From BroWiki

BinPAC compiler

Contents 1 Intro 2 Scanning and Parsing 2.1 Scanning 2.2 Parsing 3 Code generation 3.1 Preparation 3.1.1 HelperDecl 3.1.2 TypeDecl 3.1.3 RegExDecl 3.2 Generate Extern Declaration 3.3 Generate Forward Declaration 3.4 Generate Code 4 Class reference 4.1 AnalyzerElement 4.1.1 AnalyzerFlow 4.1.2 AnalyzerHelper 4.2 ID 4.3 Decl 4.3.1 TypeDecl 4.3.1.1 AnalyzerContextDecl 4.3.1.2 AnalyzerDecl 4.3.1.2.1 ConnDecl 4.3.2 Type

Intro

Doxygen HTML

Scanning and Parsing

Scanning and parsing is done by flex and bison generated code. A user can learn the formal grammar specification from aux/binpac/src/pac_scan.ll and aux/binpac/src/pac_parse.yy under the Bro source tree.

Scanning

Scanner has four start conditions:

• EC - Scanning embedded code

Input is appended into EmbeddedCode untill %}. Also parse for embedded_pac_primitive

• PP - Scanning PAC primitive
• INCL - Scanning include
• RE - Scanning a regex

Parsing

As binpac parses through input, it builds a declaration class for each construct. There are six different types of declarations:

• EnumDecl
• FuncDecl
• HelperDecl - Embbeded C code
• LetDecl
• RegExDecl - Regex defintion
• TypeDecl - analyzer, connection, flow, record definition

The generated declaration is added to decl_list_ static DeclList *decl_list_; // pac_decl.h

Code generation

void Decl::ProcessDecls(Output *out_h, Output *out_cc)

This function is the entry point to code generation. It goes through four phases: preparation, generate extern declaration, generate forward declaration, and generate code. It iterates through all declaration by the order which they occur in the PAC file.

Preparation

Before code is generated, the compiler iterate through all the declaration and prepare them for code generation. Typical action are adding Param and Fields.

HelperDecl

Do nothing

TypeDecl

RegExDecl

Add regex ID into global env as GLOBAL_VAR.

Generate Extern Declaration

Generate Forward Declaration

Generate Code

Class reference

Object is the base class which all other language elements extend.

• filename
• line_num
• location

AnalyzerElement

Used to store fields in flow and connection

AnalyzerFlow

upflow/downflow information, used in connection.

AnalyzerHelper

Store embedded code in connection and flow

ID

Contains the name and the location of an ID. Typically used to store declaration name.

• name
• anonymous_id
• locname

Decl

TypeDecl

• Type *type_

AnalyzerContextDecl

• current_analyzer_context_ - global context pointer, initialized to 0
• constructor - create new context and set current_analyzer_context_

AnalyzerDecl

ConnDecl

Connection declaration

Type

• UNDEF
• EMPTY
• BUILTIN
• PARAMETERIZED
• RECORD
• CASE
• ARRAY
• STRING
• EXTERN
• DUMMY