code> is <code>None</code>, then calling <code>expect</code> will panic.</li>
</ol>
<p>Are there any others? What if we can't find a matching city? Tools like <code>grep</code>
will return an error code, so we probably should too. So we have logic errors
specific to our problem, IO errors and CSV parsing errors. We're going to
explore two different ways to approach handling these errors.</p>
<p>I'd like to start with <code>Box<Error></code>. Later, we'll see how defining our own
error type can be useful.</p>
<a class="header" href="error-handling.html#error-handling-with-boxerror" id="error-handling-with-boxerror"><h2>Error handling with <code>Box<Error></code></h2></a>
<p><code>Box<Error></code> is nice because it <em>just works</em>. You don't need to define your own
error types and you don't need any <code>From</code> implementations. The downside is that
since <code>Box<Error></code> is a trait object, it <em>erases the type</em>, which means the
compiler can no longer reason about its underlying type.</p>
<p><a href="error-handling.html#the-limits-of-combinators">Previously</a> we started refactoring our code by
changing the type of our function from <code>T</code> to <code>Result<T, OurErrorType></code>. In
this case, <code>OurErrorType</code> is only <code>Box<Error></code>. But what's <code>T</code>? And can we add
a return type to <code>main</code>?</p>
<p>The answer to the second question is no, we can't. That means we'll need to
write a new function. But what is <code>T</code>? The simplest thing we can do is to
return a list of matching <code>Row</code> values as a <code>Vec<Row></code>. (Better code would
return an iterator, but that is left as an exercise to the reader.)</p>
<p>Let's refactor our code into its own function, but keep the calls to <code>unwrap</code>.
Note that we opt to handle the possibility of a missing population count by
simply ignoring that row.</p>
<pre><code class="language-rust ignore">use std::path::Path;
struct Row {
// This struct remains unchanged.
}
struct PopulationCount {
city: String,
country: String,
// This is no longer an `Option` because values of this type are only
// constructed if they have a population count.
count: u64,
}
fn print_usage(program: &str, opts: Options) {
println!("{}", opts.usage(&format!("Usage: {} [options] <data-path> <city>", program)));
}
fn search<P: AsRef<Path>>(file_path: P, city: &str) -> Vec<PopulationCount> {
let mut found = vec![];
let file = File::open(file_path).unwrap();
let mut rdr = csv::Reader::from_reader(file);
for row in rdr.decode::<Row>() {
let row = row.unwrap();
match row.population {
None => { } // Skip it.
Some(count) => if row.city == city {
found.push(PopulationCount {
city: row.city,
country: row.country,
count: count,
});
},
}
}
found
}
fn main() {
let args: Vec<String> = env::args().collect();
let program = &args[0];
let mut opts = Options::new();
opts.optflag("h", "help", "Show this usage message.");
let matches = match opts.parse(&args[1..]) {
Ok(m) => { m }
Err(e) => { panic!(e.to_string()) }
};
if matches.opt_present("h") {
print_usage(&program, opts);
return;
}
let data_path = &matches.free[0];
let city: &str = &matches.free[1];
for pop in search(data_path, city) {
println!("{}, {}: {:?}", pop.city, pop.country, pop.count);
}
}
</code></pre>
<p>While we got rid of one use of <code>expect</code> (which is a nicer variant of <code>unwrap</code>),
we still should handle the absence of any search results.</p>
<p>To convert this to proper error handling, we need to do the following:</p>
<ol>
<li>Change the return type of <code>search</code> to be <code>Result<Vec<PopulationCount>, Box<Error>></code>.</li>
<li>Use the <a href="error-handling.html#code-try-def"><code>try!</code> macro</a> so that errors are returned to the
caller instead of panicking the program.</li>
<li>Handle the error in <code>main</code>.</li>
</ol>
<p>Let's try it:</p>
<pre><code class="language-rust ignore">use std::error::Error;
// The rest of the code before this is unchanged.
fn search<P: AsRef<Path>>
(file_path: P, city: &str)
-> Result<Vec<PopulationCount>, Box<Error>> {
let mut found = vec![];
let file = try!(File::open(file_path));
let mut rdr = csv::Reader::from_reader(file);
for row in rdr.decode::<Row>() {
let row = try!(row);
match row.population {
None => { } // Skip it.
Some(count) => if row.city == city {
found.push(PopulationCount {
city: row.city,
country: row.country,
count: count,
});
},
}
}
if found.is_empty() {
Err(From::from("No matching cities with a population were found."))
} else {
Ok(found)
}
}
</code></pre>
<p>Instead of <code>x.unwrap()</code>, we now have <code>try!(x)</code>. Since our function returns a
<code>Result<T, E></code>, the <code>try!</code> macro will return early from the function if an
error occurs.</p>
<p>At the end of <code>search</code> we also convert a plain string to an error type
by using the <a href="../../std/convert/trait.From.html">corresponding <code>From</code> impls</a>:</p>
<pre><code class="language-rust ignore">// We are making use of this impl in the code above, since we call `From::from`
// on a `&'static str`.
impl<'a> From<&'a str> for Box<Error>
// But this is also useful when you need to allocate a new string for an
// error message, usually with `format!`.
impl From<String> for Box<Error>
</code></pre>
<p>Since <code>search</code> now returns a <code>Result<T, E></code>, <code>main</code> should use case analysis
when calling <code>search</code>:</p>
<pre><code class="language-rust ignore">...
match search(data_path, city) {
Ok(pops) => {
for pop in pops {
println!("{}, {}: {:?}", pop.city, pop.country, pop.count);
}
}
Err(err) => println!("{}", err)
}
...
</code></pre>
<p>Now that we've seen how to do proper error handling with <code>Box<Error></code>, let's
try a different approach with our own custom error type. But first, let's take
a quick break from error handling and add support for reading from <code>stdin</code>.</p>
<a class="header" href="error-handling.html#reading-from-stdin" id="reading-from-stdin"><h2>Reading from stdin</h2></a>
<p>In our program, we accept a single file for input and do one pass over the
data. This means we probably should be able to accept input on stdin. But maybe
we like the current format too—so let's have both!</p>
<p>Adding support for stdin is actually quite easy. There are only three things we
have to do:</p>
<ol>
<li>Tweak the program arguments so that a single parameter—the
city—can be accepted while the population data is read from stdin.</li>
<li>Modify the program so that an option <code>-f</code> can take the file, if it
is not passed into stdin.</li>
<li>Modify the <code>search</code> function to take an <em>optional</em> file path. When <code>None</code>,
it should know to read from stdin.</li>
</ol>
<p>First, here's the new usage:</p>
<pre><code class="language-rust ignore">fn print_usage(program: &str, opts: Options) {
println!("{}", opts.usage(&format!("Usage: {} [options] <city>", program)));
}
</code></pre>
<p>Of course we need to adapt the argument handling code:</p>
<pre><code class="language-rust ignore">...
let mut opts = Options::new();
opts.optopt("f", "file", "Choose an input file, instead of using STDIN.", "NAME");
opts.optflag("h", "help", "Show this usage message.");
...
let data_path = matches.opt_str("f");
let city = if !matches.free.is_empty() {
&matches.free[0]
} else {
print_usage(&program, opts);
return;
};
match search(&data_path, city) {
Ok(pops) => {
for pop in pops {
println!("{}, {}: {:?}", pop.city, pop.country, pop.count);
}
}
Err(err) => println!("{}", err)
}
...
</code></pre>
<p>We've made the user experience a bit nicer by showing the usage message,
instead of a panic from an out-of-bounds index, when <code>city</code>, the
remaining free argument, is not present.</p>
<p>Modifying <code>search</code> is slightly trickier. The <code>csv</code> crate can build a
parser out of
<a href="http://burntsushi.net/rustdoc/csv/struct.Reader.html#method.from_reader">any type that implements <code>io::Read</code></a>.
But how can we use the same code over both types? There's actually a
couple ways we could go about this. One way is to write <code>search</code> such
that it is generic on some type parameter <code>R</code> that satisfies
<code>io::Read</code>. Another way is to use trait objects:</p>
<pre><code class="language-rust ignore">use std::io;
// The rest of the code before this is unchanged.
fn search<P: AsRef<Path>>
(file_path: &Option<P>, city: &str)
-> Result<Vec<PopulationCount>, Box<Error>> {
let mut found = vec![];
let input: Box<io::Read> = match *file_path {
None => Box::new(io::stdin()),
Some(ref file_path) => Box::new(try!(File::open(file_path))),
};
let mut rdr = csv::Reader::from_reader(input);
// The rest remains unchanged!
}
</code></pre>
<a class="header" href="error-handling.html#error-handling-with-a-custom-type" id="error-handling-with-a-custom-type"><h2>Error handling with a custom type</h2></a>
<p>Previously, we learned how to
<a href="error-handling.html#composing-custom-error-types">compose errors using a custom error type</a>.
We did this by defining our error type as an <code>enum</code> and implementing <code>Error</code>
and <code>From</code>.</p>
<p>Since we have three distinct errors (IO, CSV parsing and not found), let's
define an <code>enum</code> with three variants:</p>
<pre><code class="language-rust ignore">#[derive(Debug)]
enum CliError {
Io(io::Error),
Csv(csv::Error),
NotFound,
}
</code></pre>
<p>And now for impls on <code>Display</code> and <code>Error</code>:</p>
<pre><code class="language-rust ignore">use std::fmt;
impl fmt::Display for CliError {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match *self {
CliError::Io(ref err) => err.fmt(f),
CliError::Csv(ref err) => err.fmt(f),
CliError::NotFound => write!(f, "No matching cities with a \
population were found."),
}
}
}
impl Error for CliError {
fn description(&self) -> &str {
match *self {
CliError::Io(ref err) => err.description(),
CliError::Csv(ref err) => err.description(),
CliError::NotFound => "not found",
}
}
fn cause(&self) -> Option<&Error> {
match *self {
CliError::Io(ref err) => Some(err),
CliError::Csv(ref err) => Some(err),
// Our custom error doesn't have an underlying cause,
// but we could modify it so that it does.
CliError::NotFound => None,
}
}
}
</code></pre>
<p>Before we can use our <code>CliError</code> type in our <code>search</code> function, we need to
provide a couple <code>From</code> impls. How do we know which impls to provide? Well,
we'll need to convert from both <code>io::Error</code> and <code>csv::Error</code> to <code>CliError</code>.
Those are the only external errors, so we'll only need two <code>From</code> impls for
now:</p>
<pre><code class="language-rust ignore">impl From<io::Error> for CliError {
fn from(err: io::Error) -> CliError {
CliError::Io(err)
}
}
impl From<csv::Error> for CliError {
fn from(err: csv::Error) -> CliError {
CliError::Csv(err)
}
}
</code></pre>
<p>The <code>From</code> impls are important because of how
<a href="error-handling.html#code-try-def"><code>try!</code> is defined</a>. In particular, if an error occurs,
<code>From::from</code> is called on the error, which in this case, will convert it to our
own error type <code>CliError</code>.</p>
<p>With the <code>From</code> impls done, we only need to make two small tweaks to our
<code>search</code> function: the return type and the “not found” error. Here it is in
full:</p>
<pre><code class="language-rust ignore">fn search<P: AsRef<Path>>
(file_path: &Option<P>, city: &str)
-> Result<Vec<PopulationCount>, CliError> {
let mut found = vec![];
let input: Box<io::Read> = match *file_path {
None => Box::new(io::stdin()),
Some(ref file_path) => Box::new(try!(File::open(file_path))),
};
let mut rdr = csv::Reader::from_reader(input);
for row in rdr.decode::<Row>() {
let row = try!(row);
match row.population {
None => { } // Skip it.
Some(count) => if row.city == city {
found.push(PopulationCount {
city: row.city,
country: row.country,
count: count,
});
},
}
}
if found.is_empty() {
Err(CliError::NotFound)
} else {
Ok(found)
}
}
</code></pre>
<p>No other changes are necessary.</p>
<a class="header" href="error-handling.html#adding-functionality" id="adding-functionality"><h2>Adding functionality</h2></a>
<p>Writing generic code is great, because generalizing stuff is cool, and
it can then be useful later. But sometimes, the juice isn't worth the
squeeze. Look at what we just did in the previous step:</p>
<ol>
<li>Defined a new error type.</li>
<li>Added impls for <code>Error</code>, <code>Display</code> and two for <code>From</code>.</li>
</ol>
<p>The big downside here is that our program didn't improve a whole lot.
There is quite a bit of overhead to representing errors with <code>enum</code>s,
especially in short programs like this.</p>
<p><em>One</em> useful aspect of using a custom error type like we've done here is that
the <code>main</code> function can now choose to handle errors differently. Previously,
with <code>Box<Error></code>, it didn't have much of a choice: just print the message.
We're still doing that here, but what if we wanted to, say, add a <code>--quiet</code>
flag? The <code>--quiet</code> flag should silence any verbose output.</p>
<p>Right now, if the program doesn't find a match, it will output a message saying
so. This can be a little clumsy, especially if you intend for the program to
be used in shell scripts.</p>
<p>So let's start by adding the flags. Like before, we need to tweak the usage
string and add a flag to the Option variable. Once we've done that, Getopts does the rest:</p>
<pre><code class="language-rust ignore">...
let mut opts = Options::new();
opts.optopt("f", "file", "Choose an input file, instead of using STDIN.", "NAME");
opts.optflag("h", "help", "Show this usage message.");
opts.optflag("q", "quiet", "Silences errors and warnings.");
...
</code></pre>
<p>Now we only need to implement our “quiet” functionality. This requires us to
tweak the case analysis in <code>main</code>:</p>
<pre><code class="language-rust ignore">use std::process;
...
match search(&data_path, city) {
Err(CliError::NotFound) if matches.opt_present("q") => process::exit(1),
Err(err) => panic!("{}", err),
Ok(pops) => for pop in pops {
println!("{}, {}: {:?}", pop.city, pop.country, pop.count);
}
}
...
</code></pre>
<p>Certainly, we don't want to be quiet if there was an IO error or if the data
failed to parse. Therefore, we use case analysis to check if the error type is
<code>NotFound</code> <em>and</em> if <code>--quiet</code> has been enabled. If the search failed, we still
quit with an exit code (following <code>grep</code>'s convention).</p>
<p>If we had stuck with <code>Box<Error></code>, then it would be pretty tricky to implement
the <code>--quiet</code> functionality.</p>
<p>This pretty much sums up our case study. From here, you should be ready to go
out into the world and write your own programs and libraries with proper error
handling.</p>
<a class="header" href="error-handling.html#the-short-story" id="the-short-story"><h1>The Short Story</h1></a>
<p>Since this section is long, it is useful to have a quick summary for error
handling in Rust. These are some good “rules of thumb." They are emphatically
<em>not</em> commandments. There are probably good reasons to break every one of these
heuristics!</p>
<ul>
<li>If you're writing short example code that would be overburdened by error
handling, it's probably fine to use <code>unwrap</code> (whether that's
<a href="../../std/result/enum.Result.html#method.unwrap"><code>Result::unwrap</code></a>,
<a href="../../std/option/enum.Option.html#method.unwrap"><code>Option::unwrap</code></a>
or preferably
<a href="../../std/option/enum.Option.html#method.expect"><code>Option::expect</code></a>).
Consumers of your code should know to use proper error handling. (If they
don't, send them here!)</li>
<li>If you're writing a quick 'n' dirty program, don't feel ashamed if you use
<code>unwrap</code>. Be warned: if it winds up in someone else's hands, don't be
surprised if they are agitated by poor error messages!</li>
<li>If you're writing a quick 'n' dirty program and feel ashamed about panicking
anyway, then use either a <code>String</code> or a <code>Box<Error></code> for your
error type.</li>
<li>Otherwise, in a program, define your own error types with appropriate
<a href="../../std/convert/trait.From.html"><code>From</code></a>
and
<a href="../../std/error/trait.Error.html"><code>Error</code></a>
impls to make the <a href="../../std/macro.try.html"><code>try!</code></a>
macro more ergonomic.</li>
<li>If you're writing a library and your code can produce errors, define your own
error type and implement the
<a href="../../std/error/trait.Error.html"><code>std::error::Error</code></a>
trait. Where appropriate, implement
<a href="../../std/convert/trait.From.html"><code>From</code></a> to make both
your library code and the caller's code easier to write. (Because of Rust's
coherence rules, callers will not be able to impl <code>From</code> on your error type,
so your library should do it.)</li>
<li>Learn the combinators defined on
<a href="../../std/option/enum.Option.html"><code>Option</code></a>
and
<a href="../../std/result/enum.Result.html"><code>Result</code></a>.
Using them exclusively can be a bit tiring at times, but I've personally
found a healthy mix of <code>try!</code> and combinators to be quite appealing.
<code>and_then</code>, <code>map</code> and <code>unwrap_or</code> are my favorites.</li>
</ul>
</div>
<!-- Mobile navigation buttons -->
<a href="concurrency.html" class="mobile-nav-chapters previous">
<i class="fa fa-angle-left"></i>
</a>
<a href="choosing-your-guarantees.html" class="mobile-nav-chapters next">
<i class="fa fa-angle-right"></i>
</a>
</div>
<a href="concurrency.html" class="nav-chapters previous" title="You can navigate through the chapters using the arrow keys">
<i class="fa fa-angle-left"></i>
</a>
<a href="choosing-your-guarantees.html" class="nav-chapters next" title="You can navigate through the chapters using the arrow keys">
<i class="fa fa-angle-right"></i>
</a>
</div>
<!-- Local fallback for Font Awesome -->
<script>
if ($(".fa").css("font-family") !== "FontAwesome") {
$('<link rel="stylesheet" type="text/css" href="_FontAwesome/css/font-awesome.css">').prependTo('head');
}
</script>
<!-- Livereload script (if served using the cli tool) -->
<script src="highlight.js"></script>
<script src="book.js"></script>
</body>
</html>
distrib
>
Mageia
>
6
>
x86_64
>
media
>
core-updates
>
by-pkgid
>
878cdd00a13d17a73c6619a777ef5d74
>
files
>
138
