mirror of
https://github.com/frankwxu/mobile-pii-discovery-agent.git
synced 2026-02-20 13:40:41 +00:00
6 lines
34 KiB
JSON
6 lines
34 KiB
JSON
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "EMAIL", "PII": ["aniulka616@hotmail.com", "josephphw3r5@outlook.com", "jreynoldscbdemo+this.is.secret@gmail.com", "jthskp131699@icloud.com", "sousiezorjrc@outlook.com", "sweeneyfrank594@gmail.com", "ottomatik1234@gmail.com"], "Num_of_PII": 7, "source_columns": ["handle.id", "handle.uncanonicalized_id", "chat.account_id", "chat.chat_identifier", "chat.account_login", "chat.last_addressed_handle", "chat.display_name", "chat.cloudkit_record_id", "attachment.filename", "attachment.transfer_name", "attachment.guid", "attachment.original_guid", "chat.guid", "chat.room_name", "chat.group_id", "chat.engram_id", "chat.server_change_token", "chat.original_group_id", "message.text", "message.service_center", "message.subject", "message.service", "message.account", "message.account_guid", "message.guid", "message.associated_message_guid", "message.balloon_bundle_id", "message.expressive_send_style_id", "message.ck_record_id", "message.ck_record_change_tag", "message.destination_caller_id", "message.reply_to_guid", "message.thread_originator_guid", "message.thread_originator_part", "message.syndication_ranges", "message.synced_syndication_ranges", "message.bia_reference_id", "message.fallback_hash"], "Raw_rows_first_100": [["aniulka616@hotmail.com"], ["josephphw3r5@outlook.com"], ["jreynoldscbdemo+this.is.secret@gmail.com"], ["jthskp131699@icloud.com"], ["sousiezorjrc@outlook.com"], ["sweeneyfrank594@gmail.com"], ["aniulka616@hotmail.com"], ["josephphw3r5@outlook.com"], ["jreynoldscbdemo+this.is.secret@gmail.com"], ["jthskp131699@icloud.com"], ["sousiezorjrc@outlook.com"], ["sweeneyfrank594@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["SMS;-;jreynoldscbdemo+this.is.secret@gmail.com"], ["iMessage;-;aniulka616@hotmail.com"], ["iMessage;-;josephphw3r5@outlook.com"], ["iMessage;-;jthskp131699@icloud.com"], ["iMessage;-;sousiezorjrc@outlook.com"], ["iMessage;-;sweeneyfrank594@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"]], "Total_raw_rows": 30, "Exploration_sql": "SELECT id AS possible_email\nFROM handle\nWHERE id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT uncanonicalized_id AS possible_email\nFROM handle\nWHERE uncanonicalized_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT account_id AS possible_email\nFROM chat\nWHERE account_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT chat_identifier AS possible_email\nFROM chat\nWHERE chat_identifier REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT account_login AS possible_email\nFROM chat\nWHERE account_login REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT last_addressed_handle AS possible_email\nFROM chat\nWHERE last_addressed_handle REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT display_name AS possible_email\nFROM chat\nWHERE display_name REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT cloudkit_record_id AS possible_email\nFROM chat\nWHERE cloudkit_record_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT filename AS possible_email\nFROM attachment\nWHERE filename REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT transfer_name AS possible_email\nFROM attachment\nWHERE transfer_name REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT guid AS possible_email\nFROM attachment\nWHERE guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT original_guid AS possible_email\nFROM attachment\nWHERE original_guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT guid AS possible_email\nFROM chat\nWHERE guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT room_name AS possible_email\nFROM chat\nWHERE room_name REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT group_id AS possible_email\nFROM chat\nWHERE group_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT engram_id AS possible_email\nFROM chat\nWHERE engram_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT server_change_token AS possible_email\nFROM chat\nWHERE server_change_token REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT original_group_id AS possible_email\nFROM chat\nWHERE original_group_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT text AS possible_email\nFROM message\nWHERE text REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT service_center AS possible_email\nFROM message\nWHERE service_center REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT subject AS possible_email\nFROM message\nWHERE subject REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT service AS possible_email\nFROM message\nWHERE service REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT account AS possible_email\nFROM message\nWHERE account REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT account_guid AS possible_email\nFROM message\nWHERE account_guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT guid AS possible_email\nFROM message\nWHERE guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT associated_message_guid AS possible_email\nFROM message\nWHERE associated_message_guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT balloon_bundle_id AS possible_email\nFROM message\nWHERE balloon_bundle_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT expressive_send_style_id AS possible_email\nFROM message\nWHERE expressive_send_style_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT ck_record_id AS possible_email\nFROM message\nWHERE ck_record_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT ck_record_change_tag AS possible_email\nFROM message\nWHERE ck_record_change_tag REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT destination_caller_id AS possible_email\nFROM message\nWHERE destination_caller_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT reply_to_guid AS possible_email\nFROM message\nWHERE reply_to_guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT thread_originator_guid AS possible_email\nFROM message\nWHERE thread_originator_guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT thread_originator_part AS possible_email\nFROM message\nWHERE thread_originator_part REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT syndication_ranges AS possible_email\nFROM message\nWHERE syndication_ranges REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT synced_syndication_ranges AS possible_email\nFROM message\nWHERE synced_syndication_ranges REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT bia_reference_id AS possible_email\nFROM message\nWHERE bia_reference_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT fallback_hash AS possible_email\nFROM message\nWHERE fallback_hash REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'", "Extraction_sql": "SELECT id AS possible_email\nFROM handle\nWHERE id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT uncanonicalized_id AS possible_email\nFROM handle\nWHERE uncanonicalized_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT account_id AS possible_email\nFROM chat\nWHERE account_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT chat_identifier AS possible_email\nFROM chat\nWHERE chat_identifier REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT account_login AS possible_email\nFROM chat\nWHERE account_login REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT last_addressed_handle AS possible_email\nFROM chat\nWHERE last_addressed_handle REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT display_name AS possible_email\nFROM chat\nWHERE display_name REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT cloudkit_record_id AS possible_email\nFROM chat\nWHERE cloudkit_record_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT filename AS possible_email\nFROM attachment\nWHERE filename REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT transfer_name AS possible_email\nFROM attachment\nWHERE transfer_name REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT guid AS possible_email\nFROM attachment\nWHERE guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT original_guid AS possible_email\nFROM attachment\nWHERE original_guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT guid AS possible_email\nFROM chat\nWHERE guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT room_name AS possible_email\nFROM chat\nWHERE room_name REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT group_id AS possible_email\nFROM chat\nWHERE group_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT engram_id AS possible_email\nFROM chat\nWHERE engram_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT server_change_token AS possible_email\nFROM chat\nWHERE server_change_token REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT original_group_id AS possible_email\nFROM chat\nWHERE original_group_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT text AS possible_email\nFROM message\nWHERE text REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT service_center AS possible_email\nFROM message\nWHERE service_center REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT subject AS possible_email\nFROM message\nWHERE subject REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT service AS possible_email\nFROM message\nWHERE service REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT account AS possible_email\nFROM message\nWHERE account REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT account_guid AS possible_email\nFROM message\nWHERE account_guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT guid AS possible_email\nFROM message\nWHERE guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT associated_message_guid AS possible_email\nFROM message\nWHERE associated_message_guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT balloon_bundle_id AS possible_email\nFROM message\nWHERE balloon_bundle_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT expressive_send_style_id AS possible_email\nFROM message\nWHERE expressive_send_style_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT ck_record_id AS possible_email\nFROM message\nWHERE ck_record_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT ck_record_change_tag AS possible_email\nFROM message\nWHERE ck_record_change_tag REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT destination_caller_id AS possible_email\nFROM message\nWHERE destination_caller_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT reply_to_guid AS possible_email\nFROM message\nWHERE reply_to_guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT thread_originator_guid AS possible_email\nFROM message\nWHERE thread_originator_guid REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT thread_originator_part AS possible_email\nFROM message\nWHERE thread_originator_part REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT syndication_ranges AS possible_email\nFROM message\nWHERE syndication_ranges REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT synced_syndication_ranges AS possible_email\nFROM message\nWHERE synced_syndication_ranges REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT bia_reference_id AS possible_email\nFROM message\nWHERE bia_reference_id REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\nUNION ALL\nSELECT fallback_hash AS possible_email\nFROM message\nWHERE fallback_hash REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'", "PII_Prompt": "a unique identifier for a destination to which electronic mail (email) can be sent and received over the internet; examples include jane.doe@example.com, john.smith@provider.net, dev-team@startup.io, and user.name+label@domain.org"}
|
||
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [], "Total_raw_rows": 0, "Exploration_sql": "SELECT id AS possible_us_phone\nFROM handle\nWHERE id REGEXP '^(\\\\+1[ .-]?|1[ .-]?)?\\\\(?[2-9][0-9]{2}\\\\)?[ .-]?[2-9][0-9]{2}[ .-]?[0-9]{4}$'\n\nUNION ALL\n\nSELECT uncanonicalized_id AS possible_us_phone\nFROM handle\nWHERE uncanonicalized_id REGEXP '^(\\\\+1[ .-]?|1[ .-]?)?\\\\(?[2-9][0-9]{2}\\\\)?[ .-]?[2-9][0-9]{2}[ .-]?[0-9]{4}$'\n\nUNION ALL\n\nSELECT chat_identifier AS possible_us_phone\nFROM chat\nWHERE chat_identifier REGEXP '^(\\\\+1[ .-]?|1[ .-]?)?\\\\(?[2-9][0-9]{2}\\\\)?[ .-]?[2-9][0-9]{2}[ .-]?[0-9]{4}$'\n\nUNION ALL\n\nSELECT last_addressed_handle AS possible_us_phone\nFROM chat\nWHERE last_addressed_handle REGEXP '^(\\\\+1[ .-]?|1[ .-]?)?\\\\(?[2-9][0-9]{2}\\\\)?[ .-]?[2-9][0-9]{2}[ .-]?[0-9]{4}$'\n\nUNION ALL\n\nSELECT text AS possible_us_phone\nFROM message\nWHERE text REGEXP '^(.*)(\\\\+?1?[ .-]?\\\\(?[2-9][0-9]{2}\\\\)?[ .-]?[2-9][0-9]{2}[ .-]?[0-9]{4})(.*)$'\n\nUNION ALL\n\nSELECT service_center AS possible_us_phone\nFROM message\nWHERE service_center REGEXP '^(\\\\+1[ .-]?|1[ .-]?)?\\\\(?[2-9][0-9]{2}\\\\)?[ .-]?[2-9][0-9]{2}[ .-]?[0-9]{4}$'\n\nUNION ALL\n\nSELECT destination_caller_id AS possible_us_phone\nFROM message\nWHERE destination_caller_id REGEXP '^(\\\\+1[ .-]?|1[ .-]?)?\\\\(?[2-9][0-9]{2}\\\\)?[ .-]?[2-9][0-9]{2}[ .-]?[0-9]{4}$';", "Extraction_sql": null, "PII_Prompt": "a US phone number is a 10-digit NANP number (area code + exchange + line) that may be written as 2023133725, 202-313-3725, (202) 313-3725, 202.313.3725, +1 202 313 3725, or 1-202-313-3725"}
|
||
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [["chat70863266579689223"], ["chat818965589567390604"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["CBE9797C-F36D-4678-960B-659CD3618AB2"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["chat70863266579689223"], ["chat818965589567390604"], ["Hola"], ["Sorry"], ["Yeah"], ["Cheers"], ["Hello"], ["Abe "], ["SMS"], ["SMS"], ["SMS"], ["SMS"], ["SMS"], ["iMessage"], ["SMS"], ["iMessage"], ["SMS"], ["SMS"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["B0690F46-EE5F-4019-ACFC-7782054C012B"], ["chat70863266579689223"], ["chat70863266579689223"], ["chat818965589567390604"], ["IMG_3525.jpeg"], ["IMG_0035.MOV"], ["Resized_20240603_213954_1717465199266.jpeg"], ["IMG_0324.HEIC"], ["Resized_20240605_162458_1717619104441.jpeg"], ["Resized_20240605_193505_1717630512670.jpeg"], ["IMG_0341.HEIC"], ["IMG_0399.PNG"], ["Resized_20240620_071946_1718895471091.jpeg"], ["Resized_20240619_112503_1718895476715.jpeg"]], "Total_raw_rows": 53, "Exploration_sql": "SELECT chat_identifier AS possible_username\nFROM chat\nWHERE chat_identifier REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND chat_identifier NOT REGEXP '@';\n\nUNION ALL\n\nSELECT account_id AS possible_username\nFROM chat\nWHERE account_id REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND account_id NOT REGEXP '@';\n\nUNION ALL\n\nSELECT account_login AS possible_username\nFROM chat\nWHERE account_login REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND account_login NOT REGEXP '@';\n\nUNION ALL\n\nSELECT last_addressed_handle AS possible_username\nFROM chat\nWHERE last_addressed_handle REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND last_addressed_handle NOT REGEXP '@';\n\nUNION ALL\n\nSELECT display_name AS possible_username\nFROM chat\nWHERE display_name REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND display_name NOT REGEXP '@';\n\nUNION ALL\n\nSELECT room_name AS possible_username\nFROM chat\nWHERE room_name REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND room_name NOT REGEXP '@';\n\nUNION ALL\n\nSELECT id AS possible_username\nFROM handle\nWHERE id REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND id NOT REGEXP '@';\n\nUNION ALL\n\nSELECT uncanonicalized_id AS possible_username\nFROM handle\nWHERE uncanonicalized_id REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND uncanonicalized_id NOT REGEXP '@';\n\nUNION ALL\n\nSELECT text AS possible_username\nFROM message\nWHERE text REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND text NOT REGEXP '@';\n\nUNION ALL\n\nSELECT service_center AS possible_username\nFROM message\nWHERE service_center REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND service_center NOT REGEXP '@';\n\nUNION ALL\n\nSELECT subject AS possible_username\nFROM message\nWHERE subject REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND subject NOT REGEXP '@';\n\nUNION ALL\n\nSELECT service AS possible_username\nFROM message\nWHERE service REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND service NOT REGEXP '@';\n\nUNION ALL\n\nSELECT account AS possible_username\nFROM message\nWHERE account REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND account NOT REGEXP '@';\n\nUNION ALL\n\nSELECT account_guid AS possible_username\nFROM message\nWHERE account_guid REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND account_guid NOT REGEXP '@';\n\nUNION ALL\n\nSELECT cache_roomnames AS possible_username\nFROM message\nWHERE cache_roomnames REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND cache_roomnames NOT REGEXP '@';\n\nUNION ALL\n\nSELECT group_title AS possible_username\nFROM message\nWHERE group_title REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND group_title NOT REGEXP '@';\n\nUNION ALL\n\nSELECT transfer_name AS possible_username\nFROM attachment\nWHERE transfer_name REGEXP '^[A-Za-z][A-Za-z0-9._-]{2,51}$'\n AND transfer_name NOT REGEXP '@';", "Extraction_sql": null, "PII_Prompt": "a username is a short textual identifier chosen by a user to represent their account or public handle within an application or service it is stored as plain text contains no whitespace does not include a domain component and is intended for human recognition rather than internal system uniqueness"}
|
||
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "PERSON_NAME", "PII": ["Anya", "William", "Eddie", "Otto", "Tracy", "Dimitra", "Anna", "Joe Church", "Sharon Oneil", "Ella Rutman Doligo", "Ed Venture", "Ron Serber", "Ronen Engler", "Abe"], "Num_of_PII": 14, "source_columns": ["chat.display_name", "chat.chat_identifier", "chat.account_login", "chat.last_addressed_handle", "handle.id", "handle.uncanonicalized_id", "message.text", "message.subject", "message.cache_roomnames"], "Raw_rows_first_100": [["aniulka616@hotmail.com"], ["chat70863266579689223"], ["chat818965589567390604"], ["josephphw3r5@outlook.com"], ["jreynoldscbdemo+this.is.secret@gmail.com"], ["jthskp131699@icloud.com"], ["sousiezorjrc@outlook.com"], ["sweeneyfrank594@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["E:ottomatik1234@gmail.com"], ["aniulka616@hotmail.com"], ["josephphw3r5@outlook.com"], ["jreynoldscbdemo+this.is.secret@gmail.com"], ["jthskp131699@icloud.com"], ["sousiezorjrc@outlook.com"], ["sweeneyfrank594@gmail.com"], ["I'm Anya, director of Gmary jewelry, Hope you like that Rose Gold bracelet, if you have any comments, please contact me, Have a nice day"], ["Hola"], ["[TikTok] 810197 is your verification code, valid for 5 minutes. To keep your account safe, never forward this code."], ["[TikTok] 122628 is your verification code, valid for 5 minutes. To keep your account safe, never forward this code."], ["[TikTok] 172994 is your verification code, valid for 5 minutes. To keep your account safe, never forward this code."], ["Your Google verification code is 326338."], ["Wrong number "], ["Your SIGNAL verification code is: 902475"], ["If someone requests this code, it is a scam. Use code 914337 only in Google Voice app to sign up. g.co/voice/help"], ["Your Viber code is: 827215.\nClose this message and enter the code into Viber to activate your account."], ["Snapchat: 264691 is your one time passcode for phone enrollment. Snapchat will never call or text you for this code"], ["Opps!, isn't this William's phone number?? I will check the phone number again"], ["My man....what's uo"], ["Finnnnnn wazzup"], ["Yo boss. Ssup?😎"], ["I was about to say is this a scam or something? lol 😂"], ["Sorry"], ["Its fine!"], ["By awhile since we did that thing in Dallas"], ["Yayza \nHow's your weekend so far?"], ["Good, no complaints....boss sending me out to do a job next week, so I'm chillin...."], ["You?"], ["Lol, mine should be good. How's Eddie V? Haven't heard from him in awhile. "], ["He's just waking up. Just busy on his other phone so maybe he's behind "], ["I hope my next job pays well....I'm \"bear\"ly getting by"], ["Uh oh, "], ["Your Uber code is 0703. Never share this code."], ["Please verify your phone # for Life360 by entering this code: 941626"], ["Ah ok"], ["FRM:93557\nSUBJ:Verify Wells Fargo Activity\nMSG:Reach +1 (844) 407.4819 if you haven't made the acquisition for $666.02 @Rite Aid WellsID: ZST2T8TALSOWZ"], ["Hi, it's AT&T. You're all set! Enjoy unlimited talk, text & data at no extra cost while traveling in 19 Latin America countries. Coverage/data speeds vary by country. Details at att.com/latamcountries (free view). "], ["AT&T Free Msg: Great news! You have AT&T International Day Pass for Business on this line. Enjoy the plan you use at work while traveling abroad in 210+ included countries for just $10/day. The $10 daily fee applies in each 24-hour period you use data, talk or send a text in an included country. Your organization will be charged only for the days used abroad. Go to att.com/idpforbiz (a free view) for details. Enjoy your trip and safe travels!"], ["AT&T Free Msg: Great news! You have AT&T International Day Pass for Business on this line. Enjoy the plan you use at work while traveling abroad in 210+ included countries for just $10/day. The $10 daily fee applies in each 24-hour period you use data, talk or send a text in an included country. Your organization will be charged only for the days used abroad. Go to att.com/idpforbiz (a free view) for details. Enjoy your trip and safe travels!"], ["AT&T Free Msg: Great news! You have AT&T International Day Pass for Business on this line. Enjoy the plan you use at work while traveling abroad in 210+ included countries for just $10/day. The $10 daily fee applies in each 24-hour period you use data, talk or send a text in an included country. Your organization will be charged only for the days used abroad. Go to att.com/idpforbiz (a free view) for details. Enjoy your trip and safe travels!"], ["The USPS package has arrived at the warehouse and cannot be delivered due to incomplete address information. Please confirm your address in the link within 12 hours.\n\nhttps://usp.usspiq.top\n\n(Please reply to Y, then exit the SMS, open the SMS activation link again, or copy the link to Safari browser and open it)\n\nThe US Postal team wishes you a wonderful day"], ["The USPS package arrived at the warehouse but could not be delivered due to incomplete address information. Please confirm your address in the link.\nhttp://u6v.cn/6oB4iT\n(Please reply Y, then exit the text message and open it again to activate the link, or copy the link and open it in your Safari browser).\nThe USPS team wishes you a wonderful day!"], ["🎁 Crypto Results 3731: Aspire to earn $5K to $15K daily? Text \"Yes\" to get strategic insights from C0INBASE'S elite crypto experts and a chance to win $500. Supercharge your portfolio now."], ["C0INBASE 3731: Race in the champions’ circuit of crypto trading! Reply \"Yes\" to train with Coinbase’s elite and sprint towards daily profits of $5K to $15K, with a pit stop for a $500 win."], ["The temporary PIN you requested is 7233. Please use this PIN to complete set up. If you received this in error, call 800-331-0500 or 611 and say representative."], ["AT&T free Msg- your PIN was changed. Have questions? Please dial 1-800-331-0500 or 611 from your wireless phone."], ["AT&T Free Message: Valued Customer, thanks for choosing us. We'll text you to ask for some feedback about the service provided by Dimitra for your wireless account. Survey texts are free."], ["1 of 5: How satisfied are you with Dimitra's ability to assist you with your call?\r\n1=Very DISSATISFIED\r\n2=Dissatisfied\r\n3=Neutral\r\n4=Satisfied\r\n5=Very SATISFIED"], ["It's Otto"], ["I'm at the marina Grill. You coming here anytime soon?"], ["You all still there?"], ["Yeah"], ["You made quite the impression. Don't blow it. "], ["Loved “You made quite the impression. Don't blow it. ”"], ["You want her number?"], ["She said you could call. "], ["Yes please 🙏 "], ["Hi. 👋 \nI’m the guy from the marina grill\nAnted to say Hi\n\nOtto"], ["What?"], ["Who are you? Do I know you?"], ["Actually, who Gave you my number?"], ["RP?"], ["Are you out?"], ["Just got back to my hotel. Saw you at dinner and "], ["Russell gave me your #"], ["I'm up to meet tomorrow. Coffee."], ["Let me know when you wake up \nI'm an early 🐦 "], ["Loved an image"], ["Oh I'm not! I'll text you"], ["Loved “Oh I'm not! I'll text you”"], ["Up and moving. Prepping for my meeting but around for coffee"], ["See you soon "], ["Going to get coffee if you're around "], ["I'm here. Drinking coffee"], ["Are you on social."], ["I'm also on FB coz I'm old "], ["qzhskp kfpckudumbasree.com"], ["Was good seeing you!!!!"], ["Same to you"], ["A crazy drunk lady just hit on me at the airport!"], ["Oh lol\n"], ["Cheers"], ["Tell that lady you're taken"], ["Oh yeah?"], ["Snapchat: You have friends on Snapchat! Add them: https://snapchat.com/t/bZc40uiQ\nReply STOP to opt out"]], "Total_raw_rows": 191, "Exploration_sql": "SELECT display_name AS possible_person_name\nFROM chat\nWHERE display_name REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT chat_identifier AS possible_person_name\nFROM chat\nWHERE chat_identifier REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT account_login AS possible_person_name\nFROM chat\nWHERE account_login REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT last_addressed_handle AS possible_person_name\nFROM chat\nWHERE last_addressed_handle REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT id AS possible_person_name\nFROM handle\nWHERE id REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT uncanonicalized_id AS possible_person_name\nFROM handle\nWHERE uncanonicalized_id REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT text AS possible_person_name\nFROM message\nWHERE text REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT subject AS possible_person_name\nFROM message\nWHERE subject REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT cache_roomnames AS possible_person_name\nFROM message\nWHERE cache_roomnames REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'", "Extraction_sql": "SELECT display_name AS possible_person_name\nFROM chat\nWHERE display_name REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT chat_identifier AS possible_person_name\nFROM chat\nWHERE chat_identifier REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT account_login AS possible_person_name\nFROM chat\nWHERE account_login REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT last_addressed_handle AS possible_person_name\nFROM chat\nWHERE last_addressed_handle REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT id AS possible_person_name\nFROM handle\nWHERE id REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT uncanonicalized_id AS possible_person_name\nFROM handle\nWHERE uncanonicalized_id REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT text AS possible_person_name\nFROM message\nWHERE text REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT subject AS possible_person_name\nFROM message\nWHERE subject REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT cache_roomnames AS possible_person_name\nFROM message\nWHERE cache_roomnames REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'", "PII_Prompt": "a loosely structured human name-like strings that typically consist of a first name, a first name and a last name, and may also include middle names, initials, prefixes (e.g., Mr., Dr.), and suffixes (e.g., Jr., Sr.)"}
|
||
{"db_path": "selectedDBs\\I3_sms.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [["message.text", "Opps!, isn't this William's phone number?? I will check the phone number again"], ["message.text", "My man....what's uo"], ["message.text", "Yayza \nHow's your weekend so far?"], ["message.text", "Lol, mine should be good. How's Eddie V? Haven't heard from him in awhile. "], ["message.text", "He's just waking up. Just busy on his other phone so maybe he's behind "], ["message.text", "Hi, it's AT&T. You're all set! Enjoy unlimited talk, text & data at no extra cost while traveling in 19 Latin America countries. Coverage/data speeds vary by country. Details at att.com/latamcountries (free view). "], ["message.text", "🎁 Crypto Results 3731: Aspire to earn $5K to $15K daily? Text \"Yes\" to get strategic insights from C0INBASE'S elite crypto experts and a chance to win $500. Supercharge your portfolio now."], ["message.text", "C0INBASE 3731: Race in the champions’ circuit of crypto trading! Reply \"Yes\" to train with Coinbase’s elite and sprint towards daily profits of $5K to $15K, with a pit stop for a $500 win."], ["message.text", "1 of 5: How satisfied are you with Dimitra's ability to assist you with your call?\r\n1=Very DISSATISFIED\r\n2=Dissatisfied\r\n3=Neutral\r\n4=Satisfied\r\n5=Very SATISFIED"], ["message.text", "It's Otto"]], "Total_raw_rows": 10, "Exploration_sql": "SELECT\n 'message.text' AS source,\n text AS value\nFROM message\nWHERE text REGEXP '(?i)\\b(?:p\\.?\\s*o\\.?\\s*box|post\\s+office\\s+box|ave\\.?|avenue|st\\.?|street|rd\\.?|road|blvd\\.?|boulevard|dr\\.?|drive|ln\\.?|lane|ct\\.?|court|pl\\.?|place|way|pkwy\\.?|parkway|cir\\.?|circle|ter\\.?|terrace|hwy\\.?|highway|trl\\.?|trail|sq\\.?|square|pike|loop|run|walk|path|byp\\.?|bypass|(?:n|s|e|w|ne|nw|se|sw)\\b)\\b'\nUNION ALL\nSELECT\n 'chat.chat_identifier' AS source,\n chat_identifier AS value\nFROM chat\nWHERE chat_identifier REGEXP '(?i)\\b(?:p\\.?\\s*o\\.?\\s*box|post\\s+office\\s+box|ave\\.?|avenue|st\\.?|street|rd\\.?|road|blvd\\.?|boulevard|dr\\.?|drive|ln\\.?|lane|ct\\.?|court|pl\\.?|place|way|pkwy\\.?|parkway|cir\\.?|circle|ter\\.?|terrace|hwy\\.?|highway|trl\\.?|trail|sq\\.?|square|pike|loop|run|walk|path|byp\\.?|bypass|(?:n|s|e|w|ne|nw|se|sw)\\b)\\b'\nUNION ALL\nSELECT\n 'chat.last_addressed_handle' AS source,\n last_addressed_handle AS value\nFROM chat\nWHERE last_addressed_handle REGEXP '(?i)\\b(?:p\\.?\\s*o\\.?\\s*box|post\\s+office\\s+box|ave\\.?|avenue|st\\.?|street|rd\\.?|road|blvd\\.?|boulevard|dr\\.?|drive|ln\\.?|lane|ct\\.?|court|pl\\.?|place|way|pkwy\\.?|parkway|cir\\.?|circle|ter\\.?|terrace|hwy\\.?|highway|trl\\.?|trail|sq\\.?|square|pike|loop|run|walk|path|byp\\.?|bypass|(?:n|s|e|w|ne|nw|se|sw)\\b)\\b'\nUNION ALL\nSELECT\n 'chat.display_name' AS source,\n display_name AS value\nFROM chat\nWHERE display_name REGEXP '(?i)\\b(?:p\\.?\\s*o\\.?\\s*box|post\\s+office\\s+box|ave\\.?|avenue|st\\.?|street|rd\\.?|road|blvd\\.?|boulevard|dr\\.?|drive|ln\\.?|lane|ct\\.?|court|pl\\.?|place|way|pkwy\\.?|parkway|cir\\.?|circle|ter\\.?|terrace|hwy\\.?|highway|trl\\.?|trail|sq\\.?|square|pike|loop|run|walk|path|byp\\.?|bypass|(?:n|s|e|w|ne|nw|se|sw)\\b)\\b'\nUNION ALL\nSELECT\n 'handle.id' AS source,\n id AS value\nFROM handle\nWHERE id REGEXP '(?i)\\b(?:p\\.?\\s*o\\.?\\s*box|post\\s+office\\s+box|ave\\.?|avenue|st\\.?|street|rd\\.?|road|blvd\\.?|boulevard|dr\\.?|drive|ln\\.?|lane|ct\\.?|court|pl\\.?|place|way|pkwy\\.?|parkway|cir\\.?|circle|ter\\.?|terrace|hwy\\.?|highway|trl\\.?|trail|sq\\.?|square|pike|loop|run|walk|path|byp\\.?|bypass|(?:n|s|e|w|ne|nw|se|sw)\\b)\\b'", "Extraction_sql": null, "PII_Prompt": "a US postal address is a street-level mailing location in the United States, commonly appearing as a street name and suffix (e.g., 'Market St') optionally with a street number (e.g., '1500 Market St'), unit, city/state, ZIP, or a PO Box (e.g., 'P.O. Box 123')"}
|