Re: New proposal: HTML_Miner

Hi Niccolo, Out of curiosity, how do you parse HTML? I once tried with the PEAR XML_HTMLSax(3) and quickly gave up; the class doesn't work and looks abandoned. What I do instead: - transform HTML into XML with tidy - extract what I want to extract by passing an Xpath expression. This is flexible and powerful. You'll be able to extract not only HTML blocks, but pretty much anything. HTH Ignatius Niccolò Venturoli thus spake on 2006-01-15 14:07:

Hello, I developed a PHP module for extracting blocks from a HTML file, I called it HTML_Miner. The class is very easy to use: you call the method parseBlock($nodeSpecs, $html) that returns the inner and outer HTML of the block you specified in $nodeSpecs. An example: parseBlock("body->table->table", file_get_contents("my.html")); Will return the flowwing array: array( 'outer' => '<table ...>CONTENT</table>', 'inner' => 'CONTENT' ) The class will return the correct block following the html tree structure. In this example the class will return the first table, nested in first table of the document's body. The grammar used for $nodeSpecs can handle little more complicated requests, for example you can have: parseBlock("body->table[3]->table", file_get_contents("my.html")); or maybe: parseBlock("body->table(class='myTable')->table(cellpadding='7')", file_get_contents("my.html")); The former will get the first table nested in the third table of the body, the second example will get the first table having cellpadding=7 in the first table of the document's body having class='myTable'. You can mix both of these options and create complex expressions for block parsing. Do you think that this class can be of any interest to the pear community? I was thinking about subscribing to pear and submitting the class properly commented and formatted. What do you suggest? Thank you. Regards, Niccolò Venturoli Rome, Italy