1 =================================================
2 Kaleidoscope: Tutorial Introduction and the Lexer
3 =================================================
8 Written by `Chris Lattner <mailto:sabre@nondot.org>`_
13 Welcome to the "Implementing a language with LLVM" tutorial. This
14 tutorial runs through the implementation of a simple language, showing
15 how fun and easy it can be. This tutorial will get you up and started as
16 well as help to build a framework you can extend to other languages. The
17 code in this tutorial can also be used as a playground to hack on other
20 The goal of this tutorial is to progressively unveil our language,
21 describing how it is built up over time. This will let us cover a fairly
22 broad range of language design and LLVM-specific usage issues, showing
23 and explaining the code for it all along the way, without overwhelming
24 you with tons of details up front.
26 It is useful to point out ahead of time that this tutorial is really
27 about teaching compiler techniques and LLVM specifically, *not* about
28 teaching modern and sane software engineering principles. In practice,
29 this means that we'll take a number of shortcuts to simplify the
30 exposition. For example, the code leaks memory, uses global variables
31 all over the place, doesn't use nice design patterns like
32 `visitors <http://en.wikipedia.org/wiki/Visitor_pattern>`_, etc... but
33 it is very simple. If you dig in and use the code as a basis for future
34 projects, fixing these deficiencies shouldn't be hard.
36 I've tried to put this tutorial together in a way that makes chapters
37 easy to skip over if you are already familiar with or are uninterested
38 in the various pieces. The structure of the tutorial is:
40 - `Chapter #1 <#language>`_: Introduction to the Kaleidoscope
41 language, and the definition of its Lexer - This shows where we are
42 going and the basic functionality that we want it to do. In order to
43 make this tutorial maximally understandable and hackable, we choose
44 to implement everything in C++ instead of using lexer and parser
45 generators. LLVM obviously works just fine with such tools, feel free
46 to use one if you prefer.
47 - `Chapter #2 <LangImpl2.html>`_: Implementing a Parser and AST -
48 With the lexer in place, we can talk about parsing techniques and
49 basic AST construction. This tutorial describes recursive descent
50 parsing and operator precedence parsing. Nothing in Chapters 1 or 2
51 is LLVM-specific, the code doesn't even link in LLVM at this point.
53 - `Chapter #3 <LangImpl3.html>`_: Code generation to LLVM IR - With
54 the AST ready, we can show off how easy generation of LLVM IR really
56 - `Chapter #4 <LangImpl4.html>`_: Adding JIT and Optimizer Support
57 - Because a lot of people are interested in using LLVM as a JIT,
58 we'll dive right into it and show you the 3 lines it takes to add JIT
59 support. LLVM is also useful in many other ways, but this is one
60 simple and "sexy" way to shows off its power. :)
61 - `Chapter #5 <LangImpl5.html>`_: Extending the Language: Control
62 Flow - With the language up and running, we show how to extend it
63 with control flow operations (if/then/else and a 'for' loop). This
64 gives us a chance to talk about simple SSA construction and control
66 - `Chapter #6 <LangImpl6.html>`_: Extending the Language:
67 User-defined Operators - This is a silly but fun chapter that talks
68 about extending the language to let the user program define their own
69 arbitrary unary and binary operators (with assignable precedence!).
70 This lets us build a significant piece of the "language" as library
72 - `Chapter #7 <LangImpl7.html>`_: Extending the Language: Mutable
73 Variables - This chapter talks about adding user-defined local
74 variables along with an assignment operator. The interesting part
75 about this is how easy and trivial it is to construct SSA form in
76 LLVM: no, LLVM does *not* require your front-end to construct SSA
78 - `Chapter #8 <LangImpl8.html>`_: Conclusion and other useful LLVM
79 tidbits - This chapter wraps up the series by talking about
80 potential ways to extend the language, but also includes a bunch of
81 pointers to info about "special topics" like adding garbage
82 collection support, exceptions, debugging, support for "spaghetti
83 stacks", and a bunch of other tips and tricks.
85 By the end of the tutorial, we'll have written a bit less than 700 lines
86 of non-comment, non-blank, lines of code. With this small amount of
87 code, we'll have built up a very reasonable compiler for a non-trivial
88 language including a hand-written lexer, parser, AST, as well as code
89 generation support with a JIT compiler. While other systems may have
90 interesting "hello world" tutorials, I think the breadth of this
91 tutorial is a great testament to the strengths of LLVM and why you
92 should consider it if you're interested in language or compiler design.
94 A note about this tutorial: we expect you to extend the language and
95 play with it on your own. Take the code and go crazy hacking away at it,
96 compilers don't need to be scary creatures - it can be a lot of fun to
102 This tutorial will be illustrated with a toy language that we'll call
103 "`Kaleidoscope <http://en.wikipedia.org/wiki/Kaleidoscope>`_" (derived
104 from "meaning beautiful, form, and view"). Kaleidoscope is a procedural
105 language that allows you to define functions, use conditionals, math,
106 etc. Over the course of the tutorial, we'll extend Kaleidoscope to
107 support the if/then/else construct, a for loop, user defined operators,
108 JIT compilation with a simple command line interface, etc.
110 Because we want to keep things simple, the only datatype in Kaleidoscope
111 is a 64-bit floating point type (aka 'double' in C parlance). As such,
112 all values are implicitly double precision and the language doesn't
113 require type declarations. This gives the language a very nice and
114 simple syntax. For example, the following simple example computes
115 `Fibonacci numbers: <http://en.wikipedia.org/wiki/Fibonacci_number>`_
119 # Compute the x'th fibonacci number.
126 # This expression will compute the 40th number.
129 We also allow Kaleidoscope to call into standard library functions (the
130 LLVM JIT makes this completely trivial). This means that you can use the
131 'extern' keyword to define a function before you use it (this is also
132 useful for mutually recursive functions). For example:
138 extern atan2(arg1 arg2);
140 atan2(sin(.4), cos(42))
142 A more interesting example is included in Chapter 6 where we write a
143 little Kaleidoscope application that `displays a Mandelbrot
144 Set <LangImpl6.html#example>`_ at various levels of magnification.
146 Lets dive into the implementation of this language!
151 When it comes to implementing a language, the first thing needed is the
152 ability to process a text file and recognize what it says. The
153 traditional way to do this is to use a
154 "`lexer <http://en.wikipedia.org/wiki/Lexical_analysis>`_" (aka
155 'scanner') to break the input up into "tokens". Each token returned by
156 the lexer includes a token code and potentially some metadata (e.g. the
157 numeric value of a number). First, we define the possibilities:
161 // The lexer returns tokens [0-255] if it is an unknown character, otherwise one
162 // of these for known things.
167 tok_def = -2, tok_extern = -3,
170 tok_identifier = -4, tok_number = -5,
173 static std::string IdentifierStr; // Filled in if tok_identifier
174 static double NumVal; // Filled in if tok_number
176 Each token returned by our lexer will either be one of the Token enum
177 values or it will be an 'unknown' character like '+', which is returned
178 as its ASCII value. If the current token is an identifier, the
179 ``IdentifierStr`` global variable holds the name of the identifier. If
180 the current token is a numeric literal (like 1.0), ``NumVal`` holds its
181 value. Note that we use global variables for simplicity, this is not the
182 best choice for a real language implementation :).
184 The actual implementation of the lexer is a single function named
185 ``gettok``. The ``gettok`` function is called to return the next token
186 from standard input. Its definition starts as:
190 /// gettok - Return the next token from standard input.
191 static int gettok() {
192 static int LastChar = ' ';
194 // Skip any whitespace.
195 while (isspace(LastChar))
196 LastChar = getchar();
198 ``gettok`` works by calling the C ``getchar()`` function to read
199 characters one at a time from standard input. It eats them as it
200 recognizes them and stores the last character read, but not processed,
201 in LastChar. The first thing that it has to do is ignore whitespace
202 between tokens. This is accomplished with the loop above.
204 The next thing ``gettok`` needs to do is recognize identifiers and
205 specific keywords like "def". Kaleidoscope does this with this simple
210 if (isalpha(LastChar)) { // identifier: [a-zA-Z][a-zA-Z0-9]*
211 IdentifierStr = LastChar;
212 while (isalnum((LastChar = getchar())))
213 IdentifierStr += LastChar;
215 if (IdentifierStr == "def") return tok_def;
216 if (IdentifierStr == "extern") return tok_extern;
217 return tok_identifier;
220 Note that this code sets the '``IdentifierStr``' global whenever it
221 lexes an identifier. Also, since language keywords are matched by the
222 same loop, we handle them here inline. Numeric values are similar:
226 if (isdigit(LastChar) || LastChar == '.') { // Number: [0-9.]+
230 LastChar = getchar();
231 } while (isdigit(LastChar) || LastChar == '.');
233 NumVal = strtod(NumStr.c_str(), 0);
237 This is all pretty straight-forward code for processing input. When
238 reading a numeric value from input, we use the C ``strtod`` function to
239 convert it to a numeric value that we store in ``NumVal``. Note that
240 this isn't doing sufficient error checking: it will incorrectly read
241 "1.23.45.67" and handle it as if you typed in "1.23". Feel free to
242 extend it :). Next we handle comments:
246 if (LastChar == '#') {
247 // Comment until end of line.
248 do LastChar = getchar();
249 while (LastChar != EOF && LastChar != '\n' && LastChar != '\r');
255 We handle comments by skipping to the end of the line and then return
256 the next token. Finally, if the input doesn't match one of the above
257 cases, it is either an operator character like '+' or the end of the
258 file. These are handled with this code:
262 // Check for end of file. Don't eat the EOF.
266 // Otherwise, just return the character as its ascii value.
267 int ThisChar = LastChar;
268 LastChar = getchar();
272 With this, we have the complete lexer for the basic Kaleidoscope
273 language (the `full code listing <LangImpl2.html#code>`_ for the Lexer
274 is available in the `next chapter <LangImpl2.html>`_ of the tutorial).
275 Next we'll `build a simple parser that uses this to build an Abstract
276 Syntax Tree <LangImpl2.html>`_. When we have that, we'll include a
277 driver so that you can use the lexer and parser together.
279 `Next: Implementing a Parser and AST <LangImpl2.html>`_