Phases of a compiler

A compiler translates a source program into target code in a sequence of phases. Two components work with every phase: the symbol table (stores names and attributes) and the error handler (reports errors with line numbers).

Running example: position = initial + rate * 60 where position, initial, rate are float.

Phase	Task	Output for the example
1. Lexical analysis	Scans characters, groups them into lexemes and produces tokens; enters identifiers into the symbol table; removes whitespace and comments	<id,1> <=> <id,2> <+> <id,3> <*> <num,60>
2. Syntax analysis	Checks token order against the grammar and builds a parse/syntax tree	Syntax tree (diagram below)
3. Semantic analysis	Type checking, declaration checks, type conversions (coercion)	60 is int, rate is float, so insert inttofloat(60)
4. Intermediate code generation	Produces machine-independent code, usually three-address code	t1 = inttofloat(60)
t2 = id3 * t1
t3 = id2 + t2
id1 = t3
5. Code optimization	Improves speed/size: constant folding, copy propagation, dead-code removal	t1 = id3 * 60.0
id1 = id2 + t1
6. Code generation	Selects instructions, allocates registers, emits target code	LDF R2, id3
MULF R2, R2, #60.0
LDF R1, id2
ADDF R1, R1, R2
STF id1, R1

Input and output summary

• Lexical analyzer: character stream to token stream.
• Syntax analyzer: tokens to syntax tree.
• Semantic analyzer: syntax tree to type-checked (annotated) tree.
• Intermediate code generator: annotated tree to three-address code.
• Optimizer: intermediate code to better intermediate code.
• Code generator: optimized intermediate code to target code.

The first three phases form the analysis (front end); the last three form the synthesis (back end).

Syntax tree (phase 2)

Tree after semantic analysis (phase 3): the integer constant is converted to float.