Git-Mirrors/brozzler
1
0
Fork 0
mirror of https://github.com/internetarchive/brozzler.git synced 2025-08-11 07:50:24 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix behaviors.yaml, switch simpledo behavior to jinja2 template

Browse source
This commit is contained in:
Noah Levitt 2017-01-06 17:19:00 -08:00 • committed by Barbara Miller
parent 3851daa27d
commit 01f5343b48
2 changed files with 88 additions and 88 deletions
Download patch file Download diff file Expand all files Collapse all files

167
brozzler/behaviors.yaml
View file

@ -17,85 +17,88 @@
# #
# first matched behavior is used, so order matters here # first matched behavior is used, so order matters here
behaviors: -
- url_regex: '^https?://(?:www\.)?facebook\.com/.*$'
url_regex: '^https?://(?:www\.)?facebook\.com/.*$' behavior_js_template: facebook.js
behavior_js_template: facebook.js.template request_idle_timeout_sec: 30
# default_parameters: -
# parameter_username: jdoe@example.com url_regex: '^https?://(?:www\.)?marquette\.edu/.*$'
# parameter_password: abcd1234 behavior_js_template: marquette_edu.js
request_idle_timeout_sec: 30 request_idle_timeout_sec: 10
- -
url_regex: '^https?://(?:www\.)?marquette\.edu/.*$' url_regex: '^https?://(?:www\.)?vimeo\.com/.*$'
behavior_js: marquette_edu.js behavior_js_template: vimeo.js
request_idle_timeout_sec: 10 request_idle_timeout_sec: 10
- -
url_regex: '^https?://(?:www\.)?vimeo\.com/.*$' url_regex: '^https?://(?:www\.)?psu24.psu.edu/.*$'
behavior_js: vimeo.js behavior_js_template: psu24.js
request_idle_timeout_sec: 10 request_idle_timeout_sec: 10
- -
url_regex: '^https?://(?:www\.)?psu24.psu.edu/.*$' url_regex: '^https?://(?:www\.)?instagram\.com/.*$'
behavior_js: psu24.js behavior_js_template: instagram.js
request_idle_timeout_sec: 10 request_idle_timeout_sec: 10
- -
url_regex: '^https?://(?:www\.)?instagram\.com/.*$' url_regex: '^https?://(?:www\.)?brooklynmuseum\.org/exhibitions/.*$'
behavior_js: instagram.js behavior_js_template: simpleclicks.js.j2
request_idle_timeout_sec: 10 default_parameters:
- click_css_selector: img.img-responsive
url_regex: '^https?://(?:www\.)?brooklynmuseum\.org/exhibitions/.*$' click_until_hard_timeout: False
behavior_js_template: simpleclicks.js.template request_idle_timeout_sec: 10
default_parameters: - # acalog https://webarchive.jira.com/browse/ARI-3775
click_css_selector: img.img-responsive url_regex: '^https?://.*[?&]catoid=[^?]*$'
request_idle_timeout_sec: 10 behavior_js_template: simpleclicks.js.j2
- # acalog https://webarchive.jira.com/browse/ARI-3775 default_parameters:
url_regex: '^https?://.*[?&]catoid=[^?]*$' click_css_selector: a[onclick]
behavior_js_template: simpleclicks.js.template click_until_hard_timeout: False
default_parameters: request_idle_timeout_sec: 10
click_css_selector: a[onclick] - # https://webarchive.jira.com/browse/ARI-3956
request_idle_timeout_sec: 10 url_regex: '^https?://(?:www\.)?usask.ca/.*$'
- # https://webarchive.jira.com/browse/ARI-3956 behavior_js_template: simpleclicks.js.j2
url_regex: '^https?://(?:www\.)?usask.ca/.*$' default_parameters:
behavior_js_template: simpleclicks.js.template click_css_selector: a[id='feature-next']
default_parameters: click_until_hard_timeout: False
click_css_selector: a[id='feature-next'] request_idle_timeout_sec: 10
request_idle_timeout_sec: 10 - # https://webarchive.jira.com/browse/AITFIVE-451
- # https://webarchive.jira.com/browse/AITFIVE-451 url_regex: '^https?://(?:www\.)?soundcloud.com/.*$'
url_regex: '^https?://(?:www\.)?soundcloud.com/.*$' behavior_js_template: simpledo.js.j2
behavior_js_template: simpledo.js.template default_parameters:
default_parameters: sdo_css_selector: button.sc-button-play, button.playButton
sdo_css_selector: button.sc-button-play, button.playButton sdo_action: click
sdo_action: click sdo_until_hard_timeout: False
request_idle_timeout_sec: 10 request_idle_timeout_sec: 10
- # https://webarchive.jira.com/browse/AITFIVE-463 - # https://webarchive.jira.com/browse/AITFIVE-463
url_regex: '^https?://(?:www\.)?christophercerrone.com/.*$' url_regex: '^https?://(?:www\.)?christophercerrone.com/.*$'
behavior_js_template: simpleclicks.js.template behavior_js_template: simpleclicks.js.j2
default_parameters: default_parameters:
click_css_selector: button.playButton.medium click_css_selector: button.playButton.medium
request_idle_timeout_sec: 10 click_until_hard_timeout: False
- # https://webarchive.jira.com/browse/ARI-4690 request_idle_timeout_sec: 10
url_regex: '^https?://(?:www\.)?youtube.com/.*$' - # https://webarchive.jira.com/browse/ARI-4690
behavior_js_template: simpleclicks.js.template url_regex: '^https?://(?:www\.)?youtube.com/.*$'
default_parameters: behavior_js_template: simpleclicks.js.j2
click_css_selector: span.load-more-text default_parameters:
request_idle_timeout_sec: 10 click_css_selector: span.load-more-text
- # https://webarchive.jira.com/browse/ARI-4725 click_until_hard_timeout: False
url_regex: '^https?://(?:www\.)?moma.org/.*$' request_idle_timeout_sec: 10
behavior_js_template: simpleclicks.js.template - # https://webarchive.jira.com/browse/ARI-4725
default_parameters: url_regex: '^https?://(?:www\.)?moma.org/.*$'
click_css_selector: button[data-more-results-bottom-button] behavior_js_template: simpleclicks.js.j2
click_until_hard_timeout: True default_parameters:
request_idle_timeout_sec: 10 click_css_selector: button[data-more-results-bottom-button]
- # https://webarchive.jira.com/browse/ARI-4692 click_until_hard_timeout: True
url_regex: '^https?://(?:www\.)?fec.gov/data/.*$' request_idle_timeout_sec: 10
behavior_js: fec_gov.js - # https://webarchive.jira.com/browse/ARI-4692
request_idle_timeout_sec: 10 url_regex: '^https?://(?:www\.)?fec.gov/data/.*$'
- url_regex: '^https?://(?:www\.)?news\.com\.au/.*$' behavior_js_template: fec_gov.js
behavior_js_template: simpledo.js.template request_idle_timeout_sec: 10
default_parameters: - url_regex: '^https?://(?:www\.)?news\.com\.au/.*$'
sdo_css_selector: .menu-item a behavior_js_template: simpledo.js.j2
sdo_action: mouseover default_parameters:
request_idle_timeout_sec: 10 sdo_css_selector: .menu-item a
- # default fallback behavior sdo_action: mouseover
url_regex: '^.*$' sdo_until_hard_timeout: False
request_idle_timeout_sec: 10 request_idle_timeout_sec: 10
behavior_js: default.js - # default fallback behavior
url_regex: '^.*$'
request_idle_timeout_sec: 10
behavior_js_template: default.js

9
brozzler/js-templates/simpledo.js.j2
View file

@ -26,12 +26,9 @@ var umbraBehavior = {
var didSomething = false; var didSomething = false;
var somethingLeftBelow = false; var somethingLeftBelow = false;
var somethingLeftAbove = false; var somethingLeftAbove = false;
var cssSelector = "${sdo_css_selector}"; var cssSelector = {{sdo_css_selector}};
var doAction = "${sdo_action}"; // currently supports click, mouseover var doAction = {{sdo_action}}; // currently supports click, mouseover
var doUntilTimeout = "${sdo_until_hard_timeout}"; var doUntilTimeout = {{sdo_until_hard_timeout}};
//handle Python to JavaScript boolean conversion
doUntilTimeout == "True" ? doUntilTimeout = true : doUntilTimeout = false;
var iframes = document.querySelectorAll("iframe"); var iframes = document.querySelectorAll("iframe");
var documents = Array(iframes.length + 1); var documents = Array(iframes.length + 1);