add babycode parser, courtesy of kaesa

This commit is contained in:
2025-05-30 22:59:21 +03:00
parent 3e9f771ad3
commit 1a37ccfd86
3 changed files with 519 additions and 122 deletions

View File

@ -3,6 +3,8 @@ local babycode = {}
local string_trim = require("lapis.util").trim
local emoji = require("lib.babycode-emoji")
local Parser = require("lib.babycode-parser")
local function s_split(s, delimiter, max_matches, trim, allow_empty)
local result = {}
if s == "" then
@ -55,116 +57,94 @@ local function s_split(s, delimiter, max_matches, trim, allow_empty)
return result
end
local function get_list_items(list_body, escape_html)
list_body = list_body:gsub(" +%s*\r?\n", "<br>")
list_body = list_body:gsub("(%S)(\r?\n\r?\n)\r?\n*", "%1\1")
local function list(tag, children)
local list_body = children:gsub(" +\n", "<br>"):gsub("\n\n+", "\1")
local list_items = s_split(list_body, "\1")
local lis = ""
for _, li in ipairs(list_items) do
local rendered = babycode.to_html(li, escape_html)
lis = lis .. "<li>" .. rendered .. "</li>"
lis = lis .. "<li>" .. li .. "</li>"
end
return lis
return "<" .. tag .. ">" .. lis .. "</" .. tag .. ">"
end
local tags = {
b = "<strong>$S</strong>",
i = "<em>$S</em>",
s = "<del>$S</del>",
img = "<div class=\"post-img-container\"><img class=\"block-img\" src=$A alt=%S></div>",
url = "<a href=\"$A\">$S</a>",
quote = "<blockquote>$S</blockquote>",
code = function(children)
local is_inline = children:match("\n") == nil
if is_inline then
return "<code class=\"inline-code\">" .. children .. "</code>"
else
local t = string_trim(children)
local button = ("<button type=button class=\"copy-code\" value=\"%s\">Copy</button>"):format(t)
return "<pre><span class=\"copy-code-container\">"..button.."</span><code>"..t.."</code></pre>"
end
end,
ul = function(children)
return list("ul", children)
end,
ol = function(children)
return list("ol", children)
end,
}
local text_only = {
code = true,
}
---renders babycode to html
---@param s string input babycode
---@param escape_html fun(s: string): string function that escapes html
function babycode.to_html(s, escape_html)
if not s or s == "" then return "" end
local text = escape_html(s)
-- extract code blocks and store them as placeholders
-- don't want to process bbcode embedded into a code block
local code_blocks = {}
local inline_codes = {}
text = text:gsub("%[code%](.-)%[/code%]", function(code)
local is_inline = code:match("\n") == nil
if is_inline then
table.insert(inline_codes, code)
return "\1ICODE:"..#inline_codes.."\1"
else
-- strip leading and trailing newlines, preserve others
local m, _ = code:gsub("^%s*(.-)%s*$", "%1")
table.insert(code_blocks, m)
return "\1CODE:"..#code_blocks.."\1"
---@param html_escape fun(s: string): string function to escape html
function babycode.to_html(s, html_escape)
-- normalize line ending chars
local subj = string_trim(html_escape(s)):gsub("\r\n", "\n"):gsub("\r", "\n")
local parser = Parser.new(subj)
parser.valid_bbcode_tags = tags
parser.valid_emotes = emoji
parser.bbcode_tags_only_text_children = text_only
local elements = parser:parse()
local out = ""
local function fold(element, nobr)
if type(element) == "string" then
if nobr then
return element
end
return element:gsub(" +\n", "<br>"):gsub("\n\n+", "<br><br>")
end
end)
text = text:gsub("%[ul%](.-)%[/ul%]", function(list_body)
return "<ul>" .. get_list_items(list_body, escape_html) .. "</ul>"
end)
text = text:gsub("%[ol%](.-)%[/ol%]", function(list_body)
return "<ol>" .. get_list_items(list_body, escape_html) .. "</ol>"
end)
-- images
local images = {}
text = text:gsub("%[img=(.-)%](.-)%[/img%]", function (img, alt)
table.insert(images, {img = img, alt = alt})
return "\1IMG:"..#images.."\1"
end)
-- normalize newlines, attempt #4
text = text:gsub(" +%s*\r?\n", "<br>")
text = text:gsub("(%S)(\r?\n\r?\n)\r?\n*", "%1<br><br>")
local url_tags = {}
-- replace `[url=https://example.com]Example[/url] tags
text = text:gsub("%[url=([^%]]+)%](.-)%[/url%]", function(url, label)
table.insert(url_tags, {url = url, label = label})
return "\1URL:"..#url_tags.."\1"
end)
-- bold, italics, strikethrough
text = text:gsub("%[b%](.-)%[/b%]", "<strong>%1</strong>")
text = text:gsub("%[i%](.-)%[/i%]", "<em>%1</em>")
text = text:gsub("%[s%](.-)%[/s%]", "<del>%1</del>")
-- these can be nested, so replace open and closed separately
text = text:gsub("%[(/?)quote%]", "<%1blockquote>")
text = text:gsub(":(.-):", function(code)
if emoji[code] then
return emoji[code]
else
return code
if element.type == "bbcode" then
local c = ""
for _, child in ipairs(element.children) do
local _nobr = element.name == "code" or element.name == "ul" or element.name == "ol"
c = c .. fold(child, _nobr)
end
local res = ""
if type(tags[element.name]) == "string" then
res = (tags[element.name]):gsub("%$S", c)
if element.attribute then
res = res:gsub("%$A", element.attribute)
end
return res
elseif type(tags[element.name]) == "function" then
res = tags[element.name](c, element.attribute)
end
return res
elseif element.type == "link" then
return "<a href=\""..element.url.."\">"..element.url.."</a>"
elseif element.type == "emote" then
return emoji[element.name]
elseif element.type == "ruler" then
return "<hr>"
end
end)
-- replace loose links
text = text:gsub("(https?://[%w-_%.%?%.:/%+=&~%@#%%]+[%w-/])", function(url)
if not text:find('<a[^>]*>'..url..'</a>') then
return '<a href="'..url..'">'..url..'</a>'
end
return url
end)
text = text:gsub("\1URL:(%d+)\1", function(n)
local url = url_tags[tonumber(n)]
return ("<a href=%s>%s</a>"):format(url.url, url.label)
end)
-- rule
text = text:gsub("\n+%-%-%-", "<hr>")
-- <div class=\"post-img-container\"><img src=%1 alt=%2></div>
text = text:gsub("\1IMG:(%d+)\1", function (n)
local img = images[tonumber(n)]
return ("<div class=\"block-img-container\"><img class=\"block-img\" src=\"%s\" alt=\"%s\"></div>"):format(img.img, img.alt)
end)
-- replace code block placeholders back with their original contents
text = text:gsub("\1CODE:(%d+)\1", function(n)
local code = code_blocks[tonumber(n)]
local button = ("<button type=button class=\"copy-code\" value=\"%s\">Copy</button>"):format(code)
return "<pre><span class=\"copy-code-container\">" .. button .. "</span><code>"..code.."</code></pre>"
end)
text = text:gsub("\1ICODE:(%d+)\1", function (n)
local code = inline_codes[tonumber(n)]
return "<code class=\"inline-code\">" .. code .. "</code>"
end)
return text
end
for _, e in ipairs(elements) do
out = out .. fold(e, false)
end
return out
end
return babycode