Merge pull request #907 from jgmdev/PR/less-hacky-tokenizer-fix

syntax: remove pattern re-ordering on optimization
This commit is contained in:
Jefferson González 2022-03-29 22:26:50 -04:00 committed by GitHub
commit 693bf2cf29
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 45 additions and 59 deletions

View File

@ -7,47 +7,22 @@ local plain_text_syntax = { name = "Plain Text", patterns = {}, symbols = {} }
function syntax.add(t)
-- the rule %s+ gives us a performance gain for the tokenizer in lines with
-- long amounts of consecutive spaces, to not affect other patterns we
-- insert it after any rule that starts with spaces to prevent conflicts
if type(t.space_handling) ~= "boolean" then t.space_handling = true end
if t.patterns then
local temp_patterns = {}
::pattern_remove_loop::
for pos, pattern in ipairs(t.patterns) do
local pattern_str = ""
local ptype = pattern.pattern
and "pattern" or (pattern.regex and "regex" or nil)
if ptype then
if type(pattern[ptype]) == "table" then
pattern_str = pattern[ptype][1]
else
pattern_str = pattern[ptype]
-- the rule %s+ gives us a performance gain for the tokenizer in lines with
-- long amounts of consecutive spaces, can be disabled by plugins where it
-- causes conflicts by declaring the table property: space_handling = false
if t.space_handling then
table.insert(t.patterns, { pattern = "%s+", type = "normal" })
end
if (ptype == "pattern" and(
pattern_str:find("^%^?%%s")
or
pattern_str:find("^%^?%s")
))
or
(ptype == "regex" and (
pattern_str:find("^%^?\\s")
or
pattern_str:find("^%^?%s")
))
then
table.insert(temp_patterns, table.remove(t.patterns, pos))
-- since we are removing from iterated table we need to start
-- from the beginning again to prevent any issues
goto pattern_remove_loop
end
end
end
for pos, pattern in ipairs(temp_patterns) do
table.insert(t.patterns, pos, pattern)
end
local pos = 1
if #temp_patterns > 0 then pos = #temp_patterns+1 end
table.insert(t.patterns, pos, { pattern = "%s+", type = "normal" })
-- this rule gives us additional performance gain by matching every word
-- that was not matched by the syntax patterns as a single token, preventing
-- the tokenizer from iterating over each character individually which is a
-- lot slower since iteration occurs in lua instead of C and adding to that
-- it will also try to match every pattern to a single char (same as spaces)
table.insert(t.patterns, { pattern = "%w+%f[%s]", type = "normal" })
end
table.insert(syntax.items, t)

View File

@ -31,7 +31,35 @@ syntax.add {
name = "Markdown",
files = { "%.md$", "%.markdown$" },
block_comment = { "<!--", "-->" },
space_handling = false, -- turn off this feature to handle it our selfs
patterns = {
---- Place patterns that require spaces at start to optimize matching speed
---- and apply the %s+ optimization immediately afterwards
-- bullets
{ pattern = "^%s*%*%s", type = "number" },
{ pattern = "^%s*%-%s", type = "number" },
{ pattern = "^%s*%+%s", type = "number" },
-- numbered bullet
{ pattern = "^%s*[0-9]+[%.%)]%s", type = "number" },
-- blockquote
{ pattern = "^%s*>+%s", type = "string" },
-- alternative bold italic formats
{ pattern = { "%s___", "___%f[%s]" }, type = "markdown_bold_italic" },
{ pattern = { "%s__", "__%f[%s]" }, type = "markdown_bold" },
{ pattern = { "%s_[%S]", "_%f[%s]" }, type = "markdown_italic" },
-- reference links
{
pattern = "^%s*%[%^()["..in_squares_match.."]+()%]: ",
type = { "function", "number", "function" }
},
{
pattern = "^%s*%[%^?()["..in_squares_match.."]+()%]:%s+.+\n",
type = { "function", "number", "function" }
},
-- optimization
{ pattern = "%s+", type = "normal" },
---- HTML rules imported and adapted from language_html
---- to not conflict with markdown rules
-- Inline JS and CSS
@ -129,14 +157,6 @@ syntax.add {
{ pattern = "^%-%-%-+\n", type = "comment" },
{ pattern = "^%*%*%*+\n", type = "comment" },
{ pattern = "^___+\n", type = "comment" },
-- bullets
{ pattern = "^%s*%*%s", type = "number" },
{ pattern = "^%s*%-%s", type = "number" },
{ pattern = "^%s*%+%s", type = "number" },
-- numbered bullet
{ pattern = "^%s*[0-9]+[%.%)]%s", type = "number" },
-- blockquote
{ pattern = "^%s*>+%s", type = "string" },
-- bold and italic
{ pattern = { "%*%*%*%S", "%*%*%*" }, type = "markdown_bold_italic" },
{ pattern = { "%*%*%S", "%*%*" }, type = "markdown_bold" },
@ -149,9 +169,6 @@ syntax.add {
{ pattern = "^___[%s%p%w]+___%s" , type = "markdown_bold_italic" },
{ pattern = "^__[%s%p%w]+__%s" , type = "markdown_bold" },
{ pattern = "^_[%s%p%w]+_%s" , type = "markdown_italic" },
{ pattern = { "%s___", "___%f[%s]" }, type = "markdown_bold_italic" },
{ pattern = { "%s__", "__%f[%s]" }, type = "markdown_bold" },
{ pattern = { "%s_[%S]", "_%f[%s]" }, type = "markdown_italic" },
-- heading with custom id
{
pattern = "^#+%s[%w%s%p]+(){()#[%w%-]+()}",
@ -186,14 +203,6 @@ syntax.add {
pattern = "%[()["..in_squares_match.."]+()%] *()%[()["..in_squares_match.."]+()%]",
type = { "function", "string", "function", "function", "number", "function" }
},
{
pattern = "^%s*%[%^()["..in_squares_match.."]+()%]: ",
type = { "function", "number", "function" }
},
{
pattern = "^%s*%[%^?()["..in_squares_match.."]+()%]:%s+.+\n",
type = { "function", "number", "function" }
},
{
pattern = "!?%[%^?()["..in_squares_match.."]+()%]",
type = { "function", "number", "function" }
@ -204,7 +213,9 @@ syntax.add {
type = "function"
},
{ pattern = "<https?://%S+>", type = "function" },
{ pattern = "https?://%S+", type = "function" }
{ pattern = "https?://%S+", type = "function" },
-- optimize consecutive dashes used in tables
{ pattern = "%-+", type = "normal" },
},
symbols = { },
}