There are three main phases in the algorithm:
-
Sanitize/format input source
-
Search for invalid blocks
-
Format invalid blocks into something meaninful
The Code frontier is a critical part of the second step
## Knowing where we’ve been
Once a code block is generated it is added onto the frontier. Then it will be sorted by indentation and frontier can be filtered. Large blocks that fully enclose a smaller block will cause the smaller block to be evicted.
CodeFrontier#<<(block) # Adds block to frontier CodeFrontier#pop # Removes block from frontier
## Knowing where we can go
Internally the frontier keeps track of “unvisited” lines which are exposed via ‘next_indent_line` when called, this method returns, a line of code with the highest indentation.
The returned line of code can be used to build a CodeBlock and then that code block is added back to the frontier. Then, the lines are removed from the “unvisited” so we don’t double-create the same block.
CodeFrontier#next_indent_line # Shows next line CodeFrontier#register_indent_block(block) # Removes lines from unvisited
## Knowing when to stop
The frontier knows how to check the entire document for a syntax error. When blocks are added onto the frontier, they’re removed from the document. When all code containing syntax errors has been added to the frontier, the document will be parsable without a syntax error and the search can stop.
CodeFrontier#holds_all_syntax_errors? # Returns true when frontier holds all syntax errors
## Filtering false positives
Once the search is completed, the frontier may have multiple blocks that do not contain the syntax error. To limit the result to the smallest subset of “invalid blocks” call:
CodeFrontier#detect_invalid_blocks
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 162
def self.combination(array)
guesses = []
1.upto(array.length).each do |size|
guesses.concat(array.combination(size).to_a)
end
guesses
end
Example:
combination([:a, :b, :c, :d]) # => [[:a], [:b], [:c], [:d], [:a, :b], [:a, :c], [:a, :d], [:b, :c], [:b, :d], [:c, :d], [:a, :b, :c], [:a, :b, :d], [:a, :c, :d], [:b, :c, :d], [:a, :b, :c, :d]]
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 53
def initialize(code_lines:, unvisited: UnvisitedLines.new(code_lines: code_lines))
@code_lines = code_lines
@unvisited = unvisited
@queue = PriorityEngulfQueue.new
@check_next = true
end
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 148
def <<(block)
@unvisited.visit_block(block)
@queue.push(block)
@check_next = true if block.invalid?
self
end
Add a block to the frontier
This method ensures the frontier always remains sorted (in indentation order) and that each code block’s lines are removed from the indentation hash so we don’t re-evaluate the same line multiple times.
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 74
def can_skip_check?
check_next = @check_next
@check_next = false
if check_next
false
else
true
end
end
Performance optimization
Parsing with ripper is expensive If we know we don’t have any blocks with invalid syntax, then we know we cannot have found the incorrect syntax yet.
When an invalid block is added onto the frontier check document state
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 61
def count
@queue.length
end
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 172
def detect_invalid_blocks
self.class.combination(@queue.to_a.select(&:invalid?)).detect do |block_array|
holds_all_syntax_errors?(block_array, can_cache: false)
end || []
end
Given that we know our syntax error exists somewhere in our frontier, we want to find the smallest possible set of blocks that contain all the syntax errors
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 111
def expand?
return false if @queue.empty?
return true if @unvisited.empty?
frontier_indent = @queue.peek.current_indent
unvisited_indent = next_indent_line.indent
if ENV["SYNTAX_SUGGEST_DEBUG"]
puts "```"
puts @queue.peek
puts "```"
puts " @frontier indent: #{frontier_indent}"
puts " @unvisited indent: #{unvisited_indent}"
end
# Expand all blocks before moving to unvisited lines
frontier_indent >= unvisited_indent
end
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 89
def holds_all_syntax_errors?(block_array = @queue, can_cache: true)
return false if can_cache && can_skip_check?
without_lines = block_array.to_a.flat_map do |block|
block.lines
end
SyntaxSuggest.valid_without?(
without_lines: without_lines,
code_lines: @code_lines
)
end
Returns true if the document is valid with all lines removed. By default it checks all blocks in present in the frontier array, but can be used for arbitrary arrays of codeblocks as well
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 107
def next_indent_line
@unvisited.peek
end
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 103
def pop
@queue.pop
end
Returns a code block with the largest indentation possible
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 140
def register_engulf_block(block)
end
When one element fully encapsulates another we remove the smaller block from the frontier. This prevents double expansions and all-around weird behavior. However this guarantee is quite expensive to maintain
# File tmp/rubies/ruby-3.4.0-preview1/lib/syntax_suggest/code_frontier.rb, line 132
def register_indent_block(block)
@unvisited.visit_block(block)
self
end
Keeps track of what lines have been added to blocks and which are not yet visited.