diff options
| -rw-r--r-- | abonnes.lemonde.fr.txt | 33 | ||||
| -rw-r--r-- | lemonde.fr.txt | 2 |
2 files changed, 35 insertions, 0 deletions
diff --git a/abonnes.lemonde.fr.txt b/abonnes.lemonde.fr.txt new file mode 100644 index 0000000..4c23307 --- /dev/null +++ b/abonnes.lemonde.fr.txt @@ -0,0 +1,33 @@ +title: //h1 + +# We can have multiple authors +author: //span[@id='publisher'] +author: //a[@class='auteur'] + +# Last edition date (if any) +date: //time[@itemprop='dateModified']/@datetime +# Publication date +date: //time[@itemprop='datePublished']/@datetime + + +body: //div[@id='articleBody']|//section[@class='contenu']//div[@class='texte'] + +# Remove highlighted quotes +strip_id_or_class: accroche +strip_id_or_class: encart + +# Remove "Lire aussi" blocks +strip: //p[@class='lire'] + +# Remove footers for Le Monde Festival +strip: //aside[@class='fenetre'] + +# Remove the insane "conjugaison.lemonde.fr" links: +find_string: <a target='_blank' onclick='return false;' class='lien_interne conjug' +replace_string: <input type='hidden' style='display:none;' + +prune: no + +test_url: http://abonnes.lemonde.fr/festival/article/2017/08/22/la-terreur-un-pan-de-l-histoire-russe_5175130_4415198.html +test_url: http://abonnes.lemonde.fr/festival/article/2017/08/18/fin-de-partie-pour-les-fraudeurs-au-carbone_5173972_4415198.html +test_url: http://abonnes.lemonde.fr/politique/article/2017/08/22/code-du-travail-pour-hollande-il-ne-faudrait-pas-demander-aux-francais-des-sacrifices-qui-ne-sont-pas-utiles_5175211_823448.html diff --git a/lemonde.fr.txt b/lemonde.fr.txt index 1994524..d5f9a9e 100644 --- a/lemonde.fr.txt +++ b/lemonde.fr.txt @@ -8,6 +8,8 @@ date: //time[@itemprop='dateModified']/@datetime # Publication date date: //time[@itemprop='datePublished']/@datetime +# Remove "Lire aussi" blocks +strip: //p[@class='lire'] body: //div[@id='articleBody'] |