From 9a7eab02d62ee194f595171ebc748431ee15f11c Mon Sep 17 00:00:00 2001 From: Barbara Miller Date: Fri, 6 Oct 2017 15:51:59 -0700 Subject: [PATCH] test private domain wix --- brozzler/behaviors.yaml | 80 ++++++++++++++++++++++++++++++++--------- 1 file changed, 63 insertions(+), 17 deletions(-) diff --git a/brozzler/behaviors.yaml b/brozzler/behaviors.yaml index 273e5bb..d3089aa 100644 --- a/brozzler/behaviors.yaml +++ b/brozzler/behaviors.yaml @@ -1,7 +1,7 @@ # # brozzler/behaviors.yaml - behavior configuration # -# Copyright (C) 2014-2016 Internet Archive +# Copyright (C) 2014-2017 Internet Archive # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -37,10 +37,26 @@ url_regex: '^https?://(?:www\.)?instagram\.com/.*$' behavior_js_template: instagram.js request_idle_timeout_sec: 10 +- + url_regex: '^https?://(?:www\.)?pm\.gc\.ca/.*$' + behavior_js_template: umbraBehavior.js.j2 + default_parameters: + actions: + - selector: div.teaser, li.pager__item a +- # https://webarchive.jira.com/browse/ARI-5389 + url_regex: '^https?://pitchfork\.com/.*$' + behavior_js_template: umbraBehavior.js.j2 + default_parameters: + actions: + - selector: div.teaser, li.pager__item a + closeSelector: .pmf-artist-modal__close-btn - url_regex: '^https?://(?:www\.)?huffingtonpost\.com/.*$' - behavior_js_template: huffpostslides.js - request_idle_timeout_sec: 10 + behavior_js_template: umbraBehavior.js.j2 + default_parameters: + actions: + - selector: .slideshow-card__overlay + - selector: .slideshow__next - url_regex: '^https?://(?:www\.)?brooklynmuseum\.org/exhibitions/.*$' behavior_js_template: simpleclicks.js.j2 @@ -55,6 +71,13 @@ click_css_selector: a[onclick] click_until_hard_timeout: False request_idle_timeout_sec: 10 +- # https://webarchive.jira.com/browse/ARI-5366 + url_regex: '^https?://(?:(ici|beta)\.)radio-canada\.ca/.*$' + behavior_js_template: simpleclicks.js.j2 + default_parameters: + click_css_selector: .bigBtnPlay + click_until_hard_timeout: False + request_idle_timeout_sec: 10 - # https://webarchive.jira.com/browse/ARI-5409 url_regex: '^https?://(?:www\.)?tuebingen.de/.*$' behavior_js_template: simpleclicks.js.j2 @@ -73,7 +96,7 @@ url_regex: '^https?://(?:www\.)?soundcloud.com/.*$' behavior_js_template: simpleclicks.js.j2 default_parameters: - click_css_selector: button.sc-button-play, button.playButton, div.compactTrackListItem + click_css_selector: button.sc-button-play, .playButton, div.compactTrackListItem click_until_hard_timeout: False request_idle_timeout_sec: 10 - # https://webarchive.jira.com/browse/AITFIVE-463 @@ -91,10 +114,10 @@ click_until_hard_timeout: False request_idle_timeout_sec: 10 - # https://webarchive.jira.com/browse/ARI-5453 - url_regex: '^https?://.*\.wixsite.com/.*$' + url_regex: '^https?://.*\.(wixsite|dorothearockburne)\.com/.*$' behavior_js_template: simpleclicks.js.j2 default_parameters: - click_css_selector: .ddm1repeaterButtonlabel + click_css_selector: .ddm1repeaterButtonlabel, .ytp-button click_until_hard_timeout: False request_idle_timeout_sec: 10 - # https://webarchive.jira.com/browse/ARI-4725 @@ -108,11 +131,32 @@ url_regex: '^https?://(?:www\.)?fec.gov/data/.*$' behavior_js_template: fec_gov.js request_idle_timeout_sec: 10 +- url_regex: '^https?://(?:www\.)?news\.com\.au/.*$' + behavior_js_template: umbraBehavior.js.j2 + default_parameters: + actions: + - selector: .menu-item a + do: mouseover - url_regex: '^https?://(?:www\.)?news\.com\.au/.*$' behavior_js_template: mouseovers.js.j2 default_parameters: - mouseover_css_selector: .menu-item a - mouseover_until_hard_timeout: False + sdo_css_selector: .menu-item a + sdo_action: mouseover + sdo_until_hard_timeout: False + request_idle_timeout_sec: 10 +- # https://webarchive.jira.com/browse/ARI-5259 + url_regex: '^https?://blog\.sina\.com\.cn/.*$' + behavior_js_template: simpleclicks.js.j2 + default_parameters: + click_css_selector: li.SG_pgnext a + click_until_hard_timeout: False + request_idle_timeout_sec: 10 +- # https://webarchive.jira.com/browse/ARI-5334 + url_regex: '^https?://(?:www\.)?google\.com/search.*$' + behavior_js_template: simpleclicks.js.j2 + default_parameters: + click_css_selector: a#pnnext + click_until_hard_timeout: True request_idle_timeout_sec: 10 - # https://webarchive.jira.com/browse/ARI-5259 url_regex: '^https?://blog\.sina\.com\.cn/.*$' @@ -135,13 +179,6 @@ click_css_selector: button#ird3-button-next click_until_hard_timeout: True request_idle_timeout_sec: 10 -- # https://webarchive.jira.com/browse/ARI-5389 - url_regex: '^https?://pitchfork\.com/.*$' - behavior_js_template: pitchfork.js -- # https://webarchive.jira.com/browse/ARI-5379 - url_regex: '^https?://(?:www\.)?pm\.gc\.ca/.*$' - behavior_js_template: pm-ca.js - request_idle_timeout_sec: 10 - # https://webarchive.jira.com/browse/ARI-4960 url_regex: '^https?://(?:www\.)?fortstjames.ca/community-events-calendar/$' behavior_js_template: simpleclicks.js.j2 @@ -149,7 +186,16 @@ click_css_selector: img#navForward1 click_until_hard_timeout: True request_idle_timeout_sec: 10 +- # https://webarchive.jira.com/browse/ARI-5210 + url_regex: '^https?://(?:www\.)?ssab.gov/Our-Work$' + behavior_js_template: simpleclicks.js.j2 + default_parameters: + click_css_selector: input[type=button][value=Next] + click_until_hard_timeout: False + request_idle_timeout_sec: 10 - # default fallback behavior url_regex: '^.*$' - request_idle_timeout_sec: 10 - behavior_js_template: default.js + behavior_js_template: umbraBehavior.js.j2 + default_parameters: + actions: + - selector: button.sc-button-play, button.playButton, div.soundItem