1 =================================================
2 Kaleidoscope: Tutorial Introduction and the Lexer
3 =================================================
11 Welcome to the "Implementing a language with LLVM" tutorial. This
12 tutorial runs through the implementation of a simple language, showing
13 how fun and easy it can be. This tutorial will get you up and started as
14 well as help to build a framework you can extend to other languages. The
15 code in this tutorial can also be used as a playground to hack on other
18 The goal of this tutorial is to progressively unveil our language,
19 describing how it is built up over time. This will let us cover a fairly
20 broad range of language design and LLVM-specific usage issues, showing
21 and explaining the code for it all along the way, without overwhelming
22 you with tons of details up front.
24 It is useful to point out ahead of time that this tutorial is really
25 about teaching compiler techniques and LLVM specifically, *not* about
26 teaching modern and sane software engineering principles. In practice,
27 this means that we'll take a number of shortcuts to simplify the
28 exposition. For example, the code uses global variables
29 all over the place, doesn't use nice design patterns like
30 `visitors <http://en.wikipedia.org/wiki/Visitor_pattern>`_, etc... but
31 it is very simple. If you dig in and use the code as a basis for future
32 projects, fixing these deficiencies shouldn't be hard.
34 I've tried to put this tutorial together in a way that makes chapters
35 easy to skip over if you are already familiar with or are uninterested
36 in the various pieces. The structure of the tutorial is:
38 - `Chapter #1 <#language>`_: Introduction to the Kaleidoscope
39 language, and the definition of its Lexer - This shows where we are
40 going and the basic functionality that we want it to do. In order to
41 make this tutorial maximally understandable and hackable, we choose
42 to implement everything in C++ instead of using lexer and parser
43 generators. LLVM obviously works just fine with such tools, feel free
44 to use one if you prefer.
45 - `Chapter #2 <LangImpl02.html>`_: Implementing a Parser and AST -
46 With the lexer in place, we can talk about parsing techniques and
47 basic AST construction. This tutorial describes recursive descent
48 parsing and operator precedence parsing. Nothing in Chapters 1 or 2
49 is LLVM-specific, the code doesn't even link in LLVM at this point.
51 - `Chapter #3 <LangImpl03.html>`_: Code generation to LLVM IR - With
52 the AST ready, we can show off how easy generation of LLVM IR really
54 - `Chapter #4 <LangImpl04.html>`_: Adding JIT and Optimizer Support
55 - Because a lot of people are interested in using LLVM as a JIT,
56 we'll dive right into it and show you the 3 lines it takes to add JIT
57 support. LLVM is also useful in many other ways, but this is one
58 simple and "sexy" way to show off its power. :)
59 - `Chapter #5 <LangImpl05.html>`_: Extending the Language: Control
60 Flow - With the language up and running, we show how to extend it
61 with control flow operations (if/then/else and a 'for' loop). This
62 gives us a chance to talk about simple SSA construction and control
64 - `Chapter #6 <LangImpl06.html>`_: Extending the Language:
65 User-defined Operators - This is a silly but fun chapter that talks
66 about extending the language to let the user program define their own
67 arbitrary unary and binary operators (with assignable precedence!).
68 This lets us build a significant piece of the "language" as library
70 - `Chapter #7 <LangImpl07.html>`_: Extending the Language: Mutable
71 Variables - This chapter talks about adding user-defined local
72 variables along with an assignment operator. The interesting part
73 about this is how easy and trivial it is to construct SSA form in
74 LLVM: no, LLVM does *not* require your front-end to construct SSA
76 - `Chapter #8 <LangImpl08.html>`_: Compiling to Object Files - This
77 chapter explains how to take LLVM IR and compile it down to object
79 - `Chapter #9 <LangImpl09.html>`_: Extending the Language: Debug
80 Information - Having built a decent little programming language with
81 control flow, functions and mutable variables, we consider what it
82 takes to add debug information to standalone executables. This debug
83 information will allow you to set breakpoints in Kaleidoscope
84 functions, print out argument variables, and call functions - all
85 from within the debugger!
86 - `Chapter #10 <LangImpl10.html>`_: Conclusion and other useful LLVM
87 tidbits - This chapter wraps up the series by talking about
88 potential ways to extend the language, but also includes a bunch of
89 pointers to info about "special topics" like adding garbage
90 collection support, exceptions, debugging, support for "spaghetti
91 stacks", and a bunch of other tips and tricks.
93 By the end of the tutorial, we'll have written a bit less than 1000 lines
94 of non-comment, non-blank, lines of code. With this small amount of
95 code, we'll have built up a very reasonable compiler for a non-trivial
96 language including a hand-written lexer, parser, AST, as well as code
97 generation support with a JIT compiler. While other systems may have
98 interesting "hello world" tutorials, I think the breadth of this
99 tutorial is a great testament to the strengths of LLVM and why you
100 should consider it if you're interested in language or compiler design.
102 A note about this tutorial: we expect you to extend the language and
103 play with it on your own. Take the code and go crazy hacking away at it,
104 compilers don't need to be scary creatures - it can be a lot of fun to
110 This tutorial will be illustrated with a toy language that we'll call
111 "`Kaleidoscope <http://en.wikipedia.org/wiki/Kaleidoscope>`_" (derived
112 from "meaning beautiful, form, and view"). Kaleidoscope is a procedural
113 language that allows you to define functions, use conditionals, math,
114 etc. Over the course of the tutorial, we'll extend Kaleidoscope to
115 support the if/then/else construct, a for loop, user defined operators,
116 JIT compilation with a simple command line interface, etc.
118 Because we want to keep things simple, the only datatype in Kaleidoscope
119 is a 64-bit floating point type (aka 'double' in C parlance). As such,
120 all values are implicitly double precision and the language doesn't
121 require type declarations. This gives the language a very nice and
122 simple syntax. For example, the following simple example computes
123 `Fibonacci numbers: <http://en.wikipedia.org/wiki/Fibonacci_number>`_
127 # Compute the x'th fibonacci number.
134 # This expression will compute the 40th number.
137 We also allow Kaleidoscope to call into standard library functions (the
138 LLVM JIT makes this completely trivial). This means that you can use the
139 'extern' keyword to define a function before you use it (this is also
140 useful for mutually recursive functions). For example:
146 extern atan2(arg1 arg2);
148 atan2(sin(.4), cos(42))
150 A more interesting example is included in Chapter 6 where we write a
151 little Kaleidoscope application that `displays a Mandelbrot
152 Set <LangImpl06.html#kicking-the-tires>`_ at various levels of magnification.
154 Lets dive into the implementation of this language!
159 When it comes to implementing a language, the first thing needed is the
160 ability to process a text file and recognize what it says. The
161 traditional way to do this is to use a
162 "`lexer <http://en.wikipedia.org/wiki/Lexical_analysis>`_" (aka
163 'scanner') to break the input up into "tokens". Each token returned by
164 the lexer includes a token code and potentially some metadata (e.g. the
165 numeric value of a number). First, we define the possibilities:
169 // The lexer returns tokens [0-255] if it is an unknown character, otherwise one
170 // of these for known things.
183 static std::string IdentifierStr; // Filled in if tok_identifier
184 static double NumVal; // Filled in if tok_number
186 Each token returned by our lexer will either be one of the Token enum
187 values or it will be an 'unknown' character like '+', which is returned
188 as its ASCII value. If the current token is an identifier, the
189 ``IdentifierStr`` global variable holds the name of the identifier. If
190 the current token is a numeric literal (like 1.0), ``NumVal`` holds its
191 value. Note that we use global variables for simplicity, this is not the
192 best choice for a real language implementation :).
194 The actual implementation of the lexer is a single function named
195 ``gettok``. The ``gettok`` function is called to return the next token
196 from standard input. Its definition starts as:
200 /// gettok - Return the next token from standard input.
201 static int gettok() {
202 static int LastChar = ' ';
204 // Skip any whitespace.
205 while (isspace(LastChar))
206 LastChar = getchar();
208 ``gettok`` works by calling the C ``getchar()`` function to read
209 characters one at a time from standard input. It eats them as it
210 recognizes them and stores the last character read, but not processed,
211 in LastChar. The first thing that it has to do is ignore whitespace
212 between tokens. This is accomplished with the loop above.
214 The next thing ``gettok`` needs to do is recognize identifiers and
215 specific keywords like "def". Kaleidoscope does this with this simple
220 if (isalpha(LastChar)) { // identifier: [a-zA-Z][a-zA-Z0-9]*
221 IdentifierStr = LastChar;
222 while (isalnum((LastChar = getchar())))
223 IdentifierStr += LastChar;
225 if (IdentifierStr == "def")
227 if (IdentifierStr == "extern")
229 return tok_identifier;
232 Note that this code sets the '``IdentifierStr``' global whenever it
233 lexes an identifier. Also, since language keywords are matched by the
234 same loop, we handle them here inline. Numeric values are similar:
238 if (isdigit(LastChar) || LastChar == '.') { // Number: [0-9.]+
242 LastChar = getchar();
243 } while (isdigit(LastChar) || LastChar == '.');
245 NumVal = strtod(NumStr.c_str(), 0);
249 This is all pretty straight-forward code for processing input. When
250 reading a numeric value from input, we use the C ``strtod`` function to
251 convert it to a numeric value that we store in ``NumVal``. Note that
252 this isn't doing sufficient error checking: it will incorrectly read
253 "1.23.45.67" and handle it as if you typed in "1.23". Feel free to
254 extend it :). Next we handle comments:
258 if (LastChar == '#') {
259 // Comment until end of line.
261 LastChar = getchar();
262 while (LastChar != EOF && LastChar != '\n' && LastChar != '\r');
268 We handle comments by skipping to the end of the line and then return
269 the next token. Finally, if the input doesn't match one of the above
270 cases, it is either an operator character like '+' or the end of the
271 file. These are handled with this code:
275 // Check for end of file. Don't eat the EOF.
279 // Otherwise, just return the character as its ascii value.
280 int ThisChar = LastChar;
281 LastChar = getchar();
285 With this, we have the complete lexer for the basic Kaleidoscope
286 language (the `full code listing <LangImpl02.html#full-code-listing>`_ for the Lexer
287 is available in the `next chapter <LangImpl02.html>`_ of the tutorial).
288 Next we'll `build a simple parser that uses this to build an Abstract
289 Syntax Tree <LangImpl02.html>`_. When we have that, we'll include a
290 driver so that you can use the lexer and parser together.
292 `Next: Implementing a Parser and AST <LangImpl02.html>`_