From 9f586aa21e46561d0b6d31562d2d1e884da59fd3 Mon Sep 17 00:00:00 2001 From: Thiago Talma Date: Mon, 19 Aug 2013 12:46:50 -0300 Subject: Create globoesporte.globo.com.txt --- globoesporte.globo.com.txt | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) create mode 100644 globoesporte.globo.com.txt (limited to 'globoesporte.globo.com.txt') diff --git a/globoesporte.globo.com.txt b/globoesporte.globo.com.txt new file mode 100644 index 0000000..4cda55b --- /dev/null +++ b/globoesporte.globo.com.txt @@ -0,0 +1,19 @@ +title: //h1[@class="entry-title"] + +body: //*[@id="materia-letra"] + +date: //abbr[@class="published"] +date: //abbr[@class="updated"] + +author: //*[@class="author"]/strong + +strip: //*[@class="foto"]//strong +strip: //*[@class="saibamais"] +strip: //*[@class="frase-materia"] + +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 -- cgit v1.3.1 From 606c9d2f8f5d07a51941a591aa0d8fc93652a213 Mon Sep 17 00:00:00 2001 From: Thiago Talma Date: Mon, 19 Aug 2013 14:00:19 -0300 Subject: Update globoesporte.globo.com.txt --- globoesporte.globo.com.txt | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'globoesporte.globo.com.txt') diff --git a/globoesporte.globo.com.txt b/globoesporte.globo.com.txt index 4cda55b..81efd2e 100644 --- a/globoesporte.globo.com.txt +++ b/globoesporte.globo.com.txt @@ -1,15 +1,17 @@ title: //h1[@class="entry-title"] -body: //*[@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: //*[@class="frase-materia"] prune: no -- cgit v1.3.1 From ebc5a52dce8c12ea1e51a9e4812d3e7f932e621c Mon Sep 17 00:00:00 2001 From: Thiago Talma Date: Mon, 19 Aug 2013 17:12:23 -0300 Subject: Some adjust... --- globoesporte.globo.com.txt | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'globoesporte.globo.com.txt') diff --git a/globoesporte.globo.com.txt b/globoesporte.globo.com.txt index 81efd2e..d463602 100644 --- a/globoesporte.globo.com.txt +++ b/globoesporte.globo.com.txt @@ -10,8 +10,9 @@ author: //*[@class="author"]/strong # quotes wrap_in(blockquote): //*[@class="frase-materia"] -strip: //*[@class="foto"]//strong +strip: //*[@class="foto"]/strong strip: //*[@class="saibamais"] +strip: //ul[@class="toolbar"] prune: no -- cgit v1.3.1 From d12078050b25afd8047f4a25a6f8dea962aca1eb Mon Sep 17 00:00:00 2001 From: Thiago Talma Date: Tue, 20 Aug 2013 16:22:56 -0300 Subject: Some adjust... Sorry for the flood of changes. As I still do not have the sandbox, let the trial and error... I believe that now hit the target. Hugs! --- globoesporte.globo.com.txt | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) (limited to 'globoesporte.globo.com.txt') 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 -- cgit v1.3.1