From 65e782d608b7686409a4dcf5b9a8861cd85d3e85 Mon Sep 17 00:00:00 2001 From: Arpit Jain <3242828+arpitjain099@users.noreply.github.com> Date: Tue, 20 May 2025 04:32:43 +0000 Subject: [PATCH 1/2] Fix some more typos --- experiments/execution-context-destroyed/over_time.csv | 2 +- experiments/execution-context-destroyed/template_script.py | 2 +- report_generators/target_url_list.py | 2 +- tests/test_snapshot_old.csv | 2 +- 4 files changed, 4 insertions(+), 4 deletions(-) diff --git a/experiments/execution-context-destroyed/over_time.csv b/experiments/execution-context-destroyed/over_time.csv index b2011522..caf6c4e3 100644 --- a/experiments/execution-context-destroyed/over_time.csv +++ b/experiments/execution-context-destroyed/over_time.csv @@ -7,7 +7,7 @@ wellnessnews.ors.nih.gov,execution_context_destroyed,execution_context_destroyed kurage.nimh.nih.gov,execution_context_destroyed,execution_context_destroyed,completed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed www.its.gov,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,completed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed listserv.osc.gov,execution_context_destroyed,execution_context_destroyed,completed,completed,execution_context_destroyed,completed,completed,completed,completed -nees.nih.gov,execution_context_destroyed,unknown_error,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,completed,execution_context_destroyed +needs.nih.gov,execution_context_destroyed,unknown_error,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,completed,execution_context_destroyed awsdemo.grantsolutions.gov,execution_context_destroyed,unknown_error,execution_context_destroyed,execution_context_destroyed,unknown_error,unknown_error,execution_context_destroyed,execution_context_destroyed,completed cdxapps.epa.gov,execution_context_destroyed,execution_context_destroyed,completed,completed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,completed,execution_context_destroyed srs.ntis.gov,execution_context_destroyed,completed,execution_context_destroyed,unknown_error,completed,completed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed diff --git a/experiments/execution-context-destroyed/template_script.py b/experiments/execution-context-destroyed/template_script.py index aef18ae8..f50296f0 100644 --- a/experiments/execution-context-destroyed/template_script.py +++ b/experiments/execution-context-destroyed/template_script.py @@ -8,7 +8,7 @@ def get_df(db_name: str): """ This function could be set up to pull data from a local copy of the postgres - database. It could also be reworked to pull from the publically hosted snapshot CSV + database. It could also be reworked to pull from the publicly hosted snapshot CSV if we move from weekly to daily snapshot generation. """ pass diff --git a/report_generators/target_url_list.py b/report_generators/target_url_list.py index 9308c916..7833cd55 100644 --- a/report_generators/target_url_list.py +++ b/report_generators/target_url_list.py @@ -16,7 +16,7 @@ def generate_report(self): 'How many urls are sourced from DAP?': self.num_true('source_list_dap'), 'How many urls are sourced from the other websites list?': self.num_true('source_list_other'), 'How many blank cells are there in the target URL list?': self.num_blank(), - 'How many blank cells are there in the target URL list (exluding omb_idea_public column)?': self.num_blank_without_omb_idea_public(), + 'How many blank cells are there in the target URL list (excluding omb_idea_public column)?': self.num_blank_without_omb_idea_public(), 'How many urls are on the omb_idea_public list?': self.num_omb_idea_public(), 'How many omb_idea_public urls are flagged as public?': self.num_true('omb_idea_public'), 'How many omb_idea_public urls are not flagged as public?': self.num_false('omb_idea_public'), diff --git a/tests/test_snapshot_old.csv b/tests/test_snapshot_old.csv index 998479bc..3f70613f 100644 --- a/tests/test_snapshot_old.csv +++ b/tests/test_snapshot_old.csv @@ -1,2 +1,2 @@ "target_url","target_url_domain","target_url_top_level_domain","target_url_redirects","final_url","final_url_domain","final_url_top_level_domain","final_url_website","final_url_live","final_url_status_code","final_url_media_type","final_url_same_domain","final_url_same_website","target_url_agency_owner","target_url_bureau_owner","target_url_branch","target_url_404_test","source_list","scan_date","primary_scan_status","dns_scan_status","not_found_scan_status","robots_txt_scan_status","sitemap_xml_scan_status","ipv6","hostname","cms","cloud_dot_gov_pages","login_provider","login","site_search","search_dot_gov","dap","dap_parameters","third_party_service_domains","third_party_service_count","cookie_domains","viewport_meta_tag","required_links_url","required_links_text","og_title","og_description","og_article_published","og_article_modified","canonical_link","main_element_present","robots_txt_detected","robots_txt_target_url_redirects","robots_txt_final_url","robots_txt_final_url_live","robots_txt_final_url_status_code","robots_txt_final_url_media_type","robots_txt_final_url_filesize","robots_txt_crawl_delay","robots_txt_sitemap_locations","sitemap_xml_detected","sitemap_xml_target_url_redirects","sitemap_xml_final_url","sitemap_xml_final_url_live","sitemap_xml_final_url_status_code","sitemap_xml_final_url_media_type","sitemap_xml_final_url_filesize","sitemap_xml_count","sitemap_xml_pdf_count","uswds_favicon","uswds_favicon_in_css","uswds_publicsans_font","uswds_inline_css","uswds_usa_classes","uswds_string","uswds_string_in_css","uswds_semantic_version","uswds_version","uswds_count" -"test.gov","test.gov","gov",true,"https://www.test.gov/index.html","test.gov","gov","test.gov",true,200,"text/html",true,true,"Department of Testing","","Executive","[gov]","2022-10-19T12:37:31.492Z","timeout","completed","completed","completed","completed",false,false,true,false,0,0,0,26,5,0,1,20,,0,,,"HHS Accessibility & Section 508","HHS Accessibility & Section 508",true,"https://www.hhs.gov/web/508/robots.txt",false,false,true,404,"text/html",,,,"https://www.hhs.gov/web/508/sitemap.xml",false,false,true,404,"text/html",,,,"[""6282609.global.r2.siteimproveanalytics.io"",""ajax.googleapis.com"",""ampcid.google.com"",""analytics.google.com"",""dap.digitalgov.gov"",""gov1.siteintercept.qualtrics.com"",""pro.ip-api.com"",""script.crazyegg.com"",""siteimproveanalytics.com"",""stats.g.doubleclick.net"",""unpkg.com"",""www.google-analytics.com"",""www.google.com"",""www.googletagmanager.com"",""www.youtube.com"",""zn8nyvgbt59q9merx-hhsgovfedramp.gov1.siteintercept.qualtrics.com""]",16,,true,,,,,,,,,,,"hhs",,"true","24","true",,,,,,,,"dimension1|1",,"dimension2|2","dimension5|5","dimension3|3","true","UA-36351725-9","hhs.gov","drupal 9 ","health","sitesearchtracking,q",,,"aspa",,,,,,"true" +"test.gov","test.gov","gov",true,"https://www.test.gov/index.html","test.gov","gov","test.gov",true,200,"text/html",true,true,"Department of Testing","","Executive","[gov]","2022-10-19T12:37:31.492Z","timeout","completed","completed","completed","completed",false,false,true,false,0,0,0,26,5,0,1,20,,0,,,"HHS Accessibility & Section 508","HHS Accessibility & Section 508",true,"https://www.hhs.gov/web/508/robots.txt",false,false,true,404,"text/html",,,,"https://www.hhs.gov/web/508/sitemap.xml",false,false,true,404,"text/html",,,,"[""6282609.global.r2.siteimproveanalytics.io"",""ajax.googleapis.com"",""ampcid.google.com"",""analytics.google.com"",""dap.digitalgov.gov"",""gov1.siteintercept.qualtrics.com"",""pro.ip-api.com"",""script.crazyegg.com"",""siteimproveanalytics.com"",""stats.g.double-click.net"",""unpkg.com"",""www.google-analytics.com"",""www.google.com"",""www.googletagmanager.com"",""www.youtube.com"",""zn8nyvgbt59q9merx-hhsgovfedramp.gov1.siteintercept.qualtrics.com""]",16,,true,,,,,,,,,,,"hhs",,"true","24","true",,,,,,,,"dimension1|1",,"dimension2|2","dimension5|5","dimension3|3","true","UA-36351725-9","hhs.gov","drupal 9 ","health","sitesearchtracking,q",,,"aspa",,,,,,"true" From 2d25ee7b58882c5279b5bdeaa1842cd1a7eb678c Mon Sep 17 00:00:00 2001 From: Arpit Jain <3242828+arpitjain099@users.noreply.github.com> Date: Tue, 20 May 2025 04:35:48 +0000 Subject: [PATCH 2/2] revert wrong change --- experiments/execution-context-destroyed/over_time.csv | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/experiments/execution-context-destroyed/over_time.csv b/experiments/execution-context-destroyed/over_time.csv index caf6c4e3..b2011522 100644 --- a/experiments/execution-context-destroyed/over_time.csv +++ b/experiments/execution-context-destroyed/over_time.csv @@ -7,7 +7,7 @@ wellnessnews.ors.nih.gov,execution_context_destroyed,execution_context_destroyed kurage.nimh.nih.gov,execution_context_destroyed,execution_context_destroyed,completed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed www.its.gov,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,completed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed listserv.osc.gov,execution_context_destroyed,execution_context_destroyed,completed,completed,execution_context_destroyed,completed,completed,completed,completed -needs.nih.gov,execution_context_destroyed,unknown_error,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,completed,execution_context_destroyed +nees.nih.gov,execution_context_destroyed,unknown_error,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,completed,execution_context_destroyed awsdemo.grantsolutions.gov,execution_context_destroyed,unknown_error,execution_context_destroyed,execution_context_destroyed,unknown_error,unknown_error,execution_context_destroyed,execution_context_destroyed,completed cdxapps.epa.gov,execution_context_destroyed,execution_context_destroyed,completed,completed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed,completed,execution_context_destroyed srs.ntis.gov,execution_context_destroyed,completed,execution_context_destroyed,unknown_error,completed,completed,execution_context_destroyed,execution_context_destroyed,execution_context_destroyed