转载请联系作者：Email: pengwuwang@21cn.com darlingpeng@sina.com
在工作中,经常需要对网页上的表格内容进行处理,但是,由于表格内容制作过程中的随意性,跨行跨列经常发生,所以我作了这几个函数,以获取表格的内容,程序中重要的地方已作了注解,所以在此不再重复说明,经过测试,非常成功.所以拿出来供大家共享.
<?php
//作者: 王朋武
//日期: 2005,3,31
//目的: 获取网页上(任意跨行跨列的)表格中的内容
function fun_proc_rowspan($l_str)
{
$l_a = explode("\n", $l_str);
for($i=count($l_a)-1; $i>=0; $i--) {
$l_str = trim($l_a[$i]);
if(empty($l_str)) continue;
$l_str = eregi_replace("<td", "<b><td", $l_str);
$l_b = explode("<b>", $l_str); //如有n个<b>,则分成(n+1)个组,最开始一项为空.
for($j=0; $j<count($l_b); $j++) { /*不能反过来(即从右开始)，否则在有些情况下会有错误，因为在表格的一行中如有几个连续的格跨行，则必须从左开始，否则下一行的列数就可能不够计算，即计算跨行时保存的列有可能大于下一行的总列数, 则操作就不会成功. 实测也证明.
如 +----+----+----+----+----+----+----+
+----+----+----+----+----+----+----+此一行保存7格, 前6格都有rowspan=2
+-----------------------------+----+此一行只保存2格
+----------------------------------+此一行只保存1格
*/
$l_str = trim($l_b[$j]);
if(eregi("rowspan", $l_str)) {
$rowspan = preg_replace("/^<td.+rowspan=[\"\']?(\d{1,2})[\"\']?.*/", "\1", $l_str);
$rowcont = preg_replace("/^<td.+rowspan=[\"\']?\d{1,2}[\"\']?.*>(.+)<\/td>/", "\1", $l_str);
$l_a = fun_add_row_td($l_a, $i, $j, $rowspan, $rowcont);
$l_str = implode("\n", $l_a);
return $l_str;
}
}//end of for j
}//end of for i
return $l_str;
}
function fun_add_row_td($l_a, $r, $l, $add_n, $add_cont) //l_a是数组, r是插入的开始行,l是开始的列, add_n是跨多少行, add_cont是加的内容
{
for($i=$r; $i<$r+$add_n; $i++) {
if($i == $r) {
$l_str = eregi_replace("<td", "<b><td", $l_a[$i]);
$l_b = explode("<b>", $l_str);
$l_c = "";
for($j=1; $j<count($l_b); $j++) {
if($j == $l) $l_c .= eregi_replace("rowspan", "", $l_b[$j]); //去掉rowspan
else $l_c .= $l_b[$j];
}
$l_a[$i] = $l_c;
continue;
}
$l_str = eregi_replace("<td", "<b><td", $l_a[$i]); //加<b>分隔符
$l_b = explode("<b>", $l_str); //以<b>分组
$l_c = "";
for($j=1; $j<$l+$add_n; $j++) { //略过开始的空项,从1开始
if($j == $l) {
$l_c .= "<td>".$add_cont."</td>";
}
$l_c .= $l_b[$j];
}//end of for j
$l_a[$i] = $l_c; //更新后的新内容
}//end of for i
return $l_a;
}
function fun_proc_colspan($l_str)
{
$l_a = explode("\n", $l_str);
for($i=0; $i<count($l_a); $i++) {
$l_str = trim($l_a[$i]);
if(empty($l_str)) continue;
$l_str = eregi_replace("<td", "<b><td", $l_str);
$l_b = explode("<b>", $l_str); //如有n个<b>,则分成(n+1)个组,最开始一项为空.
for($j=1; $j<count($l_b); $j++) { //此处不能反过来，必须从左到右取表格内容
$l_str = trim($l_b[$j]);
if(eregi("<td.+colspan", $l_str)) {
$colspan = preg_replace("/^<td.+colspan=[\"\']?(\d{1,2})[\"\']?.*/", "\1", $l_str);
$colcont = preg_replace("/^<td.+colspan=[\"\']?\d{1,2}[\"\']?.*>(.+)<\/td>/", "\1", $l_str);
$l_a = fun_add_col_td($l_a, $i, $j, $colspan, $colcont);
$l_str = implode("\n", $l_a);
return $l_str;
}
}//end of for j
}//end of for i
return $l_str;
}
function fun_add_col_td($l_a, $r, $l, $add_n, $add_cont) //l_a是数组, r是插入的开始行,l是开始的列, add_n是跨多少列, add_cont是加的内容
{
$l_str = eregi_replace("<td", "<b><td", $l_a[$r]); //加<b>分隔符
$l_b = explode("<b>", $l_str); //以<b>分组
$l_c = "";
for($j=1; $j<=count($l_b); $j++) { //略过开始的空项,从1开始
if($j == $l) {
$l_c .= eregi_replace("colspan", "", $l_b[$j]); //去掉colspan
continue;
}
if($j == $l+1) {
for($k=0; $k<$add_n-1; $k++)
$l_c .= "<td>".$add_cont."</td>";
}
$l_c .= $l_b[$j];
}//end of for j
$l_a[$r] = $l_c; //更新后的新内容
return $l_a;
}
$l_str = file_get_contents("test.htm"); //获取网页内容
$l_str = eregi_replace(".*<table", "<table", $l_str);
$l_str = eregi_replace("</table>.*", "", $l_str);
$l_str = eregi_replace("\r", "", $l_str); //去掉\r
$l_str = eregi_replace("\n", "", $l_str); //去掉\n
$l_str = eregi_replace("</tr>", "\n", $l_str); //使表格的一行成为一行数据
$l_str = strip_tags($l_str, "<td>"); //只保留<td>,</td>HTML表记
while(eregi("<td.+colspan", $l_str)) $l_str = fun_proc_colspan($l_str);
while(eregi("<td.+rowspan", $l_str)) $l_str = fun_proc_rowspan($l_str);
/////至此,表格处理已完成.
$l_str = eregi_replace("</td>", "<br>", $l_str); //给每一格之间加上标记
$l_str = strip_tags($l_str, "<br>");
$l_a = explode("\n", $l_str);
for($i=0; $i<count($l_a); $i++) {
$l_str = trim($l_a[$i]);
if(empty($l_str)) continue;
$l_b = explode("<br>", $l_str);
foreach($l_b as $val) {
echo $val." "; //表格每列内容间加空格
}
echo "<br>"; //表格的一行显示为一行
}
//附测试用例和测试结果:
/***********测试用表格*************************************************************
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<title>Untitled Document</title>
<meta http-equiv="Content-Type" content="text/html; charset=gb2312">
</head>
<body>
<table width="200" border="1" cellspacing="0" cellpadding="0">
<tr>
<td>1</td>
<td>2</td>
<td>3</td>
<td>4</td>
<td>5</td>
<td>6</td>
<td>7</td>
</tr>
<tr>
<td>a1</td>
<td rowspan="7">22</td>
<td>a2</td>
<td colspan="4">a3</td>
</tr>
<tr>
<td>b1</td>
<td colspan="2">b2</td>
<td>b3</td>
<td>b4</td>
<td rowspan="4">33</td>
</tr>
<tr>
<td>c1</td>
<td>c2</td>
<td>c3</td>
<td>c4</td>
<td rowspan="2">44</td>
</tr>
<tr>
<td>d1</td>
<td>d2</td>
<td>d3</td>
<td>d4</td>
</tr>
<tr>
<td>e1</td>
<td>e2</td>
<td>e3</td>
<td>e4</td>
<td>e5</td>
</tr>
<tr>
<td>f1</td>
<td>f2</td>
<td colspan="4">f3</td>
</tr>
<tr>
<td>g1</td>
<td colspan="5">g2</td>
</tr>
<tr>
<td colspan="7">h1</td>
</tr>
<tr>
<td>i1</td>
<td colspan="3">i2</td>
<td>i3</td>
<td>i4</td>
<td>i5</td>
</tr>
<tr>
<td>j1</td>
<td colspan="3">j2</td>
<td>j3</td>
<td colspan="2">j4</td>
</tr>
<tr>
<td colspan="6">k1</td>
<td>k2</td>
</tr>
<tr>
<td colspan="6">l1</td>
<td>l2</td>
</tr>
<tr>
<td colspan="5">m1</td>
<td>m2</td>
<td>m3</td>
</tr>
</table>
</body>
</html>
*********************************************************************************/
/**************操作结果的表格***********************************************
<table border='1'>
<tr> <td>1</td> <td>2</td> <td>3</td> <td>4</td> <td>5</td> <td>6</td> <td>7</td> </tr>
<tr><td>a1</td> <td ="7">22</td> <td>a2</td> <td ="4">a3</td> <td>a3</td><td>a3</td><td>a3</td></tr>
<tr><td>b1</td> <td>22</td><td ="2">b2</td> <td>b2</td><td>b3</td> <td>b4</td> <td ="4">33</td> </tr>
<tr><td>c1</td> <td>22</td><td>c2</td> <td>c3</td> <td>c4</td> <td ="2">44</td> <td>33</td></tr>
<tr><td>d1</td> <td>22</td><td>d2</td> <td>d3</td> <td>d4</td> <td>44</td><td>33</td></tr>
<tr><td>e1</td> <td>22</td><td>e2</td> <td>e3</td> <td>e4</td> <td>e5</td> <td>33</td></tr>
<tr><td>f1</td> <td>22</td><td>f2</td> <td ="4">f3</td> <td>f3</td><td>f3</td><td>f3</td></tr>
<tr><td>g1</td> <td>22</td><td ="5">g2</td> <td>g2</td><td>g2</td><td>g2</td><td>g2</td></tr>
<tr><td ="7">h1</td> <td>h1</td><td>h1</td><td>h1</td><td>h1</td><td>h1</td><td>h1</td></tr>
<tr><td>i1</td> <td ="3">i2</td> <td>i2</td><td>i2</td><td>i3</td> <td>i4</td> <td>i5</td> </tr>
<tr><td>j1</td> <td ="3">j2</td> <td>j2</td><td>j2</td><td>j3</td> <td ="2">j4</td> <td>j4</td></tr>
<tr><td ="6">k1</td> <td>k1</td><td>k1</td><td>k1</td><td>k1</td><td>k1</td><td>k2</td> </tr>
<tr><td ="6">l1</td> <td>l1</td><td>l1</td><td>l1</td><td>l1</td><td>l1</td><td>l2</td> </tr>
<tr><td ="5">m1</td> <td>m1</td><td>m1</td><td>m1</td><td>m1</td><td>m2</td> <td>m3</td> </tr>
</table>
******************************************************************/
测试用表：
1
2
3
4
5
6
7
a1
22
a2
a3
b1
b2
b3
b4
33
c1
c2
c3
c4
44
d1
d2
d3
d4
e1
e2
e3
e4
e5
f1
f2
f3
g1
g2
h1
i1
i2
i3
i4
i5
j1
j2
j3
j4
k1
k2
l1
l2
m1
m2
m3
测试结果的表：
1
2
3
4
5
6
7
a1
22
a2
a3
a3
a3
a3
b1
22
b2
b2
b3
b4
33
c1
22
c2
c3
c4
44
33
d1
22
d2
d3
d4
44
33
e1
22
e2
e3
e4
e5
33
f1
22
f2
f3
f3
f3
f3
g1
22
g2
g2
g2
g2
g2
h1
h1
h1
h1
h1
h1
h1
i1
i2
i2
i2
i3
i4
i5
j1
j2
j2
j2
j3
j4
j4
k1
k1
k1
k1
k1
k1
k2
l1
l1
l1
l1
l1
l1
l2
m1
m1
m1
m1
m1
m2
m3
