Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 95ce0d61ab7051847cc8d562d9e5301c8948d658 / . / docs / tutorial / LangImpl1.html
blob: ee097ff6bfd6fdc7ac4e7d321992212affcbbd90 [file] [log] [blame]
Chris Lattnerc38ef542007-10-22 04:32:37 +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>
Chris Lattner95ce0d62007-11-06 05:02:48 +0000[diff] [blame^]6 <title>Kaleidoscope: Tutorial Introduction and the Lexer</title>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]7 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
8 <meta name="author" content="Chris Lattner">
9 <link rel="stylesheet" href="../llvm.css" type="text/css">
10</head>
11
12<body>
13
Chris Lattner95ce0d62007-11-06 05:02:48 +0000[diff] [blame^]14<div class="doc_title">Kaleidoscope: Tutorial Introduction and the Lexer</div>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]15
Chris Lattner128eb862007-11-05 19:06:59 +0000[diff] [blame]16<ul>
Chris Lattner0e555b12007-11-05 20:04:56 +0000[diff] [blame]17<li><a href="index.html">Up to Tutorial Index</a></li>
Chris Lattner128eb862007-11-05 19:06:59 +0000[diff] [blame]18<li>Chapter 1
19 <ol>
20 <li><a href="#intro">Tutorial Introduction</a></li>
21 <li><a href="#language">The Basic Language</a></li>
22 <li><a href="#lexer">The Lexer</a></li>
23 </ol>
24</li>
Chris Lattner0e555b12007-11-05 20:04:56 +0000[diff] [blame]25<li><a href="LangImpl2.html">Chapter 2</a>: Implementing a Parser and AST</li>
Chris Lattner128eb862007-11-05 19:06:59 +0000[diff] [blame]26</ul>
27
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]28<div class="doc_author">
29 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a></p>
30</div>
31
32<!-- *********************************************************************** -->
33<div class="doc_section"><a name="intro">Tutorial Introduction</a></div>
34<!-- *********************************************************************** -->
35
36<div class="doc_text">
37
38<p>Welcome to the "Implementing a language with LLVM" tutorial. This tutorial
Chris Lattner95ce0d62007-11-06 05:02:48 +0000[diff] [blame^]39runs through the implementation of a simple language, showing how fun and
Chris Lattner619bc0a2007-11-05 20:13:56 +0000[diff] [blame]40easy it can be. This tutorial will get you up and started as well as help to
Chris Lattner95ce0d62007-11-06 05:02:48 +0000[diff] [blame^]41build a framework you can extend to other languages, allowing you to use this
42as a way to start playing with other LLVM specific things.
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]43</p>
44
Chris Lattner95ce0d62007-11-06 05:02:48 +0000[diff] [blame^]45<p>
46The goal of this tutorial is to progressively unveil our language, describing
47how it is built up over time. This will let us cover a fairly broad range of
48language design and LLVM-specific usage issues, showing and explaining the code
49for it all along the way, without overwhelming you with tons of details up
50front.</p>
51
52<p>It is useful to point out ahead of time that this tutorial is really about
53teaching compiler techniques and LLVM specifically, <em>not</em> about teaching
54modern and sane software engineering principles. In practice, this means that
55we'll take a number of shortcuts to simplify the exposition. For example, the
56code leaks memory, uses global variables all over the place, doesn't use nice
57design patterns like visitors, etc... but it is very simple. If you dig in and
58use the code as a basis for future projects, fixing these deficiencies shouldn't
59be hard.</p>
60
61<p>I've tried to put this tutorial together in a way that makes chapters easy to
62skip over if you are already familiar or are uninterested with various pieces.
63The structure of the tutorial is:
64</p>
65
66<ul>
67<li><b><a href="#language">Chapter #1</a>: Introduction to the Kaleidoscope
68language, and the definition of its Lexer</b> - This shows where we are going
69and the basic functionality that we want it to do. In order to make this
70tutorial maximally understandable and hackable, we choose to implement
71everything in C++ instead of using lexer and parser generators. LLVM obviously
72works just fine with such tools, feel free to use one if you prefer.</li>
73<li><b><a href="LangImpl2.html">Chapter #2</a>: Implementing a Parser and
74AST</b> - With the lexer in place, we can talk about parsing techniques and
75basic AST construction. This tutorial describes recursive descent parsing and
76operator precedence parsing. Nothing in Chapters 1 or 2 is LLVM-specific,
77the code doesn't even link in LLVM at this point. :)</li>
78<li><b><a href="LangImpl3.html">Chapter #3</a>: Code generation to LLVM IR</b> -
79With the AST ready, we can show off how easy generation of LLVM IR really
80is.</li>
81<li><b><a href="LangImpl4.html">Chapter #4</a>: Adding JIT and Optimizer
82Support</b> - Because a lot of people are interested in using LLVM as a JIT,
83we'll dive right into it and show you the 3 lines it takes to add JIT support.
84LLVM is also useful in many other ways, but this is one simple and "sexy" way
85that shows off its power. :)</li>
86<li><b><a href="LangImpl5.html">Chapter #5</a>: Extending the Language: Control
87Flow</b> - With the language up and running, we show how to extend it with
88control flow operations (if/then/else and a for loop). This gives us a chance
89to talk about simple SSA construction and control flow.</li>
90<li><b><a href="LangImpl6.html">Chapter #6</a>: Extending the Language:
91User-defined Operators</b> - This is a silly but fun chapter that talks about
92extending the language to let the user program define their own arbitrary
93unary and binary operators (with assignable precedence!). This lets us build a
94significant piece of the "language" as library routines.</li>
95<li><b><a href="LangImpl7.html">Chapter #7</a>: Extending the Language: Mutable
96Variables</b> - This chapter talks about adding user-defined local variables
97along with variable assignment operator. The interesting part about this is how
98easy and trivial it is to construct SSA form in LLVM (no, LLVM does <em>not</em>
99require your front-end to construct SSA form!).</li>
100<li><b><a href="LangImpl8.html">Chapter #8</a>: Conclusion and other useful LLVM
101tidbits</b> - This chapter wraps up the series by talking about potential
102ways to extend the language, but also includes a bunch of pointers to info about
103"special topics" like adding garbage collection support, exceptions, debugging,
104support for "spaghetti stacks", and a bunch of other tips and tricks.</li>
105
106</ul>
107
108<p>By the end of the tutorial, we'll have written about 700 lines of
109non-comment, non-blank lines of code. With this small amount of code, we'll
110have built up a very reasonable compiler for a non-trivial language including
111a hand-written lexer, parser, AST, as well as code generation support with a JIT
112compiler. While other systems may have interesting "hello world" tutorials,
113I think the breadth of this tutorial is a great testament to the strengths of
114LLVM and why you should consider it if you're interested in language or compiler
115design.</p>
116
117<p>A note about this tutorial: we expect you to extend the language and play
118with it on your own. Take the code and go crazy hacking away at it. It can be
119a lot of fun to play with languages! In any case, lets get into the code!</p>
120
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]121</div>
122
123<!-- *********************************************************************** -->
Chris Lattner128eb862007-11-05 19:06:59 +0000[diff] [blame]124<div class="doc_section"><a name="language">The Basic Language</a></div>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]125<!-- *********************************************************************** -->
126
127<div class="doc_text">
128
129<p>This tutorial will be illustrated with a toy language that we'll call
130"<a href="http://en.wikipedia.org/wiki/Kaleidoscope">Kaleidoscope</a>".
131Kaleidoscope is a procedural language that allows you to define functions, use
132conditionals, math, etc. Over the course of the tutorial, we'll extend
Chris Lattner619bc0a2007-11-05 20:13:56 +0000[diff] [blame]133Kaleidoscope to support the if/then/else construct, a for loop, user defined
134operators, JIT compilation with a simple command line interface, etc.</p>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]135
Chris Lattner619bc0a2007-11-05 20:13:56 +0000[diff] [blame]136<p>Because we want to keep things simple, the only datatype in Kaleidoscope is a
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]13764-bit floating point type (aka 'double' in C parlance). As such, all values
138are implicitly double precision and the language doesn't require type
139declarations. This gives the language a very nice and simple syntax. For
Chris Lattner619bc0a2007-11-05 20:13:56 +0000[diff] [blame]140example, the following simple example computes <a
141href="http://en.wikipedia.org/wiki/Fibonacci_number">Fibonacci numbers:</a></p>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]142
143<div class="doc_code">
144<pre>
145# Compute the x'th fibonacci number.
146def fib(x)
Chris Lattnere6c91042007-10-22 06:34:15 +0000[diff] [blame]147 if x &lt; 3 then
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]148 1
149 else
150 fib(x-1)+fib(x-2)
151
152# This expression will compute the 40th number.
153fib(40)
154</pre>
155</div>
156
Duncan Sands72261ff2007-11-05 16:04:58 +0000[diff] [blame]157<p>We also allow Kaleidoscope to call into standard library functions (the LLVM
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]158JIT makes this completely trivial). This means that you can use the 'extern'
159keyword to define a function before you use it (this is also useful for mutually
160recursive functions). For example:</p>
161
162<div class="doc_code">
163<pre>
164extern sin(arg);
165extern cos(arg);
166extern atan2(arg1 arg2);
167
168atan2(sin(.4), cos(42))
169</pre>
170</div>
171
Chris Lattner95ce0d62007-11-06 05:02:48 +0000[diff] [blame^]172<p>A more interesting example is included in Chapter 6 where we show the code
173used to <a href="LangImpl6.html#example">implement a Mandelbrot Set viewer</a>
174in Kaleidoscope.</p>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]175
176</div>
177
178<!-- *********************************************************************** -->
Chris Lattner128eb862007-11-05 19:06:59 +0000[diff] [blame]179<div class="doc_section"><a name="lexer">The Lexer</a></div>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]180<!-- *********************************************************************** -->
181
182<div class="doc_text">
183
184<p>When it comes to implementing a language, the first thing needed is
185the ability to process a text file and recognize what it says. The traditional
186way to do this is to use a "<a
187href="http://en.wikipedia.org/wiki/Lexical_analysis">lexer</a>" (aka 'scanner')
188to break the input up into "tokens". Each token returned by the lexer includes
189a token code and potentially some metadata (e.g. the numeric value of a number).
190First, we define the possibilities:
191</p>
192
193<div class="doc_code">
194<pre>
195// The lexer returns tokens [0-255] if it is an unknown character, otherwise one
196// of these for known things.
197enum Token {
198 tok_eof = -1,
199
200 // commands
201 tok_def = -2, tok_extern = -3,
202
203 // primary
204 tok_identifier = -4, tok_number = -5,
205};
206
207static std::string IdentifierStr; // Filled in if tok_identifier
208static double NumVal; // Filled in if tok_number
209</pre>
210</div>
211
212<p>Each token returned by our lexer will either be one of the Token enum values
Chris Lattner619bc0a2007-11-05 20:13:56 +0000[diff] [blame]213or it will be an 'unknown' character like '+', which is returned as its ascii
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]214value. If the current token is an identifier, the <tt>IdentifierStr</tt>
215global variable holds the name of the identifier. If the current token is a
216numeric literal (like 1.0), <tt>NumVal</tt> holds its value. Note that we use
217global variables for simplicity, this is not the best choice for a real language
218implementation :).
219</p>
220
Chris Lattner619bc0a2007-11-05 20:13:56 +0000[diff] [blame]221<p>The actual implementation of the lexer is a single function named
222<tt>gettok</tt>. The <tt>gettok</tt> function is called to return the next token
223from standard input. Its definition starts as:</p>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]224
225<div class="doc_code">
226<pre>
227/// gettok - Return the next token from standard input.
228static int gettok() {
229 static int LastChar = ' ';
230
231 // Skip any whitespace.
232 while (isspace(LastChar))
233 LastChar = getchar();
234</pre>
235</div>
236
237<p>
238<tt>gettok</tt> works by calling the C <tt>getchar()</tt> function to read
239characters one at a time from standard input. It eats them as it recognizes
Chris Lattner619bc0a2007-11-05 20:13:56 +0000[diff] [blame]240them and stores the last character read, but not processed, in LastChar. The
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]241first thing that it has to do is ignore whitespace between tokens. This is
242accomplished with the loop above.</p>
243
Chris Lattner619bc0a2007-11-05 20:13:56 +0000[diff] [blame]244<p>The next thing <tt>gettok</tt> needs to do is recognize identifiers and
245specific keywords like "def". Kaleidoscope does this with this simple loop:</p>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]246
247<div class="doc_code">
248<pre>
249 if (isalpha(LastChar)) { // identifier: [a-zA-Z][a-zA-Z0-9]*
250 IdentifierStr = LastChar;
251 while (isalnum((LastChar = getchar())))
252 IdentifierStr += LastChar;
253
254 if (IdentifierStr == "def") return tok_def;
255 if (IdentifierStr == "extern") return tok_extern;
256 return tok_identifier;
257 }
258</pre>
259</div>
260
Chris Lattner619bc0a2007-11-05 20:13:56 +0000[diff] [blame]261<p>Note that this code sets the '<tt>IdentifierStr</tt>' global whenever it
262lexes an identifier. Also, since language keywords are matched by the same
263loop, we handle them here inline. Numeric values are similar:</p>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]264
265<div class="doc_code">
266<pre>
267 if (isdigit(LastChar) || LastChar == '.') { // Number: [0-9.]+
268 std::string NumStr;
269 do {
270 NumStr += LastChar;
271 LastChar = getchar();
272 } while (isdigit(LastChar) || LastChar == '.');
273
274 NumVal = strtod(NumStr.c_str(), 0);
275 return tok_number;
276 }
277</pre>
278</div>
279
280<p>This is all pretty straight-forward code for processing input. When reading
281a numeric value from input, we use the C <tt>strtod</tt> function to convert it
282to a numeric value that we store in <tt>NumVal</tt>. Note that this isn't doing
Duncan Sands72261ff2007-11-05 16:04:58 +0000[diff] [blame]283sufficient error checking: it will incorrectly read "1.23.45.67" and handle it as
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]284if you typed in "1.23". Feel free to extend it :). Next we handle comments:
285</p>
286
287<div class="doc_code">
288<pre>
289 if (LastChar == '#') {
290 // Comment until end of line.
291 do LastChar = getchar();
292 while (LastChar != EOF &amp;&amp; LastChar != '\n' &amp; LastChar != '\r');
293
294 if (LastChar != EOF)
295 return gettok();
296 }
297</pre>
298</div>
299
Chris Lattner71155212007-11-06 01:39:12 +0000[diff] [blame]300<p>We handle comments by skipping to the end of the line and then return the
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]301next comment. Finally, if the input doesn't match one of the above cases, it is
Duncan Sands72261ff2007-11-05 16:04:58 +0000[diff] [blame]302either an operator character like '+' or the end of the file. These are handled with
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]303this code:</p>
304
305<div class="doc_code">
306<pre>
307 // Check for end of file. Don't eat the EOF.
308 if (LastChar == EOF)
309 return tok_eof;
310
311 // Otherwise, just return the character as its ascii value.
312 int ThisChar = LastChar;
313 LastChar = getchar();
314 return ThisChar;
315}
316</pre>
317</div>
318
Chris Lattner619bc0a2007-11-05 20:13:56 +0000[diff] [blame]319<p>With this, we have the complete lexer for the basic Kaleidoscope language
320(the <a href="LangImpl2.html#code">full code listing</a> for the Lexer is
321available in the <a href="LangImpl2.html">next chapter</a> of the tutorial).
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]322Next we'll <a href="LangImpl2.html">build a simple parser that uses this to
Chris Lattnere6c91042007-10-22 06:34:15 +0000[diff] [blame]323build an Abstract Syntax Tree</a>. When we have that, we'll include a driver
324so that you can use the lexer and parser together.
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]325</p>
326
327</div>
328
329<!-- *********************************************************************** -->
330<hr>
331<address>
332 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
333 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
334 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner8eef4b22007-10-23 06:30:50 +0000[diff] [blame]335 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Chris Lattnerc38ef542007-10-22 04:32:37 +0000[diff] [blame]336
337 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
338 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
339 Last modified: $Date: 2007-10-17 11:05:13 -0700 (Wed, 17 Oct 2007) $
340</address>
341</body>
342</html>