diff --git a/src/data/.medical_procedures.csv.swp b/src/data/.medical_procedures.csv.swp new file mode 100644 index 0000000..661751b Binary files /dev/null and b/src/data/.medical_procedures.csv.swp differ diff --git a/src/data/database b/src/data/database index 8062965..5c76342 100644 Binary files a/src/data/database and b/src/data/database differ diff --git a/src/data/diseases.csv b/src/data/diseases.csv new file mode 100644 index 0000000..8709165 --- /dev/null +++ b/src/data/diseases.csv @@ -0,0 +1,108 @@ +AIDS +alkhurma +anaplasmosis +anthrax +arenavirus +babesiosis +bordetella +pertussis +borreliosis +botulism +brucellosis +campylobacteriosis +chickenpox +varicella +chikungunya +chlamydia +cholera +ciguatera +clostridium +rubella +syphilis +coronavirus +covid +cowpox +coxsackievirus +cjd +cchf +cryptosporidiosis +cutaneous warts +dengue +dhf +diphtheria +ebola +echinococcosis +enteric +ehec +enterovirus +typhus +ecoli +flu +measles +rubella +giardiasis +gonorrhoea +haemophilus +haemorrhagic +haemorrhagic +hantavirus +hepatitis +hiv +hpv +haemophilus +meningococcal +pneumococcal +encephalitis +lassa +legionnaires’ +leishmaniasis +leptospirosis +listeriosis +lyme +borreliosis +lymphogranuloma +malaria +meningococcal +mpox +monkeypox +mumps +nephropathia +nipah +norovirus +paratyphoid +pertussis +piroplasmosis +plague +pneumococcal +poliomyelitis +rabies +rickettsiosis +rotavirus +rubella +salmonellosis +sars-cov-2 +schmallenberg +sbv +sars +shigellosis +sindbis +smallpox +streptococcus +syphilis +tetanus +encephalitis +tbe +tbrf +toscana +toxoplasmosis +trichinellosis +tuberculosis +b +tularaemia +typhoid +paratyphoid +vcjd +varicella +hepatitis +yersiniosis +zoonosis diff --git a/src/data/medical_procedures.csv b/src/data/medical_procedures.csv new file mode 100644 index 0000000..9f1ef23 --- /dev/null +++ b/src/data/medical_procedures.csv @@ -0,0 +1,74 @@ +mastectomy +biopsy +cesarean +abortion +miscarriage +hysterectomy +hysteroscopy +colectomy +prostatectomy +colonoscopy +cystoscopy +electrocardiography +electrocorticography +electroencephalography +electromyography +electroneuronography +electronystagmography +electrooculography +electroretinography +endoscopy +colonoscopy +colposcopy +cystoscopy +gastroscopy +laparoscopy +laryngoscopy +ophthalmoscopy +otoscopy +sigmoidoscopy +magnetoencephalography +angiography +aortography +lymphangiography +ventriculography +computed tomography +echocardiography +fluoroscopy +radiography +scintillography +spect +ultrasonography +thermography +neuroimaging +posturography +politzerization +hemodialysis +hemofiltration +plasmapheresis +apheresis +immunotherapy +chemotherapy +phototerapy +hydrotherapy +electrotherapy +psychotherapy +acupuncture +antivenom +intubation +nebulization +ablation +amputation +biopsy +cryosurgery +hemilaminectomy +image-guided surgery +laminectomy +lithotomy +lithotriptor +lobotomy +neovaginoplasty +radiosurgery +radiosurgery +vaginoplasty +xenotransplantation diff --git a/src/getNewPrompt.mjs b/src/getNewPrompt.mjs index 72b18bb..e197295 100644 --- a/src/getNewPrompt.mjs +++ b/src/getNewPrompt.mjs @@ -1,7 +1,11 @@ import { db } from "./db.mjs" -const badWords = db('bad_words') - .select('word') +const blocklist = db.union([ + db('bad_words').select('word'), + db('medical_procedures').select('name'), + db('diseases').select('name') +]) + export default async function getNewPrompt({ minCount = 200000, maxCount = 30000000, rarityBias = 0.5 }) { @@ -14,7 +18,7 @@ export default async function getNewPrompt({ minCount = 200000, maxCount = 30000 }) .andWhere('count', '<', maxCount) .andWhere('count', '>', minCount) - .andWhere('word', 'not in', badWords) + .andWhere('word', 'not in', blocklist) .whereNotNull('pronunciation') .orderByRaw('count desc') @@ -42,5 +46,5 @@ export default async function getNewPrompt({ minCount = 200000, maxCount = 30000 //console.dir(await getNewPrompt({})) - +console.log(await blocklist)