read_excel.lua

-- Requires slaxdom and slaxml from https://github.com/Phrogz/SLAXML
local slaxdom = require'slaxdom'

local xlsxNs = 'http://schemas.openxmlformats.org/spreadsheetml/2006/main'
local relNs = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships'
local function filteredNext(ns, element)
  local function iter(table, i)
    i = i + 1
    local elem = table[i]
    if not elem then
      return nil
    end
    if elem.nsURI == ns and elem.name == element then
      return i, elem
    end
    return iter(table, i)
  end
  return iter
end
local function elementIter(parent, ns, element)
  return filteredNext(ns, element), parent.el, 0
end
local function findAttr(element, ns, attr)
  if not attr and ns then ns, attr = attr, ns end
  for _,attribute in ipairs(element.attr) do
    if attribute.nsURI == ns and attribute.name == attr then
      return attribute.value
    end
  end
end

local function readDom(f, msg)
  if not f then return f, msg end
  local data = f:read'*a'
  f:close()
  return slaxdom:dom(data)
end

local f = assert(zip.open'how_should_i_know_how_you_call_your_files.xlsx')
local sheets = {}
local relationships do
  local ns = 'http://schemas.openxmlformats.org/package/2006/relationships'
  local dom = assert(readDom(f:open'xl/_rels/workbook.xml.rels')).root
  relationships = {}
  for _, rel in elementIter(dom, ns, 'Relationship') do
    relationships[rel.attr.Id] = rel.attr.Target
  end
end
local workbook = assert(readDom(f:open'xl/workbook.xml')).root
local worksheets = {}
for _, sheets in elementIter(workbook, xlsxNs, 'sheets') do
  for _, sheet in elementIter(sheets, xlsxNs, 'sheet') do
    local worksheet = assert(readDom(f:open('xl/' .. relationships[findAttr(sheet, relNs, 'id')]))).root
    assert(worksheet.nsURI == xlsxNs and worksheet.name == 'worksheet')
    local sheetData
    for _, element in elementIter(worksheet, xlsxNs, 'sheetData') do
      assert(not sheetData)
      sheetData = element
    end
    assert(sheetData)
    local rows = {}
    for i, rowElement in elementIter(sheetData, xlsxNs, 'row') do
      local row = {}
      for j, colElem in elementIter(rowElement, xlsxNs, 'c') do
        for _, entry in elementIter(colElem, xlsxNs, 'v') do
          local value = entry.kids
          if not value or #value == 0 then
            value = nil
          elseif #value == 1 and value[1].type == 'text' then
            value = value[1].value
            value = tonumber(value) or value
          else
            error'???'
          end
          row[j] = value
          break
        end
      end
      rows[i] = row
    end
    worksheets[sheet.attr.name] = rows
  end
  break
end

print(worksheets.Sheet1[5][2])