-- Parsers for the sampler caches.
--
-- Every function takes the file *contents* as a string rather than reading the
-- file itself, so the parsers are testable against fixtures without mocking the
-- filesystem.

local M = {}

-- Read a whole file, returning nil if it cannot be read, so a missing cache
-- degrades to nil instead of raising.
function M.slurp(path)
  local f = io.open(path, 'r')
  if not f then return nil end
  local s = f:read('*a')
  f:close()
  return s
end

-- Human-readable size to bytes: '1.1G' -> 1181116006.
--
-- Needed because du -sh is what the sampler runs (its output is also what the
-- card displays), and 'biggest' has to be decided numerically: sorted as
-- strings, '245M' beats '1.1G'.
local SUFFIX = { K = 1024, M = 1024^2, G = 1024^3, T = 1024^4, P = 1024^5 }

function M.human_bytes(s)
  if type(s) ~= 'string' then return nil end
  local n, suf = s:match('^%s*([%d%.]+)%s*([KMGTP]?)')
  n = tonumber(n)
  if not n then return nil end
  return n * (SUFFIX[suf] or 1)
end

-- Parse `du -sh` output: the first line is the total, the rest are children
-- already sorted largest-first by the sampler. Returns nil when there is
-- nothing usable, so the widget can draw its "no data" state.
function M.du_parse(text)
  if type(text) ~= 'string' or text == '' then return nil end
  local lines = {}
  for line in text:gmatch('[^\n]+') do lines[#lines + 1] = line end
  if #lines == 0 then return nil end

  local function entry(line)
    local label, path = line:match('^(%S+)%s+(.+)$')
    if not label then return nil end
    return { label = label, bytes = M.human_bytes(label) or 0,
             name = path:match('([^/]+)/?$') or path, path = path }
  end

  local total = entry(lines[1])
  if not total then return nil end
  local items = {}
  for i = 2, #lines do
    local e = entry(lines[i])
    if e then items[#items + 1] = e end
  end
  return { total = total, items = items }
end

-- Parse 'key value' lines into a table of strings.
--
-- Values stay strings and are converted by the caller. The sampler writes
-- epochs as integers and the widget decides whether an age reads as hours or
-- days; a parser that guessed would have to know that.
--
-- Only the first space separates, so a value keeps its own spaces:
-- 'version Slackware 15.0+' yields 'Slackware 15.0+', not 'Slackware'.
function M.kv_parse(text)
  local out = {}
  if type(text) ~= 'string' then return out end
  for line in text:gmatch('[^\n]+') do
    local k, v = line:match('^(%S+)%s+(.*)$')
    if k and v ~= '' then out[k] = v end
  end
  return out
end

-- The calendar cache written by bin/calendar-sample.sh.
--
-- Returns { today = { y, m, d }, colors = { <calendar> = <khal colour name> },
-- events = { { y, m, d, start, finish, calendar, title }, ... } }, or nil when
-- the cache is absent or carries no 'generated' key. nil means "the sampler
-- never ran", which the card names; an empty events list means "nothing in the
-- next week", which is a different and equally valid state.
--
-- Dates arrive in khal's own `longdateformat`, whatever the user set it to, so
-- the three numbers are pulled out by position and assigned by magnitude: the
-- 4-digit group is the year, and of the remaining two the one that cannot be a
-- month is the day. An unambiguous pair (03.04.2026) is read as day-first,
-- matching khal's default and the European formats its config ships. This is a
-- heuristic, and it is the honest one available: khal will not emit ISO.
function M.calendar_parse(text)
  if type(text) ~= 'string' then return nil end

  local out = { colors = {}, events = {} }
  local generated = false

  for line in text:gmatch('[^\n]+') do
    local kind, rest = line:match('^(%S+)%s+(.*)$')
    if kind == 'generated' then
      generated = true
    elseif kind == 'today' then
      local y, m, d = rest:match('^(%d+)-(%d+)-(%d+)$')
      if y then out.today = { tonumber(y), tonumber(m), tonumber(d) } end
    elseif kind == 'color' then
      local name, colour = rest:match('^(%S+)%s+(.*)$')
      if name and colour ~= '' then out.colors[name] = colour end
    elseif kind == 'event' then
      -- Split on the first four pipes only: a title may contain one.
      local date, st, en, cal, title =
        rest:match('^([^|]*)|([^|]*)|([^|]*)|([^|]*)|(.*)$')
      local ev = date and M.calendar_date(date)
      if ev then
        out.events[#out.events + 1] = {
          y = ev[1], m = ev[2], d = ev[3],
          start = st ~= '' and st or nil,
          finish = en ~= '' and en or nil,
          calendar = cal ~= '' and cal or nil,
          title = title,
        }
      end
    end
  end

  if not generated then return nil end
  return out
end

-- Three numbers out of a formatted date, as { year, month, day }.
-- See calendar_parse for why this is positional rather than a format string.
function M.calendar_date(s)
  if type(s) ~= 'string' then return nil end
  local nums = {}
  for n in s:gmatch('%d+') do nums[#nums + 1] = n end
  if #nums < 3 then return nil end

  -- The year is the 4-digit group. A 2-digit year is not handled: khal's
  -- shipped formats all use %Y, and guessing a century from two digits would
  -- be a second heuristic stacked on the first.
  local yi
  for i = 1, 3 do
    if #nums[i] == 4 then yi = i break end
  end
  if not yi then return nil end

  local rest = {}
  for i = 1, 3 do
    if i ~= yi then rest[#rest + 1] = tonumber(nums[i]) end
  end
  local a, b = rest[1], rest[2]

  local day, month
  if a > 12 then day, month = a, b
  elseif b > 12 then month, day = a, b
  else day, month = a, b end  -- ambiguous: day-first, as khal's defaults are

  if month < 1 or month > 12 or day < 1 or day > 31 then return nil end
  return { tonumber(nums[yi]), month, day }
end

-- Which days of a given month carry an event, and whose calendar owns each.
-- Returns { [day] = <calendar name> }. First event of a day wins, so a day
-- with two calendars takes the earlier one rather than blending into a colour
-- that means neither.
function M.calendar_month_days(events, year, month)
  local out = {}
  if type(events) ~= 'table' then return out end
  for _, e in ipairs(events) do
    if e.y == year and e.m == month and not out[e.d] then
      out[e.d] = e.calendar or true
    end
  end
  return out
end

-- Days in a month, and the weekday its 1st falls on.
--
-- os.time/os.date rather than a leap-year rule: the C library already knows,
-- and a hand-rolled rule is one more thing to get wrong in a century year.
-- Normalised to Monday = 1 .. Sunday = 7, because khal's firstweekday = 0
-- means Monday while os.date's wday means Sunday = 1.
function M.month_shape(year, month)
  local first = os.date('*t', os.time({ year = year, month = month, day = 1,
                                        hour = 12 }))
  -- Day 0 of the next month is the last day of this one, which os.time
  -- normalises for us across the December boundary.
  local last = os.date('*t', os.time({ year = year, month = month + 1, day = 0,
                                       hour = 12 }))
  return last.day, (first.wday + 5) % 7 + 1
end

-- The breaktimer cache written by bin/breaktimer-sample.sh.
--
-- Returns a table describing the daemon, or nil when the cache is absent or
-- carries no 'generated' key (the sampler never ran). The display strings are
-- derived here rather than in the widget so the phase-to-word and
-- phase-to-colour decisions can be tested without Cairo:
--
--   { running, state, phase, frozen, remain,
--     countdown = '24:31' or '--',
--     phase_label = 'working' | 'micro-pausa' | 'pausa lunga' | 'in pausa'
--                   | 'outside working hours' | 'fermo',
--     next = 'pausa' | 'lavoro' | '--',
--     role = 'ok' | 'heading' | 'highlight' | 'warning' | 'label' }
--
-- `frozen` is the sampler's read of whether the clock is ticking: the phase is
-- 'working' but the remain file has stopped being rewritten. Outside the work
-- window that is normal and reads as such; a hung daemon looks the same, and
-- should.
--
-- The countdown leads to the next phase, but the daemon keeps the cycle count
-- in memory and never writes it, so whether the NEXT break is a micro or a long
-- one is not knowable here. 'pausa' is the honest answer; the phase word carries
-- the rest.
function M.breaktimer_parse(text)
  if type(text) ~= 'string' then return nil end
  local kv = M.kv_parse(text)
  if not kv.generated then return nil end

  local running = kv.running == 'yes'
  local state = kv.state or 'stopped'
  local phase = kv.phase or 'stopped'
  local remain = tonumber(kv.remain) or 0
  if remain < 0 then remain = 0 end

  local out = {
    running = running,
    state = state,
    phase = phase,
    frozen = kv.frozen == 'yes',
    remain = remain,
  }

  -- Not running and stopped are the same thing on the card. The sampler's
  -- liveness check is what turns a dead PID into this reading: the state file
  -- can still say 'running' after a crash or a kill -9.
  if not running or state == 'stopped' then
    out.countdown = '--'
    out.phase_label = 'fermo'
    out.next = '--'
    out.role = 'label'
    return out
  end

  out.countdown = string.format('%d:%02d', math.floor(remain / 60), remain % 60)

  if state == 'paused' then
    out.phase_label = 'in pausa'
    out.role = 'warning'
  elseif out.frozen then
    out.phase_label = 'outside working hours'
    out.role = 'label'
  else
    local labels = { working = 'working', breaking = 'micro-pausa',
                     longbreak = 'pausa lunga' }
    local roles = { working = 'ok', breaking = 'heading',
                    longbreak = 'highlight' }
    out.phase_label = labels[phase] or phase
    out.role = roles[phase] or 'value'
  end

  out.next = (phase == 'working') and 'pausa' or 'lavoro'
  return out
end

-- The Immich memories index written by bin/immich-fetch.sh.
--
-- Returns a list of { id, year, taken = { y, m, d, H, M }, place }, plus
-- .url, the server's address for a click-through, and .week, true when the
-- photos are the past week's fallback rather than today's, or nil when
-- the index is absent or carries no 'generated' key. nil means "the fetch never
-- ran"; an empty list means "no memories today", which the card names
-- differently. `place` is nil when the key lacks asset.read or the photo has no
-- location, so the card can print '--'.
function M.immich_parse(text)
  if type(text) ~= 'string' then return nil end
  local out, generated = {}, false
  for line in text:gmatch('[^\n]+') do
    local kind, rest = line:match('^(%S+)%s+(.*)$')
    if kind == 'generated' then
      generated = true
    elseif kind == 'url' then
      out.url = rest
    elseif kind == 'scope' then
      out.week = rest == 'week'
    elseif kind == 'photo' then
      -- The place is last and takes the rest of the line, so a pipe in a
      -- place name cannot shift a field.
      local id, year, taken, place = rest:match('^([^|]+)|(%d*)|([^|]*)|(.*)$')
      if id then
        local y, m, d, H, Mi = taken:match('^(%d+)-(%d+)-(%d+)T(%d+):(%d+)')
        out[#out + 1] = {
          id = id,
          year = tonumber(year),
          taken = y and { tonumber(y), tonumber(m), tonumber(d),
                          tonumber(H), tonumber(Mi) } or nil,
          place = place ~= '' and place or nil,
        }
      end
    end
  end
  if not generated then return nil end
  return out
end

-- nvtable's plain table (`nvtable -C -E`), as bin/updates-sample.sh caches it:
--
--   [personal] foo-bin   1.14.9 -> 1.14.9   !! ~  1.14.10
--   <repo>     <name>  <installed> -> <shipped>  <marker> [~] [upstream]
--
-- followed by a summary line with the counts. The marker is '!!' (installed
-- behind upstream), 'ok' or '-' (not installed); '~' flags the shipped .info
-- as behind. Upstream appears only when shipped does not already show it.
--
-- Returns the rows as { repo, name, installed, shipped, upstream, behind,
-- bump } with '--' turned into nil, plus .upgrade/.bump/.current/.missing
-- counts, or nil when the text has neither rows nor a summary.
function M.nvtable_parse(text)
  if type(text) ~= 'string' then return nil end
  local out, seen = {}, false
  local function v(s) return s ~= '--' and s or nil end
  for line in text:gmatch('[^\n]+') do
    local repo, name, inst, ship, rest =
      line:match('^%s*%[(.-)%]%s+(%S+)%s+(%S+)%s+%->%s+(%S+)%s*(.*)$')
    if repo then
      local mark, tail = rest:match('^(%S*)%s*(.*)$')
      local bump = tail:match('^~') ~= nil
      if bump then tail = tail:gsub('^~%s*', '') end
      out[#out + 1] = {
        repo = v(repo), name = name, installed = v(inst), shipped = v(ship),
        upstream = tail:match('^(%S+)'), behind = mark == '!!', bump = bump,
      }
      seen = true
    elseif line:match('to upgrade') then
      out.upgrade = tonumber(line:match('(%d+) to upgrade'))
      out.bump    = tonumber(line:match('(%d+) repos? to bump'))
      out.current = tonumber(line:match('(%d+) current'))
      out.missing = tonumber(line:match('(%d+) not installed')) or 0
      seen = true
    end
  end
  return seen and out or nil
end

return M