标题: 折腾FLEX（词法分析器） [打印本页]

---

作者: 51hei不    时间: 2016-6-19 01:42
标题: 折腾FLEX（词法分析器）
上周折腾了半天FLEX词法分析器，N多年前玩过，现在全忘记了，捡起来重新折腾了一下，小计之，下次捡起来快一点。
FLEX是什么？ 把文本按一定规则(正则表达式定义)把文本格式化输出。循环版本的正则表达式。
干什么用的？   通常用于数据格式化，比如解析源代码，解析结构化的文本等等。
怎么玩？           配置一个.l模板文件，描叙规则，然后用 flex.exe 生成解析C源码，复制到工程里去，完事。
存在的问题？   中文比较麻烦，目前没需求，以后有需求再继续折腾
价值收益？       开脑洞，又多了一个偷懒神器

模板范例，从sgml解析htm里抄来的，简单跑跑很舒爽。
test.l：
%{


   #include "stdio.h"
   #include "stdlib.h"


   int num_num=0,num_id=0;

   #define fileno _fileno
   int bShow = 0;
%}

/* $Id: sgml.l,v 1.9 1996/02/07 15:32:28 connolly Exp $ */
/* sgml.l -- a lexical analyzer for Basic+/- SGML Documents
* See: "A Lexical Analyzer for HTML and Baisc SGML"
*/

/*
* NOTE: We assume the locale used by lex and the C compiler
* agrees with ISO-646-IRV; for example: '1' == 0x31.
*/


/* Figure 1 -- Character Classes: Abstract Syntax */

Digit        [0-9]
LCLetter    [a-z]
Special        ['()_,\-\./:=?]
UCLetter    [A-Z]

/* Figure 2 -- Character Classes: Concrete Syntax */

LCNMCHAR    [\.-]
/* LCNMSTRT    [] */
UCNMCHAR    [\.-]
/* UCNMSTRT    [] */
/* @# hmmm. sgml spec says \015 */
RE        \n
/* @# hmmm. sgml spec says \012 */
RS        \r
SEPCHAR        \011
SPACE        \040

/* Figure 3 -- Reference Delimiter Set: General */

COM    "--"
CRO    "&#"
DSC    "]"
DSO    "["
ERO    "&"
ETAGO  "</"
LIT    \"
LITA   "'"
MDC    ">"
MDO    "<!"
MSC    "]]"
NET    "/"
PERO   "%"
PIC    ">"
PIO    "<?"
REFC   ";"
STAGO  "<"
TAGC   ">"

/* 9.2.1 SGML Character */

/*name_start_character    {LCLetter}|{UCLetter}|{LCNMSTRT}|{UCNMSTRT}*/
name_start_character    {LCLetter}|{UCLetter}
name_character        {name_start_character}|{Digit}|{LCNMCHAR}|{UCNMCHAR}|[\xa1-\xff]

/* 9.3 Name */

name        {name_start_character}{name_character}*
number        {Digit}+
number_token    {Digit}{name_character}*
name_token    {name_character}+

/* 6.2.1 Space */
s        {SPACE}|{RE}|{RS}|{SEPCHAR}
ps        ({SPACE}|{RE}|{RS}|{SEPCHAR})+

/* trailing white space */
ws        ({SPACE}|{RE}|{RS}|{SEPCHAR})*

/* 9.4.5 Reference End */
reference_end    ({REFC}|{RE})

/*
* 10.1.2 Parameter Literal
* 7.9.3  Attribute Value Literal
* (we leave recognition of character references and entity references,
*  and whitespace compression to further processing)
*
* @# should split this into minimum literal, parameter literal,
* @# and attribute value literal.
*/
literal        ({LIT}[^\"]*{LIT})|({LITA}[^\']*{LITA})



/* 9.6.1 Recognition modes */

/*
* Recognition modes are represented here by start conditions.
* The default start condition, INITIAL, represents the
* CON recognition mode. This condition is used to detect markup
* while parsing normal data charcters (mixed content).
*
* The CDATA start condition represents the CON recognition
* mode with the restriction that only end-tags are recognized,
* as in elements with CDATA declared content.
* (@# no way to activate it yet: need hook to parser.)
*
* The TAG recognition mode is split into two start conditions:
* ATTR, for recognizing attribute value list sub-tokens in
* start-tags, and TAG for recognizing the TAGC (">") delimiter
* in end-tags.
*
* The MD start condition is used in markup declarations. The COM
* start condition is used for comment declarations.
*
* The DS condition is an approximation of the declaration subset
* recognition mode in SGML. As we only use this condition after signalling
* an error, it is merely a recovery device.
*
* The CXT, LIT, PI, and REF recognition modes are not separated out
* as start conditions, but handled within the rules of other start
* conditions. The GRP mode is not represented here.
*/

/* EXCERPT ACTIONS: START */

/* %x CON == INITIAL */
%x CDATA

%x TAG
%x ATTR
%x ATTRVAL
%x NETDATA
%x ENDTAG
/* this is only to be permissive with bad end-tags: */
%x JUNKTAG

%x MD
%x COM
%x DS

/* EXCERPT ACTIONS: STOP */

%%

  int *types = NULL;
  char **strings = NULL;
  size_t *lengths = NULL;
  int qty = 0;

      /*
       * See sgml_lex.c for description of
       *   ADD, CALLBACK, ERROR, TOK macros.
       */

  /* <name -- start tag */
{STAGO}{name}{ws}        {
            //printf("TAG:[%s]\r\n",yytext);
                                  //ADDCASE(SGML_START, yytext, yyleng);
                  //BEGIN(ATTR);
                 //Sleep(200);
                }

  /* <a ^href = "xxx"> -- attribute name */
{name}{s}*={ws}        {
            if (stricmp(yytext,"href=")==0)
            {
                printf("ATTR:[%s]\r\n",yytext);
                bShow = 1;
            }
                }

  /* <a name = ^xyz> -- name token */
{name_token}{ws}        {
            //printf("ATTR2:[%s]\r\n",yytext);
                                }

  /* <a href = ^"a b c"> -- literal */
{literal}{ws}                {
            if (bShow)
            {
                printf("VALUE:[%s]\r\n",yytext);
                bShow = 0;
            }
                                }
                                                

.|\r|\n {}

%%

int main(int argc, char* argv[])
{
   yyin=fopen("./test.html","r");
   yylex();
   printf("num=%d,id=%d/n",num_num,num_id);
   return 0;
}

int yywrap()//此函数必须由用户提供
{
    return 1;
}

---

欢迎光临 (http://www.51hei.com/bbs/)

Powered by Discuz! X3.1