From cf6b2923812d491d8a0fdc1dbd93464d06200443 Mon Sep 17 00:00:00 2001 From: Andrei Isvoran Date: Wed, 6 Mar 2024 11:39:32 +0200 Subject: [PATCH] RED-8645 - Update RM rules --- .../resources/drools/acceptance_rules.drl | 196 +------------ .../drools/all_redact_manager_rules.drl | 207 +------------ .../resources/all_redact_manager_rules.drl | 275 ++---------------- 3 files changed, 45 insertions(+), 633 deletions(-) diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/acceptance_rules.drl b/redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/acceptance_rules.drl index d9f65c3f..76e84325 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/acceptance_rules.drl +++ b/redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/acceptance_rules.drl @@ -83,23 +83,13 @@ rule "SYN.1.0: Recommend CTL/BL laboratory that start with BL or CTL" //------------------------------------ CBI rules ------------------------------------ // Rule unit: CBI.0 -rule "CBI.0.0: Redact CBI Authors (non vertebrate Study)" +rule "CBI.0.0: Redact CBI Authors" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $entity: TextEntity(type() == "CBI_author", dictionaryEntry) then $entity.redact("CBI.0.0", "Author found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "CBI.0.1: Redact CBI Authors (vertebrate Study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $entity: TextEntity(type() == "CBI_author", dictionaryEntry) - then - $entity.redact("CBI.0.1", "Author found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - - // Rule unit: CBI.1 rule "CBI.1.0: Do not redact CBI Address (non vertebrate Study)" when @@ -300,23 +290,13 @@ rule "CBI.20.1: Redact between \"PERFORMING LABORATORY\" and \"LABORATORY PROJEC //------------------------------------ PII rules ------------------------------------ // Rule unit: PII.0 -rule "PII.0.0: Redact all PII (non vertebrate study)" +rule "PII.0.0: Redact all PII" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $pii: TextEntity(type() == "PII", dictionaryEntry) then $pii.redact("PII.0.0", "Personal Information found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "PII.0.1: Redact all PII (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $pii: TextEntity(type() == "PII", dictionaryEntry) - then - $pii.redact("PII.0.1", "Personal Information found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - - // Rule unit: PII.1 rule "PII.1.0: Redact Emails by RegEx (Non vertebrate study)" when @@ -327,20 +307,11 @@ rule "PII.1.0: Redact Emails by RegEx (Non vertebrate study)" .forEach(emailEntity -> emailEntity.redact("PII.1.0", "Found by Email Regex", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.1.1: Redact Emails by RegEx (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsString("@")) - then - entityCreationService.byRegex("\\b([A-Za-z0-9._%+\\-]+@[A-Za-z0-9.\\-]+\\.[A-Za-z\\-]{1,23}[A-Za-z])\\b", "PII", EntityType.ENTITY, 1, $section) - .forEach(emailEntity -> emailEntity.redact("PII.1.1", "Found by Email Regex", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end // Rule unit: PII.2 -rule "PII.2.0: Redact Phone and Fax by RegEx (non vertebrate study)" +rule "PII.2.0: Redact Phone and Fax by RegEx" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(containsString("Contact") || containsString("Telephone") || containsString("Phone") || @@ -356,49 +327,23 @@ rule "PII.2.0: Redact Phone and Fax by RegEx (non vertebrate study)" .forEach(contactEntity -> contactEntity.redact("PII.2.0", "Found by Phone and Fax Regex", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.2.1: Redact Phone and Fax by RegEx (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsString("Contact") || - containsString("Telephone") || - containsString("Phone") || - containsString("Ph.") || - containsString("Fax") || - containsString("Tel") || - containsString("Ter") || - containsString("Mobile") || - containsString("Fel") || - containsString("Fer")) - then - entityCreationService.byRegexIgnoreCase("\\b(contact|telephone|phone|ph\\.|fax|tel|ter|mobile|fel|fer)[a-zA-Z\\s]{0,10}[:.\\s]{0,3}([\\+\\d\\(][\\s\\d\\(\\)\\-\\/\\.]{4,100}\\d)\\b", "PII", EntityType.ENTITY, 2, $section) - .forEach(contactEntity -> contactEntity.redact("PII.2.1", "Found by Phone and Fax Regex", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end + // Rule unit: PII.3 -rule "PII.3.0: Redact telephone numbers by RegEx (Non vertebrate study)" +rule "PII.3.0: Redact telephone numbers by RegEx" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(matchesRegex("[+]\\d{1,}")) then entityCreationService.byRegex("((([+]\\d{1,3} (\\d{7,12})\\b)|([+]\\d{1,3}(\\d{3,12})\\b|[+]\\d{1,3}([ -]\\(?\\d{1,6}\\)?){2,4})|[+]\\d{1,3} ?((\\d{2,6}\\)?)([ -]\\d{2,6}){1,4}))(-\\d{1,3})?\\b)", "PII", EntityType.ENTITY, 1, $section) .forEach(entity -> entity.redact("PII.3.0", "Telephone number found by regex", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.3.1: Redact telephone numbers by RegEx (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(matchesRegex("[+]\\d{1,}")) - then - entityCreationService.byRegex("((([+]\\d{1,3} (\\d{7,12})\\b)|([+]\\d{1,3}(\\d{3,12})\\b|[+]\\d{1,3}([ -]\\(?\\d{1,6}\\)?){2,4})|[+]\\d{1,3} ?((\\d{2,6}\\)?)([ -]\\d{2,6}){1,4}))(-\\d{1,3})?\\b)", "PII", EntityType.ENTITY, 1, $section) - .forEach(entity -> entity.redact("PII.3.1", "Telephone number found by regex", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end // Rule unit: PII.4 -rule "PII.4.0: Redact line after contact information keywords (non vertebrate study)" +rule "PII.4.0: Redact line after contact information keywords" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $contactKeyword: String() from List.of("Contact point:", "Contact:", "Alternative contact:", @@ -424,39 +369,11 @@ rule "PII.4.0: Redact line after contact information keywords (non vertebrate st .forEach(contactEntity -> contactEntity.redact("PII.4.0", "Found after \"" + $contactKeyword + "\" contact keyword", "Reg (EC) No 1107/2009 Art. 63 (2e)")); end -rule "PII.4.1: Redact line after contact information keywords (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $contactKeyword: String() from List.of("Contact point:", - "Contact:", - "Alternative contact:", - "European contact:", - "No:", - "Contact:", - "Tel.:", - "Tel:", - "Telephone number:", - "Telephone No:", - "Telephone:", - "Phone No.", - "Phone:", - "Fax number:", - "Fax:", - "E-mail:", - "Email:", - "e-mail:", - "E-mail address:") - $section: Section(containsString($contactKeyword)) - then - entityCreationService.lineAfterString($contactKeyword, "PII", EntityType.ENTITY, $section) - .forEach(contactEntity -> contactEntity.redact("PII.4.1", "Found after \"" + $contactKeyword + "\" contact keyword", "Reg (EC) No 1107/2009 Art. 63 (2e)")); - end // Rule unit: PII.5 -rule "PII.5.0: Redact line after contact information keywords reduced (non vertebrate study)" +rule "PII.5.0: Redact line after contact information keywords reduced" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $contactKeyword: String() from List.of("Contact point:", "Contact:", "Alternative contact:", @@ -467,24 +384,9 @@ rule "PII.5.0: Redact line after contact information keywords reduced (non verte .forEach(contactEntity -> contactEntity.redact("PII.5.0", "Found after \"" + $contactKeyword + "\" contact keyword", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.5.1: Redact line after contact information keywords reduced (Vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $contactKeyword: String() from List.of("Contact point:", - "Contact:", - "Alternative contact:", - "European contact:") - $section: Section(containsString($contactKeyword)) - then - entityCreationService.lineAfterString($contactKeyword, "PII", EntityType.ENTITY, $section) - .forEach(contactEntity -> contactEntity.redact("PII.5.1", "Found after \"" + $contactKeyword + "\" contact keyword", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.6 -rule "PII.6.0: Redact line between contact keywords (non vertebrate study)" +rule "PII.6.0: Redact line between contact keywords" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section((containsString("No:") && containsString("Fax")) || (containsString("Contact:") && containsString("Tel"))) then Stream.concat( @@ -494,23 +396,9 @@ rule "PII.6.0: Redact line between contact keywords (non vertebrate study)" .forEach(contactEntity -> contactEntity.redact("PII.6.0", "Found between contact keywords", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.6.1: Redact line between contact keywords (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section((containsString("No:") && containsString("Fax")) || (containsString("Contact:") && containsString("Tel"))) - then - Stream.concat( - entityCreationService.betweenStrings("No:", "Fax", "PII", EntityType.ENTITY, $section), - entityCreationService.betweenStrings("Contact:", "Tel", "PII", EntityType.ENTITY, $section) - ) - .forEach(contactEntity -> contactEntity.redact("PII.6.1", "Found between contact keywords", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.7 -rule "PII.7.0: Redact contact information if applicant is found (non vertebrate study)" +rule "PII.7.0: Redact contact information if applicant is found" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(getHeadline().containsString("applicant") || getHeadline().containsString("Primary contact") || getHeadline().containsString("Alternative contact") || @@ -526,29 +414,9 @@ rule "PII.7.0: Redact contact information if applicant is found (non vertebrate .forEach(entity -> entity.redact("PII.7.0", "Applicant information was found", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.7.1: Redact contact information if applicant is found (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(getHeadline().containsString("applicant") || - getHeadline().containsString("Primary contact") || - getHeadline().containsString("Alternative contact") || - containsString("Applicant") || - containsString("Telephone number:")) - then - Stream.concat(entityCreationService.lineAfterStrings(List.of("Contact point:", "Contact:", "Alternative contact:", "European contact:", "No:", "Contact:", "Tel.:", "Tel:", "Telephone number:", - "Telephone No:", "Telephone:", "Phone No.", "Phone:", "Fax number:", "Fax:", "E-mail:", "Email:", "e-mail:", "E-mail address:"), "PII", EntityType.ENTITY, $section), - Stream.concat( - entityCreationService.betweenStrings("No:", "Fax", "PII", EntityType.ENTITY, $section), - entityCreationService.betweenStrings("Contact:", "Tel", "PII", EntityType.ENTITY, $section) - )) - .forEach(entity -> entity.redact("PII.7.1", "Applicant information was found", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.8 -rule "PII.8.0: Redact contact information if producer is found (non vertebrate study)" +rule "PII.8.0: Redact contact information if producer is found" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(containsStringIgnoreCase("producer of the plant protection") || containsStringIgnoreCase("producer of the active substance") || containsStringIgnoreCase("manufacturer of the active substance") || @@ -564,45 +432,15 @@ rule "PII.8.0: Redact contact information if producer is found (non vertebrate s .forEach(entity -> entity.redact("PII.8.0", "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)")); end -rule "PII.8.1: Redact contact information if producer is found (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsStringIgnoreCase("producer of the plant protection") || - containsStringIgnoreCase("producer of the active substance") || - containsStringIgnoreCase("manufacturer of the active substance") || - containsStringIgnoreCase("manufacturer:") || - containsStringIgnoreCase("Producer or producers of the active substance")) - then - Stream.concat(entityCreationService.lineAfterStrings(List.of("Contact point:", "Contact:", "Alternative contact:", "European contact:", "No:", "Contact:", "Tel.:", "Tel:", "Telephone number:", - "Telephone No:", "Telephone:", "Phone No.", "Phone:", "Fax number:", "Fax:", "E-mail:", "Email:", "e-mail:", "E-mail address:"), "PII", EntityType.ENTITY, $section), - Stream.concat( - entityCreationService.betweenStrings("No:", "Fax", "PII", EntityType.ENTITY, $section), - entityCreationService.betweenStrings("Contact:", "Tel", "PII", EntityType.ENTITY, $section) - )) - .forEach(entity -> entity.redact("PII.8.1", "Producer was found", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.9 -rule "PII.9.0: Redact between \"AUTHOR(S)\" and \"(STUDY) COMPLETION DATE\" (non vertebrate study)" +rule "PII.9.0: Redact between \"AUTHOR(S)\" and \"(STUDY) COMPLETION DATE\"" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $document: Document(containsStringIgnoreCase("AUTHOR(S)"), containsAnyStringIgnoreCase("COMPLETION DATE", "STUDY COMPLETION DATE")) then entityCreationService.shortestBetweenAnyStringIgnoreCase(List.of("AUTHOR(S)", "AUTHOR(S):"), List.of("COMPLETION DATE", "COMPLETION DATE:", "STUDY COMPLETION DATE", "STUDY COMPLETION DATE:"), "PII", EntityType.ENTITY, $document) .forEach(authorEntity -> authorEntity.redact("PII.9.0", "AUTHOR(S) was found", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.9.1: Redact between \"AUTHOR(S)\" and \"(STUDY) COMPLETION DATE\" (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $document: Document(containsStringIgnoreCase("AUTHOR(S)"), containsAnyStringIgnoreCase("COMPLETION DATE", "STUDY COMPLETION DATE")) - then - entityCreationService.shortestBetweenAnyStringIgnoreCase(List.of("AUTHOR(S)", "AUTHOR(S):"), List.of("COMPLETION DATE", "COMPLETION DATE:", "STUDY COMPLETION DATE", "STUDY COMPLETION DATE:"), "PII", EntityType.ENTITY, $document) - .forEach(authorEntity -> authorEntity.redact("PII.9.1", "AUTHOR(S) was found", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.10 rule "PII.10.0: Redact study director abbreviation" when @@ -656,23 +494,15 @@ rule "ETC.0.0: Purity Hint" // Rule unit: ETC.2 -rule "ETC.2.0: Redact signatures (non vertebrate study)" +rule "ETC.2.0: Redact signatures" when + not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $signature: Image(imageType == ImageType.SIGNATURE) then $signature.redact("ETC.2.0", "Signature Found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "ETC.2.1: Redact signatures (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $signature: Image(imageType == ImageType.SIGNATURE) - then - $signature.redact("ETC.2.1", "Signature Found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - - // Rule unit: ETC.3 rule "ETC.3.0: Skip logos (non vertebrate study)" when diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/all_redact_manager_rules.drl b/redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/all_redact_manager_rules.drl index 1396c534..009a5166 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/all_redact_manager_rules.drl +++ b/redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/all_redact_manager_rules.drl @@ -96,23 +96,13 @@ rule "SYN.1.0: Recommend CTL/BL laboratory that start with BL or CTL" //------------------------------------ CBI rules ------------------------------------ // Rule unit: CBI.0 -rule "CBI.0.0: Redact CBI Authors (non vertebrate Study)" +rule "CBI.0.0: Redact CBI Authors" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $entity: TextEntity(type() == "CBI_author", dictionaryEntry) then $entity.redact("CBI.0.0", "Author found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "CBI.0.1: Redact CBI Authors (vertebrate Study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $entity: TextEntity(type() == "CBI_author", dictionaryEntry) - then - $entity.redact("CBI.0.1", "Author found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - - // Rule unit: CBI.1 rule "CBI.1.0: Do not redact CBI Address (non vertebrate Study)" when @@ -702,23 +692,13 @@ rule "CBI.22.0: Redact Addresses in Reference Tables for vertebrate studies in n //------------------------------------ PII rules ------------------------------------ // Rule unit: PII.0 -rule "PII.0.0: Redact all PII (non vertebrate study)" +rule "PII.0.0: Redact all PII" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $pii: TextEntity(type() == "PII", dictionaryEntry) then $pii.redact("PII.0.0", "Personal Information found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "PII.0.1: Redact all PII (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $pii: TextEntity(type() == "PII", dictionaryEntry) - then - $pii.redact("PII.0.1", "Personal Information found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - - // Rule unit: PII.1 rule "PII.1.0: Redact Emails by RegEx (Non vertebrate study)" when @@ -729,14 +709,6 @@ rule "PII.1.0: Redact Emails by RegEx (Non vertebrate study)" .forEach(emailEntity -> emailEntity.redact("PII.1.0", "Found by Email Regex", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.1.1: Redact Emails by RegEx (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsString("@")) - then - entityCreationService.byRegex("\\b([A-Za-z0-9._%+\\-]+@[A-Za-z0-9.\\-]+\\.[A-Za-z\\-]{1,23}[A-Za-z])\\b", "PII", EntityType.ENTITY, 1, $section) - .forEach(emailEntity -> emailEntity.redact("PII.1.1", "Found by Email Regex", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end rule "PII.1.2: Redact typoed Emails with indicator" when @@ -748,9 +720,8 @@ rule "PII.1.2: Redact typoed Emails with indicator" // Rule unit: PII.2 -rule "PII.2.0: Redact Phone and Fax by RegEx (non vertebrate study)" +rule "PII.2.0: Redact Phone and Fax by RegEx" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(containsString("Contact") || containsString("Telephone") || containsString("Phone") || @@ -766,23 +737,7 @@ rule "PII.2.0: Redact Phone and Fax by RegEx (non vertebrate study)" .forEach(contactEntity -> contactEntity.redact("PII.2.0", "Found by Phone and Fax Regex", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.2.1: Redact Phone and Fax by RegEx (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsString("Contact") || - containsString("Telephone") || - containsString("Phone") || - containsString("Ph.") || - containsString("Fax") || - containsString("Tel") || - containsString("Ter") || - containsString("Mobile") || - containsString("Fel") || - containsString("Fer")) - then - entityCreationService.byRegexIgnoreCase("\\b(contact|telephone|phone|ph\\.|fax|tel|ter|mobile|fel|fer)[a-zA-Z\\s]{0,10}[:.\\s]{0,3}([\\+\\d\\(][\\s\\d\\(\\)\\-\\/\\.]{4,100}\\d)\\b", "PII", EntityType.ENTITY, 2, $section) - .forEach(contactEntity -> contactEntity.redact("PII.2.1", "Found by Phone and Fax Regex", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end + rule "PII.2.2: Redact phone numbers without indicators" when @@ -794,29 +749,19 @@ rule "PII.2.2: Redact phone numbers without indicators" // Rule unit: PII.3 -rule "PII.3.0: Redact telephone numbers by RegEx (Non vertebrate study)" +rule "PII.3.0: Redact telephone numbers by RegEx" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(matchesRegex("[+]\\d{1,}")) then entityCreationService.byRegex("((([+]\\d{1,3} (\\d{7,12})\\b)|([+]\\d{1,3}(\\d{3,12})\\b|[+]\\d{1,3}([ -]\\(?\\d{1,6}\\)?){2,4})|[+]\\d{1,3} ?((\\d{2,6}\\)?)([ -]\\d{2,6}){1,4}))(-\\d{1,3})?\\b)", "PII", EntityType.ENTITY, 1, $section) .forEach(entity -> entity.redact("PII.3.0", "Telephone number found by regex", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.3.1: Redact telephone numbers by RegEx (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(matchesRegex("[+]\\d{1,}")) - then - entityCreationService.byRegex("((([+]\\d{1,3} (\\d{7,12})\\b)|([+]\\d{1,3}(\\d{3,12})\\b|[+]\\d{1,3}([ -]\\(?\\d{1,6}\\)?){2,4})|[+]\\d{1,3} ?((\\d{2,6}\\)?)([ -]\\d{2,6}){1,4}))(-\\d{1,3})?\\b)", "PII", EntityType.ENTITY, 1, $section) - .forEach(entity -> entity.redact("PII.3.1", "Telephone number found by regex", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end // Rule unit: PII.4 -rule "PII.4.0: Redact line after contact information keywords (non vertebrate study)" +rule "PII.4.0: Redact line after contact information keywords" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $contactKeyword: String() from List.of("Contact point:", "Contact:", "Alternative contact:", @@ -842,39 +787,11 @@ rule "PII.4.0: Redact line after contact information keywords (non vertebrate st .forEach(contactEntity -> contactEntity.redact("PII.4.0", "Found after \"" + $contactKeyword + "\" contact keyword", "Reg (EC) No 1107/2009 Art. 63 (2e)")); end -rule "PII.4.1: Redact line after contact information keywords (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $contactKeyword: String() from List.of("Contact point:", - "Contact:", - "Alternative contact:", - "European contact:", - "No:", - "Contact:", - "Tel.:", - "Tel:", - "Telephone number:", - "Telephone No:", - "Telephone:", - "Phone No.", - "Phone:", - "Fax number:", - "Fax:", - "E-mail:", - "Email:", - "e-mail:", - "E-mail address:") - $section: Section(containsString($contactKeyword)) - then - entityCreationService.lineAfterString($contactKeyword, "PII", EntityType.ENTITY, $section) - .forEach(contactEntity -> contactEntity.redact("PII.4.1", "Found after \"" + $contactKeyword + "\" contact keyword", "Reg (EC) No 1107/2009 Art. 63 (2e)")); - end // Rule unit: PII.5 -rule "PII.5.0: Redact line after contact information keywords reduced (non vertebrate study)" +rule "PII.5.0: Redact line after contact information keywords reduced" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $contactKeyword: String() from List.of("Contact point:", "Contact:", "Alternative contact:", @@ -885,24 +802,9 @@ rule "PII.5.0: Redact line after contact information keywords reduced (non verte .forEach(contactEntity -> contactEntity.redact("PII.5.0", "Found after \"" + $contactKeyword + "\" contact keyword", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.5.1: Redact line after contact information keywords reduced (Vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $contactKeyword: String() from List.of("Contact point:", - "Contact:", - "Alternative contact:", - "European contact:") - $section: Section(containsString($contactKeyword)) - then - entityCreationService.lineAfterString($contactKeyword, "PII", EntityType.ENTITY, $section) - .forEach(contactEntity -> contactEntity.redact("PII.5.1", "Found after \"" + $contactKeyword + "\" contact keyword", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.6 -rule "PII.6.0: Redact line between contact keywords (non vertebrate study)" +rule "PII.6.0: Redact line between contact keywords" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section((containsString("No:") && containsString("Fax")) || (containsString("Contact:") && containsString("Tel"))) then Stream.concat( @@ -912,23 +814,9 @@ rule "PII.6.0: Redact line between contact keywords (non vertebrate study)" .forEach(contactEntity -> contactEntity.redact("PII.6.0", "Found between contact keywords", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.6.1: Redact line between contact keywords (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section((containsString("No:") && containsString("Fax")) || (containsString("Contact:") && containsString("Tel"))) - then - Stream.concat( - entityCreationService.betweenStrings("No:", "Fax", "PII", EntityType.ENTITY, $section), - entityCreationService.betweenStrings("Contact:", "Tel", "PII", EntityType.ENTITY, $section) - ) - .forEach(contactEntity -> contactEntity.redact("PII.6.1", "Found between contact keywords", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.7 -rule "PII.7.0: Redact contact information if applicant is found (non vertebrate study)" +rule "PII.7.0: Redact contact information if applicant is found" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(getHeadline().containsString("applicant") || getHeadline().containsString("Primary contact") || getHeadline().containsString("Alternative contact") || @@ -944,29 +832,9 @@ rule "PII.7.0: Redact contact information if applicant is found (non vertebrate .forEach(entity -> entity.redact("PII.7.0", "Applicant information was found", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.7.1: Redact contact information if applicant is found (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(getHeadline().containsString("applicant") || - getHeadline().containsString("Primary contact") || - getHeadline().containsString("Alternative contact") || - containsString("Applicant") || - containsString("Telephone number:")) - then - Stream.concat(entityCreationService.lineAfterStrings(List.of("Contact point:", "Contact:", "Alternative contact:", "European contact:", "No:", "Contact:", "Tel.:", "Tel:", "Telephone number:", - "Telephone No:", "Telephone:", "Phone No.", "Phone:", "Fax number:", "Fax:", "E-mail:", "Email:", "e-mail:", "E-mail address:"), "PII", EntityType.ENTITY, $section), - Stream.concat( - entityCreationService.betweenStrings("No:", "Fax", "PII", EntityType.ENTITY, $section), - entityCreationService.betweenStrings("Contact:", "Tel", "PII", EntityType.ENTITY, $section) - )) - .forEach(entity -> entity.redact("PII.7.1", "Applicant information was found", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.8 -rule "PII.8.0: Redact contact information if producer is found (non vertebrate study)" +rule "PII.8.0: Redact contact information if producer is found" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(containsStringIgnoreCase("producer of the plant protection") || containsStringIgnoreCase("producer of the active substance") || containsStringIgnoreCase("manufacturer of the active substance") || @@ -982,45 +850,15 @@ rule "PII.8.0: Redact contact information if producer is found (non vertebrate s .forEach(entity -> entity.redact("PII.8.0", "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)")); end -rule "PII.8.1: Redact contact information if producer is found (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsStringIgnoreCase("producer of the plant protection") || - containsStringIgnoreCase("producer of the active substance") || - containsStringIgnoreCase("manufacturer of the active substance") || - containsStringIgnoreCase("manufacturer:") || - containsStringIgnoreCase("Producer or producers of the active substance")) - then - Stream.concat(entityCreationService.lineAfterStrings(List.of("Contact point:", "Contact:", "Alternative contact:", "European contact:", "No:", "Contact:", "Tel.:", "Tel:", "Telephone number:", - "Telephone No:", "Telephone:", "Phone No.", "Phone:", "Fax number:", "Fax:", "E-mail:", "Email:", "e-mail:", "E-mail address:"), "PII", EntityType.ENTITY, $section), - Stream.concat( - entityCreationService.betweenStrings("No:", "Fax", "PII", EntityType.ENTITY, $section), - entityCreationService.betweenStrings("Contact:", "Tel", "PII", EntityType.ENTITY, $section) - )) - .forEach(entity -> entity.redact("PII.8.1", "Producer was found", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.9 -rule "PII.9.0: Redact between \"AUTHOR(S)\" and \"(STUDY) COMPLETION DATE\" (non vertebrate study)" +rule "PII.9.0: Redact between \"AUTHOR(S)\" and \"(STUDY) COMPLETION DATE\"" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $document: Document(containsStringIgnoreCase("AUTHOR(S)"), containsAnyStringIgnoreCase("COMPLETION DATE", "STUDY COMPLETION DATE")) then entityCreationService.shortestBetweenAnyStringIgnoreCase(List.of("AUTHOR(S)", "AUTHOR(S):"), List.of("COMPLETION DATE", "COMPLETION DATE:", "STUDY COMPLETION DATE", "STUDY COMPLETION DATE:"), "PII", EntityType.ENTITY, $document) .forEach(authorEntity -> authorEntity.redact("PII.9.0", "AUTHOR(S) was found", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.9.1: Redact between \"AUTHOR(S)\" and \"(STUDY) COMPLETION DATE\" (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $document: Document(containsStringIgnoreCase("AUTHOR(S)"), containsAnyStringIgnoreCase("COMPLETION DATE", "STUDY COMPLETION DATE")) - then - entityCreationService.shortestBetweenAnyStringIgnoreCase(List.of("AUTHOR(S)", "AUTHOR(S):"), List.of("COMPLETION DATE", "COMPLETION DATE:", "STUDY COMPLETION DATE", "STUDY COMPLETION DATE:"), "PII", EntityType.ENTITY, $document) - .forEach(authorEntity -> authorEntity.redact("PII.9.1", "AUTHOR(S) was found", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.10 rule "PII.10.0: Redact study director abbreviation" when @@ -1094,23 +932,15 @@ rule "ETC.1.0: Redact Purity" // Rule unit: ETC.2 -rule "ETC.2.0: Redact signatures (non vertebrate study)" +rule "ETC.2.0: Redact signatures" when + not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $signature: Image(imageType == ImageType.SIGNATURE) then $signature.redact("ETC.2.0", "Signature Found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "ETC.2.1: Redact signatures (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $signature: Image(imageType == ImageType.SIGNATURE) - then - $signature.redact("ETC.2.1", "Signature Found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - - // Rule unit: ETC.3 rule "ETC.3.0: Skip logos (non vertebrate study)" when @@ -1199,22 +1029,13 @@ rule "ETC.7.0: Guidelines FileAttributes" // Rule unit: ETC.8 -rule "ETC.8.0: Redact formulas (vertebrate study)" +rule "ETC.8.0: Redact formulas" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $logo: Image(imageType == ImageType.FORMULA) then $logo.redact("ETC.8.0", "Logo Found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "ETC.8.1: Redact formulas (non vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $logo: Image(imageType == ImageType.FORMULA) - then - $logo.redact("ETC.8.1", "Logo Found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - // Rule unit: ETC.9 rule "ETC.9.0: Redact skipped impurities" diff --git a/redaction-service-v1/rules-management/src/main/resources/all_redact_manager_rules.drl b/redaction-service-v1/rules-management/src/main/resources/all_redact_manager_rules.drl index 527faf09..0b078ff7 100644 --- a/redaction-service-v1/rules-management/src/main/resources/all_redact_manager_rules.drl +++ b/redaction-service-v1/rules-management/src/main/resources/all_redact_manager_rules.drl @@ -96,23 +96,13 @@ rule "SYN.1.0: Recommend CTL/BL laboratory that start with BL or CTL" //------------------------------------ CBI rules ------------------------------------ // Rule unit: CBI.0 -rule "CBI.0.0: Redact CBI Authors (non vertebrate Study)" +rule "CBI.0.0: Redact CBI Authors" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $entity: TextEntity(type() == "CBI_author", dictionaryEntry) then $entity.redact("CBI.0.0", "Author found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "CBI.0.1: Redact CBI Authors (vertebrate Study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $entity: TextEntity(type() == "CBI_author", dictionaryEntry) - then - $entity.redact("CBI.0.1", "Author found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - - // Rule unit: CBI.1 rule "CBI.1.0: Do not redact CBI Address (non vertebrate Study)" when @@ -556,10 +546,9 @@ rule "CBI.15.1: Redact CBI_author and CBI_address if row contains \"determinatio // Rule unit: CBI.16 -rule "CBI.16.0: Add CBI_author with \"et al.\" RegEx (non vertebrate study)" +rule "CBI.16.0: Add CBI_author with \"et al.\" RegEx" agenda-group "LOCAL_DICTIONARY_ADDS" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(containsString("et al.")) then entityCreationService.byRegex("\\b([A-ZÄÖÜ][^\\s\\.,]+( [A-ZÄÖÜ]{1,2}\\.?)?( ?[A-ZÄÖÜ]\\.?)?) et al\\.?", "CBI_author", EntityType.ENTITY, 1, $section) @@ -569,19 +558,6 @@ rule "CBI.16.0: Add CBI_author with \"et al.\" RegEx (non vertebrate study)" }); end -rule "CBI.16.1: Add CBI_author with \"et al.\" RegEx (vertebrate study)" - agenda-group "LOCAL_DICTIONARY_ADDS" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsString("et al.")) - then - entityCreationService.byRegex("\\b([A-ZÄÖÜ][^\\s\\.,]+( [A-ZÄÖÜ]{1,2}\\.?)?( ?[A-ZÄÖÜ]\\.?)?) et al\\.?", "CBI_author", EntityType.ENTITY, 1, $section) - .forEach(entity -> { - entity.redact("CBI.16.1", "Author found by \"et al\" regex", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - dictionary.recommendEverywhere(entity); - }); - end - // Rule unit: CBI.17 rule "CBI.17.0: Add recommendation for Addresses in Test Organism sections, without colon" @@ -635,10 +611,9 @@ rule "CBI.19.0: Expand CBI_author entities with salutation prefix" // Rule unit: CBI.20 -rule "CBI.20.0: Redact between \"PERFORMING LABORATORY\" and \"LABORATORY PROJECT ID:\" (non vertebrate study)" +rule "CBI.20.0: Redact between \"PERFORMING LABORATORY\" and \"LABORATORY PROJECT ID:\"" agenda-group "LOCAL_DICTIONARY_ADDS" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(!hasTables(), containsString("PERFORMING LABORATORY:"), containsString("LABORATORY PROJECT ID:")) then entityCreationService.betweenStrings("PERFORMING LABORATORY:", "LABORATORY PROJECT ID:", "CBI_address", EntityType.ENTITY, $section) @@ -648,23 +623,9 @@ rule "CBI.20.0: Redact between \"PERFORMING LABORATORY\" and \"LABORATORY PROJEC }); end -rule "CBI.20.1: Redact between \"PERFORMING LABORATORY\" and \"LABORATORY PROJECT ID:\" (vertebrate study)" - agenda-group "LOCAL_DICTIONARY_ADDS" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(!hasTables(), containsString("PERFORMING LABORATORY:"), containsString("LABORATORY PROJECT ID:")) - then - entityCreationService.betweenStrings("PERFORMING LABORATORY:", "LABORATORY PROJECT ID:", "CBI_address", EntityType.ENTITY, $section) - .forEach(laboratoryEntity -> { - laboratoryEntity.redact("CBI.20.1", "PERFORMING LABORATORY was found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - dictionary.recommendEverywhere(laboratoryEntity); - }); - end - // Rule unit: CBI.21 -rule "CBI.21.0: Redact short Authors section (non vertebrate study)" +rule "CBI.21.0: Redact short Authors section" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(containsAnyStringIgnoreCase("author(s)", "author", "authors"), length() < 50, getTreeId().get(0) <= 20) //TODO: evaluate the reason of this rule not TextEntity(type() == "CBI_author", engines contains Engine.NER) from $section.getEntities() then @@ -674,18 +635,6 @@ rule "CBI.21.0: Redact short Authors section (non vertebrate study)" }); end -rule "CBI.21.1: Redact short Authors section (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsAnyStringIgnoreCase("author(s)", "author", "authors"), length() < 50, getTreeId().get(0) <= 20) //TODO: evaluate the reason of this rule - not TextEntity(type() == "CBI_author", engines contains Engine.NER) from $section.getEntities() - then - entityCreationService.byRegexIgnoreCase("(?<=author\\(?s\\)?\\s\\n?)([\\p{Lu}\\p{L} ]{5,15}(,|\\n)?){1,3}", "CBI_author", EntityType.ENTITY, $section) - .forEach(entity -> { - entity.redact("CBI.21.1", "AUTHOR(S) was found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - }); - end - // Rule unit: CBI.22 rule "CBI.22.0: Redact Addresses in Reference Tables for vertebrate studies in non-vertebrate documents" @@ -702,23 +651,13 @@ rule "CBI.22.0: Redact Addresses in Reference Tables for vertebrate studies in n //------------------------------------ PII rules ------------------------------------ // Rule unit: PII.0 -rule "PII.0.0: Redact all PII (non vertebrate study)" +rule "PII.0.0: Redact all PII" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $pii: TextEntity(type() == "PII", dictionaryEntry) then $pii.redact("PII.0.0", "Personal Information found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "PII.0.1: Redact all PII (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $pii: TextEntity(type() == "PII", dictionaryEntry) - then - $pii.redact("PII.0.1", "Personal Information found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - - // Rule unit: PII.1 rule "PII.1.0: Redact Emails by RegEx (Non vertebrate study)" when @@ -729,14 +668,6 @@ rule "PII.1.0: Redact Emails by RegEx (Non vertebrate study)" .forEach(emailEntity -> emailEntity.redact("PII.1.0", "Found by Email Regex", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.1.1: Redact Emails by RegEx (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsString("@")) - then - entityCreationService.byRegex("\\b([A-Za-z0-9._%+\\-]+@[A-Za-z0-9.\\-]+\\.[A-Za-z\\-]{1,23}[A-Za-z])\\b", "PII", EntityType.ENTITY, 1, $section) - .forEach(emailEntity -> emailEntity.redact("PII.1.1", "Found by Email Regex", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end rule "PII.1.2: Redact typoed Emails with indicator" when @@ -748,9 +679,8 @@ rule "PII.1.2: Redact typoed Emails with indicator" // Rule unit: PII.2 -rule "PII.2.0: Redact Phone and Fax by RegEx (non vertebrate study)" +rule "PII.2.0: Redact Phone and Fax by RegEx" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(containsString("Contact") || containsString("Telephone") || containsString("Phone") || @@ -766,23 +696,7 @@ rule "PII.2.0: Redact Phone and Fax by RegEx (non vertebrate study)" .forEach(contactEntity -> contactEntity.redact("PII.2.0", "Found by Phone and Fax Regex", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.2.1: Redact Phone and Fax by RegEx (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsString("Contact") || - containsString("Telephone") || - containsString("Phone") || - containsString("Ph.") || - containsString("Fax") || - containsString("Tel") || - containsString("Ter") || - containsString("Mobile") || - containsString("Fel") || - containsString("Fer")) - then - entityCreationService.byRegexIgnoreCase("\\b(contact|telephone|phone|ph\\.|fax|tel|ter|mobile|fel|fer)[a-zA-Z\\s]{0,10}[:.\\s]{0,3}([\\+\\d\\(][\\s\\d\\(\\)\\-\\/\\.]{4,100}\\d)\\b", "PII", EntityType.ENTITY, 2, $section) - .forEach(contactEntity -> contactEntity.redact("PII.2.1", "Found by Phone and Fax Regex", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end + rule "PII.2.2: Redact phone numbers without indicators" when @@ -794,28 +708,18 @@ rule "PII.2.2: Redact phone numbers without indicators" // Rule unit: PII.3 -rule "PII.3.0: Redact telephone numbers by RegEx (Non vertebrate study)" +rule "PII.3.0: Redact telephone numbers by RegEx" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(matchesRegex("[+]\\d{1,}")) then entityCreationService.byRegex("((([+]\\d{1,3} (\\d{7,12})\\b)|([+]\\d{1,3}(\\d{3,12})\\b|[+]\\d{1,3}([ -]\\(?\\d{1,6}\\)?){2,4})|[+]\\d{1,3} ?((\\d{2,6}\\)?)([ -]\\d{2,6}){1,4}))(-\\d{1,3})?\\b)", "PII", EntityType.ENTITY, 1, $section) .forEach(entity -> entity.redact("PII.3.0", "Telephone number found by regex", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.3.1: Redact telephone numbers by RegEx (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(matchesRegex("[+]\\d{1,}")) - then - entityCreationService.byRegex("((([+]\\d{1,3} (\\d{7,12})\\b)|([+]\\d{1,3}(\\d{3,12})\\b|[+]\\d{1,3}([ -]\\(?\\d{1,6}\\)?){2,4})|[+]\\d{1,3} ?((\\d{2,6}\\)?)([ -]\\d{2,6}){1,4}))(-\\d{1,3})?\\b)", "PII", EntityType.ENTITY, 1, $section) - .forEach(entity -> entity.redact("PII.3.1", "Telephone number found by regex", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end // Rule unit: PII.4 -rule "PII.4.0: Redact line after contact information keywords (non vertebrate study)" +rule "PII.4.0: Redact line after contact information keywords" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $contactKeyword: String() from List.of("Contact point:", "Contact:", "Alternative contact:", @@ -841,39 +745,11 @@ rule "PII.4.0: Redact line after contact information keywords (non vertebrate st .forEach(contactEntity -> contactEntity.redact("PII.4.0", "Found after \"" + $contactKeyword + "\" contact keyword", "Reg (EC) No 1107/2009 Art. 63 (2e)")); end -rule "PII.4.1: Redact line after contact information keywords (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $contactKeyword: String() from List.of("Contact point:", - "Contact:", - "Alternative contact:", - "European contact:", - "No:", - "Contact:", - "Tel.:", - "Tel:", - "Telephone number:", - "Telephone No:", - "Telephone:", - "Phone No.", - "Phone:", - "Fax number:", - "Fax:", - "E-mail:", - "Email:", - "e-mail:", - "E-mail address:") - $section: Section(containsString($contactKeyword)) - then - entityCreationService.lineAfterString($contactKeyword, "PII", EntityType.ENTITY, $section) - .forEach(contactEntity -> contactEntity.redact("PII.4.1", "Found after \"" + $contactKeyword + "\" contact keyword", "Reg (EC) No 1107/2009 Art. 63 (2e)")); - end // Rule unit: PII.5 -rule "PII.5.0: Redact line after contact information keywords reduced (non vertebrate study)" +rule "PII.5.0: Redact line after contact information keywords reduced" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $contactKeyword: String() from List.of("Contact point:", "Contact:", "Alternative contact:", @@ -884,24 +760,9 @@ rule "PII.5.0: Redact line after contact information keywords reduced (non verte .forEach(contactEntity -> contactEntity.redact("PII.5.0", "Found after \"" + $contactKeyword + "\" contact keyword", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.5.1: Redact line after contact information keywords reduced (Vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $contactKeyword: String() from List.of("Contact point:", - "Contact:", - "Alternative contact:", - "European contact:") - $section: Section(containsString($contactKeyword)) - then - entityCreationService.lineAfterString($contactKeyword, "PII", EntityType.ENTITY, $section) - .forEach(contactEntity -> contactEntity.redact("PII.5.1", "Found after \"" + $contactKeyword + "\" contact keyword", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.6 -rule "PII.6.0: Redact line between contact keywords (non vertebrate study)" +rule "PII.6.0: Redact line between contact keywords" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section((containsString("No:") && containsString("Fax")) || (containsString("Contact:") && containsString("Tel"))) then Stream.concat( @@ -911,23 +772,9 @@ rule "PII.6.0: Redact line between contact keywords (non vertebrate study)" .forEach(contactEntity -> contactEntity.redact("PII.6.0", "Found between contact keywords", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.6.1: Redact line between contact keywords (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section((containsString("No:") && containsString("Fax")) || (containsString("Contact:") && containsString("Tel"))) - then - Stream.concat( - entityCreationService.betweenStrings("No:", "Fax", "PII", EntityType.ENTITY, $section), - entityCreationService.betweenStrings("Contact:", "Tel", "PII", EntityType.ENTITY, $section) - ) - .forEach(contactEntity -> contactEntity.redact("PII.6.1", "Found between contact keywords", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.7 -rule "PII.7.0: Redact contact information if applicant is found (non vertebrate study)" +rule "PII.7.0: Redact contact information if applicant is found" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(getHeadline().containsString("applicant") || getHeadline().containsString("Primary contact") || getHeadline().containsString("Alternative contact") || @@ -943,29 +790,9 @@ rule "PII.7.0: Redact contact information if applicant is found (non vertebrate .forEach(entity -> entity.redact("PII.7.0", "Applicant information was found", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.7.1: Redact contact information if applicant is found (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(getHeadline().containsString("applicant") || - getHeadline().containsString("Primary contact") || - getHeadline().containsString("Alternative contact") || - containsString("Applicant") || - containsString("Telephone number:")) - then - Stream.concat(entityCreationService.lineAfterStrings(List.of("Contact point:", "Contact:", "Alternative contact:", "European contact:", "No:", "Contact:", "Tel.:", "Tel:", "Telephone number:", - "Telephone No:", "Telephone:", "Phone No.", "Phone:", "Fax number:", "Fax:", "E-mail:", "Email:", "e-mail:", "E-mail address:"), "PII", EntityType.ENTITY, $section), - Stream.concat( - entityCreationService.betweenStrings("No:", "Fax", "PII", EntityType.ENTITY, $section), - entityCreationService.betweenStrings("Contact:", "Tel", "PII", EntityType.ENTITY, $section) - )) - .forEach(entity -> entity.redact("PII.7.1", "Applicant information was found", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.8 -rule "PII.8.0: Redact contact information if producer is found (non vertebrate study)" +rule "PII.8.0: Redact contact information if producer is found" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $section: Section(containsStringIgnoreCase("producer of the plant protection") || containsStringIgnoreCase("producer of the active substance") || containsStringIgnoreCase("manufacturer of the active substance") || @@ -981,45 +808,15 @@ rule "PII.8.0: Redact contact information if producer is found (non vertebrate s .forEach(entity -> entity.redact("PII.8.0", "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)")); end -rule "PII.8.1: Redact contact information if producer is found (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $section: Section(containsStringIgnoreCase("producer of the plant protection") || - containsStringIgnoreCase("producer of the active substance") || - containsStringIgnoreCase("manufacturer of the active substance") || - containsStringIgnoreCase("manufacturer:") || - containsStringIgnoreCase("Producer or producers of the active substance")) - then - Stream.concat(entityCreationService.lineAfterStrings(List.of("Contact point:", "Contact:", "Alternative contact:", "European contact:", "No:", "Contact:", "Tel.:", "Tel:", "Telephone number:", - "Telephone No:", "Telephone:", "Phone No.", "Phone:", "Fax number:", "Fax:", "E-mail:", "Email:", "e-mail:", "E-mail address:"), "PII", EntityType.ENTITY, $section), - Stream.concat( - entityCreationService.betweenStrings("No:", "Fax", "PII", EntityType.ENTITY, $section), - entityCreationService.betweenStrings("Contact:", "Tel", "PII", EntityType.ENTITY, $section) - )) - .forEach(entity -> entity.redact("PII.8.1", "Producer was found", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.9 -rule "PII.9.0: Redact between \"AUTHOR(S)\" and \"(STUDY) COMPLETION DATE\" (non vertebrate study)" +rule "PII.9.0: Redact between \"AUTHOR(S)\" and \"(STUDY) COMPLETION DATE\"" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $document: Document(containsStringIgnoreCase("AUTHOR(S)"), containsAnyStringIgnoreCase("COMPLETION DATE", "STUDY COMPLETION DATE")) then entityCreationService.shortestBetweenAnyStringIgnoreCase(List.of("AUTHOR(S)", "AUTHOR(S):"), List.of("COMPLETION DATE", "COMPLETION DATE:", "STUDY COMPLETION DATE", "STUDY COMPLETION DATE:"), "PII", EntityType.ENTITY, $document) .forEach(authorEntity -> authorEntity.redact("PII.9.0", "AUTHOR(S) was found", "Article 39(e)(3) of Regulation (EC) No 178/2002")); end -rule "PII.9.1: Redact between \"AUTHOR(S)\" and \"(STUDY) COMPLETION DATE\" (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $document: Document(containsStringIgnoreCase("AUTHOR(S)"), containsAnyStringIgnoreCase("COMPLETION DATE", "STUDY COMPLETION DATE")) - then - entityCreationService.shortestBetweenAnyStringIgnoreCase(List.of("AUTHOR(S)", "AUTHOR(S):"), List.of("COMPLETION DATE", "COMPLETION DATE:", "STUDY COMPLETION DATE", "STUDY COMPLETION DATE:"), "PII", EntityType.ENTITY, $document) - .forEach(authorEntity -> authorEntity.redact("PII.9.1", "AUTHOR(S) was found", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - - // Rule unit: PII.10 rule "PII.10.0: Redact study director abbreviation" when @@ -1051,16 +848,6 @@ rule "PII.12.0: Expand PII entities with salutation prefix" end -// Rule unit: PII.12 -rule "PII.12.1: Expand PII entities with salutation prefix" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $entityToExpand: TextEntity(type() == "PII", anyMatch(textBefore, "\\b(Mrs?|Ms|Miss|Sir|Madame?|Mme)\\s?\\.?\\s*")) - then - entityCreationService.byPrefixExpansionRegex($entityToExpand, "\\b(Mrs?|Ms|Miss|Sir|Madame?|Mme)\\s?\\.?\\s*") - .ifPresent(expandedEntity -> expandedEntity.apply("PII.12.1", "Expanded PII with salutation prefix", "Article 39(e)(2) of Regulation (EC) No 178/2002")); - end - // Rule unit: PII.13 rule "PII.13.0: Add recommendation for PII after Contact Person" when @@ -1103,23 +890,15 @@ rule "ETC.1.0: Redact Purity" // Rule unit: ETC.2 -rule "ETC.2.0: Redact signatures (non vertebrate study)" +rule "ETC.2.0: Redact signatures" when + not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $signature: Image(imageType == ImageType.SIGNATURE) then $signature.redact("ETC.2.0", "Signature Found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "ETC.2.1: Redact signatures (vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $signature: Image(imageType == ImageType.SIGNATURE) - then - $signature.redact("ETC.2.1", "Signature Found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - - // Rule unit: ETC.3 rule "ETC.3.0: Skip logos (non vertebrate study)" when @@ -1208,22 +987,13 @@ rule "ETC.7.0: Guidelines FileAttributes" // Rule unit: ETC.8 -rule "ETC.8.0: Redact formulas (vertebrate study)" +rule "ETC.8.0: Redact formulas" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $logo: Image(imageType == ImageType.FORMULA) then $logo.redact("ETC.8.0", "Logo Found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "ETC.8.1: Redact formulas (non vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $logo: Image(imageType == ImageType.FORMULA) - then - $logo.redact("ETC.8.1", "Logo Found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - // Rule unit: ETC.9 rule "ETC.9.0: Redact skipped impurities" when @@ -1262,22 +1032,13 @@ rule "ETC.11.0: Recommend first line in table cell with name and address of owne // Rule unit: ETC.12 -rule "ETC.12.0: Redact dossier_redaction (Non vertebrate study)" +rule "ETC.12.0: Redact dossier_redaction" when - not FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") $dossierRedaction: TextEntity(type() == "dossier_redaction") then $dossierRedaction.redact("ETC.12.0", "Dossier dictionary entry found", "Article 39(e)(3) of Regulation (EC) No 178/2002"); end -rule "ETC.12.1: Redact dossier_redaction (Vertebrate study)" - when - FileAttribute(label == "Vertebrate Study", value soundslike "Yes" || value.toLowerCase() == "y") - $dossierRedaction: TextEntity(type() == "dossier_redaction") - then - $dossierRedaction.redact("ETC.12.1", "Dossier dictionary entry found", "Article 39(e)(2) of Regulation (EC) No 178/2002"); - end - //------------------------------------ AI rules ------------------------------------ // Rule unit: AI.0