From 73df8d400e8cb77ba2d7e533195d59a2686a2af4 Mon Sep 17 00:00:00 2001
From: Robin Malley <rmalley@cock.li>
Date: Sun, 4 Apr 2021 05:13:09 +0000
Subject: [PATCH] Updates to imageboard parser

---
 spec/parser_imageboard_spec.lua | 157 +++++++++++++++++++++++++++++++-
 src/lua/parser_imageboard.lua   |  17 ++--
 2 files changed, 164 insertions(+), 10 deletions(-)

diff --git a/spec/parser_imageboard_spec.lua b/spec/parser_imageboard_spec.lua
index 4756489..e747520 100644
--- a/spec/parser_imageboard_spec.lua
+++ b/spec/parser_imageboard_spec.lua
@@ -1,5 +1,9 @@
-
-describe("smr imageboard parser",function()
+function assertf(stmt, fmt, ...)
+	if not stmt then
+		error(string.format(fmt,...))
+	end
+end
+describe("smr imageboard parser #parsers",function()
 	it("should load without error",function()
 		local parser = require("parser_imageboard")
 	end)
@@ -9,4 +13,153 @@ describe("smr imageboard parser",function()
 		local output = parser(input)
 		assert(type(output) == "string","Expected string, got: %s",type(output))
 	end)
+	it("should spoiler text in asterisks ",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello, **world**!"
+		local output = parser(input)
+		local expected = [[<p>Hello, <span class="spoiler">world</span>!</p> ]]
+		assertf(output == expected, "Expected\n%s\ngot\n%s\n", expected, output)
+	end)
+	it("should spoiler text in [spoiler] tags",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello, [spoiler]world[/spoiler]!"
+		local output = parser(input)
+		local expected = [[<p>Hello, <span class="spoiler2">world</span>!</p> ]]
+		assertf(output == expected, "Expected\n%s\ngot\n%s\n", expected, output)
+	end)
+	it("should italicize words in double single quotes ('')",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello, ''world''!"
+		local output = parser(input)
+		local expected = [[<p>Hello, <i>world</i>!</p> ]]
+		assertf(output == expected, "Expected\n%s\ngot\n%s\n", expected, output)
+	end)
+	it("should bold words in tripple single quotes (''')",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello, '''world'''!"
+		local output = parser(input)
+		local expected = [[<p>Hello, <b>world</b>!</p> ]]
+		assertf(output == expected, "Expected\n%s\ngot\n%s\n", expected, output)
+	end)
+	it("should underline words in double underscores (__)",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello, __world__!"
+		local output = parser(input)
+		local expected = [[<p>Hello, <u>world</u>!</p> ]]
+		assertf(output == expected, "Expected\n%s\ngot\n%s\n", expected, output)
+	end)
+	it("should make a heading out of things in double equals(==)",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello, ==world==!"
+		local output = parser(input)
+		local expected = [[<p>Hello, <h2>world</h2>!</p> ]]
+		assertf(output == expected, "Expected\n%s\ngot\n%s\n", expected, output)
+	end)
+	it("should strikethrough words in double tildes (~~)",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello, ~~world~~!"
+		local output = parser(input)
+		local expected = [[<p>Hello, <s>world</s>!</p> ]]
+		assertf(output == expected, "Expected\n%s\ngot\n%s\n", expected, output)
+	end)
+	it("should codify words in [code] tags",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello, [code]world[/code]!"
+		local output = parser(input)
+		local expected = [[<p>Hello, <pre><code>world</code></pre>!</p> ]]
+		assertf(output == expected, "Expected\n%s\ngot\n%s\n", expected, output)
+	end)
+	it("should greentext lines that start with >",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello,\n> world!"
+		local output = parser(input)
+		local expected = [[<p>Hello,</p> <p><span class="greentext">&gt; world!</span></p> ]]
+		assertf(output == expected, "Expected\n%s\ngot\n%s\n", expected, output)
+	end)
+	it("should pinktext lines that start with <",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello,\n< world!"
+		local output = parser(input)
+		local expected = [[<p>Hello,</p> <p><span class="pinktext">&lt; world!</span></p> ]]
+		assertf(output == expected, "Expected\n%s\ngot\n%s\n", expected, output)
+	end)
+	it("should allow for bold+italic text",function()
+		local parser = require("parser_imageboard")
+		local input = "Hello,'''''world!'''''"
+		local output = parser(input)
+		local expected = [[<p>Hello,<i><b>world!</b></i></p> ]]
+	end)
+	local formatting = {
+		{"**","**"},
+		{"[spoiler]","[/spoiler]"},
+		{"''","''"},
+		{"'''","'''"},
+		{"__","__"},
+		{"==","=="},
+		{"~~","~~"},
+		{"[code]","[/code]"}
+	}
+	local formatting_line = {"> ", "< "}
+	for k,v in pairs(formatting) do
+		for i = 1, 50 do
+			it("should not break with " .. i .. " " .. v[1] .. " indicators in a row ",function()
+				local parser = require("parser_imageboard")
+				local input = "Hello, " .. string.rep(v[1],i) .. " world!"
+				local start_time = os.clock()
+				local output = parser(input)
+				local end_time = os.clock()
+				print(end_time - start_time)
+				assert(end_time - start_time < 1, "Took too long")
+			end)
+		end
+	end
+
+	for i = 1, 80 do
+		it("Should withstand a random string of " .. i .. " formatters and words. ",function()
+			local parser = require("parser_imageboard")
+			local input = {}
+			local function random_text()
+				if math.random() > 0.5 then
+					return "Hello"
+				else
+					return "world"
+				end
+			end
+			local function random_wrap(text)
+				local rngwrap = formatting[math.random(#formatting)]
+				return rngwrap[1] .. text .. rngwrap[2]
+			end
+			local function random_text_recursive(i)
+				if i == 0 then
+					return ""
+				end
+				local j = math.random()
+				if j < 0.33 then
+					return random_text_recursive(i-1) .. random_wrap(random_text())
+				elseif j < 0.66 then
+					return random_wrap(random_text() .. random_text_recursive(i-1)) .. random_wrap(random_text())
+				else
+					return random_wrap(random_text() .. random_text_recursive(i - 1))
+				end
+			end
+			input = random_text_recursive(i)
+			print("input is:",input)
+			local start_time = os.clock()
+			local output = parser(input)
+			print("output is:",output)
+			local end_time = os.clock()
+			print(end_time - start_time)
+			assert(end_time - start_time < 1, "Took too long")
+		end)
+	end
+	for _,file_name in ipairs{
+		"Beauty_and_the_Banchou_1"
+	} do
+		it("should parser " .. file_name,function()
+			local parser = require("parser_imageboard")
+			local input = require("spec.parser_tests." .. file_name)
+			local output = parser(input)
+			--print("output:",output)
+		end)
+	end
 end)
diff --git a/src/lua/parser_imageboard.lua b/src/lua/parser_imageboard.lua
index 791bffa..3c14b60 100644
--- a/src/lua/parser_imageboard.lua
+++ b/src/lua/parser_imageboard.lua
@@ -62,7 +62,7 @@ end
 ]]
 
 local function wrap(seq,format,s)
-	return P(seq) * Cs((((V"marked" - s) + word + P"\n"))^0) * P(seq) / function(a)
+	return P(seq) * Cs(((s + word + P"\n"))^0) * P(seq) / function(a)
 		return string.format(format,a)
 	end
 end
@@ -77,16 +77,17 @@ local function tag(name,format)
 	end
 end
 
+--local grammar = P(require('pegdebug').trace({
 local grammar = P{
 	"chunk";
 	--regular
-	spoiler = wrap("**",[[<span class="spoiler">%s</span>]],V"spoiler"),
-	spoiler2 = tag("spoiler",[[<span class="spoiler2">%s</span>]]),
-	italic = wrap("''",[[<i>%s</i>]], V"italic"),
-	bold = wrap("'''",[[<b>%s</b>]], V"bold"),
-	underline = wrap("__",[[<u>%s</u>]], V"underline"),
-	heading = wrap("==",[[<h2>%s</h2>]], V"heading"),
-	strike = wrap("~~",[[<s>%s</s>]], V"strike"),
+	heading = wrap("==",[[<h2>%s</h2>]], V"underline" + V"strike" + V"italic"),
+	bold = wrap("'''",[[<b>%s</b>]], V"italic" + V"underline" + V"strike"),
+	italic = wrap("''",[[<i>%s</i>]], V"underline" + V"strike"),
+	underline = wrap("__",[[<u>%s</u>]], V"strike"),
+	strike = wrap("~~",[[<s>%s</s>]], P("blah")),
+	spoiler = wrap("**",[[<span class="spoiler">%s</span>]],V"spoiler2" + V"bold" + V"italic" + V"underline" + V"strike"),
+	spoiler2 = tag("spoiler",[[<span class="spoiler2">%s</span>]],V"spoiler" + V"bold" + V"italic" + V"underline" + V"strike"),
 	code = tag("code",[[<pre><code>%s</code></pre>]]),
 	greentext = P">" * (B"\n>" + B">") * Cs((V"marked" + word)^0) / function(a)
 		return string.format([[<span class="greentext">&gt;%s</span>]],a)