Tool to generate a Solidity smart contract for a given regular expression.
npm install -g solregex
Provide optional --name
parameter and regex as argument.
$ solregex --name EmailRegex '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-_]+\.[a-zA-Z]{2,}'
solregex
prints the contents of a standalone Solidity source file (.sol
)
to your terminal's standard out.
You may want to:
solregex 'ab*c?' > Regex.sol
)solregex '.+abc.*' | pbcopy
on macOS, orThere are many different workflows for developing / deploying applications with Ethereum.
Generally the process follows these steps:
solregex
with a regular expression (generates Solidity smart contract)solregex
supports generating Graphviz DOT output for a given regular
expression's DFA (deterministic finite automaton).
To generate DOT output instead of Solidity, pass the --dot
parameter.
Sample DOT Output: solregex --dot '[a-f]x|[d-i]y|[g-l]z' | dot -Tsvg > sample-regex.svg
A contract to match email addresses is deployed at
0x537837D00047C874D19B68E94ADbA107674C21b8
(Etherscan)
A contract to match Ethereum addresses is deployed at
0x62C8b4aC2aEF3Ed13B929cA9FB20caCB222E3fA6
(Etherscan)
Compiling a regular expression to Solidity is done via several steps:
Parse regex using regjsparser
Build NFA (non-deterministic finite automaton) from parse result. Use graph.js for underlying state machine data.
Split overlapping character class ranges into non-overlapping subset ranges
(e.g. [a-f]
, [d-i]
become [a-c]
, [d-f]
, [g-i]
) using interval trees
Ref. Graphviz output above for example to highlight this behavior.
Use powerset construction to convert NFA to DFA (deterministic finite automaton)
Convert DFA into Solidity source using a handlebars template.
Supports disjunctions |
, alternations (e.g. ab
, concatenation), quantifiers
(+
, *
, ?
, {n}
, {n,m}
, {n,}
), wildcard matching (.
), quantified
groups ((...)*
, etc.), character classes (positive, negative, ranges)
Supports true
/false
result for string matching against a regex.
^
, $
for start/end). Currently "enabled" by default.(a*)(b*)
indicating a/b groups in input string)(a*)\1
)Quantifiers using numeric literals (e.g. a{40}
) generate numerous resulting
DFA states. This makes the output code very large very fast.
It may be possible to add support for compressing mostly-identical states into a single state with parameters, to avoid so much output code.
Feel free to contact me in the Gitter channel for this repository with any comments, concerns, questions. Let me know if anything is unclear about usage or if you encounter any problems!
If you are interested in helping improve the state of efficient string pattern matching on the EVM, get in touch or open a pull request! Feedback, fixes, and improvements of all kinds are most appreciated :). Thank you!