fix!: lookaround assertions (#39)

BREAKING CHANGES: - Depends on tree-sitter-regex [v0.20.0](https://github.com/tree-sitter/tree-sitter-regex/pull/15/files) * chore: simplify test runner, ci, and makefile * test: fixup lookbehind cases * fix: lookbehind
bennypowers · Aug 3, 2023 · aeacf14 · aeacf14
1 parent 49ade10
commit aeacf14
Show file tree

Hide file tree

Showing 8 changed files with 107 additions and 89 deletions.
diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
@@ -23,18 +23,12 @@ jobs:
       - name: checkout
         uses: actions/checkout@v2
 
-      - name: Install Packages
-        run: |
-          git clone --depth 1 https://github.com/nvim-lua/plenary.nvim           ~/.local/share/nvim/site/pack/vendor/start/plenary.nvim
-          git clone --depth 1 https://github.com/MunifTanjim/nui.nvim            ~/.local/share/nvim/site/pack/vendor/start/nui.nvim
-          git clone --depth 1 https://github.com/nvim-treesitter/nvim-treesitter ~/.local/share/nvim/site/pack/vendor/start/nvim-treesitter.nvim
-
       - name: Install Neovim
         uses: rhysd/action-setup-vim@v1
         with:
           neovim: true
           version: ${{ matrix.nvim-versions }}
 
       - name: Test
-        run: make ci
+        run: make test
 
diff --git a/.gitignore b/.gitignore
@@ -1 +1,3 @@
 vendor
+.tests
+
diff --git a/Makefile b/Makefile
@@ -2,39 +2,21 @@ SHELL:=/usr/bin/env bash
 .PHONY: test run_tests watch ci unload
 
 clean:
-	@rm -rf vendor/plenary.nvim
-	@rm -rf vendor/nvim-treesitter
-	@rm -rf vendor/nui.nvim
-
-unload:
-	@pgrep -f 'nvim --headless' | xargs kill -s KILL;
-
-run_tests:
-	@REGEXPLAINER_DEBOUNCE=false \
-		nvim \
-		--headless \
-		-u tests/mininit.lua \
-		-c "lua require'plenary.test_harness'.test_directory('tests/regexplainer/', {minimal_init='tests/mininit.lua',sequential=true})"\
-	  -c "qa!"
+	@rm -rf vendor
 
 watch:
 	@echo "Testing..."
 	@find . \
 		-type f \
 		-name '*.lua' \
 		-o -name '*.js' \
-		! -path "./vendor/**/*" | entr -d make run_tests
+		! -path "./.tests/**/*" | entr -d make run_tests
 
 test:
-	@make unload
-	@make run_tests
-
-ci:
-	@nvim --version
-	@nvim \
-		--headless \
-		--noplugin \
-		-u tests/mininit.lua \
-		-c "TSUpdateSync javascript typescript regex" \
-		-c "qa!"
-	@make run_tests
+	@REGEXPLAINER_DEBOUNCE=false \
+		nvim \
+			--headless \
+			--noplugin \
+			-u tests/mininit.lua \
+			-c "lua require'plenary.test_harness'.test_directory('tests/regexplainer/', {minimal_init='tests/mininit.lua',sequential=true})"\
+			-c "qa!"
diff --git a/lua/regexplainer/component/init.lua b/lua/regexplainer/component/init.lua
@@ -15,7 +15,8 @@ local get_node_text = vim.treesitter.get_node_text or vim.treesitter.query.get_n
 ---@field zero_or_more?   boolean                   # a regexp component marked with `*`
 ---@field one_or_more?    boolean                   # a regexp component marked with `+`
 ---@field lazy?           boolean                   # a regexp quantifier component marked with `?`
----@field negative?       boolean                   # when it's a negative lookbehind
+---@field negative?       boolean                   # when it's a negative lookaround
+---@field direction?      'ahead'|'behind'          # when it's a lookaround, is it a lookahead or a lookbehind
 ---@field error?          any                       # parsing error
 
 ---@class RegexplainerParentComponent               : RegexplainerBaseComponent
@@ -34,7 +35,7 @@ local get_node_text = vim.treesitter.get_node_text or vim.treesitter.query.get_n
 ---| 'control_escape',
 ---| 'decimal_escape',
 ---| 'identity_escape',
----| 'lookahead_assertion'
+---| 'lookaround_assertion'
 ---| 'pattern'
 ---| 'pattern_character'
 ---| 'term'
@@ -55,7 +56,7 @@ local component_types = {
   'control_escape',
   'decimal_escape',
   'identity_escape',
-  'lookahead_assertion',
+  'lookaround_assertion',
   'pattern',
   'pattern_character',
   'term',
@@ -113,10 +114,18 @@ end
 ---@param component RegexplainerComponent
 ---@return boolean
 --
-function M.is_look_assertion(component)
-  return component.type:find('^look%a+_assertion') ~= nil
+function M.is_lookaround_assertion(component)
+  return component.type:find('^lookaround_assertion') ~= nil
 end
 
+---@param component RegexplainerComponent
+---@return boolean
+--
+function M.is_lookbehind_assertion(component)
+  return component.type:find('^lookaround_assertion') ~= nil
+end
+
+
 -- Does a container component contain nothing by pattern_characters?
 ---@param component RegexplainerComponent
 ---@return boolean
@@ -370,11 +379,12 @@ function M.make_components(node, parent, root_regex_node)
             end
           end
 
-          if node_pred.is_look_assertion(child) then
-            local _, _, sign   = string.find(text, '%(%?<?([=!])')
+          if node_pred.is_lookaround_assertion(child) then
+            local _, _, behind, sign   = string.find(text, '%(%?(<?)([=!])')
             component.type     = type
             component.negative = sign == '!'
             component.depth    = (parent and parent.depth or 0) + 1
+            component.direction = behind == '<' and 'behind' or 'ahead'
           end
 
           -- once state has been set above, process the children

diff --git a/lua/regexplainer/renderers/narrative/narrative.lua b/lua/regexplainer/renderers/narrative/narrative.lua
@@ -189,13 +189,16 @@ local function get_narrative_clause(component, options, state)
 
   end
 
-  if comp.is_look_assertion(component) then
-    if component.type == 'lookbehind_assertion' then
+  if comp.is_lookaround_assertion(component) then
+    if comp.direction == 'behind' then
       state.lookbehind_found = true
     end
 
     local negation = component.negative and 'NOT ' or ''
-    local direction = comp.is_lookahead_assertion(component) and 'followed by' or 'preceeding'
+    local direction = 'followed by'
+    if component.direction == 'behind' then
+      direction = 'preceeding'
+    end
     prefix = '**' .. negation .. direction .. ' ' .. '**'
 
     local sublines, sep = get_sublines(component, options, state)
@@ -209,7 +212,7 @@ local function get_narrative_clause(component, options, state)
   end
 
   if not comp.is_capture_group(component)
-      and not comp.is_look_assertion(component) then
+      and not comp.is_lookaround_assertion(component) then
     suffix = get_suffix(component)
   end
 
@@ -249,7 +252,7 @@ function M.recurse(components, options, state)
         last = last,
       }))
 
-    if comp.is_lookahead_assertion(component) then
+    if comp.is_lookaround_assertion(component) then
       if not clauses[#clauses] then
         table.insert(clauses, next_clause)
       else

diff --git a/lua/regexplainer/utils/treesitter.lua b/lua/regexplainer/utils/treesitter.lua
@@ -17,8 +17,7 @@ local node_types = {
   'document',
   'group_name',
   'identity_escape',
-  'lookahead_assertion',
-  'lookbehind_assertion',
+  'lookaround_assertion',
   'named_capturing_group',
   'non_capturing_group',
   'one_or_more',
@@ -74,8 +73,7 @@ function M.is_container(node)
     return type == 'anonymous_capturing_group'
         or type == 'alternation'
         or type == 'character_class'
-        or type == 'lookahead_assertion'
-        or type == 'lookbehind_assertion'
+        or type == 'lookaround_assertion'
         or type == 'named_capturing_group'
         or type == 'non_capturing_group'
         or type == 'pattern'
@@ -135,15 +133,9 @@ function M.is_upwards_stop(node)
   return node and node:type() == 'pattern' or M.is_document(node)
 end
 
--- Is it a lookahead or lookbehind assertion?
-function M.is_look_assertion(node)
-  ---@see https://github.com/tree-sitter/tree-sitter-regex/issues/13
-  if node:type() == 'ERROR' then
-    local text = get_node_text(node, 0)
-    return text:match [[^%(%<]]
-  else
-    return require 'regexplainer.component'.is_look_assertion { type = node:type() }
-  end
+-- Is it a lookaround assertion?
+function M.is_lookaround_assertion(node)
+  return require 'regexplainer.component'.is_lookaround_assertion { type = node:type() }
 end
 
 function M.is_modifier(node)

diff --git a/tests/fixtures/narrative/09 Lookaround.js b/tests/fixtures/narrative/09 Lookaround.js
@@ -37,23 +37,14 @@
 /@(?=g(?:raph)?ql)@/;
 
 /**
- * ⚠️ **Lookbehinds are poorly supported**
- * ⚠️ results may not be accurate
- * ⚠️ See https://github.com/tree-sitter/tree-sitter-regex/issues/13
- *
  * **preceeding **:
  *   `it's the `
  * `attack of the killer tomatos`
  */
 /(?<=it's the )attack of the killer tomatos/;
 
 /**
- * ⚠️ **Lookbehinds are poorly supported**
- * ⚠️ results may not be accurate
- * ⚠️ See https://github.com/tree-sitter/tree-sitter-regex/issues/13
- *
- * `x`
- * **NOT preceeding **:
+ * `x` **NOT preceeding **:
  *   `u`
  * `@`
  */
@@ -65,10 +56,6 @@
 
 
 /**
- * ⚠️ **Lookbehinds are poorly supported**
- * ⚠️ results may not be accurate
- * ⚠️ See https://github.com/tree-sitter/tree-sitter-regex/issues/13
- *
  * **preceeding **:
  *   `g`
  * `\``
@@ -79,10 +66,6 @@
 /(?<=g)`(.*)`/mg;
 
 /**
- * ⚠️ **Lookbehinds are poorly supported**
- * ⚠️ results may not be accurate
- * ⚠️ See https://github.com/tree-sitter/tree-sitter-regex/issues/13
- *
  * **preceeding **:
  *   `g`
  *   non-capturing group (_optional_):

diff --git a/tests/mininit.lua b/tests/mininit.lua
@@ -1,12 +1,64 @@
-vim.cmd([[
-  set rtp+=.
-  set noswapfile
-  filetype on
-  packloadall
-  runtime plugin/regexplainer.vim
-]])
-
-local did, configs = pcall(require, 'nvim-treesitter.configs')
-if not did then print(configs) end
-
-configs.setup {}
+local M = {}
+
+function M.root(root)
+  local f = debug.getinfo(1, "S").source:sub(2)
+  return vim.fn.fnamemodify(f, ":p:h:h") .. "/" .. (root or "")
+end
+
+---@param plugin string
+function M.load(plugin)
+  local name = plugin:match(".*/(.*)")
+  local package_root = M.root(".tests/site/pack/deps/start/")
+  if not vim.loop.fs_stat(package_root .. name) then
+    print("Installing " .. plugin)
+    vim.fn.mkdir(package_root, "p")
+    vim.fn.system({
+      "git",
+      "clone",
+      "--depth=1",
+      "https://github.com/" .. plugin .. ".git",
+      package_root .. "/" .. name,
+    })
+  end
+end
+
+local langs = {
+  'html',
+  'javascript',
+  'typescript',
+  'regex',
+}
+
+function M.setup()
+  vim.cmd([[
+    set noswapfile
+    filetype on
+    set runtimepath=$VIMRUNTIME
+    runtime plugin/regexplainer.vim
+  ]])
+
+  vim.opt.runtimepath:append(M.root())
+  vim.opt.packpath = { M.root(".tests/site") }
+
+  M.load("MunifTanjim/nui.nvim")
+  M.load("nvim-lua/plenary.nvim")
+  M.load("nvim-treesitter/nvim-treesitter")
+
+  local parser_install_dir = M.root(".tests/share/treesitter");
+  vim.opt.runtimepath:append(parser_install_dir)
+
+  vim.cmd[[packloadall]]
+
+  require 'nvim-treesitter.configs'.setup {
+    parser_install_dir = parser_install_dir,
+  }
+
+  for _, lang in ipairs(langs) do
+    if not require'nvim-treesitter.parsers'.has_parser(lang) then
+      vim.cmd('TSInstallSync ' .. lang)
+    end
+  end
+end
+
+M.setup()
+