Lark Cheat Sheet

A Programming › Python cheat sheet in English by erezsh

Cheatsheet for the Lark parsing toolkit

23,707 views 3,181 downloads 26 May 2018 History

PRs

3.7 avg · 0 ratings

Lark Options

`parser="earley"`	Earley - default
`parser="lalr"`	LALR(1)
`debug=True`	Enable debug prints
`lexer="standard"`	Revert to simple lexer
`ambiguity='explicit'`	Return all derivations for Earley
`start="foo"`	Set starting rule
`cache=True`	Enable grammar caching
`transformer=...`	Apply transformer to tree (for LALR)
`propagate_positions`	Fill tree instances with line number information
`maybe_placeholders`	`[]` returns `None` when not matched
`keep_all_tokens=True`	Don't remove unnamed terminals
`postlex`	Provide a wrapper for the lexer
`tree_class`	Provide an alternative for `Tree`
`regex=True`	Use the `regex` module

Tree Reference

`tree.data`	Rule name
`tree.children`	Rule matches
`tree.meta`	Positional information, if enabled
`print(tree.pretty())`
`tree.iter_subtrees()`	Iterate all subtrees
`tree.find_data("foo")`	Find by rule
`tree.find_pred(...)`	Find by predicate
`tree1 == tree2`

Token Reference

`token.type`	Terminal name
`token.value`	Matched text
`token.pos_in_stream`	Index in source text
`token.line`
`token.column`
`token.end_line`
`token.end_column`
`token.end_pos`
`len(token)`

Tokens inherit from str, so all string operations are valid (such as token.upper()).

Grammar Definitions

`rule: ...`	Define a rule
`TERM: ...`	Define a terminal
`rule.n: ...`	Rule with priority n
`TERM.n: ...`	Terminal with priority n
`// text`	Comment
`%ignore ...`	Ignore terminal in input
`%import ...`	Import terminal from file
`%declare TERM`	Declare a terminal without a pattern (used for postlex)
`t{p1, p2}: ...`	Define template
`rule: t{foo, bar}`	Use template

Rules consist of values, other rules and terminals.
Terminals only consist of values and other terminals.

Grammar Patterns

`foo bar`	Match sequence
`(foo bar)`	Group together (for operations)
`foo \| bar`	Match one or the other
`foo?`	Match 0 or 1 instances
`[foo bar]`	Match 0 or 1 instances
`foo*`	Match 0 or more instances
`foo+`	Match 1 or more instances
`foo~3`	Match exactly 3 instances
`foo~3..5`	Match between 3 to 5 instances

Terminal Atoms

`"string"`	String to match
`"string"i`	Case-insensitive string
`/regexp/`	Regular Expression
`/re/imslux`	Regular Expression with flags
`"a".."z"`	Literal range

Tree Shaping

`rule: "foo" BAR`	"foo" will be filtered out
`!rule: "foo" BAR`	"foo" will be kept
`rule: /foo/ BAR`	/foo/ will be kept
`_TERM`	Filter out this terminal
`_rule`	Always inline this rule
`?rule: ...`	Inline if matched 1 child
`foo bar -> new_name`	Rename this derivation

Rules are a branch (node) in the resulting tree, and its children are its matches, in the order of matching.
Terminals (tokens) are always values in the tree, never branches.
Inlining rules means removing their branch and replacing it with their children.

Examples

// Define template for comma-separated list
cs_list{item}: item ("," item)*

// Use template to make a list of numbers
number_list: cs_list{ number }

// Example of a terminal for a Python comment
PY_COMMENT: /#[^\n]*/

// Example of a terminal for C comment
C_COMMENT: "/*" /.*?/s "*/"

Badges

Nice Cheat Sheet Downloaded from Unusual Locations × 2

Comments

No comments yet.