mirror of
https://github.com/frankwxu/mobile-pii-discovery-agent.git
synced 2026-02-20 13:40:41 +00:00
6 lines
13 KiB
JSON
6 lines
13 KiB
JSON
{"db_path": "selectedDBs\\A5_SBrowser2.db", "PII_type": "EMAIL", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [], "Total_raw_rows": 0, "Exploration_sql": "SELECT\n URL AS possible_email_source\nFROM BOOKMARKS\nWHERE URL REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n SURL AS possible_email_source\nFROM BOOKMARKS\nWHERE SURL REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n TITLE AS possible_email_source\nFROM BOOKMARKS\nWHERE TITLE REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n TAGS AS possible_email_source\nFROM BOOKMARKS\nWHERE TAGS REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n SOURCEID AS possible_email_source\nFROM BOOKMARKS\nWHERE SOURCEID REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n ACCOUNT_NAME AS possible_email_source\nFROM BOOKMARKS\nWHERE ACCOUNT_NAME REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n ACCOUNT_TYPE AS possible_email_source\nFROM BOOKMARKS\nWHERE ACCOUNT_TYPE REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n DEVICE_NAME AS possible_email_source\nFROM BOOKMARKS\nWHERE DEVICE_NAME REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n SYNC1 AS possible_email_source\nFROM BOOKMARKS\nWHERE SYNC1 REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n SYNC2 AS possible_email_source\nFROM BOOKMARKS\nWHERE SYNC2 REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n SYNC3 AS possible_email_source\nFROM BOOKMARKS\nWHERE SYNC3 REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n SYNC4 AS possible_email_source\nFROM BOOKMARKS\nWHERE SYNC4 REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n SYNC5 AS possible_email_source\nFROM BOOKMARKS\nWHERE SYNC5 REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n keyword AS possible_email_source\nFROM BOOKMARKS\nWHERE keyword REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'\n\nUNION ALL\n\nSELECT\n description AS possible_email_source\nFROM BOOKMARKS\nWHERE description REGEXP '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}'", "Extraction_sql": null, "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\\A5_SBrowser2.db", "PII_type": "PHONE", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [], "Total_raw_rows": 0, "Exploration_sql": "SELECT URL AS possible_phone\nFROM BOOKMARKS\nWHERE URL REGEXP '\\+?1?[- .]?\\(?[2-9][0-9]{2}\\)?[- .]?[2-9][0-9]{2}[- .]?[0-9]{4}'\n\nUNION ALL\n\nSELECT SURL AS possible_phone\nFROM BOOKMARKS\nWHERE SURL REGEXP '\\+?1?[- .]?\\(?[2-9][0-9]{2}\\)?[- .]?[2-9][0-9]{2}[- .]?[0-9]{4}'\n\nUNION ALL\n\nSELECT TITLE AS possible_phone\nFROM BOOKMARKS\nWHERE TITLE REGEXP '\\+?1?[- .]?\\(?[2-9][0-9]{2}\\)?[- .]?[2-9][0-9]{2}[- .]?[0-9]{4}'\n\nUNION ALL\n\nSELECT TAGS AS possible_phone\nFROM BOOKMARKS\nWHERE TAGS REGEXP '\\+?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\\A5_SBrowser2.db", "PII_type": "USERNAME", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [["Sharon's S21"], ["Sharon's S21"], ["Sharon's S21"], ["Sharon's S21"]], "Total_raw_rows": 4, "Exploration_sql": "SELECT ACCOUNT_NAME AS possible_username\nFROM BOOKMARKS\nWHERE ACCOUNT_NAME REGEXP '\\b[a-zA-Z][a-zA-Z0-9._-]{2,51}\\b'\n\nUNION ALL\n\nSELECT DEVICE_NAME AS possible_username\nFROM BOOKMARKS\nWHERE DEVICE_NAME REGEXP '\\b[a-zA-Z][a-zA-Z0-9._-]{2,51}\\b'\n\nUNION ALL\n\nSELECT SOURCEID AS possible_username\nFROM BOOKMARKS\nWHERE SOURCEID REGEXP '\\b[a-zA-Z][a-zA-Z0-9._-]{2,51}\\b'\n\nUNION ALL\n\nSELECT TAGS AS possible_username\nFROM BOOKMARKS\nWHERE TAGS REGEXP '\\b[a-zA-Z][a-zA-Z0-9._-]{2,51}\\b'\n\nUNION ALL\n\nSELECT keyword AS possible_username\nFROM BOOKMARKS\nWHERE keyword REGEXP '\\b[a-zA-Z][a-zA-Z0-9._-]{2,51}\\b'\n\nUNION ALL\n\nSELECT description AS possible_username\nFROM BOOKMARKS\nWHERE description REGEXP '\\b[a-zA-Z][a-zA-Z0-9._-]{2,51}\\b'\n\nUNION ALL\n\nSELECT guid AS possible_username\nFROM BOOKMARKS\nWHERE guid REGEXP '\\b[a-zA-Z][a-zA-Z0-9._-]{2,51}\\b'", "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\\A5_SBrowser2.db", "PII_type": "PERSON_NAME", "PII": ["Sharon"], "Num_of_PII": 1, "source_columns": ["BOOKMARKS.TITLE", "BOOKMARKS.TAGS", "BOOKMARKS.ACCOUNT_NAME", "BOOKMARKS.DEVICE_NAME", "BOOKMARKS.keyword", "BOOKMARKS.description"], "Raw_rows_first_100": [[{"__bytes_b64__": "BAGBHAIGAQABLGwCAOx/Yz8pWMaZgKzK9gidY1JbWlHtbmZoPVRvHa8L1Qa5v77osj/b9cxf6AihGfAj9/oMzNgzwG4TEa1ocxhMgTw="}], [{"__bytes_b64__": "BAGBHAIGAQABLGwCADg08Y0rLftwtKLewxQylCXdY9YcWUe01qQU3juIWvYiZTepFwA/AGQEV8xammaotcT+LNCNEQBvYK49zZmdniQ="}], [{"__bytes_b64__": "BAGBXgIHAQABLIEsAgAx0od46RvJj3ZMpmE/r++aNuxFfDaTafncAW0XddYSB0WkI7vryB4lplJKYOFVJvriCJ54MlN56/+r6ZnwIIKx3TF4rvHTUNRIc9LKLCfh5hqUwF5ebYY+wMqebfiqMSY="}], [{"__bytes_b64__": "BAGBXgIHAQABLIEsAgAN/bPhb+tihkAfwO8uIxTzh9YWVp1U/kZ3oy/IzxRDnWH35ch+01jfwXwTtCmEj+LyGgSjHzzowXmue9uLyUvfpmVNbt3JCOqCl0EyBAo5+xpVCiij7EFhenIbvW/5cN0="}], [{"__bytes_b64__": "BAGBfgIHAQABLIFMAgAxMZf3KoVtVYHMTrajimnorfRebkfgH/qYhDfOJ2RRzWYOGMxn9xTFfrGxr/Cg60A6HnjLCVgpgftWYAW/bKRQc57iUOY3/35T/FmViNlLg/0T1xmtyNMJL85eTN6Ty8FnisaqUE3+iEK+drFSfnJt"}], [{"__bytes_b64__": "BAGBXgIHAQABLIEsAgDferB3B+clQ8gWXYeFleQ01pIC6VzQ+FxtlyoVYuGZ/sO3K+0Bwv1BqzJe5oM8CyJQtis/qRu/Li9LMYVDQxRCkfClKVoXFf6qVbzv+0XrXKADvEyQD083AV/geGnfwx8="}], ["Sharon's S21"], ["Sharon's S21"], ["Sharon's S21"], ["Sharon's S21"]], "Total_raw_rows": 10, "Exploration_sql": "SELECT TITLE AS possible_person_name\nFROM BOOKMARKS\nWHERE TITLE REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT TAGS AS possible_person_name\nFROM BOOKMARKS\nWHERE TAGS REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT ACCOUNT_NAME AS possible_person_name\nFROM BOOKMARKS\nWHERE ACCOUNT_NAME REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT DEVICE_NAME AS possible_person_name\nFROM BOOKMARKS\nWHERE DEVICE_NAME REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT keyword AS possible_person_name\nFROM BOOKMARKS\nWHERE keyword REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\n\nUNION ALL\n\nSELECT description AS possible_person_name\nFROM BOOKMARKS\nWHERE description REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'", "Extraction_sql": "SELECT TITLE AS possible_person_name\nFROM BOOKMARKS\nWHERE TITLE REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT TAGS AS possible_person_name\nFROM BOOKMARKS\nWHERE TAGS REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT ACCOUNT_NAME AS possible_person_name\nFROM BOOKMARKS\nWHERE ACCOUNT_NAME REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT DEVICE_NAME AS possible_person_name\nFROM BOOKMARKS\nWHERE DEVICE_NAME REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT keyword AS possible_person_name\nFROM BOOKMARKS\nWHERE keyword REGEXP '[A-Za-z][A-Za-z\\\\s\\\\.\\\\-]{1,50}'\nUNION ALL\nSELECT description AS possible_person_name\nFROM BOOKMARKS\nWHERE description 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\\A5_SBrowser2.db", "PII_type": "POSTAL_ADDRESS", "PII": [], "Num_of_PII": 0, "source_columns": [], "Raw_rows_first_100": [[{"__bytes_b64__": "BAGBHAIGAQABLGwCAFgwiN0XYaRNwN6JFtBRdWfQw5YoTxjvsswMtBRYpQ4bzjZW6jjqVulUtvEdo3wPiKvtT0SzbxZ9i5yPZb27tpI="}], [{"__bytes_b64__": "BAGEPgIHAQABLIQMAgABgxFFetbBD88wsavVtVjy+KN6LQaF52tl9Ztc+fzXT8MhsjknjCsDwrr+9tZQ1uZZNjvrZf13/vt7Nn58Xe1+sdDXmFRvJGfComqfeWyTQONx07TYcFCxWvZNWaecNUuUw+EC87PJColRgnZwE0JLLZ8Zcjsh8AjPicnUGgdkOfN8I1TaXKxMGebpzElMuqtVIhlph5EGeSm7pg0v/ELp/VeGLuxGeKbA6tXvUoiGaT51XBQebcsvu7qUegGuoIrZLf8Od7Q3JpKpH9miAs8O6o+o4zZXEI5LBv1suoFxmU6MRC7lTPGYgU6e+O1Wd9cSg4F2f7U2fAOBGDPYg3BrCZONls5ungK74IoKY48a5A=="}], [{"__bytes_b64__": "BAGBPgIHAQABLIEMAgD443e4SeU/jxbaN7+6lk4CFvovH1ui72l0PE6Mr20DtdnV/jQxlqAi9jhvXfYxTbDjh9dKl6yNgZm/19GPdqzMV0aYY9i3A05cPRL/VKYMpw=="}], [{"__bytes_b64__": "BAGDHgIHAQABLIJsAgCKKvrixPJ+dMqD7ppuoI0MyZaiuUvIgtJ4WHgnxrQGklMS6lrCMckJrtZnmVGBcHWPgbEIH/JKEmczg7KSJqaMfK+BKSyUC+v3ydvl2m/UC0JEA37crOZHBfy5z4B7KF1j7b1u7DDYe4DzyPE0+JyV0dfHonPF0zGA+yDZUAOoReeMJ9yHcuCKnASWELWEEvSukVLIZBmZ2OhTv4KNnbCyLBcA0Xz3SQ32AC3/+ef77BgKgSzNArHW3R5aUBmC7EE="}], [{"__bytes_b64__": "BAGDHgIHAQABLIJsAgDa5rn5/h/uPxLOY8KfuGKu45sg2E1RRV79AuDrwZAO/CdOjK94CeByOdIR/dW/LnMwaCJ/Fzsktoej/DoH9XIfnD+9F+UZYDRgfpb9Ie0a/0TZyy2zL4HhatC455OfG5sGqlu7ZvodLztLBH1iqnEr3fC5hFGYgiAIbKmbwXITQdJ6efZ2J6a4Ne5dEK7nSq8Bky9lsF2g/R6Yug7kd5m+AHc1BEZoXFDer3Gm30trqQgKEgafKEJftuh+Cf8/Wic="}], [{"__bytes_b64__": "BAGEPgIHAQABLIQMAgBQ1ryPaiEATOXfVZTPg8FP3oJkx7RwFyn+2sKiKDDTGe/zN3pxBhO+t0++nKxLvCGlZ5/q8uMRVU0e9XsXOeH4uR4SDyxAajG0+/6OmOaOjtXwZGmhkY8d0UaJSyT1OukC02GESg3KjDFOeQsNQsfWEdvhNUWozvT4y7TUo1fiKMzWpKvIVK3NRaVFs8Dxgs0dNirE73Feyeps2GRq6CrF1hisNwoWNAp6H62mYO4OSgpbH8vIo7HKytAfYok6JzGpW2KUnXjsI/aLOg+OVz5a7WbtNPtX2EZ93WfLmRflqbD03VPN73Bgz6ENlzB79bNuPDr6EXSiBjUg4lRre7kwsO7iIo2Hv70JZFsyhJfQ2A=="}], [{"__bytes_b64__": "BAGBPgIHAQABLIEMAgB/6+uZMDK39vfkGeXZ94x8HT2s4mGPQAcYJhlvjrZsx/FALnRkp5fEspNZKtEkku2zkIZzFtG4mDnV4b0hLpq/HEyEikPekmGfVSgDniYNpg=="}], [{"__bytes_b64__": "BAGDHgIHAQABLIJsAgBwvy2V6oP1GFaG5qu3Y+17nx7mxsU35WfB3ALBxs71iPNCuztsZV3ef7wmH7hMfFHfLo6gAnxafQkx7W+yh3d7gqtqsxxDhl8oRIGXZ8O8mIJQN87jXx/OzIV7jpVX4vIW2kP773vllT1cTagB4rpWx2sk8kEWCBKPRadIjgJjBNAh3e3alczuqrxeraLCXzC3wsmaCq3EHAkGC8KXuMdBJkH/s1cLUHAa6Ke0jYIhjwXKuEOIBu3bICdrPZR1UAE="}], [{"__bytes_b64__": "BAGBXgIHAQABLIEsAgAx0od46RvJj3ZMpmE/r++aNuxFfDaTafncAW0XddYSB0WkI7vryB4lplJKYOFVJvriCJ54MlN56/+r6ZnwIIKx3TF4rvHTUNRIc9LKLCfh5hqUwF5ebYY+wMqebfiqMSY="}]], "Total_raw_rows": 9, "Exploration_sql": "SELECT URL AS possible_address\nFROM BOOKMARKS\nWHERE URL 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 SURL AS possible_address\nFROM BOOKMARKS\nWHERE SURL 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 TITLE AS possible_address\nFROM BOOKMARKS\nWHERE TITLE 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 TAGS AS possible_address\nFROM BOOKMARKS\nWHERE TAGS 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 description AS possible_address\nFROM BOOKMARKS\nWHERE description 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')"}
|