November
10th,
2011
Ragel是一个状态机代码生成器，支持多种语言C, C++, Objective-C, D, Java, Ruby。一般用来写正则表达式解析协议数据，然后编译成相应代码嵌入到其他语言中。
Ragel的介绍和入门可以看这里：http://blog.dccmx.com/2011/01/ragel-intro-1/
解析http头的例子可以看这里：http://yaoweibin2008.blog.163.com/blog/static/110313920098211472956/，下面的例子抄了部分URL的定义。
所谓流式解析就是限制buffer使用，保持内存为一个常量。例如，要解析4k的数据，如果不使用流式解析，就必须申请4k的buffer。如果使用流式解析，可以每次读取1k数据，解析之后丢弃掉。
Ragel很好的支持这种模式，只需要在解析之后保存cs（自动机状态），解析之前设置cs，就能从上次断点继续解析。
#include <stdio.h>
#include <string.h>
%%{
machine range_url;
action rts_act {
range_t->total_start = range_t->total_start * 10 + *fpc - '0';
}
action rte_act {
range_t->total_end = range_t->total_end * 10 + *fpc - '0';
}
action rss_act {
range_t->seg_start = range_t->seg_start * 10 + *fpc - '0';
}
action rse_act {
range_t->seg_end = range_t->seg_end * 10 + *fpc - '0';
}
action url_act {
range_t->url[range_t->url_idx] = *fpc;
range_t->url_idx++;
}
CTL = (cntrl | 127);
safe = ("$" | "-" | "_" | ".");
extra = ("!" | "*" | "'" | "(" | ")" | ",");
reserved = (";" | "/" | "?" | ":" | "@" | "&" | "=" | "+");
unsafe = (CTL | " " | "\"" | "#" | "%" | "<" | ">");
national = any -- (alpha | digit | reserved | extra | safe | unsafe);
unreserved = (alpha | digit | safe | extra | national);
escape = ("%" xdigit xdigit);
uchar = (unreserved | escape);
range_total_start = ([0-9]+ );
range_total_end = ([0-9]+ );
range_seg_start = ([0-9]+);
range_seg_end = ([0-9]+);
range_total = (range_total_start $ rts_act '-' range_total_end $ rte_act );
range_seg = (range_seg_start $ rss_act '-' range_seg_end $ rse_act );
scheme = ( alpha | "+" | "-" | "." )* ;
url = (uchar | reserved )* $ url_act;
line := range_total ':' range_seg ':' url "\n";
}%%
%%write data;
typedef struct {
int cs;
off_t total_start;
off_t total_end;
off_t seg_start;
off_t seg_end;
char url[1024];
int url_idx;
} range_url_t;
int parser_init(range_url_t *range_t) {
int cs = 0;
%%write init;
range_t->cs = cs; /* init */
range_t->total_start = 0;
range_t->total_end = 0;
range_t->seg_start = 0;
range_t->seg_end = 0;
memset(range_t->url, 0, 1024);
range_t->url_idx = 0;
return 0;
}
void parser(char *buffer, range_url_t *range_t) {
char *p = buffer;
char *pe = p + strlen(buffer);
int cs = range_t->cs; /* set cs */
%%write exec;
range_t->cs = cs; /* save cs */
}
int main() {
char s[100] = "1000-2000:3000-4000:http://aslfjsd/skfjdks/sdf28347?p=2834&f=123abc\n";
char s1[100];
char s2[100];
int i;
range_url_t range_t;
for (i = 0;i<=strlen(s);i++) {
snprintf(s1, i + 1, "%s", s);
sprintf(s2, "%s", s+i);
printf("==%s,%s\n",s1,s2);
parser_init(&range_t);
parser(s1, &range_t);
parser(s2, &range_t);
printf("=%ld, %ld\n", range_t.total_start, range_t.total_end);
printf("=%ld, %ld\n", range_t.seg_start, range_t.seg_end);
printf("=%s\n", range_t.url);
}
return 0;
}