require('strict')
local p = {}

--[[--------------------------< S H A R E D   H E L P E R S >--------------------------------]]

local function resolve_id_pool(val)
    if type(val) == "table" then
        local count = 0
        for _ in ipairs(val) do
            count = count + 1
        end
        if count == 0 then return nil end
        local index = math.random(1, count)
        return val[index]
    end
    return val
end

local function get_safe_ia_id(raw_id)
    raw_id = resolve_id_pool(raw_id)
    if not raw_id or raw_id == "" then return nil end
    if mw.ustring.match(raw_id, "^http") then
        local extracted = mw.ustring.match(raw_id, "/details/([^/?#]+)")
        if extracted then return extracted end
        return mw.ustring.match(raw_id, "([^/]+)$")
    end
    return raw_id
end

local function get_range_start(val)
    if not val then return nil end
    return mw.ustring.match(val, "^([%w_]+)") or mw.text.trim(val)
end

-- Converts Roman numerals to integers for Google Books PR pages
local function roman_to_int(roman)
    local map = {I=1, V=5, X=10, L=50, C=100, D=500, M=1000}
    roman = string.upper(roman)
    local total = 0
    local prev_val = 0
    
    -- Iterate backwards through the roman numeral
    for i = #roman, 1, -1 do
        local char = string.sub(roman, i, i)
        local val = map[char]
        if not val then return nil end -- Failsafe: Not a valid roman numeral character
        
        if val < prev_val then
            total = total - val
        else
            total = total + val
        end
        prev_val = val
    end
    
    return total
end

local function resolve_ia_search(explicit_search, display_text, user_search, config_search)
    if explicit_search and explicit_search ~= "" then return explicit_search end
    local override = user_search
    if not override or override == "" then override = config_search end
    local implicit_search = mw.ustring.match(display_text, "([%w%.]+)") or display_text
    implicit_search = mw.ustring.gsub(implicit_search, "[f%+–—%-%.]+$", "")
    if override then
        if override == 'none' then return nil end
        if override == 'pagenum' then return implicit_search end
        return override
    end
    return implicit_search
end

-- Extracts the first wikilink target and returns it along with the plain text display string
local function unpack_wikilink(text)
    if not text or type(text) ~= 'string' or not mw.ustring.match(text, "%[%[") then 
        return nil, text 
    end
    
    local target = mw.ustring.match(text, "%[%[([^|%]]+)%|?[^%]]*%]%]")
    local clean_text = mw.ustring.gsub(text, "%[%[([^|%]]+)%|?([^%]]*)%]%]", function(t, d)
        return (d and mw.text.trim(d) ~= "") and d or t
    end)
    
    return target, clean_text
end

--[[--------------------------< P A R S E R >------------------------------------------------]]

local function parse_segment(segment)
    if mw.ustring.match(segment, "^%[http") then
        segment = mw.ustring.match(segment, "%s+([^%]]+)%]%s*$") or segment
    end

    local search = segment:match("%[(.-)%]")
    local clean_segment = segment:gsub("%s*%[.-%]%s*", "") 
    
    -- 1. Try matching a quoted machine ID first (for Wikisource chapters with spaces/hyphens like @"Chapter 1")
    local display, machine_id, rest = clean_segment:match("^(.-)@(\"[^\"]+\")(.*)$")
    
    if not display then
        -- 2. Standard match: Alphanumerics and dots. It will STOP immediately at a hyphen.
        display, machine_id, rest = clean_segment:match("^(.-)@([%w%.]+)(.*)$")
    end
    
    if display then
        local clean_display = mw.text.trim(display .. rest)
        return clean_display, mw.text.trim(machine_id), search
    else
        return mw.text.trim(clean_segment), nil, search
    end
end

local function process_page_string(page_str, engine, host_data, frame_args, is_chapter)
    page_str = mw.ustring.gsub(page_str, ";", ",")
    local segments = mw.text.split(page_str, ",")
    local processed_segments = {}
    local clean_segments = {}
    local first_bare_url = nil
    local made_link = false
    
    for _, segment in ipairs(segments) do
        local trimmed = mw.text.trim(segment)
        if trimmed ~= "" then
            local display, machine, search = parse_segment(trimmed)
            table.insert(clean_segments, display) 
            
            if type(engine.build_link) == "function" then
                local formatted_link, bare_url = engine.build_link(host_data, display, machine, search, frame_args)
                table.insert(processed_segments, formatted_link)
                if bare_url then
                    made_link = true
                    if not first_bare_url then first_bare_url = bare_url end
                end
            else
                table.insert(processed_segments, display) 
            end
        end
    end
    
    if is_chapter then return first_bare_url, made_link end
    
    return table.concat(processed_segments, ", "), made_link, first_bare_url, table.concat(clean_segments, ", ")
end

--[[--------------------------< T H E   E N G I N E S >--------------------------------------]]
-- Placed before the Orchestrator to ensure tripwires can safely assign p.Physical 

-- 1. INTERNET ARCHIVE (IA)
p.IA = {
    build_link = function(host_data, display, machine, search, frame_args)
        local active_id = get_safe_ia_id(host_data.id or host_data.ia_id)
        if not active_id then return display, nil end
        
        local final_target = machine or get_range_start(display)
        if not final_target then return display, nil end
        
        local u_search = frame_args['ia-search']
        local c_search = host_data.iasearch
        local search_term = resolve_ia_search(search, display, u_search, c_search)

        local display_mode = frame_args['ia-display'] or host_data.iadisplay
        local endpoint = (display_mode == "full screen") and "stream" or "details"
        
        local url = string.format("https://archive.org/%s/%s/page/%s/mode/2up", endpoint, active_id, final_target)
        if search_term and search_term ~= "" then
            url = url .. "?q=" .. mw.uri.encode(search_term, "QUERY")
        end
        
        return string.format("[%s %s]", url, display), url
    end,

    build_base_url = function(host_data, frame_args)
        local active_id = get_safe_ia_id(host_data.id or host_data.ia_id)
        local active_query = (host_data.query and host_data.query ~= "") and host_data.query or nil

        if active_id then
            local display_mode = frame_args['ia-display'] or host_data.iadisplay
            local endpoint = (display_mode == "full screen") and "stream" or "details"
            local url = "https://archive.org/" .. endpoint .. "/" .. active_id .. "/"
            
            local fallback_search = frame_args['ia-search'] or host_data.iasearch
            if fallback_search and fallback_search ~= "none" and fallback_search ~= "pagenum" and fallback_search ~= "" then
                url = url .. "?q=" .. mw.uri.encode(fallback_search, "QUERY")
            end
            return url, nil
            
        elseif active_query then
            return "https://archive.org/search?query=" .. mw.uri.encode(active_query, "QUERY"), nil
        end
        return nil, nil
    end
}

-- 2. PROJECT GUTENBERG
p.Gutenberg = {
    build_link = function(host_data, display, machine, search, frame_args)
        local id = resolve_id_pool(host_data.id)
        if not id or id == "" then return display, nil end
        
        local raw_target = machine or display
        local target = get_range_start(raw_target)
        if not target then return display, nil end
        
        target = mw.ustring.gsub(target, "^u(%d+)", "%1")
        
        if mw.ustring.match(target, "^%d+$") then
            target = "page-" .. target
        end
        
        local url = string.format("https://gutenberg.org/cache/epub/%s/pg%s-images.html#%s", id, id, target)
        return string.format("[%s %s]", url, display), url
    end,

    build_base_url = function(host_data, frame_args)
        local id = resolve_id_pool(host_data.id)
        if not id or id == "" then return nil, nil end
        return "https://www.gutenberg.org/ebooks/" .. id, nil
    end
}

-- 3. HATHITRUST
p.Hathi = {
    build_link = function(host_data, display, machine, search, frame_args)
        local id = resolve_id_pool(host_data.id)
        if not id or id == "" then return display, nil end
        
        -- STRICT SSTS ENFORCEMENT: HathiTrust requires a machine sequence (seq)
        if not machine then return display, nil end
        
        local target = get_range_start(machine)
        target = mw.ustring.gsub(target, "^s(%d+)", "%1")
        
        local url = string.format("https://babel.hathitrust.org/cgi/pt?id=%s&view=1up&seq=%s", id, target)
        
        if search and search ~= "" then
            url = url .. "&q1=" .. mw.uri.encode(search, "QUERY")
        end
        
        return string.format("[%s %s]", url, display), url
    end,

    build_base_url = function(host_data, frame_args)
        local id = resolve_id_pool(host_data.id)
        if not id or id == "" then return nil, nil end
        return "https://hdl.handle.net/2027/" .. id, nil
    end
}

-- 4. WIKISOURCE
p.Wikisource = {
    build_link = function(host_data, display, machine, search, frame_args)
        local id = resolve_id_pool(host_data.id)
        if not id or id == "" then return display, nil end
        
        -- STRICT SSTS ENFORCEMENT: Only link if an explicit @machine leaf is provided
        if not machine then return display, nil end
        
        local target = get_range_start(machine)
        target = mw.ustring.gsub(target, '^"(.-)"$', "%1")
        target = mw.ustring.gsub(target, "^w(.+)", "%1")
        target = mw.ustring.gsub(target, " ", "_")
        
        local file_id = mw.ustring.gsub(id, " ", "_")
        local bare_url = string.format("https://en.wikisource.org/wiki/Page:%s/%s", file_id, target)
        local formatted_link = string.format("[[s:Page:%s/%s|%s]]", file_id, target, display)
        
        return formatted_link, bare_url
    end,

    build_base_url = function(host_data, frame_args)
        local id = resolve_id_pool(host_data.id)
        if not id or id == "" then return nil, nil end
        
        local clean_id = mw.ustring.gsub(id, " ", "_")
        clean_id = mw.ustring.gsub(clean_id, "^Page:", "")
        clean_id = mw.ustring.gsub(clean_id, "%.%l%l%l%l?$", "")
        
        return "https://en.wikisource.org/wiki/" .. clean_id, nil
    end
}

-- 5. GOOGLE BOOKS
p.GBook = {
    build_link = function(host_data, display, machine, search, frame_args)
        local id = resolve_id_pool(host_data.id)
        if not id or id == "" then return display, nil end
        
        local raw_target = machine or display
        local target = get_range_start(raw_target)
        if not target then return display, nil end
        
        if machine then
            target = mw.ustring.gsub(target, "^g([%w_]+)", "%1")
        else
            if mw.ustring.match(target, "^%d+$") then
                -- Pure Arabic numeral (e.g., "42")
                target = "PA" .. target
            elseif mw.ustring.match(string.lower(target), "^[ivxlcdm]+$") then
                -- Pure Roman numeral (e.g., "xiii")
                local int_val = roman_to_int(target)
                if int_val then
                    target = "PR" .. tostring(int_val)
                else
                    target = "PA" .. target -- Failsafe
                end
            else
                -- Unrecognized text or ranges (e.g., "42-45" or "cover"). 
                -- Extract the first digit for the base URL parameter.
                local first_num = mw.ustring.match(target, "^%d+")
                if first_num then
                    target = "PA" .. first_num
                end
            end
        end
        
        local url = string.format("https://books.google.com/books?id=%s&pg=%s", id, target)
        
        if search and search ~= "" then
            url = url .. "&dq=" .. mw.uri.encode(search, "QUERY")
        end
        
        return string.format("[%s %s]", url, display), url
    end,

    build_base_url = function(host_data, frame_args)
        local id = resolve_id_pool(host_data.id)
        if not id or id == "" then return nil, nil end
        return "https://books.google.com/books?id=" .. id, nil
    end
}

-- 6. STATIC WEB LINK
p.Web = {
    build_link = function(host_data, display, machine, search, frame_args)
        return display, nil
    end,

    build_base_url = function(host_data, frame_args)
        if not host_data.url or host_data.url == "" then return nil, nil end
        return host_data.url, host_data.access
    end
}

-- 7. PHYSICAL BOOK (NO DIGITAL HOST)
p.Physical = {
    build_link = function(host_data, display, machine, search, frame_args)
        return display, nil
    end,

    build_base_url = function(host_data, frame_args)
        return nil, nil
    end
}

--[[--------------------------< O R C H E S T R A T O R   H E L P E R S >--------------------]]

local function extract_core_isbn(isbn_str)
    if not isbn_str then return nil end
    local clean = mw.ustring.gsub(isbn_str, "[^%dX%x]", "")
    clean = mw.ustring.upper(clean)
    -- If it's a 13-digit ISBN starting with 978, the core is the next 9 digits
    local core13 = mw.ustring.match(clean, "^978(%d%d%d%d%d%d%d%d%d)")
    if core13 then return core13 end
    -- Otherwise grab the first 9 digits
    return mw.ustring.match(clean, "^(%d%d%d%d%d%d%d%d%d)")
end

local function evaluate_mismatches(citeArgs, frame_args)
    local mismatch_detected = false
    -- citeArgs was loaded with the shard's cite_params in the router
    local shard_year = citeArgs['year']
    local user_year = frame_args['year'] or frame_args['date']
    
    -- 1. Check for Year conflicts
    if shard_year and user_year then
        local s_yr = mw.ustring.match(shard_year, "%d%d%d%d")
        local u_yr = mw.ustring.match(user_year, "%d%d%d%d")
        if s_yr and u_yr and s_yr ~= u_yr then
            mismatch_detected = true
        end
    end
    
    -- 2. Check for ISBN conflicts
    local shard_isbn = citeArgs['isbn']
    local user_isbn = frame_args['isbn']
    if user_isbn then
        if shard_isbn then
            local s_core = extract_core_isbn(shard_isbn)
            local u_core = extract_core_isbn(user_isbn)
            if s_core and u_core and s_core ~= u_core then
                mismatch_detected = true
            end
        else
            -- User provided an ISBN, but shard has none. This is a later reprint.
            mismatch_detected = true
        end
    end
    
    return mismatch_detected, shard_year
end

local function apply_mismatch_rescues(citeArgs, mismatch_detected, shard_year)
    if not mismatch_detected then return end
    
    -- 1. Push shard year to orig-year so historical info isn't lost
    local user_year = citeArgs['year'] or citeArgs['date']
    if shard_year and not citeArgs['orig-year'] and not citeArgs['orig-date'] then
        local s_yr = mw.ustring.match(shard_year, "%d%d%d%d")
        local u_yr = user_year and mw.ustring.match(user_year, "%d%d%d%d")
        
        -- Only push to orig-year if the user explicitly provided a DIFFERENT year
        if u_yr and s_yr ~= u_yr then
            citeArgs['orig-year'] = shard_year
        end
    end

    -- We intentionally DO NOT delete the year to appease pre-1970 ISBN CS1 errors.
    -- Deleting the year also destroys the CITEREF Harvard anchors used by {{sfn}}.
end

local function process_page_arguments(engine, host_data, citeArgs, frame_args)
    local internal_keys = {
        ['ia-search'] = true,
        ['ia-display'] = true,
        ['chapter-page'] = true,
        ['title-page'] = true  
    }
    local has_deep_link = false

    -- CS1 Collision Shield: If the user provides a specific in-source locator, 
    -- instantly delete the shard's default locators to prevent red errors.
    if frame_args['page'] or frame_args['pages'] or frame_args['p'] or frame_args['pp'] or frame_args['at'] or frame_args['sheet'] or frame_args['sheets'] then
        citeArgs['page'] = nil
        citeArgs['pages'] = nil
        citeArgs['at'] = nil
        citeArgs['sheet'] = nil
        citeArgs['sheets'] = nil
    end

    for k, v in pairs(frame_args) do
        local val = mw.text.trim(v)
        if val ~= "" then
            local k_lower = string.lower(k)
            
            if k_lower == 'page' or k_lower == 'pages' or k_lower == 'p' or k_lower == 'pp' then
                local apply_article_url = (citeArgs['_template'] == 'cite encyclopedia' or citeArgs['_template'] == 'cite journal') and not host_data.ws_base
                local formatted_str, link_generated, bare_url, clean_str = process_page_string(val, engine, host_data, frame_args, false)
                local final_key = (k_lower == 'page' or k_lower == 'p') and 'page' or 'pages'
                
                if apply_article_url then
                    if bare_url then
                        citeArgs['url'] = bare_url
                        has_deep_link = true
                    end
                    
                    if not citeArgs[final_key] or k_lower == final_key then
                        citeArgs[final_key] = formatted_str
                    end
                else
                    if not citeArgs[final_key] or k_lower == final_key then
                        citeArgs[final_key] = formatted_str
                    end
                    if link_generated then has_deep_link = true end
                end
                
            elseif k_lower == 'chapter-page' then
                local bare_url, _ = process_page_string(val, engine, host_data, frame_args, true)
                if bare_url then citeArgs['chapter-url'] = bare_url end
                
            elseif k_lower == 'title-page' then
                local bare_url, _ = process_page_string(val, engine, host_data, frame_args, true)
                if bare_url then citeArgs['url'] = bare_url end
                
            elseif not internal_keys[k_lower] then
                -- Protect shard data: prevent routing integers from clobbering defined volumes/editions
                if (k_lower == 'volume' or k_lower == 'edition') and (citeArgs[k] or citeArgs[k_lower]) then
                    -- Preserve the shard's text, ignore the user's routing input
                else
                    citeArgs[k] = val 
                    -- CS1 Collision Prevention: Clean up legacy shard dates
                    if k_lower == 'date' then citeArgs['year'] = nil end
                    if k_lower == 'year' then citeArgs['date'] = nil end
                end
            end
        end
    end

    -- Auto-infer chapter-url
    local user_chap = citeArgs['chapter'] or citeArgs['article'] or citeArgs['entry']
    if user_chap and not citeArgs['chapter-url'] and not citeArgs['article-url'] and not citeArgs['entry-url'] then
        local pgs = frame_args['page'] or frame_args['pages'] or frame_args['p'] or frame_args['pp']
        if pgs then
            local auto_chapter_page = mw.ustring.match(pgs, "^%s*([%w%._]+)")
            if auto_chapter_page then
                local bare_url, _ = process_page_string(auto_chapter_page, engine, host_data, frame_args, true)
                if bare_url then citeArgs['chapter-url'] = bare_url end
            end
        end
    end

    -- CS1 Encyclopedia URL Shift
    if citeArgs['_template'] == 'cite encyclopedia' and citeArgs['url'] and user_chap and not citeArgs['title'] then
        if not citeArgs['chapter-url'] then
            citeArgs['chapter-url'] = citeArgs['url']
        end
        citeArgs['url'] = nil
    end

    return has_deep_link
end

local function apply_wikisource_formatting(host_data, citeArgs)
    if host_data.ws_base then
        if citeArgs['title'] then
            citeArgs['title'] = "[[s:" .. host_data.ws_base .. "/" .. citeArgs['title'] .. "|" .. citeArgs['title'] .. "]]"
            
            if citeArgs['encyclopedia'] and not mw.ustring.match(citeArgs['encyclopedia'], "%[%[") then
                citeArgs['encyclopedia'] = "[[s:" .. host_data.ws_base .. "|" .. citeArgs['encyclopedia'] .. "]]"
            end
            
            if citeArgs['journal'] and not mw.ustring.match(citeArgs['journal'], "%[%[") then
                citeArgs['journal'] = "[[s:" .. host_data.ws_base .. "|" .. citeArgs['journal'] .. "]]"
            end
            
            citeArgs['title-link'] = nil
            citeArgs['url'] = nil -- Prevent CS1 from throwing "extraneous wikilink" error
        else
            citeArgs['title-link'] = "s:" .. host_data.ws_base
            citeArgs['url'] = nil -- FIX: Prevent URL-wikilink conflict
        end
    end
end

--[[--------------------------< M A I N   O R C H E S T R A T O R >--------------------------]]

function p.process(engine, host_data, citeArgs, frame_args)
    -- ==========================================================
    -- 1. SAFETY TRIPWIRES (Detect Mismatches)
    -- ==========================================================
    local mismatch_detected, shard_year = evaluate_mismatches(citeArgs, frame_args)
    if mismatch_detected then
        host_data = {} -- Safe empty table to sever links instead of mutating read-only table
        engine = p.Physical
    end
    
    -- ==========================================================
    -- 1.5. ENCYCLOPEDIA ROUTING & DOWNGRADE
    -- ==========================================================
    if citeArgs['_template'] == 'cite encyclopedia' then
        local user_article = frame_args['article'] or frame_args['entry'] or frame_args['title'] or frame_args['chapter']
        local has_article = user_article and mw.text.trim(user_article) ~= ""

        if has_article then
            if citeArgs['title'] and not citeArgs['encyclopedia'] then
                if citeArgs['title-link'] then
                    citeArgs['encyclopedia'] = "[[" .. citeArgs['title-link'] .. "|" .. citeArgs['title'] .. "]]"
                    citeArgs['title-link'] = nil
                else
                    citeArgs['encyclopedia'] = citeArgs['title']
                end
                citeArgs['title'] = nil
            end
        else
            citeArgs['_template'] = 'cite book'
            if citeArgs['encyclopedia'] and not citeArgs['title'] then
                local encyc_link, clean_encyc = unpack_wikilink(citeArgs['encyclopedia'])
                citeArgs['title'] = clean_encyc
                if encyc_link and not citeArgs['title-link'] then
                    citeArgs['title-link'] = encyc_link
                end
                citeArgs['encyclopedia'] = nil
            end
        end
    end
    -- ==========================================================
    -- 2. UNPACK WIKILINKED TITLES
    -- ==========================================================
    local extracted_link, clean_title = unpack_wikilink(citeArgs['title'])
    if extracted_link then
        if not citeArgs['title-link'] then citeArgs['title-link'] = extracted_link end
        citeArgs['title'] = clean_title
    end

    -- ==========================================================
    -- 3. ROUTE ARGUMENTS & DEEP LINKS
    -- ==========================================================
    local has_deep_link = process_page_arguments(engine, host_data, citeArgs, frame_args)

    -- ==========================================================
    -- 4. BASE URL FALLBACK
    -- ==========================================================
    if not citeArgs['url'] and not citeArgs['chapter-url'] and not citeArgs['title-link'] then
        if not has_deep_link and type(engine.build_base_url) == "function" then
            local url, access = engine.build_base_url(host_data, frame_args)
            if url then citeArgs['url'] = url end
            if access then citeArgs['url-access'] = access end
        end
    end

    -- ==========================================================
    -- 5. WIKISOURCE FORMATTING
    -- ==========================================================
    apply_wikisource_formatting(host_data, citeArgs)
    
    -- ==========================================================
    -- 6. SAFETY TRIPWIRE RESCUES
    -- ==========================================================
    apply_mismatch_rescues(citeArgs, mismatch_detected, shard_year)
    
    -- ==========================================================
    -- 7. CS1 CONFLICT RESOLUTION
    -- ==========================================================
    if citeArgs['url'] and citeArgs['title-link'] then
        citeArgs['title-link'] = nil
    end

    return citeArgs
end

return p