add

users-events/cf.py

@@ -0,0 +1,62 @@
+# /// script
+# dependencies = [
+#   "cloudflare"
+# ]
+# ///
+
+from cloudflare import Cloudflare
+
+CLOUDFLARE_ACCOUNT_ID = '5436b62470020c04b434ad31c3e4cf4e'
+CLOUDFLARE_API_TOKEN = 'gFndkBJCzH4pRX7mKXokdWfw1xhm8-9FHfvLfhwa'
+
+
+client = Cloudflare(api_token=CLOUDFLARE_API_TOKEN)
+
+assistant = """
+You are a data analysis assistant specialized in identifying Brazilian
+personal data from CSV files.
+
+These CSV files may or may not include headers.
+
+Your task is to analyze the content and identify only three possible
+data types: 'name', 'cpf', and 'email'.
+
+Ignore all other fields.
+"""
+
+csv_content = """
+,RICARDO GALLES BONET,ricardo.bonet@fanucamerica.com,424.430.528-93,NR-10 (RECICLAGEM)
+,RULIO SIEFERT SERA,rulio.sera@fanucamerica.com,063.916.859-08,NR-10 (RECICLAGEM)
+,MACIEL FERREIRA BOMFIM,maciel.bomfim@fanucamerica.com,334.547.088-85,NR-10 (RECICLAGEM)
+,JAIME EDUARDO GALVEZ AVILES,jaime.galvez@fanucamerica.com,280.238.818-50,NR-12
+,JAIME EDUARDO GALVEZ AVILES,jaime.galvez@fanucamerica.com,280.238.818-50,NR-35 (RECICLAGEM)
+,HIGOR MACHADO SILVA,higor.silva@fanucamerica.com,419.879.878-88,NR-12
+,LÁZARO SOUZA DIAS,lazaro.dias@fanucamerica.com,067.179.825-19,NR-12
+,JOÃO PEDRO AGUIAR GALASSO,joao.pedro@fanucamerica.com,570.403.588-40,NR-12
+"""
+
+prompt = f"""
+Here is a CSV sample:
+
+{csv_content}
+
+Your task is to:
+- Detect which columns most likely contain "name", "cpf", or "email".
+- Skip any category that is not present in the data.
+- Return ONLY a valid Python list of tuples, like:
+[('name', index), ('cpf', index), ('email', index)]
+- Use the column index that most likely matches each data type,
+based on frequency and data format.
+- Don't include explanations, code, or any additional text.
+"""
+
+r = client.ai.run(
+    model_name='@cf/meta/llama-3-8b-instruct',
+    account_id=CLOUDFLARE_ACCOUNT_ID,
+    messages=[
+        {'role': 'system', 'content': assistant},
+        {'role': 'user', 'content': prompt},
+    ],
+)
+
+print(r)