fix(lsp): small bugs in snippet-parser #18998

This fixes the following bugs:
`${1:else_text}` -> format with if_text: "else_text"
`${1:-else_text}` -> format with if_text: "else_text"
`${1:}` in `format` (eg. empty else_text) -> error.
`${1:}` (eg. empty placeholder) -> error.

Thanks hrsh7th :)

2 files changed, 154 insertions, 37 deletions

diff --git a/runtime/lua/vim/lsp/_snippet.lua b/runtime/lua/vim/lsp/_snippet.lua
index 28064f36e9..910deba556 100644
--- a/runtime/lua/vim/lsp/_snippet.lua
+++ b/runtime/lua/vim/lsp/_snippet.lua
@@ -156,10 +156,10 @@ P.seq = function(...)
   return function(input, pos)
     local values = {}
     local new_pos = pos
-    for _, parser in ipairs(parsers) do
+    for i, parser in ipairs(parsers) do
       local result = parser(input, new_pos)
       if result.parsed then
-        table.insert(values, result.value)
+        values[i] = result.value
         new_pos = result.pos
       else
         return P.unmatch(pos)
@@ -272,22 +272,48 @@ S.format = P.any(
       S.open,
       S.int,
       S.colon,
-      P.any(
-        P.seq(S.question, P.take_until({ ':' }, { '\\' }), S.colon, P.take_until({ '}' }, { '\\' })),
-        P.seq(S.plus, P.take_until({ '}' }, { '\\' })),
-        P.seq(S.minus, P.take_until({ '}' }, { '\\' }))
-      ),
+      P.seq(S.question, P.opt(P.take_until({ ':' }, { '\\' })), S.colon, P.opt(P.take_until({ '}' }, { '\\' }))),
       S.close
     ),
     function(values)
       return setmetatable({
         type = Node.Type.FORMAT,
         capture_index = values[3],
-        if_text = values[5][2].esc,
-        else_text = (values[5][4] or {}).esc,
+        if_text = values[5][2] and values[5][2].esc or '',
+        else_text = values[5][4] and values[5][4].esc or '',
       }, Node)
     end
-  )
+  ),
+  P.map(
+    P.seq(S.dollar, S.open, S.int, S.colon, P.seq(S.plus, P.opt(P.take_until({ '}' }, { '\\' }))), S.close),
+    function(values)
+      return setmetatable({
+        type = Node.Type.FORMAT,
+        capture_index = values[3],
+        if_text = values[5][2] and values[5][2].esc or '',
+        else_text = '',
+      }, Node)
+    end
+  ),
+  P.map(
+    P.seq(S.dollar, S.open, S.int, S.colon, S.minus, P.opt(P.take_until({ '}' }, { '\\' })), S.close),
+    function(values)
+      return setmetatable({
+        type = Node.Type.FORMAT,
+        capture_index = values[3],
+        if_text = '',
+        else_text = values[6] and values[6].esc or '',
+      }, Node)
+    end
+  ),
+  P.map(P.seq(S.dollar, S.open, S.int, S.colon, P.opt(P.take_until({ '}' }, { '\\' })), S.close), function(values)
+    return setmetatable({
+      type = Node.Type.FORMAT,
+      capture_index = values[3],
+      if_text = '',
+      else_text = values[5] and values[5].esc or '',
+    }, Node)
+  end)
 )
 
 S.transform = P.map(
@@ -333,12 +359,19 @@ S.tabstop = P.any(
 
 S.placeholder = P.any(
   P.map(
-    P.seq(S.dollar, S.open, S.int, S.colon, P.many(P.any(S.toplevel, S.text({ '$', '}' }, { '\\' }))), S.close),
+    P.seq(S.dollar, S.open, S.int, S.colon, P.opt(P.many(P.any(S.toplevel, S.text({ '$', '}' }, { '\\' })))), S.close),
     function(values)
       return setmetatable({
         type = Node.Type.PLACEHOLDER,
         tabstop = values[3],
-        children = values[5],
+        -- insert empty text if opt did not match.
+        children = values[5] or {
+          setmetatable({
+            type = Node.Type.TEXT,
+            raw = '',
+            esc = '',
+          }, Node),
+        },
       }, Node)
     end
   )
diff --git a/test/functional/plugin/lsp/snippet_spec.lua b/test/functional/plugin/lsp/snippet_spec.lua
index 4e127743eb..7903885420 100644
--- a/test/functional/plugin/lsp/snippet_spec.lua
+++ b/test/functional/plugin/lsp/snippet_spec.lua
@@ -19,9 +19,9 @@ describe('vim.lsp._snippet', function()
         {
           type = snippet.NodeType.TEXT,
           raw = 'TE\\$\\}XT',
-          esc = 'TE$}XT'
-        }
-      }
+          esc = 'TE$}XT',
+        },
+      },
     }, parse('TE\\$\\}XT'))
   end)
 
@@ -36,8 +36,8 @@ describe('vim.lsp._snippet', function()
         {
           type = snippet.NodeType.TABSTOP,
           tabstop = 2,
-        }
-      }
+        },
+      },
     }, parse('$1${2}'))
   end)
 
@@ -56,7 +56,7 @@ describe('vim.lsp._snippet', function()
                 {
                   type = snippet.NodeType.TEXT,
                   raw = 'TE\\$\\}XT',
-                  esc = 'TE$}XT'
+                  esc = 'TE$}XT',
                 },
                 {
                   type = snippet.NodeType.TABSTOP,
@@ -73,21 +73,21 @@ describe('vim.lsp._snippet', function()
                       {
                         type = snippet.NodeType.FORMAT,
                         capture_index = 1,
-                        modifier = 'upcase'
-                      }
-                    }
+                        modifier = 'upcase',
+                      },
+                    },
                   },
                 },
                 {
                   type = snippet.NodeType.TEXT,
                   raw = 'TE\\$\\}XT',
-                  esc = 'TE$}XT'
+                  esc = 'TE$}XT',
                 },
-              }
-            }
-          }
+              },
+            },
+          },
         },
-      }
+      },
     }, parse('${1:${2:TE\\$\\}XT$3${1/regex/${1:/upcase}/i}TE\\$\\}XT}}'))
   end)
 
@@ -110,8 +110,8 @@ describe('vim.lsp._snippet', function()
             {
               type = snippet.NodeType.TABSTOP,
               tabstop = 1,
-            }
-          }
+            },
+          },
         },
         {
           type = snippet.NodeType.VARIABLE,
@@ -124,11 +124,11 @@ describe('vim.lsp._snippet', function()
                 type = snippet.NodeType.FORMAT,
                 capture_index = 1,
                 modifier = 'upcase',
-              }
-            }
-          }
+              },
+            },
+          },
         },
-      }
+      },
     }, parse('$VAR${VAR}${VAR:$1}${VAR/regex/${1:/upcase}/}'))
   end)
 
@@ -141,12 +141,96 @@ describe('vim.lsp._snippet', function()
           tabstop = 1,
           items = {
             ',',
-            '|'
-          }
-        }
-      }
+            '|',
+          },
+        },
+      },
     }, parse('${1|\\,,\\||}'))
   end)
 
-end)
+  it('should parse format', function()
+    eq({
+      type = snippet.NodeType.SNIPPET,
+      children = {
+        {
+          type = snippet.NodeType.VARIABLE,
+          name = 'VAR',
+          transform = {
+            type = snippet.NodeType.TRANSFORM,
+            pattern = 'regex',
+            format = {
+              {
+                type = snippet.NodeType.FORMAT,
+                capture_index = 1,
+                modifier = 'upcase',
+              },
+              {
+                type = snippet.NodeType.FORMAT,
+                capture_index = 1,
+                if_text = 'if_text',
+                else_text = '',
+              },
+              {
+                type = snippet.NodeType.FORMAT,
+                capture_index = 1,
+                if_text = '',
+                else_text = 'else_text',
+              },
+              {
+                type = snippet.NodeType.FORMAT,
+                capture_index = 1,
+                else_text = 'else_text',
+                if_text = 'if_text',
+              },
+              {
+                type = snippet.NodeType.FORMAT,
+                capture_index = 1,
+                if_text = '',
+                else_text = 'else_text',
+              },
+            },
+          },
+        },
+      },
+    }, parse('${VAR/regex/${1:/upcase}${1:+if_text}${1:-else_text}${1:?if_text:else_text}${1:else_text}/}'))
+  end)
 
+  it('should parse empty strings', function()
+    eq({
+      children = {
+        {
+          children = { {
+            esc = '',
+            raw = '',
+            type = 7,
+          } },
+          tabstop = 1,
+          type = 2,
+        },
+        {
+          esc = ' ',
+          raw = ' ',
+          type = 7,
+        },
+        {
+          name = 'VAR',
+          transform = {
+            format = {
+              {
+                capture_index = 1,
+                else_text = '',
+                if_text = '',
+                type = 6,
+              },
+            },
+            option = 'g',
+            pattern = 'erg',
+            type = 5,
+          },
+          type = 3,
+        },
+      },
+      type = 0,
+    }, parse('${1:} ${VAR/erg/${1:?:}/g}'))
+  end)
+end)