API Documentation

Bloom filter

Generate a Bloom filter

class clkhash.bloomfilter.NgramEncodings

Bases: enum.Enum

The available schemes for encoding n-grams.

BLAKE_HASH = functools.partial(<function blake_encode_ngrams>)

uses the BLAKE2 hash function, which is one of the fastest modern hash functions, and does less hash function calls compared to the DOUBLE_HASH based schemes. It avoids one of the exploitable weaknesses of the DOUBLE_HASH scheme. Also see blake_encode_ngrams()

DOUBLE_HASH = functools.partial(<function double_hash_encode_ngrams>)

the initial encoding scheme as described in Schnell, R., Bachteler, T., & Reiher, J. (2011). A Novel Error-Tolerant Anonymous Linking Code. Also see double_hash_encode_ngrams()

DOUBLE_HASH_NON_SINGULAR = functools.partial(<function double_hash_encode_ngrams_non_singular>)

very similar to DOUBLE_HASH, but avoids singularities in the encoding. Also see double_hash_encode_ngrams_non_singular()

clkhash.bloomfilter.blake_encode_ngrams(ngrams, keys, k, l, encoding)

Computes the encoding of the provided ngrams using the BLAKE2 hash function.

We deliberately do not use the double hashing scheme as proposed in [Schnell2011], because this would introduce an exploitable structure into the Bloom filter. For more details on the weakness, see [Kroll2015].

In short, the double hashing scheme only allows for \(l^2\) different encodings for any possible n-gram, whereas the use of \(k\) different independent hash functions gives you \(\sum_{j=1}^{k}{\binom{l}{j}}\) combinations.

Our construction

It is advantageous to construct Bloom filters using a family of hash functions with the property of k-independence to compute the indices for an entry. This approach minimises the change of collisions.

An informal definition of k-independence of a family of hash functions is, that if selecting a function at random from the family, it guarantees that the hash codes of any designated k keys are independent random variables.

Our construction utilises the fact that the output bits of a cryptographic hash function are uniformly distributed, independent, binary random variables (well, at least as close to as possible. See [Kaminsky2011] for an analysis). Thus, slicing the output of a cryptographic hash function into k different slices gives you k independent random variables.

We chose Blake2 as the cryptographic hash function mainly for two reasons:

• it is fast.
• in keyed hashing mode, Blake2 provides MACs with just one hash function call instead of the two calls in the HMAC construction used in the double hashing scheme.

Warning

Please be aware that, although this construction makes the attack of [Kroll2015] infeasible, it is most likely not enough to ensure security. Or in their own words:

However, we think that using independent hash functions alone will not be sufficient to ensure security, since in this case other approaches (maybe related to or at least inspired through work from the area of Frequent Itemset Mining) are promising to detect at least the most frequent atoms automatically.

Parameters:

• ngrams – list of n-grams to be encoded
• keys – secret key for blake2 as bytes
• k – number of hash functions to use per element of the ngrams
• l – length of the output bitarray (has to be a power of 2)
• encoding – the encoding to use when turning the ngrams to bytes

Returns:

bitarray of length l with the bits set which correspond to the encoding of the ngrams

clkhash.bloomfilter.crypto_bloom_filter(record, tokenizers, fields, keys, hash_properties)

Computes the composite Bloom filter encoding of a record.

Using the method from http://www.record-linkage.de/-download=wp-grlc-2011-02.pdf

Parameters:

• record – plaintext record tuple. E.g. (index, name, dob, gender)
• tokenizers – A list of tokenizers. A tokenizer is a function that returns tokens from a string.
• fields – A list of FieldSpec. One for each field.
• keys – Keys for the hash functions as a tuple of lists of bytes.
• hash_properties – Global hashing properties.

Returns:

3-tuple: - bloom filter for record as a bitarray - first element of record (usually an index) - number of bits set in the bloomfilter

clkhash.bloomfilter.double_hash_encode_ngrams(ngrams, keys, k, l, encoding)

Computes the double hash encoding of the provided ngrams with the given keys.

Using the method from http://www.record-linkage.de/-download=wp-grlc-2011-02.pdf

Parameters:

• ngrams – list of n-grams to be encoded
• keys – hmac secret keys for md5 and sha1 as bytes
• k – number of hash functions to use per element of the ngrams
• l – length of the output bitarray
• encoding – the encoding to use when turning the ngrams to bytes

Returns:

bitarray of length l with the bits set which correspond to the encoding of the ngrams

clkhash.bloomfilter.double_hash_encode_ngrams_non_singular(ngrams, keys, k, l, encoding)

computes the double hash encoding of the provided n-grams with the given keys.

The original construction of [Schnell2011] displays an abnormality for certain inputs:

An n-gram can be encoded into just one bit irrespective of the number of k.

Their construction goes as follows: the \(k\) different indices \(g_i\) of the Bloom filter for an n-gram \(x\) are defined as:

\[g_{i}(x) = (h_1(x) + i h_2(x)) \mod l\]

with \(0 \leq i < k\) and \(l\) is the length of the Bloom filter. If the value of the hash of \(x\) of the second hash function is a multiple of \(l\), then

\[h_2(x) = 0 \mod l\]

and thus

\[g_i(x) = h_1(x) \mod l,\]

irrespective of the value \(i\). A discussion of this potential flaw can be found here.

Parameters:

• ngrams – list of n-grams to be encoded
• keys – tuple with (key_sha1, key_md5). That is, (hmac secret keys for sha1 as bytes, hmac secret keys for md5 as bytes)
• k – number of hash functions to use per element of the ngrams
• l – length of the output bitarray
• encoding – the encoding to use when turning the ngrams to bytes

Returns:

bitarray of length l with the bits set which correspond to the encoding of the ngrams

clkhash.bloomfilter.fold_xor(bloomfilter, folds)

Performs XOR folding on a Bloom filter.

If the length of the original Bloom filter is n and we perform r folds, then the length of the resulting filter is n / 2 ** r.

Parameters:

• bloomfilter – Bloom filter to fold
• folds – number of folds

Returns:

folded bloom filter

clkhash.bloomfilter.int_from_bytes()

int.from_bytes(bytes, byteorder, *, signed=False) -> int

Return the integer represented by the given array of bytes.

The bytes argument must be a bytes-like object (e.g. bytes or bytearray).

The byteorder argument determines the byte order used to represent the integer. If byteorder is ‘big’, the most significant byte is at the beginning of the byte array. If byteorder is ‘little’, the most significant byte is at the end of the byte array. To request the native byte order of the host system, use `sys.byteorder’ as the byte order value.

The signed keyword-only argument indicates whether two’s complement is used to represent the integer.

clkhash.bloomfilter.serialize_bitarray(ba)

Serialize a bitarray (bloomfilter)

clkhash.bloomfilter.stream_bloom_filters(dataset, keys, schema)

Compute composite Bloom filters (CLKs) for every record in an iterable dataset.

Parameters:

• dataset – An iterable of indexable records.
• schema – An instantiated Schema instance
• keys – A tuple of two lists of secret keys used in the HMAC.

Returns:

Generator yielding bloom filters as 3-tuples

CLK

Generate CLK from data.

clkhash.clk.chunks(seq, chunk_size)

Split seq into chunk_size-sized chunks.

Parameters:

• seq – A sequence to chunk.
• chunk_size – The size of chunk.

clkhash.clk.generate_clk_from_csv(input_f, keys, schema, validate=True, header=True, progress_bar=True)

Generate Bloom filters from CSV file, then serialise them.

This function also computes and outputs the Hamming weight (a.k.a popcount – the number of bits set to high) of the generated Bloom filters.

Parameters:

• input_f – A file-like object of csv data to hash.
• keys – A tuple of two lists of secret keys.
• schema – Schema specifying the record formats and hashing settings.
• validate – Set to False to disable validation of data against the schema. Note that this will silence warnings whose aim is to keep the hashes consistent between data sources; this may affect linkage accuracy.
• header – Set to False if the CSV file does not have a header. Set to ‘ignore’ if the CSV file does have a header but it should not be checked against the schema.
• progress_bar (bool) – Set to False to disable the progress bar.

Returns:

A list of serialized Bloom filters and a list of corresponding popcounts.

clkhash.clk.generate_clks(pii_data, schema, keys, validate=True, callback=None)

clkhash.clk.hash_and_serialize_chunk(chunk_pii_data, keys, schema)

Generate Bloom filters (ie hash) from chunks of PII then serialize the generated Bloom filters. It also computes and outputs the Hamming weight (or popcount) – the number of bits set to one – of the generated Bloom filters.

Parameters:

• chunk_pii_data – An iterable of indexable records.
• keys – A tuple of two lists of secret keys used in the HMAC.
• schema (Schema) – Schema specifying the entry formats and hashing settings.

Returns:

A list of serialized Bloom filters and a list of corresponding popcounts

key derivation

clkhash.key_derivation.generate_key_lists(master_secrets, num_identifier, key_size=64, salt=None, info=None, kdf='HKDF', hash_algo='SHA256')

Generates a derived key for each identifier for each master secret using a key derivation function (KDF).

The only supported key derivation function for now is ‘HKDF’.

The previous key usage can be reproduced by setting kdf to ‘legacy’. This is highly discouraged, as this strategy will map the same n-grams in different identifier to the same bits in the Bloom filter and thus does not lead to good results.

Parameters:

• master_secrets – a list of master secrets (either as bytes or strings)
• num_identifier – the number of identifiers
• key_size – the size of the derived keys
• salt – salt for the KDF as bytes
• info – optional context and application specific information as bytes
• kdf – the key derivation function algorithm to use
• hash_algo – the hashing algorithm to use (ignored if kdf is not ‘HKDF’)

Returns:

The derived keys. First dimension is of size num_identifier, second dimension is the same as master_secrets. A key is represented as bytes.

clkhash.key_derivation.hkdf(master_secret, num_keys, hash_algo='SHA256', salt=None, info=None, key_size=64)

Executes the HKDF key derivation function as described in rfc5869 to derive num_keys keys of size key_size from the master_secret.

Parameters:

• master_secret – input keying material
• num_keys – the number of keys the kdf should produce
• hash_algo – The hash function used by HKDF for the internal HMAC calls. The choice of hash function defines the maximum length of the output key material. Output bytes <= 255 * hash digest size (in bytes).
• salt –

  HKDF is defined to operate with and without random salt. This is done to accommodate applications where a salt value is not available. We stress, however, that the use of salt adds significantly to themstrength of HKDF, ensuring independence between different uses of the hash function, supporting “source-independent” extraction, and strengthening the analytical results that back the HKDF design.

  Random salt differs fundamentally from the initial keying

  material in two ways: it is non-secret and can be re-used.

  Ideally, the salt value is a random (or pseudorandom) string

  of the length HashLen. Yet, even a salt value of less quality (shorter in size or with limited entropy) may still make a significant contribution to the security of the output keying material.

• info – While the ‘info’ value is optional in the definition of HKDF, it is often of great importance in applications. Its main objective is to bind the derived key material to application- and context-specific information. For example, ‘info’ may contain a protocol number, algorithm identifiers, user identities, etc. In particular, it may prevent the derivation of the same keying material for different contexts (when the same input key material (IKM) is used in such different contexts). It may also accommodate additional inputs to the key expansion part, if so desired (e.g., an application may want to bind the key material to its length L, thus making L part of the ‘info’ field). There is one technical requirement from ‘info’: it should be independent of the input key material value IKM.
• key_size – the size of the produced keys

Returns:

Derived keys

random names

Module to produce a dataset of names, genders and dates of birth and manipulate that list

Currently very simple and not realistic. Additional functions for manipulating the list of names - producing reordered and subset lists with a specific overlap

ClassList class - generate a list of length n of [id, name, dob, gender] lists

TODO: Get age distribution right by using a mortality table TODO: Get first name distributions right by using distributions TODO: Generate realistic errors TODO: Add RESTfull api to generate reasonable name data as requested

class clkhash.randomnames.NameList(n)

Bases: object

Randomly generated PII records.

SCHEMA = <Schema (v1): 4 fields>

generate_random_person(n)

Generator that yields details on a person with plausible name, sex and age.

Yields:Generated data for one person tuple - (id: int, name: str(‘First Last’), birthdate: str(‘DD/MM/YYYY’), sex: str(‘M’ | ‘F’) )

generate_subsets(sz, overlap=0.8, subsets=2)

Return random subsets with nonempty intersection.

The random subsets are of specified size. If an element is common to two subsets, then it is common to all subsets. This overlap is controlled by a parameter.

Parameters:

• sz – size of subsets to generate
• overlap – size of the intersection, as fraction of the subset length
• subsets – number of subsets to generate

Raises:

ValueError –

if there aren’t sufficiently many names in the list to satisfy the request; more precisely, raises if (1 - subsets) * floor(overlap * sz)

• subsets * sz > len(self.names).

Returns:

tuple of subsets

load_names()

Loads a name database from package data

Uses data files sourced from http://www.quietaffiliate.com/free-first-name-and-last-name-databases-csv-and-sql/

schema_types

clkhash.randomnames.load_csv_data(resource_name)

Loads first column of specified CSV file from package data.

clkhash.randomnames.random_date(start, end)

Generate a random datetime between two datetime objects.

Parameters:

• start – datetime of start
• end – datetime of end

Returns:

random datetime between start and end

clkhash.randomnames.save_csv(data, headers, file)

Output generated data to file as CSV with header.

Parameters:

• data – An iterable of tuples containing raw data.
• headers – Iterable of feature names
• file – A writeable stream in which to write the CSV

schema

Schema loading and validation.

class clkhash.schema.GlobalHashingProperties(k, l, hash_type, kdf_type, xor_folds=0, hash_prevent_singularity=None, kdf_hash='SHA256', kdf_info=None, kdf_salt=None, kdf_key_size=64)

Bases: object

Stores global hashing properties.

Parameters:

• k – The number of bits of the hash to set per ngram.
• l – The length of the resulting hash in bits. This is the length after XOR folding.
• xor_folds – The number of XOR folds to perform on the hash.
• hash_type – The hashing function to use. Choices are ‘doubleHash’ and ‘blakeHash’.
• hash_prevent_singularity – Ignored unless hash_type is ‘doubleHash’. Prevents bloom filter collisions in certain cases when True.
• kdf_type – The key derivation function to use. Currently, the only permitted value is ‘HKDF’.
• kdf_hash – The hash function to use in key derivation. The options are ‘SHA256’ and ‘SHA512’.
• kdf_info – The info for key derivation. See documentation of hkdf for details.
• kdf_salt – The salt for key derivation. See documentation of hkdf for details.
• kdf_key_size – The size of the derived keys in bytes.

classmethod from_json_dict(properties_dict)

Make a GlobalHashingProperties object from a dictionary.

Parameters:properties_dict – The dictionary must have a ‘type’ key and a ‘config’ key. The ‘config’ key must map to a dictionary containing a ‘kdf’ key, which itself maps to a dictionary. That dictionary must have ‘type’, ‘hash’, ‘keySize’, ‘salt’, and ‘type’ keys. Returns:The resulting GlobalHashingProperties object.

exception clkhash.schema.MasterSchemaError

Bases: Exception

Master schema missing? Corrupted? Otherwise surprising? This is the exception for you!

class clkhash.schema.Schema(version, hashing_globals, fields)

Bases: object

Linkage Schema which describes how to encode plaintext identifiers.

Variables:

• version – Version for the schema. Needed to keep behaviour consistent between clkhash versions for the same schema.
• hashing_globals – Configuration affecting hashing of all fields. For example cryptographic salt material, bloom filter length.
• fields – Information and configuration specific to each field. For example how to validate and tokenize a phone number.

classmethod from_json_dict(schema_dict, validate=True)

Make a Schema object from a dictionary.

Parameters:

• schema_dict – This dictionary must have a ‘features’ key specifying the columns of the dataset. It must have a ‘version’ key containing the master schema version that this schema conforms to. It must have a ‘hash’ key with all the globals.
• validate – (default True) Raise an exception if the schema does not conform to the master schema.

Returns:

The resulting Schema object.

classmethod from_json_file(schema_file, validate=True)

Load a Schema object from a json file.

Parameters:

• schema_file – A JSON file containing the schema.
• validate – (default True) Raise an exception if the schema does not conform to the master schema.

Raises:

SchemaError – When the schema is invalid.

Returns:

The resulting Schema object.

exception clkhash.schema.SchemaError

Bases: Exception

The user-defined schema is invalid.

clkhash.schema.get_master_schema(version)

Loads the master schema of given version as bytes.

Parameters:version – The version of the master schema whose path we wish to retrieve. Raises:SchemaError – When the schema version is unknown. This usually means that either (a) clkhash is out of date, or (b) the schema version listed is incorrect. Returns:Bytes of the schema.

clkhash.schema.validate_schema_dict(schema)

Validate the schema.

This raises iff either the schema or the master schema are invalid. If it’s successful, it returns nothing.

Parameters:

schema – The schema to validate, as parsed by json.

Raises:

• SchemaError – When the schema is invalid.
• MasterSchemaError – When the master schema is invalid.

field_formats

Classes that specify the requirements for each column in a dataset. They take care of validation, and produce the settings required to perform the hashing.

class clkhash.field_formats.DateSpec(identifier, hashing_properties, format, description=None)

Bases: clkhash.field_formats.FieldSpec

Represents a field that holds dates.

Dates are specified as full-dates in a format that can be described as a strptime() (C89 standard) compatible format string. E.g.: the format for the standard internet format RFC3339 (e.g. 1996-12-19) is ‘%Y-%m-%d’.

ivar str format:  The format of the date.

OUTPUT_FORMAT = '%Y%m%d'

classmethod from_json_dict(json_dict)

Make a DateSpec object from a dictionary containing its properties.

Parameters:

• json_dict (dict) – This dictionary must contain a ‘format’ key. In addition, it must contain a ‘hashing’ key, whose contents are passed to FieldHashingProperties.
• json_dict – The properties dictionary.

validate(str_in)

Validates an entry in the field.

Raises InvalidEntryError iff the entry is invalid.

An entry is invalid iff (1) the string does not represent a date in the correct format; or (2) the date it represents is invalid (such as 30 February).

Parameters:

str_in (str) – String to validate.

Raises:

• InvalidEntryError – Iff entry is invalid.
• ValueError – When self.format is unrecognised.

class clkhash.field_formats.EnumSpec(identifier, hashing_properties, values, description=None)

Bases: clkhash.field_formats.FieldSpec

Represents a field that holds an enum.

The finite collection of permitted values must be specified.

Variables:values – The set of permitted values.

classmethod from_json_dict(json_dict)

Make a EnumSpec object from a dictionary containing its properties.

Parameters:json_dict (dict) – This dictionary must contain an ‘enum’ key specifying the permitted values. In addition, it must contain a ‘hashing’ key, whose contents are passed to FieldHashingProperties.

validate(str_in)

Validates an entry in the field.

Raises InvalidEntryError iff the entry is invalid.

An entry is invalid iff it is not one of the permitted values.

Parameters:str_in (str) – String to validate. Raises:InvalidEntryError – When entry is invalid.

class clkhash.field_formats.FieldHashingProperties(ngram, encoding='utf-8', weight=1, positional=False, missing_value=None)

Bases: object

Stores the settings used to hash a field. This includes the encoding and tokenisation parameters.

Variables:

• encoding (str) – The encoding to use when converting the string to bytes. Refer to Python’s documentation for possible values.
• ngram (int) – The n in n-gram. Possible values are 0, 1, and 2.
• positional (bool) – Controls whether the n-grams are positional.
• weight (float) – Controls the weight of the field in the Bloom filter.

classmethod from_json_dict(json_dict)

Make a FieldHashingProperties object from a dictionary.

Parameters:json_dict (dict) – The dictionary must have have an ‘ngram’ key. It may have ‘positional’ and ‘weight’ keys; if these are missing, then they are filled with the default values. The encoding is always set to the default value. Returns:A FieldHashingProperties instance.

replace_missing_value(str_in)

returns ‘str_in’ if it is not equals to the ‘sentinel’ as defined in the missingValue section of the schema. Else it will return the ‘replaceWith’ value.

Parameters:str_in – Returns:str_in or the missingValue replacement value

class clkhash.field_formats.FieldSpec(identifier, hashing_properties, description=None)

Bases: object

Abstract base class representing the specification of a column in the dataset. Subclasses validate entries, and modify the hashing_properties ivar to customise hashing procedures.

Variables:

• identifier (str) – The name of the field.
• description (str) – Description of the field format.
• hashing_properties (FieldHashingProperties) – The properties for hashing.

format_value(str_in)

formats the value ‘str_in’ for hashing according to this field’s spec.

There are several reasons why this might be necessary:

1. This field contains missing values which have to be replaced by some other string
2. There are several different ways to describe a specific value for this field, e.g.: all of ‘+65’, ‘ 65’, ‘65’ are valid representations of the integer 65.
3. Entries of this field might contain elements with no entropy, e.g. dates might be formatted as yyyy-mm-dd, thus all dates will have ‘-‘ at the same place. These artifacts have no value for entity resolution and should be removed.

Parameters:str_in (str) – the string to format Returns:a string representation of ‘str_in’ which is ready to be hashed

classmethod from_json_dict(field_dict)

Initialise a FieldSpec object from a dictionary of properties.

Parameters:field_dict (dict) – The properties dictionary to use. Must contain a ‘hashing’ key that meets the requirements of FieldHashingProperties. Subclasses may require Raises:InvalidSchemaError – When the properties dictionary contains invalid values. Exactly what that means is decided by the subclasses.

is_missing_value(str_in)

tests if ‘str_in’ is the sentinel value for this field

Parameters:str_in (str) – String to test if it stands for missing value Returns:True if a missing value is defined for this field and str_in matches this value

validate(str_in)

Validates an entry in the field.

Raises InvalidEntryError iff the entry is invalid.

Subclasses must override this method with their own validation. They should call the parent’s validate method via super.

Parameters:str_in (str) – String to validate. Raises:InvalidEntryError – When entry is invalid.

class clkhash.field_formats.Ignore(identifier=None)

Bases: clkhash.field_formats.FieldSpec

represent a field which will be ignored throughout the clk processing.

validate(str_in)

Validates an entry in the field.

Raises InvalidEntryError iff the entry is invalid.

Subclasses must override this method with their own validation. They should call the parent’s validate method via super.

Parameters:str_in (str) – String to validate. Raises:InvalidEntryError – When entry is invalid.

class clkhash.field_formats.IntegerSpec(identifier, hashing_properties, description=None, minimum=None, maximum=None, **kwargs)

Bases: clkhash.field_formats.FieldSpec

Represents a field that holds integers.

Minimum and maximum values may be specified.

Variables:

• minimum (int) – The minimum permitted value.
• maximum (int) – The maximum permitted value or None.

classmethod from_json_dict(json_dict)

Make a IntegerSpec object from a dictionary containing its properties.

Parameters:

• json_dict (dict) – This dictionary may contain ‘minimum’ and ‘maximum’ keys. In addition, it must contain a ‘hashing’ key, whose contents are passed to FieldHashingProperties.
• json_dict – The properties dictionary.

validate(str_in)

Validates an entry in the field.

Raises InvalidEntryError iff the entry is invalid.

An entry is invalid iff (1) the string does not represent a base-10 integer; (2) the integer is not between self.minimum and self.maximum, if those exist; or (3) the integer is negative.

Parameters:str_in (str) – String to validate. Raises:InvalidEntryError – When entry is invalid.

exception clkhash.field_formats.InvalidEntryError

Bases: ValueError

An entry in the data file does not conform to the schema.

field_spec = None

exception clkhash.field_formats.InvalidSchemaError

Bases: ValueError

The schema is not valid.

This exception is raised if, for example, a regular expression included in the schema is not syntactically correct.

class clkhash.field_formats.MissingValueSpec(sentinel, replace_with=None)

Bases: object

Stores the information about how to find and treat missing values.

Variables:

• sentinel (str) – sentinel is the string that identifies a missing value e.g.: ‘N/A’, ‘’. the sentinel will not be validated against the feature format definition
• replaceWith (str) – defines the string which replaces the sentinel whenever present, can be ‘None’, then sentinel will not be replaced.

classmethod from_json_dict(json_dict)

class clkhash.field_formats.StringSpec(identifier, hashing_properties, description=None, regex=None, case='mixed', min_length=0, max_length=None)

Bases: clkhash.field_formats.FieldSpec

Represents a field that holds strings.

One way to specify the format of the entries is to provide a regular expression that they must conform to. Another is to provide zero or more of: minimum length, maximum length, casing (lower, upper, mixed).

Each string field also specifies an encoding used when turning characters into bytes. This is stored in hashing_properties since it is needed for hashing.

Variables:

• regex – Compiled regular expression that entries must conform to. Present only if the specification is regex- -based.
• case (str) – The casing of the entries. One of ‘lower’, ‘upper’, or ‘mixed’. Default is ‘mixed’. Present only if the specification is not regex-based.
• min_length (int) – The minimum length of the string. None if there is no minimum length. Present only if the specification is not regex-based.
• max_length (int) – The maximum length of the string. None if there is no maximum length. Present only if the specification is not regex-based.

classmethod from_json_dict(json_dict)

Make a StringSpec object from a dictionary containing its properties.

Parameters:json_dict (dict) – This dictionary must contain an ‘encoding’ key associated with a Python-conformant encoding. It must also contain a ‘hashing’ key, whose contents are passed to FieldHashingProperties. Permitted keys also include ‘pattern’, ‘case’, ‘minLength’, and ‘maxLength’. Raises:InvalidSchemaError – When a regular expression is provided but is not a valid pattern.

validate(str_in)

Validates an entry in the field.

Raises InvalidEntryError iff the entry is invalid.

An entry is invalid iff (1) a pattern is part of the specification of the field and the string does not match it; (2) the string does not match the provided casing, minimum length, or maximum length; or (3) the specified encoding cannot represent the string.

Parameters:

str_in (str) – String to validate.

Raises:

• InvalidEntryError – When entry is invalid.
• ValueError – When self.case is not one of the permitted values (‘lower’, ‘upper’, or ‘mixed’).

clkhash.field_formats.spec_from_json_dict(json_dict)

Turns a dictionary into the appropriate object.

Parameters:json_dict (dict) – A dictionary with properties. Returns:An initialised instance of the appropriate FieldSpec subclass.

tokenizer

Functions to tokenize words (PII)

clkhash.tokenizer.get_tokenizer(hash_settings)

Get tokeniser function from the hash settings.

This function takes a FieldHashingProperties object. It returns a function that takes a string and tokenises based on those properties.

clkhash.tokenizer.tokenize(n, positional, word, ignore=None)

Produce n-grams of word.

Parameters:

• n – Length of n-grams.
• positional – If True, then include the index of the substring with the n-gram.
• word – The string to tokenize.
• ignore – The substring whose occurrences we remove from word before tokenization.

Raises:

ValueError – When n is negative.

Returns:

Tuple of n-gram strings.