Metacrafter
Python command line tool and python engine to label table fields and fields in data files. It could help to find meaningful data in your tables and data files or to find Personal identifable information (PII).
Installation
To install Python library use pip install metacrafter via pip or python setup.py install
Features
Metacrafter is a rule based tool that helps to label fields of the tables in databases. It scans table and finds person names, surnames, midnames, PII data, basic identifiers like UUID/GUID. These rules written as .yaml files and could be easily extended.
File formats supported:
- CSV
- JSON lines
- JSON (array of dicts)
- BSON
- Parquet
Databases support:
- Any database supported by SQLAlchemy
- MongoDB
Metacrafter key features:
- 25 basic and PII rules.
- all labels metadata collected into Metacrafter registry public repository -
- 312 date detection rules/patterns, date detection using qddate, "quick and dirty" date detection library
- extendable set of rules using PyParsing, exact text match and validation functions
- support any database supported by SQLAlchemy
- advanced context and language management. You could apply only rules relevant to certain data of choosen language
- built-in API server
- commercial support and additional rules available
Command line examples
File analysis examples
# Scan CSV file
$ metacrafter scan-file --format short somefile.csv
# Scan CSV file with delimiter ';' and windows-1251 encoding
$ metacrafter scan-file --format short --encoding windows-1251 --delimiter ';' somefile.csv
# Scan JSON lines file, output results as stats table to file file
$ metacrafter scan-file --format stats -o somefile_result.json somefile.jsonl
Result example of 'full' type of formatting
key ftype tags matches
-------------------- ------- ------ -------------------------------
name str uniq
addressresidence str uniq address 59.80
addressactivities str address 50.98
addressobjects str address 28.00
bin int ogrn 99.02
inn str inn 100.00,inn 99.02
purposeaudit str runpa 8.82
dateregistration str datetime 94.12 (dt:date:date_2)
expirydate str datetime 18.63 (dt:date:date_2)
startdateactivity str datetime 28.43 (dt:date:date_2)
othergrounds str dict
startdateaudit str datetime 65.69 (dt:date:date_2)
workdays int dict
workhours str dict
formaudit str dict
namestatecontrol str
assignment decree str dict
effectivedate str dict
Inspectionenddate str empty
riskcategory str dict
expirationdate str empty
startupnotifications str empty
daylastcheck str empty
otherreasonsrefusal str empty
numbersystem str empty
Database analysis examples
# Scan MongoDB database 'fns', save results as result.json and format output as 'stats'
$ metacrafter scan-mongodb --dbname fns -o result.json -f full
# Scan Postgres database 'dbname', with schema 'public'.
$ metacrafter scan-db --schema public --connstr postgresql+psycopg2://username:[email protected]:15432/dbname
Rules
All rules described as YAML files and by default rules loaded from directory 'rules' or from list of directories provided in .metacrafter file with YAML format
All rules could be applied to fields or data .
Compare engines defined in match parameter in rule description:
- text - scan text for exact match to one of text values. Text values delimited by comma (',')
- ppr - scan text for PyParsing. PyParsing rule defined as Python code with PyParsing objects like Word(nums, exact=4)
- func - scan text using Python function provided. Function shoud accept one string parameter and shoud return True or False
How to write rules
Function (func)
Example Russian administrative legal act/law matched by custom function
runpabyfunc:
key: runpa
name: Russian legal act / law
maxlen: 500
minlen: 3
priority: 1
match: func
type: data
rule: metacrafter.rules.ru.gov.is_ru_law
Exact text match (text)
Example midname matching by exact field name
midname:
key: person_midname
name: Person midname by known
rule: midname,secondname,middlename,mid_name,middle_name
type: field
match: text
PyParsing rule (ppr)
Example Russian cadastral number
rukadastr:
key: rukadastr
name: Russian land territory cadastral identifier
rule: Word(nums, min=1, max=2) + Literal(':').suppress() + Word(nums, min=1, max=2) + Literal(':').suppress() + Word(nums, min=6, max=7) + Literal(':').suppress() + Word(nums, min=1, max=6)
maxlen: 20
minlen: 12
priority: 1
match: ppr
type: data
Commercial support
Please write [email protected] or [email protected] to request beta access to commercial API. Commercial API support 195 fields and data rules and provided with dedicated support.
3 Oct 14, 2022
20 Dec 21, 2022
5 Nov 22, 2022
2 Jan 05, 2022
13.3k Jan 07, 2023
16 May 29, 2022
4 Oct 16, 2022
142 Dec 26, 2022
7 Aug 13, 2022
6 Dec 31, 2021
3 Feb 14, 2022
11 Sep 06, 2022
8 May 06, 2022
3 Nov 28, 2021
3 Dec 14, 2022
21 Jan 03, 2023
1 Sep 11, 2022
9 Mar 01, 2022
129 Dec 26, 2022
3 Mar 23, 2022