Gitiles
Code Review Sign In
gerrit-public.fairphone.software / toolchain / llvm-project / 46df6a252a8f8f36381a50a548264f453233d122 / . / llvm / docs / tutorial / OCamlLangImpl1.html
blob: aa2bd8760043f701417b2311c60873d882e1c115 [file] [log] [blame]
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3
4<html>
5<head>
6 <title>Kaleidoscope: Tutorial Introduction and the Lexer</title>
7 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
8 <meta name="author" content="Chris Lattner">
9 <meta name="author" content="Erick Tryzelaar">
10 <link rel="stylesheet" href="../llvm.css" type="text/css">
11</head>
12
13<body>
14
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]15<h1>Kaleidoscope: Tutorial Introduction and the Lexer</h1>
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]16
17<ul>
18<li><a href="index.html">Up to Tutorial Index</a></li>
19<li>Chapter 1
20 <ol>
21 <li><a href="#intro">Tutorial Introduction</a></li>
22 <li><a href="#language">The Basic Language</a></li>
23 <li><a href="#lexer">The Lexer</a></li>
24 </ol>
25</li>
26<li><a href="OCamlLangImpl2.html">Chapter 2</a>: Implementing a Parser and
27AST</li>
28</ul>
29
30<div class="doc_author">
31 <p>
32 Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
33 and <a href="mailto:idadesub@users.sourceforge.net">Erick Tryzelaar</a>
34 </p>
35</div>
36
37<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]38<h2><a name="intro">Tutorial Introduction</a></h2>
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]39<!-- *********************************************************************** -->
40
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]41<div>
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]42
43<p>Welcome to the "Implementing a language with LLVM" tutorial. This tutorial
44runs through the implementation of a simple language, showing how fun and
45easy it can be. This tutorial will get you up and started as well as help to
46build a framework you can extend to other languages. The code in this tutorial
47can also be used as a playground to hack on other LLVM specific things.
48</p>
49
50<p>
51The goal of this tutorial is to progressively unveil our language, describing
52how it is built up over time. This will let us cover a fairly broad range of
53language design and LLVM-specific usage issues, showing and explaining the code
54for it all along the way, without overwhelming you with tons of details up
55front.</p>
56
57<p>It is useful to point out ahead of time that this tutorial is really about
58teaching compiler techniques and LLVM specifically, <em>not</em> about teaching
59modern and sane software engineering principles. In practice, this means that
60we'll take a number of shortcuts to simplify the exposition. For example, the
61code leaks memory, uses global variables all over the place, doesn't use nice
62design patterns like <a
63href="http://en.wikipedia.org/wiki/Visitor_pattern">visitors</a>, etc... but it
64is very simple. If you dig in and use the code as a basis for future projects,
65fixing these deficiencies shouldn't be hard.</p>
66
67<p>I've tried to put this tutorial together in a way that makes chapters easy to
68skip over if you are already familiar with or are uninterested in the various
69pieces. The structure of the tutorial is:
70</p>
71
72<ul>
73<li><b><a href="#language">Chapter #1</a>: Introduction to the Kaleidoscope
74language, and the definition of its Lexer</b> - This shows where we are going
75and the basic functionality that we want it to do. In order to make this
76tutorial maximally understandable and hackable, we choose to implement
77everything in Objective Caml instead of using lexer and parser generators.
78LLVM obviously works just fine with such tools, feel free to use one if you
79prefer.</li>
80<li><b><a href="OCamlLangImpl2.html">Chapter #2</a>: Implementing a Parser and
81AST</b> - With the lexer in place, we can talk about parsing techniques and
82basic AST construction. This tutorial describes recursive descent parsing and
83operator precedence parsing. Nothing in Chapters 1 or 2 is LLVM-specific,
84the code doesn't even link in LLVM at this point. :)</li>
85<li><b><a href="OCamlLangImpl3.html">Chapter #3</a>: Code generation to LLVM
86IR</b> - With the AST ready, we can show off how easy generation of LLVM IR
87really is.</li>
88<li><b><a href="OCamlLangImpl4.html">Chapter #4</a>: Adding JIT and Optimizer
89Support</b> - Because a lot of people are interested in using LLVM as a JIT,
90we'll dive right into it and show you the 3 lines it takes to add JIT support.
91LLVM is also useful in many other ways, but this is one simple and "sexy" way
92to shows off its power. :)</li>
93<li><b><a href="OCamlLangImpl5.html">Chapter #5</a>: Extending the Language:
94Control Flow</b> - With the language up and running, we show how to extend it
95with control flow operations (if/then/else and a 'for' loop). This gives us a
96chance to talk about simple SSA construction and control flow.</li>
97<li><b><a href="OCamlLangImpl6.html">Chapter #6</a>: Extending the Language:
98User-defined Operators</b> - This is a silly but fun chapter that talks about
99extending the language to let the user program define their own arbitrary
100unary and binary operators (with assignable precedence!). This lets us build a
101significant piece of the "language" as library routines.</li>
102<li><b><a href="OCamlLangImpl7.html">Chapter #7</a>: Extending the Language:
103Mutable Variables</b> - This chapter talks about adding user-defined local
104variables along with an assignment operator. The interesting part about this
105is how easy and trivial it is to construct SSA form in LLVM: no, LLVM does
106<em>not</em> require your front-end to construct SSA form!</li>
107<li><b><a href="OCamlLangImpl8.html">Chapter #8</a>: Conclusion and other
108useful LLVM tidbits</b> - This chapter wraps up the series by talking about
109potential ways to extend the language, but also includes a bunch of pointers to
110info about "special topics" like adding garbage collection support, exceptions,
111debugging, support for "spaghetti stacks", and a bunch of other tips and
112tricks.</li>
113
114</ul>
115
116<p>By the end of the tutorial, we'll have written a bit less than 700 lines of
117non-comment, non-blank, lines of code. With this small amount of code, we'll
118have built up a very reasonable compiler for a non-trivial language including
119a hand-written lexer, parser, AST, as well as code generation support with a JIT
120compiler. While other systems may have interesting "hello world" tutorials,
121I think the breadth of this tutorial is a great testament to the strengths of
122LLVM and why you should consider it if you're interested in language or compiler
123design.</p>
124
125<p>A note about this tutorial: we expect you to extend the language and play
126with it on your own. Take the code and go crazy hacking away at it, compilers
127don't need to be scary creatures - it can be a lot of fun to play with
128languages!</p>
129
130</div>
131
132<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]133<h2><a name="language">The Basic Language</a></h2>
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]134<!-- *********************************************************************** -->
135
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]136<div>
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]137
138<p>This tutorial will be illustrated with a toy language that we'll call
139"<a href="http://en.wikipedia.org/wiki/Kaleidoscope">Kaleidoscope</a>" (derived
140from "meaning beautiful, form, and view").
141Kaleidoscope is a procedural language that allows you to define functions, use
142conditionals, math, etc. Over the course of the tutorial, we'll extend
143Kaleidoscope to support the if/then/else construct, a for loop, user defined
144operators, JIT compilation with a simple command line interface, etc.</p>
145
146<p>Because we want to keep things simple, the only datatype in Kaleidoscope is a
14764-bit floating point type (aka 'float' in O'Caml parlance). As such, all
148values are implicitly double precision and the language doesn't require type
149declarations. This gives the language a very nice and simple syntax. For
150example, the following simple example computes <a
151href="http://en.wikipedia.org/wiki/Fibonacci_number">Fibonacci numbers:</a></p>
152
153<div class="doc_code">
154<pre>
155# Compute the x'th fibonacci number.
156def fib(x)
157 if x &lt; 3 then
158 1
159 else
160 fib(x-1)+fib(x-2)
161
162# This expression will compute the 40th number.
163fib(40)
164</pre>
165</div>
166
167<p>We also allow Kaleidoscope to call into standard library functions (the LLVM
168JIT makes this completely trivial). This means that you can use the 'extern'
169keyword to define a function before you use it (this is also useful for mutually
170recursive functions). For example:</p>
171
172<div class="doc_code">
173<pre>
174extern sin(arg);
175extern cos(arg);
176extern atan2(arg1 arg2);
177
178atan2(sin(.4), cos(42))
179</pre>
180</div>
181
182<p>A more interesting example is included in Chapter 6 where we write a little
183Kaleidoscope application that <a href="OCamlLangImpl6.html#example">displays
184a Mandelbrot Set</a> at various levels of magnification.</p>
185
186<p>Lets dive into the implementation of this language!</p>
187
188</div>
189
190<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]191<h2><a name="lexer">The Lexer</a></h2>
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]192<!-- *********************************************************************** -->
193
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]194<div>
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]195
196<p>When it comes to implementing a language, the first thing needed is
197the ability to process a text file and recognize what it says. The traditional
198way to do this is to use a "<a
199href="http://en.wikipedia.org/wiki/Lexical_analysis">lexer</a>" (aka 'scanner')
200to break the input up into "tokens". Each token returned by the lexer includes
201a token code and potentially some metadata (e.g. the numeric value of a number).
202First, we define the possibilities:
203</p>
204
205<div class="doc_code">
206<pre>
207(* The lexer returns these 'Kwd' if it is an unknown character, otherwise one of
208 * these others for known things. *)
209type token =
210 (* commands *)
211 | Def | Extern
212
213 (* primary *)
214 | Ident of string | Number of float
215
216 (* unknown *)
217 | Kwd of char
218</pre>
219</div>
220
221<p>Each token returned by our lexer will be one of the token variant values.
Erick Tryzelaara2eba032008-03-30 19:14:31 +0000[diff] [blame]222An unknown character like '+' will be returned as <tt>Token.Kwd '+'</tt>. If
223the curr token is an identifier, the value will be <tt>Token.Ident s</tt>. If
224the current token is a numeric literal (like 1.0), the value will be
225<tt>Token.Number 1.0</tt>.
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]226</p>
227
228<p>The actual implementation of the lexer is a collection of functions driven
Erick Tryzelaara2eba032008-03-30 19:14:31 +0000[diff] [blame]229by a function named <tt>Lexer.lex</tt>. The <tt>Lexer.lex</tt> function is
230called to return the next token from standard input. We will use
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]231<a href="http://caml.inria.fr/pub/docs/manual-camlp4/index.html">Camlp4</a>
232to simplify the tokenization of the standard input. Its definition starts
233as:</p>
234
235<div class="doc_code">
236<pre>
237(*===----------------------------------------------------------------------===
238 * Lexer
239 *===----------------------------------------------------------------------===*)
240
241let rec lex = parser
242 (* Skip any whitespace. *)
243 | [&lt; ' (' ' | '\n' | '\r' | '\t'); stream &gt;] -&gt; lex stream
244</pre>
245</div>
246
247<p>
Erick Tryzelaara2eba032008-03-30 19:14:31 +0000[diff] [blame]248<tt>Lexer.lex</tt> works by recursing over a <tt>char Stream.t</tt> to read
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]249characters one at a time from the standard input. It eats them as it recognizes
Erick Tryzelaara2eba032008-03-30 19:14:31 +0000[diff] [blame]250them and stores them in in a <tt>Token.token</tt> variant. The first thing that
251it has to do is ignore whitespace between tokens. This is accomplished with the
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]252recursive call above.</p>
253
Erick Tryzelaara2eba032008-03-30 19:14:31 +0000[diff] [blame]254<p>The next thing <tt>Lexer.lex</tt> needs to do is recognize identifiers and
Torok Edwin49c28e42008-10-29 16:32:06 +0000[diff] [blame]255specific keywords like "def". Kaleidoscope does this with a pattern match
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]256and a helper function.<p>
257
258<div class="doc_code">
259<pre>
260 (* identifier: [a-zA-Z][a-zA-Z0-9] *)
261 | [&lt; ' ('A' .. 'Z' | 'a' .. 'z' as c); stream &gt;] -&gt;
262 let buffer = Buffer.create 1 in
263 Buffer.add_char buffer c;
264 lex_ident buffer stream
265
266...
267
268and lex_ident buffer = parser
269 | [&lt; ' ('A' .. 'Z' | 'a' .. 'z' | '0' .. '9' as c); stream &gt;] -&gt;
270 Buffer.add_char buffer c;
271 lex_ident buffer stream
272 | [&lt; stream=lex &gt;] -&gt;
273 match Buffer.contents buffer with
274 | "def" -&gt; [&lt; 'Token.Def; stream &gt;]
275 | "extern" -&gt; [&lt; 'Token.Extern; stream &gt;]
276 | id -&gt; [&lt; 'Token.Ident id; stream &gt;]
277</pre>
278</div>
279
Gabor Greif605c7ca2009-03-11 20:04:08 +0000[diff] [blame]280<p>Numeric values are similar:</p>
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]281
282<div class="doc_code">
283<pre>
284 (* number: [0-9.]+ *)
285 | [&lt; ' ('0' .. '9' as c); stream &gt;] -&gt;
286 let buffer = Buffer.create 1 in
287 Buffer.add_char buffer c;
288 lex_number buffer stream
289
290...
291
292and lex_number buffer = parser
293 | [&lt; ' ('0' .. '9' | '.' as c); stream &gt;] -&gt;
294 Buffer.add_char buffer c;
295 lex_number buffer stream
296 | [&lt; stream=lex &gt;] -&gt;
297 [&lt; 'Token.Number (float_of_string (Buffer.contents buffer)); stream &gt;]
298</pre>
299</div>
300
301<p>This is all pretty straight-forward code for processing input. When reading
302a numeric value from input, we use the ocaml <tt>float_of_string</tt> function
Erick Tryzelaara2eba032008-03-30 19:14:31 +0000[diff] [blame]303to convert it to a numeric value that we store in <tt>Token.Number</tt>. Note
304that this isn't doing sufficient error checking: it will raise <tt>Failure</tt>
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]305if the string "1.23.45.67". Feel free to extend it :). Next we handle
306comments:
307</p>
308
309<div class="doc_code">
310<pre>
311 (* Comment until end of line. *)
312 | [&lt; ' ('#'); stream &gt;] -&gt;
313 lex_comment stream
314
315...
316
317and lex_comment = parser
318 | [&lt; ' ('\n'); stream=lex &gt;] -&gt; stream
319 | [&lt; 'c; e=lex_comment &gt;] -&gt; e
320 | [&lt; &gt;] -&gt; [&lt; &gt;]
321</pre>
322</div>
323
324<p>We handle comments by skipping to the end of the line and then return the
325next token. Finally, if the input doesn't match one of the above cases, it is
326either an operator character like '+' or the end of the file. These are handled
327with this code:</p>
328
329<div class="doc_code">
330<pre>
331 (* Otherwise, just return the character as its ascii value. *)
332 | [&lt; 'c; stream &gt;] -&gt;
333 [&lt; 'Token.Kwd c; lex stream &gt;]
334
335 (* end of stream. *)
336 | [&lt; &gt;] -&gt; [&lt; &gt;]
337</pre>
338</div>
339
340<p>With this, we have the complete lexer for the basic Kaleidoscope language
341(the <a href="OCamlLangImpl2.html#code">full code listing</a> for the Lexer is
342available in the <a href="OCamlLangImpl2.html">next chapter</a> of the
343tutorial). Next we'll <a href="OCamlLangImpl2.html">build a simple parser that
344uses this to build an Abstract Syntax Tree</a>. When we have that, we'll
345include a driver so that you can use the lexer and parser together.
346</p>
347
348<a href="OCamlLangImpl2.html">Next: Implementing a Parser and AST</a>
349</div>
350
351<!-- *********************************************************************** -->
352<hr>
353<address>
354 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
355 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
356 <a href="http://validator.w3.org/check/referer"><img
357 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
358
359 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
360 <a href="mailto:idadesub@users.sourceforge.net">Erick Tryzelaar</a><br>
NAKAMURA Takumica46f5a2011-04-09 02:13:37 +0000[diff] [blame]361 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
Dan Gohmanb44d2ab2010-02-03 17:27:31 +0000[diff] [blame]362 Last modified: $Date$
Erick Tryzelaara5e13fa2008-03-27 08:18:07 +0000[diff] [blame]363</address>
364</body>
365</html>