base = "$(homedir())/text-fabric-data/etcbc/bhsa/tf/c"
feature = "g_word_utf8"
featurePath = "$base/$feature.tf"

function error(msg)
    write(STDERR, "$msg\n")
end

function showResults(errors, data)
  if errors == 0
      maxNode = maximum(keys(data))
      print("$(length(data)) results, last node $maxNode\n")
      print("$(data[1])\n")
      print("$(data[2])\n")
      print("$(data[maxNode])\n")
  else
      print("$errors errors")
  end
end

function valueFromTf(tf)
  join([replace(replace(x, "\\t", "\t"), "\\n", "\n") for x in split(tf, "\\\\")], "\\")
end

function setFromSpec(spec)::Set{UInt32}
  covered = Set{UInt32}()
  for r_str in split(spec, ",")
    bounds = split(r_str, "-")
    if length(bounds) == 1
      push!(covered, parse(UInt32, r_str))
    else
      b = parse(UInt32, bounds[1])
      e = parse(UInt32, bounds[2])
      if e < b
        (b, e) = (e, b)
      end
      for n in b:e
        push!(covered, n)
      end
    end
  end
  covered
end

function readFile(path)
  if !isfile(path)
    error("TF reading: feature file '$path' does not exist")
    return false
  end
  contents = open(path) do fh
    read(fh, String)
  end
  lines = split(contents, "\n")
  if lines[end] == ""
    pop!(lines)
  end
  i::UInt32 = 0
  for line in lines
    i += 1
    if startswith(line, "@")
      continue
    else
      if line != ""
        error("Line $i: missing blank line after metadata")
        return false
      else
        break
      end
    end
  end
  i += 1
  (lines, i)
end

function readTf(path)
  if !isfile(path)
    error("TF reading: feature file '$path' does not exist")
    return false
  end
  fh = open(path)
  i = 0
  for line in eachline(fh)
    i += 1
    text = rstrip(line)
    if startswith(text, "@")
      continue
    else
      if text != ""
        error("Line $i: missing blank line after metadata")
        close(fh)
        return false
      else
        break
      end
    end
  end
  result = readDataTf(fh, i)
  close(fh)
  result
end

function readDataTf(fh, firstI)
  i = firstI
  implicit_node = 1
  data = Dict{Integer, String}()
  normFields = 2
  isNum = false
  errors = 0
  for line in eachline(fh)
    i += 1
    fields = split(rstrip(line, '\n'), "\t")
    lfields = length(fields)
    if lfields > normFields
      error("$(i) : wrongFields")
      errors += 1
      continue
    end
    if lfields == normFields
      nodes = setFromSpec(fields[1])
      valTf = fields[end]
    else
      nodes = Set([implicit_node])
      if lfields == 1
        valTf = fields[1]
      else
        valTf = ""
      end
    end
    implicit_node = maximum(nodes) + 1
    value = (
        valTf == "" ?
          (isNum && valTf != "") ? parse(Int, valTf) : (isNum ? nothing : "") :
          valueFromTf(valTf)
    )
    for n in nodes
      if value !== nothing
        data[n] = value
      end
    end
  end
  (errors, data)
end

function readTfList(path)
  if !isfile(path)
    error("TF reading: feature file '$path' does not exist")
    return false
  end
  fh = open(path)
  i = 0
  for line in eachline(fh)
    i += 1
    text = rstrip(line)
    if startswith(text, "@")
      continue
    else
      if text != ""
        error("Line $i: missing blank line after metadata")
        close(fh)
        return false
      else
        break
      end
    end
  end
  result = readDataTfList(fh, i)
  close(fh)
  result
end

function readDataTfList(fh, firstI)
  i = firstI
  implicit_node = 1
  data = Array{String, 1}()
  normFields = 2
  isNum = false
  errors = 0
  for line in eachline(fh)
    i += 1
    fields = split(rstrip(line, '\n'), "\t")
    lfields = length(fields)
    if lfields > normFields
      error("$(i) : wrongFields")
      errors += 1
      continue
    end
    if lfields == normFields
      nodes = setFromSpec(fields[1])
      valTf = fields[end]
    else
      nodes = Set([implicit_node])
      if lfields == 1
        valTf = fields[1]
      else
        valTf = ""
      end
    end
    implicit_node = maximum(nodes) + 1
    value = (
        valTf == "" ?
          (isNum && valTf != "") ? parse(Int, valTf) : (isNum ? nothing : "") :
          valueFromTf(valTf)
    )
    for n in nodes
      if value !== nothing
        push!(data, value)
      end
    end
  end
  (errors, data)
end

function readTfSlurp(path)
  if !isfile(path)
    error("TF reading: feature file '$path' does not exist")
    return false
  end
  contents = open(path) do fh
    read(fh, String)
  end
  lines = split(contents, "\n")
  if lines[end] == ""
    pop!(lines)
  end
  i = 0
  for line in lines
    i += 1
    if startswith(line, "@")
      continue
    else
      if line != ""
        error("Line $i: missing blank line after metadata")
        return false
      else
        break
      end
    end
  end
  result = readDataTfSlurp(lines, i + 1)
  result
end

function readDataTfSlurp(lines, firstI)
  i = firstI
  implicit_node = 1
  data = Dict{Integer, String}()
  normFields = 2
  isNum = false
  errors = 0
  for line in lines[firstI:end]
    i += 1
    fields = split(line, "\t")
    lfields = length(fields)
    if lfields > normFields
      error("$(i) : wrongFields")
      errors += 1
      continue
    end
    if lfields == normFields
      nodes = setFromSpec(fields[1])
      valTf = fields[end]
    else
      nodes = Set([implicit_node])
      if lfields == 1
        valTf = fields[1]
      else
        valTf = ""
      end
    end
    implicit_node = maximum(nodes) + 1
    value = (
        valTf == "" ?
          (isNum && valTf != "") ? parse(Int, valTf) : (isNum ? nothing : "") :
          valueFromTf(valTf)
    )
    for n in nodes
      if value !== nothing
        data[n] = value
      end
    end
  end
  (errors, data)
end

function readDataTfSlurpOpt(lines, firstI::UInt32)
  i::UInt32 = firstI
  implicit_node::UInt32 = 1
  data = Dict{UInt32, SubString{String}}()
  normFields::UInt8 = 2
  isNum::Bool = false
  errors::UInt32 = 0
  for line in lines[firstI:end]
    i += 1
    fields = split(line, "\t")
    lfields::UInt8 = length(fields)
    if lfields > normFields
      error("$(i) : wrongFields")
      errors += 1
      continue
    end
    if lfields == normFields
      nodes::Set{UInt32} = setFromSpec(fields[1])
      valTf = fields[end]
    else
      nodes = Set{UInt32}([implicit_node])
      if lfields == 1
        valTf = fields[1]
      else
        valTf = ""
      end
    end
    implicit_node = maximum(nodes) + 1
    value = (
        valTf == "" ?
          (isNum && valTf != "") ? parse(Int, valTf) : (isNum ? nothing : "") :
          valueFromTf(valTf)
    )
    for n::UInt32 in nodes
      if value !== nothing
        data[n] = value
      end
    end
  end
  (errors, data)
end

(errors, data) = readTf(featurePath)

showResults(errors, data)

(errors, data) = readTfList(featurePath)

showResults(errors, data)

(errors, data) = readTfSlurp(featurePath)

showResults(errors, data)

(lines, first) = readFile(featurePath)

(errors, data) = readDataTfSlurpOpt(lines, first)

showResults(errors, data)