truncate.lua -- html truncate for luvit lang.

truncate.lua

--
-- Created by: Cyril.
-- Created at: 15/6/25 上午12:40
-- Email: houshoushuai@gmail.com
--

local string = require("string")
local find = string.find
_G.indexOf = function(t, x)
    if type(t) == 'string' then
        return find(t, x, true)
    end
    for k, v in pairs(t) do
        if v == x then
            return k
        end
    end
    return nil
end

_G.lastIndexOf = function(t, x)
    local idx
    if type(t) == "string" then
        local i=t:match(".*"..x.."()")
        if i==nil then return nil else return i-1 end
    end
    for k, v in pairs(t) do
        if v == x then
            idx = k
        end
    end
    return idx
end

_G.pop = function(t)
    return table.remove(t)
end


_G.truncate = function(html, option)
    option = option or {}
    local tolerance = option.tolerance
    local image

    if type(option.image) == 'boolean' then
        image = option.image
    else
        image = true
    end

    local maxInt = option.max or 250
    local ellipsis = option.ellipsis or "... ..."
    local shouldAddEllipsis = false

    local tagPartten = "<(%/?)([A-Za-z0-9]+)[^%>]*>"
    local escapeTags = {"area", "base", "br", "col", "command", "embed", "hr", "img", "input", "keygen", "link", "meta", "param", "source", "track", "wbr"}
    local startTagArray = {}

    html = html:gsub(tagPartten, function(a, b)
        if b == "script" then
            return ""
        end
        if not(image) then
            if b == "img" or b == "figure" then
                return ""
            end
        end
    end)

    local htmlLen = #html
    local i = 0
    local bareHtmlLen = 0
    while i <= htmlLen and maxInt - bareHtmlLen > 0 do
        local afterfix = html:sub(i+1, htmlLen) or ""
        local start, stop = afterfix:find(tagPartten)
        if not(start or stop) then
            bareHtmlLen = maxInt
            break
        end
        local bareHtml = (afterfix:sub(0, start-1) or ""):gsub(tagPartten, function()return "" end)
        local _, utfCount = string.gsub(bareHtml, "[^\128-\193]", "")
        if maxInt < bareHtmlLen + utfCount then
            string.gsub(bareHtml, "([%z\1-\127\194-\244][\128-\191]*)", function(a)
                if bareHtmlLen < maxInt then
                    bareHtmlLen = bareHtmlLen + 1
                    if #a == 3 then
                        i = i+3
                    else
                        i = i+1
                    end
                end
            end)
            shouldAddEllipsis = true
        else
            i = stop + i
            bareHtmlLen = bareHtmlLen + utfCount
        end
    end

    html = html:sub(0, i)

    local bare = (html or ""):gsub(tagPartten, function(a, b)
        if a == "/" then -- End Tag
            if indexOf(escapeTags, b) then
            else
                if startTagArray[#startTagArray] == b then
                    _G.pop(startTagArray)
                else
                    if tolerance then
                        -- fault-tolerance
                        local lastIndex = lastIndexOf(startTagArray, b)
                        if  #startTagArray - lastIndex <= 1 then
                            _G.pop(startTagArray)
                        end
                    end
                end
            end
        else -- Start Tag
            if indexOf(escapeTags, b) then
            else
                table.insert(startTagArray, b)
            end
        end
        return ""
    end)
    -- fix end tag
    local endTagString = ""
    for _, v in pairs(startTagArray) do
        endTagString = "</".. v .. ">" .. endTagString
    end
    if #startTagArray and shouldAddEllipsis then
        html = html .. " " ..ellipsis
    end
    html = html .. endTagString
    return  html
end

-- usage truncate("<p>Hello world <a href="#">link... ...</p></a>", {image = true, ellipsis = "<br/> Read More"})