diff options
| -rw-r--r-- | globoesporte.globo.com.txt | 15 |
1 files changed, 9 insertions, 6 deletions
diff --git a/globoesporte.globo.com.txt b/globoesporte.globo.com.txt index d463602..fd8e70f 100644 --- a/globoesporte.globo.com.txt +++ b/globoesporte.globo.com.txt @@ -1,22 +1,25 @@ title: //h1[@class="entry-title"] -body: ///div[@class='materia-titulo']/h2 | //*[@id="materia-letra"] +body: //div[@class='materia-titulo']/h2 | //*[@id="materia-letra"] date: //abbr[@class="published"] date: //abbr[@class="updated"] author: //*[@class="author"]/strong -# quotes -wrap_in(blockquote): //*[@class="frase-materia"] - -strip: //*[@class="foto"]/strong -strip: //*[@class="saibamais"] +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 vídeos 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 |