昨天使用shell脚本来抽取html数据的时候，碰到了一个问题，如果要抽取的数据成了如下的情形时，数据的抽取就会出现不一致，有一些记录会没有数据，只显示"未开售" 这个时候如果还是按照原来的思路来抽取就会出现数据混乱的情况，比如根据第一列抽取数据一共有75 行，但是根据右边的赔率只能得到74行，有一行的数据混乱，后面的数据就全乱了。

+1	5.80↑ 4.40 1.38↓
	2.58↑ 3.55↑ 2.18↓
2	未开售
	1.55↓ 4.30 4.00↑

大体的Html代码如下：
可以看到对应的div FM2,FHMW如果都有数据，都含有3行对应的数据，如果div FMW中只显示“未开售”的时候，只有一行。
这样数据明显的不对应。
需要找到一定的规律来有条件的抽取和过滤。

                                                         <div class="selection">
                                                             <div class="FMW">
                                                                 <a class="homewin btn" op="w"><span class="num">6.00<span class="hoverArea" oncontextmenu="retur
 n false;"><cite></cite></a>
                                                                 <a class="draw btn" op="d"><span class="num">4.30<span class="hoverArea" oncontextmenu="return f
 alse;"><cite></cite></a>
                                                                 <a class="awaywin btn" op="l"><span class="num">1.38<span class="hoverArea" oncontextmenu="retur
 n false;"><cite></cite></a>
                                                            
                                                             <div class="FHMW">
                                                                 <a class="homewin btn" op="hdw"><span class="num">2.55<span class="hoverArea" oncontextmenu="ret
 urn false;"><cite></cite></a>
                                                                     <a class="draw btn" op="hdd"><span class="num">3.50<span class="hoverArea" oncontextmenu="re
 turn false;"><cite></cite></a>
                                                                     <a class="awaywin btn" op="hdl"><span class="num">2.22<span class="hoverArea" oncontextmenu=
 "return false;"><cite></cite></a>
                                                            
 
                                                      <span class="handicap td">
                                                             <em class="num nh">0</em>
                                                             <em class="num h"><b class="w0 td1 num h-pt">-2</b></em>
                                                        
                                                         <div class="odds-area td no-select">
                                                         <div class="selection">
                                                             <div class="FMW">
                                                                 <em class="no-sale">未开售</em>
                                                                                                                                 
                                                            
                                                             <div class="FHMW">
                                                                 <a class="homewin btn" op="hdw"><span class="num">1.53<span class="hoverArea" oncontextmenu="return false;"><cite></cite></a>
                                                                     <a class="draw btn" op="hdd"><span class="num">4.45<span class="hoverArea" oncontextmenu="return false;"><cite></cite></a>
                                                                     <a class="awaywin btn" op="hdl"><span class="num">4.00<span class="hoverArea" oncontextmenu="return false;"><cite></cite></a>
                                                                 


现在来做一个改进，按照div来抽取。
grep "em class=\"vs\"" *e|awk -F"score-text\">" '{print $2}'|awk -F"<" '{print $1}' > vs.lst
grep  -A4 "div class=\"selection\"" *e|grep -A3 "FMW" |awk '{ if($2~/no-sale/) {print "\"num\">0< \n \"num\">0< \n \"num\">0< \n" } else {print $5$6}}'|awk -F"num\">" '{print $2}'|awk -F"<" '{print $1}' |awk -v RS= '{print $1" " $2" " $3}' > fmw.lst
grep  -A9 "div class=\"selection\"" *e|grep -A3 "FHMW" |awk '{ if($2~/no-sale/) {print "\"num\">0< \n \"num\">0< \n \"num\">0< \n" } else {print $5$6}}'|awk -F"num\">" '{print $2}'|awk -F"<" '{print $1}' |awk -v RS= '{print $1" " $2" " $3}' > hfmw.lst

抽取后的数据就会很自然的显示出来，我们把“未开售”的记录设置为0
数据抽取过滤后，结果就会如下所示，数据就不会乱了。
1,3.40,3.35,1.88,1.71,3.65,3.70
2,0,0,0,2.85,4.20,1.85
1,4.90,3.95,1.50,2.24,3.60,2.47
1,7.10,4.80,1.29,2.95,3.75,1.91
1,5.30,3.85,1.48,2.26,3.35,2.58
1,5.00,4.00,1.49,2.25,3.55,2.48
1,3.20,3.40,1.93,1.68,3.75,3.75