Merge pull request #238 from noirbizarre/fix/code-blocks

fix(code-blocks): Fixes multiple language parsing cases (tilde, spaces before/after, directives...)

lua/markview/extras/editor.lua

@@ -44,7 +44,8 @@ editor.configuraton = {
 			local start, col, stop, _ = TSNode:range();
 			local lines = vim.api.nvim_buf_get_lines(buffer, start, stop, false);
 
-			local ft = lines[1]:match("```(%S*)") or "lua";
+			local _, info = languages.get_fence(lines[1])
+			local ft = languages.info(info);
 			local _l = {};
 
 			table.remove(lines, 1);
@@ -54,15 +55,16 @@ editor.configuraton = {
 				table.insert(_l, line:sub(col, #line))
 			end
 
-			return ft:gsub("[%{%}]", ""), _l, start + 1, stop - 1;
+			return ft, _l, start + 1, stop - 1;
 		end
 	},
 	appliers = {
 		["fenced_code_block"] = function (buffer, TSNode, lines)
 			local start, _, _, _ = TSNode:range();
 			local delimiter = vim.api.nvim_buf_get_lines(buffer, start, start + 1, false)[1];
 
-			local before = delimiter:match("^(.-)```");
+			local fence = languages.get_fence(delimiter)
+			local before = delimiter:match("^(.-)" .. fence);
 
 			for l, line in ipairs(lines) do
 				lines[l] = (before or "") .. line
@@ -89,23 +91,6 @@ editor.configuraton = {
 	end
 }
 
---- Gets the filetype from an info string
----@param delim string
----@return string
-local get_ft = function (delim)
-	local ft = "";
-
-	if delim:match("^```%{%{(.-)%}%}") then
-		ft = languages.get_ft(delim:match("^```%{%{(.-)%}%}"));
-	elseif delim:match("^```%{(.-)%}") then
-		ft = languages.get_ft(delim:match("^```%{(.-)%}"));
-	elseif delim:match("^```(%S+)") then
-		ft = languages.get_ft(delim:match("^```(%S+)"));
-	end
-
-	return ft;
-end
-
 --- Creates a new buffer when not available.
 --- Otherwise, returns the current editor buffer.
 ---@return integer
@@ -400,10 +385,11 @@ editor.create = function ()
 				end
 
 				start_delim = tostring(input);
-				vim.bo[editor.buffer].filetype = get_ft(start_delim);
-				local icon, hl = languages.get_icon(get_ft(start_delim));
-
-				ft = get_ft(start_delim);
+				local fence, info = languages.get_fence(start_delim);
+				ft = languages.info(info);
+				end_delim = fence or end_delim;
+				vim.bo[editor.buffer].filetype = ft
+				local icon, hl = languages.get_icon(ft);
 
 				hl = hl .. "Fg";
 

lua/markview/languages.lua

@@ -395,6 +395,28 @@ languages.patterns = {
 	["html"] = "HTML"
 };
 
+--- Known language info string patterns
+---@type string[]
+languages.info_patterns = {
+	-- {{ lang }} params
+	"%{%{([^%}]*)%}%}%s*(.*)$",
+	-- Myst code blocks (code, code-block, code-cell) with language
+	-- https://mystmd.org/guide/code#code-blocks
+	"%{code%S*%}%s*(%S+)$",
+	-- Other {}-wrapped directive with unknown processing
+	"%{([^%}]*)%}%s*(.*)$",
+	-- Language string and additional info
+	-- https://spec.commonmark.org/0.31.2/#example-143
+	"(%S-)%s+(.*)$",
+	-- Language string without additional info or no language
+	-- https://spec.commonmark.org/0.31.2/#example-143
+	"(%S*)%s*$",
+}
+
+--- Known code-block fences
+---@type string[]
+languages.fences = {"`", "~"}
+
 --- Gets the language name from a string
 ---@param name string
 ---@return string
@@ -434,4 +456,32 @@ languages.get_icon = function (ft)
 	return languages.icons[ft] or languages.icons.default, hl, sign;
 end
 
+--- Extract fenced code block header
+---@param line string
+---@return string|nil fence the matched fence string
+---@return string info the matched info string
+languages.get_fence = function(line)
+	for _, char in pairs(languages.fences) do
+		--- Match any supported fence, optionnaly indented or quoted
+		local fence, info = line:match("^>*%s*(" .. string.rep(char, 3) .. "+)%s*(.-)%s*$");
+		if fence ~= nil then
+			return fence, info;
+		end
+	end
+	return nil, ""
+end
+
+--- Extract language and parameters from an infostring
+---@param info string the info string to parse
+---@return string language the extracted language
+---@return string|nil # Some optional extra data
+languages.info = function (info)
+	for _, pattern in pairs(languages.info_patterns) do
+		local lang, extra = info:match(pattern);
+		if lang then
+			return lang, extra;
+		end
+	end
+end
+
 return languages;

lua/markview/parser.lua

@@ -127,6 +127,7 @@ parser.filter_lines = function (buffer, from, to)
 
 	local code_block_indent = 0;
 	local desc_indent = 0;
+	local current_fence = "";
 
 	local start = 0;
 
@@ -181,11 +182,14 @@ parser.filter_lines = function (buffer, from, to)
 			parent_marker = line:match("^%s*(%d+[%)%.])");
 		end
 
-		if line:match("(```)") and withinCodeBlock ~= true then
+		local fence = lang.get_fence(line)
+		if fence and withinCodeBlock ~= true then
 			withinCodeBlock = true;
+			current_fence = fence;
 			code_block_indent = spaces_before;
-		elseif line:match("(```)") and withinCodeBlock == true then
+		elseif withinCodeBlock == true and line:match(current_fence) then
 			withinCodeBlock = false;
+			current_fence = "";
 		elseif withinCodeBlock == true then
 			spaces_before = code_block_indent;
 			goto withinElement;
@@ -203,7 +207,7 @@ parser.filter_lines = function (buffer, from, to)
 		then
 			spaces_before = math.max(0, spaces_before - vim.fn.strchars((parent_marker or "") .. " "));
 
-			if line:match("(```)") then
+			if fence then
 				code_block_indent = spaces_before;
 			elseif insideDescription == true then
 				align_spaces[l] = 2;
@@ -381,20 +385,8 @@ parser.md = function (buffer, TStree, from, to)
 
 			local block_start = vim.api.nvim_buf_get_lines(buffer, row_start, row_start + 1, false)[1];
 
-			local language_string, additional_info = "", nil;
-
-			if block_start:match("%s*```%{%{([^%}]*)%}%}") then
-				language_string = block_start:match("%s*```%{%{([^%}]*)%}%}");
-				additional_info = block_start:match("%s*```%{%{[^%}]*%}%}%s*(.*)$");
-			elseif block_start:match("%s*```%{([^%}]*)%}") then
-				language_string = block_start:match("%s*```%{([^%}]*)%}");
-				additional_info = block_start:match("%s*```%{[^%}]*%}%s*(.*)$");
-			elseif block_start:match("%s*```(%S*)$") then
-				language_string = block_start:match("%s*```(%S*)$");
-			elseif block_start:match("%s*```(%S*)%s*") then
-				language_string = block_start:match("%s*```(%S*)%s");
-				additional_info = block_start:match("%s*```%S*%s+(.*)$");
-			end
+			local _, info = lang.get_fence(block_start);
+			local language_string, additional_info = lang.info(info)
 
 			local code_lines = vim.api.nvim_buf_get_lines(buffer, row_start + 1, row_end - 1, false);