fix(lsp): refactor escaping snippet text (#25611)

2 files changed, 72 insertions, 42 deletions

diff --git a/runtime/lua/vim/lsp/_snippet_grammar.lua b/runtime/lua/vim/lsp/_snippet_grammar.lua
index 86f00fea51..eb72efdf39 100644
--- a/runtime/lua/vim/lsp/_snippet_grammar.lua
+++ b/runtime/lua/vim/lsp/_snippet_grammar.lua
@@ -20,11 +20,15 @@ local format_capture = Cg(int / tonumber, 'capture')
 local format_modifier = Cg(P('upcase') + P('downcase') + P('capitalize'), 'modifier')
 local tabstop = Cg(int / tonumber, 'tabstop')
 
+-- These characters are always escapable in text nodes no matter the context.
+local escapable = '$}\\'
+
 --- Returns a function that unescapes occurrences of "special" characters.
 ---
---- @param special string
+--- @param special? string
 --- @return fun(match: string): string
 local function escape_text(special)
+  special = special or escapable
   return function(match)
     local escaped = match:gsub('\\(.)', function(c)
       return special:find(c) and c or '\\' .. c
@@ -33,25 +37,33 @@ local function escape_text(special)
   end
 end
 
--- Text nodes match "any character", but $, \, and } must be escaped.
-local escapable = '$}\\'
-local text = (backslash * S(escapable)) + (P(1) - S(escapable))
-local text_0, text_1 = (text ^ 0) / escape_text(escapable), text ^ 1
+--- Returns a pattern for text nodes. Will match characters in `escape` when preceded by a backslash,
+--- and will stop with characters in `stop_with`.
+---
+--- @param escape string
+--- @param stop_with? string
+--- @return vim.lpeg.Pattern
+local function text(escape, stop_with)
+  stop_with = stop_with or escape
+  return (backslash * S(escape)) + (P(1) - S(stop_with))
+end
+
+-- For text nodes inside curly braces. It stops parsing when reaching an escapable character.
+local braced_text = (text(escapable) ^ 0) / escape_text()
+
 -- Within choice nodes, \ also escapes comma and pipe characters.
-local choice_text = C(((backslash * S(escapable .. ',|')) + (P(1) - S(escapable .. ',|'))) ^ 1)
-  / escape_text(escapable .. ',|')
-local if_text, else_text = Cg(text_0, 'if_text'), Cg(text_0, 'else_text')
+local choice_text = C(text(escapable .. ',|') ^ 1) / escape_text(escapable .. ',|')
+
 -- Within format nodes, make sure we stop at /
-local format_text = C(((backslash * S(escapable)) + (P(1) - S(escapable .. '/'))) ^ 1)
-  / escape_text(escapable)
+local format_text = C(text(escapable, escapable .. '/') ^ 1) / escape_text()
+
+local if_text, else_text = Cg(braced_text, 'if_text'), Cg(braced_text, 'else_text')
+
 -- Within ternary condition format nodes, make sure we stop at :
-local if_till_colon_text = Cg(
-  C(((backslash * S(escapable)) + (P(1) - S(escapable .. ':'))) ^ 1) / escape_text(escapable),
-  'if_text'
-)
+local if_till_colon_text = Cg(C(text(escapable, escapable .. ':') ^ 1) / escape_text(), 'if_text')
 
 -- Matches the string inside //, allowing escaping of the closing slash.
-local regex = Cg(((backslash * slash) + (P(1) - slash)) ^ 1, 'regex')
+local regex = Cg(text('/') ^ 1, 'regex')
 
 -- Regex constructor flags (see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/RegExp/RegExp#parameters).
 local options = Cg(S('dgimsuvy') ^ 0, 'options')
@@ -94,11 +106,11 @@ local G = P({
   snippet = Ct(Cg(
     Ct((
       V('any') +
-      (Ct(Cg(text_1 / escape_text(escapable), 'text')) / node(Type.Text))
+      (Ct(Cg((text(escapable, '$') ^ 1) / escape_text(), 'text')) / node(Type.Text))
     ) ^ 1), 'children'
-  )) / node(Type.Snippet),
-  any_or_text = V('any') + (Ct(Cg(text_0 / escape_text(escapable), 'text')) / node(Type.Text)),
+  ) * -P(1)) / node(Type.Snippet),
   any = V('placeholder') + V('tabstop') + V('choice') + V('variable'),
+  any_or_text = V('any') + (Ct(Cg(braced_text, 'text')) / node(Type.Text)),
   tabstop = Ct(dollar * (tabstop + (l_brace * tabstop * r_brace))) / node(Type.Tabstop),
   placeholder = Ct(dollar * l_brace * tabstop * colon * Cg(V('any_or_text'), 'value') * r_brace) / node(Type.Placeholder),
   choice = Ct(dollar *
diff --git a/test/functional/plugin/lsp/snippet_spec.lua b/test/functional/plugin/lsp/snippet_spec.lua
index 13df861b91..ba8bc7fe04 100644
--- a/test/functional/plugin/lsp/snippet_spec.lua
+++ b/test/functional/plugin/lsp/snippet_spec.lua
@@ -1,5 +1,6 @@
 local helpers = require('test.functional.helpers')(after_each)
 local snippet = require('vim.lsp._snippet_grammar')
+local type = snippet.NodeType
 
 local eq = helpers.eq
 local exec_lua = helpers.exec_lua
@@ -15,28 +16,28 @@ describe('vim.lsp._snippet_grammar', function()
 
   it('parses only text', function()
     eq({
-      { type = snippet.NodeType.Text, data = { text = 'TE$}XT' } },
+      { type = type.Text, data = { text = 'TE$}XT' } },
     }, parse('TE\\$\\}XT'))
   end)
 
   it('parses tabstops', function()
     eq({
-      { type = snippet.NodeType.Tabstop, data = { tabstop = 1 } },
-      { type = snippet.NodeType.Tabstop, data = { tabstop = 2 } },
+      { type = type.Tabstop, data = { tabstop = 1 } },
+      { type = type.Tabstop, data = { tabstop = 2 } },
     }, parse('$1${2}'))
   end)
 
   it('parses nested placeholders', function()
     eq({
       {
-        type = snippet.NodeType.Placeholder,
+        type = type.Placeholder,
         data = {
           tabstop = 1,
           value = {
-            type = snippet.NodeType.Placeholder,
+            type = type.Placeholder,
             data = {
               tabstop = 2,
-              value = { type = snippet.NodeType.Tabstop, data = { tabstop = 3 } },
+              value = { type = type.Tabstop, data = { tabstop = 3 } },
             },
           },
         },
@@ -46,24 +47,24 @@ describe('vim.lsp._snippet_grammar', function()
 
   it('parses variables', function()
     eq({
-      { type = snippet.NodeType.Variable, data = { name = 'VAR' } },
-      { type = snippet.NodeType.Variable, data = { name = 'VAR' } },
+      { type = type.Variable, data = { name = 'VAR' } },
+      { type = type.Variable, data = { name = 'VAR' } },
       {
-        type = snippet.NodeType.Variable,
+        type = type.Variable,
         data = {
           name = 'VAR',
-          default = { type = snippet.NodeType.Tabstop, data = { tabstop = 1 } },
+          default = { type = type.Tabstop, data = { tabstop = 1 } },
         },
       },
       {
-        type = snippet.NodeType.Variable,
+        type = type.Variable,
         data = {
           name = 'VAR',
           regex = 'regex',
           options = '',
           format = {
             {
-              type = snippet.NodeType.Format,
+              type = type.Format,
               data = { capture = 1, modifier = 'upcase' },
             },
           },
@@ -75,7 +76,7 @@ describe('vim.lsp._snippet_grammar', function()
   it('parses choice', function()
     eq({
       {
-        type = snippet.NodeType.Choice,
+        type = type.Choice,
         data = { tabstop = 1, values = { ',', '|' } },
       },
     }, parse('${1|\\,,\\||}'))
@@ -85,30 +86,30 @@ describe('vim.lsp._snippet_grammar', function()
     eq(
       {
         {
-          type = snippet.NodeType.Variable,
+          type = type.Variable,
           data = {
             name = 'VAR',
             regex = 'regex',
             options = '',
             format = {
               {
-                type = snippet.NodeType.Format,
+                type = type.Format,
                 data = { capture = 1, modifier = 'upcase' },
               },
               {
-                type = snippet.NodeType.Format,
+                type = type.Format,
                 data = { capture = 1, if_text = 'if_text' },
               },
               {
-                type = snippet.NodeType.Format,
+                type = type.Format,
                 data = { capture = 1, else_text = 'else_text' },
               },
               {
-                type = snippet.NodeType.Format,
+                type = type.Format,
                 data = { capture = 1, if_text = 'if_text', else_text = 'else_text' },
               },
               {
-                type = snippet.NodeType.Format,
+                type = type.Format,
                 data = { capture = 1, else_text = 'else_text' },
               },
             },
@@ -124,24 +125,24 @@ describe('vim.lsp._snippet_grammar', function()
   it('parses empty strings', function()
     eq({
       {
-        type = snippet.NodeType.Placeholder,
+        type = type.Placeholder,
         data = {
           tabstop = 1,
-          value = { type = snippet.NodeType.Text, data = { text = '' } },
+          value = { type = type.Text, data = { text = '' } },
         },
       },
       {
-        type = snippet.NodeType.Text,
+        type = type.Text,
         data = { text = ' ' },
       },
       {
-        type = snippet.NodeType.Variable,
+        type = type.Variable,
         data = {
           name = 'VAR',
           regex = 'erg',
           format = {
             {
-              type = snippet.NodeType.Format,
+              type = type.Format,
               data = { capture = 1, if_text = '' },
             },
           },
@@ -150,4 +151,21 @@ describe('vim.lsp._snippet_grammar', function()
       },
     }, parse('${1:} ${VAR/erg/${1:+}/g}'))
   end)
+
+  it('parses closing curly brace as text', function()
+    eq(
+      {
+        { type = type.Text, data = { text = 'function ' } },
+        { type = type.Tabstop, data = { tabstop = 1 } },
+        { type = type.Text, data = { text = '() {\n  ' } },
+        { type = type.Tabstop, data = { tabstop = 0 } },
+        { type = type.Text, data = { text = '\n}' } },
+      },
+      parse(table.concat({
+        'function $1() {',
+        '  $0',
+        '}',
+      }, '\n'))
+    )
+  end)
 end)