From 4e067ceabd705201a16b4c92cf4b23f3b990326c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= Date: Sun, 13 Jul 2014 10:15:40 +0200 Subject: updated specific configuration for parsing --- .../standard/globoesporte.globo.com.txt | 25 ++++++++++++++++++++++ 1 file changed, 25 insertions(+) create mode 100755 inc/3rdparty/site_config/standard/globoesporte.globo.com.txt (limited to 'inc/3rdparty/site_config/standard/globoesporte.globo.com.txt') diff --git a/inc/3rdparty/site_config/standard/globoesporte.globo.com.txt b/inc/3rdparty/site_config/standard/globoesporte.globo.com.txt new file mode 100755 index 00000000..fd8e70ff --- /dev/null +++ b/inc/3rdparty/site_config/standard/globoesporte.globo.com.txt @@ -0,0 +1,25 @@ +title: //h1[@class="entry-title"] + +body: //div[@class='materia-titulo']/h2 | //*[@id="materia-letra"] + +date: //abbr[@class="published"] +date: //abbr[@class="updated"] + +author: //*[@class="author"]/strong + +strip: //div[contains(@class,'foto')]/strong +strip: //div[contains(@class,'frase-materia')]/div[@class='autor'] +strip: //div[contains(@class,'saibamais')] +strip: //*[contains(text(),'Clique aqui e veja mais')]/ancestor::p +strip: //ul[@class="toolbar"] + +# quotes +wrap_in(blockquote): //div[@id='materia-letra']//div[contains(@class,'frase-materia')]/div[@class='frase'] + +prune: no + +replace_string([Clique aqui e veja mais vĂ­deos do Fluminense]): [] + +test_url: http://globoesporte.globo.com/atletismo/noticia/2013/08/michael-johnson-diz-que-bolt-e-melhor-da-historia-nao-ha-duvidas.html +test_url: http://globoesporte.globo.com/futebol/futebol-internacional/futebol-espanhol/noticia/2013/08/barca-atropela-levante-e-neymar-passa-em-branco-em-estreia-oficial.html +test_url: http://globoesporte.globo.com/futebol/times/fluminense/noticia/2013/08/poupado-no-sabado-felipe-se-diz-pronto-para-ser-titular-contra-o-goias.html -- cgit v1.2.3