refactor(lsp): use LPeg for watchfiles matching (#23788)

1 files changed, 70 insertions, 21 deletions

diff --git a/test/functional/plugin/lsp/watchfiles_spec.lua b/test/functional/plugin/lsp/watchfiles_spec.lua
index c5d6803a7f..a8260e0c98 100644
--- a/test/functional/plugin/lsp/watchfiles_spec.lua
+++ b/test/functional/plugin/lsp/watchfiles_spec.lua
@@ -2,7 +2,6 @@ local helpers = require('test.functional.helpers')(after_each)
 
 local eq = helpers.eq
 local exec_lua = helpers.exec_lua
-local has_err = require('luassert').has.errors
 
 describe('vim.lsp._watchfiles', function()
   before_each(helpers.clear)
@@ -17,21 +16,31 @@ describe('vim.lsp._watchfiles', function()
       eq(true, match('', ''))
       eq(false, match('', 'a'))
       eq(true, match('a', 'a'))
+      eq(true, match('/', '/'))
       eq(true, match('abc', 'abc'))
       eq(false, match('abc', 'abcdef'))
       eq(false, match('abc', 'a'))
+      eq(false, match('abc', 'bc'))
       eq(false, match('a', 'b'))
       eq(false, match('.', 'a'))
       eq(true, match('$', '$'))
+      eq(true, match('/dir', '/dir'))
+      eq(true, match('dir/', 'dir/'))
+      eq(true, match('dir/subdir', 'dir/subdir'))
+      eq(false, match('dir/subdir', 'subdir'))
       eq(false, match('dir/subdir', 'dir/subdir/file'))
+      eq(true, match('🤠', '🤠'))
     end)
 
     it('should match * wildcards', function()
-      -- eq(false, match('*', '')) -- TODO: this fails
+      eq(false, match('*', ''))
       eq(true, match('*', 'a'))
+      eq(false, match('*', '/'))
       eq(false, match('*', '/a'))
       eq(false, match('*', 'a/'))
       eq(true, match('*', 'aaa'))
+      eq(true, match('*a', 'aa'))
+      eq(true, match('*a', 'abca'))
       eq(true, match('*.txt', 'file.txt'))
       eq(false, match('*.txt', 'file.txtxt'))
       eq(false, match('*.txt', 'dir/file.txt'))
@@ -40,9 +49,31 @@ describe('vim.lsp._watchfiles', function()
       eq(false, match('*.dir', 'test.dir/file'))
       eq(true, match('file.*', 'file.txt'))
       eq(false, match('file.*', 'not-file.txt'))
+      eq(true, match('*/file.txt', 'dir/file.txt'))
+      eq(false, match('*/file.txt', 'dir/subdir/file.txt'))
+      eq(false, match('*/file.txt', '/dir/file.txt'))
+      eq(true, match('dir/*', 'dir/file.txt'))
+      eq(false, match('dir/*', 'dir'))
       eq(false, match('dir/*.txt', 'file.txt'))
       eq(true, match('dir/*.txt', 'dir/file.txt'))
       eq(false, match('dir/*.txt', 'dir/subdir/file.txt'))
+      eq(false, match('dir/*/file.txt', 'dir/file.txt'))
+      eq(true, match('dir/*/file.txt', 'dir/subdir/file.txt'))
+      eq(false, match('dir/*/file.txt', 'dir/subdir/subdir/file.txt'))
+
+      -- TODO: The spec does not describe this, but VSCode only interprets ** when it's by
+      -- itself in a path segment, and otherwise interprets ** as consecutive * directives.
+      -- The following tests show how this behavior should work, but is not yet fully implemented.
+      -- Currently, "a**" parses incorrectly as "a" "**" and "**a" parses correctly as "*" "*" "a".
+      -- see: https://github.com/microsoft/vscode/blob/eef30e7165e19b33daa1e15e92fa34ff4a5df0d3/src/vs/base/common/glob.ts#L112
+      eq(true, match('a**', 'abc')) -- '**' should parse as two '*'s when not by itself in a path segment
+      eq(true, match('**c', 'abc'))
+      -- eq(false, match('a**', 'ab')) -- each '*' should still represent at least one character
+      eq(false, match('**c', 'bc'))
+      eq(true, match('a**', 'abcd'))
+      eq(true, match('**d', 'abcd'))
+      -- eq(false, match('a**', 'abc/d'))
+      eq(false, match('**d', 'abc/d'))
     end)
 
     it('should match ? wildcards', function()
@@ -58,52 +89,64 @@ describe('vim.lsp._watchfiles', function()
     it('should match ** wildcards', function()
       eq(true, match('**', ''))
       eq(true, match('**', 'a'))
+      eq(true, match('**', '/'))
       eq(true, match('**', 'a/'))
       eq(true, match('**', '/a'))
       eq(true, match('**', 'C:/a'))
       eq(true, match('**', 'a/a'))
       eq(true, match('**', 'a/a/a'))
-      eq(false, match('a**', ''))
-      eq(true, match('a**', 'a'))
-      eq(true, match('a**', 'abcd'))
-      eq(false, match('a**', 'ba'))
-      eq(false, match('a**', 'a/b'))
-      eq(false, match('**a', ''))
-      eq(true, match('**a', 'a'))
-      eq(true, match('**a', 'dcba'))
-      eq(false, match('**a', 'ab'))
-      eq(false, match('**a', 'b/a'))
+      eq(false, match('/**', '')) -- /** matches leading / literally
+      eq(true, match('/**', '/'))
+      eq(true, match('/**', '/a/b/c'))
+      eq(true, match('**/', '')) -- **/ absorbs trailing /
+      eq(true, match('**/', '/a/b/c'))
+      eq(true, match('**/**', ''))
+      eq(true, match('**/**', 'a'))
       eq(false, match('a/**', ''))
-      eq(true, match('a/**', 'a'))
+      eq(false, match('a/**', 'a'))
       eq(true, match('a/**', 'a/b'))
+      eq(true, match('a/**', 'a/b/c'))
       eq(false, match('a/**', 'b/a'))
       eq(false, match('a/**', '/a'))
       eq(false, match('**/a', ''))
       eq(true, match('**/a', 'a'))
       eq(false, match('**/a', 'a/b'))
       eq(true, match('**/a', '/a'))
+      eq(true, match('**/a', '/b/a'))
+      eq(true, match('**/a', '/c/b/a'))
+      eq(true, match('**/a', '/a/a'))
+      eq(true, match('**/a', '/abc/a'))
       eq(false, match('a/**/c', 'a'))
       eq(false, match('a/**/c', 'c'))
       eq(true, match('a/**/c', 'a/c'))
       eq(true, match('a/**/c', 'a/b/c'))
       eq(true, match('a/**/c', 'a/b/b/c'))
-      eq(true, match('**/a/**', 'a'))
-      eq(true, match('**/a/**', '/dir/a'))
+      eq(false, match('**/a/**', 'a'))
+      eq(true, match('**/a/**', 'a/'))
+      eq(false, match('**/a/**', '/dir/a'))
+      eq(false, match('**/a/**', 'dir/a'))
+      eq(true, match('**/a/**', 'dir/a/'))
       eq(true, match('**/a/**', 'a/dir'))
       eq(true, match('**/a/**', 'dir/a/dir'))
       eq(true, match('**/a/**', '/a/dir'))
       eq(true, match('**/a/**', 'C:/a/dir'))
-      -- eq(false, match('**/a/**', 'a.txt')) -- TODO: this fails
+      eq(false, match('**/a/**', 'a.txt'))
     end)
 
     it('should match {} groups', function()
-      eq(false, match('{}', ''))
-      eq(true, match('{,}', ''))
+      eq(true, match('{}', ''))
       eq(false, match('{}', 'a'))
+      eq(true, match('a{}', 'a'))
+      eq(true, match('{}a', 'a'))
+      eq(true, match('{,}', ''))
+      eq(true, match('{a,}', ''))
+      eq(true, match('{a,}', 'a'))
       eq(true, match('{a}', 'a'))
       eq(false, match('{a}', 'aa'))
       eq(false, match('{a}', 'ab'))
+      eq(true, match('{a?c}', 'abc'))
       eq(false, match('{ab}', 'a'))
+      eq(false, match('{ab}', 'b'))
       eq(true, match('{ab}', 'ab'))
       eq(true, match('{a,b}', 'a'))
       eq(true, match('{a,b}', 'b'))
@@ -112,11 +155,11 @@ describe('vim.lsp._watchfiles', function()
       eq(false, match('{ab,cd}', 'a'))
       eq(true, match('{ab,cd}', 'cd'))
       eq(true, match('{a,b,c}', 'c'))
-      eq(false, match('{a,{b,c}}', 'c')) -- {} can't nest
+      eq(true, match('{a,{b,c}}', 'c'))
     end)
 
     it('should match [] groups', function()
-      eq(true, match('[]', ''))
+      eq(true, match('[]', '[]')) -- empty [] is a literal
       eq(false, match('[a-z]', ''))
       eq(true, match('[a-z]', 'a'))
       eq(false, match('[a-z]', 'ab'))
@@ -141,7 +184,7 @@ describe('vim.lsp._watchfiles', function()
     end)
 
     it('should match [!...] groups', function()
-      has_err(function() match('[!]', '') end) -- not a valid pattern
+      eq(true, match('[!]', '[!]')) -- [!] is a literal
       eq(false, match('[!a-z]', ''))
       eq(false, match('[!a-z]', 'a'))
       eq(false, match('[!a-z]', 'z'))
@@ -159,11 +202,17 @@ describe('vim.lsp._watchfiles', function()
     it('should match complex patterns', function()
       eq(false, match('**/*.{c,h}', ''))
       eq(false, match('**/*.{c,h}', 'c'))
+      eq(false, match('**/*.{c,h}', 'file.m'))
       eq(true, match('**/*.{c,h}', 'file.c'))
       eq(true, match('**/*.{c,h}', 'file.h'))
       eq(true, match('**/*.{c,h}', '/file.c'))
       eq(true, match('**/*.{c,h}', 'dir/subdir/file.c'))
       eq(true, match('**/*.{c,h}', 'dir/subdir/file.h'))
+      eq(true, match('**/*.{c,h}', '/dir/subdir/file.c'))
+      eq(true, match('**/*.{c,h}', 'C:/dir/subdir/file.c'))
+      eq(true, match('/dir/**/*.{c,h}', '/dir/file.c'))
+      eq(false, match('/dir/**/*.{c,h}', 'dir/file.c'))
+      eq(true, match('/dir/**/*.{c,h}', '/dir/subdir/subdir/file.c'))
 
       eq(true, match('{[0-9],[a-z]}', '0'))
       eq(true, match('{[0-9],[a-z]}', 'a'))