transmogrify.htmlcontentextractor 1.0b5

This blueprint extracts out title, description and body from html either via xpath or by automatic cluster analysis

Introduction

transmogrify.htmlcontentextractor

This blueprint extracts out title, description and body from html either via xpath, TAL or by automatic cluster analysis

HTMLContentExtractor

This blueprint extracts out fields from html either via xpath rules or by automatic cluster analysis

transmogrify.htmlcontentextractor

You can define a series of rules which will get applied to the to the ‘_text’ of the input item. The rules use a XPATH expression or a TAL expression to extract html or text out of the html and adds it as key to the outputted item.

Each option of the blueprint is a rule of the following form

(N-)field = (optional)(text|html|delete|optional) xpath

OR

(N-)field = (optional)tal tal-expression

“field” is the attribute that will be set with the results of the xpath

“format” is what to do with the results of the xpath. “optional” means the same as “delete” but won’t cause the group to not match. if the format is delete or optional then the field name doesn’t matter but will still need to be unique

“xpath’ is an xpath expression

If the format is ‘tal’ then instead of an XPath use can use a TAL expression. TAL expression is evaluated on the item object AFTER the XPath expressions have been applied.

For example

[template]
blueprint = transmogrify.htmlcontentextractor
title = text //div[@class='body']//h1[1]
_permalink = text //div[@class='body']//a[@class='headerlink']
_text = html //div[@class='body']
_label = optional //p[contains(@class,'admonition-title')]
description = optional //div[contains(@class,'admonition-description')]/p[@class='last']/text()
_remove_useless_links = optional //div[@id = 'indices-and-tables']
mimetype = tal string:text/html
text = tal python:item['_text'].replace('id="blah"','')

You can delete a number of parts of the html by extracting content to fields such as _permalink and _label. These items won’t get used be set used to set any properties on the final content so are effective as a means of deleting parts of the html. TAL expressions are evaluated after XPath expressions so we can post process the _text XPath to produce a text stripped of a certain id.

N is the group number. Groups are run in order of group number. If any rule doesn’t match (unless its marked optional) then the next group will be tried instead. Group numbers are optional.

Instead of groups you can also chain several blueprints togeather. The blueprint will set ‘_template’ on the item. If another blueprint finds the ‘_template’ key in an item it will ignore that item.

The ‘_template’ field is the remainder of the html once all the content selected by the XPATH expressions have been applied.

transmogrify.htmlcontentextractor.auto

This blueprint will analyse the html and attempt to discover the rules to extract out the title, description and body of the html.

If the logger output is in DEBUG mode then the XPaths used by the auto extrator will be output to the logger.

Changelog

1.0b5 (2011-06-29)

• include docs

• now can use TAL expressions

1.0b4 (2011-02-06)

• handle ‘/text()’ in xpaths

• new ‘optionaltext’ rule format

1.0b3 (2010-12-13)

• simpler autogenerated xpath

• better logging

1.0b2 (2010-11-09)

• Put condition on autofinder so can be turned off

1.0b1 (2010-11-03)

• ignore already found items. better debug [“Dylan Jay”]

• skip templates if item already parsed [“Dylan Jay”]

• print automaticly found XPaths [“Dylan Jay”]

• make text fields strip tail text [“Vitaliy Podoba”]

1.0dev (2010-03-22)

• split the auto templatefinder out to it’s own blueprint [“Dylan Jay”]