gistfile1.py

class TypeTree(object):
  def __init__(self, name, *children, append=False):
    self._name = name
    self._children = children or []
    self._append = append

  def GetAllTags(self):
    result = set([self._name])
    for child in self._children:
      if not self._append:
        result.update(child.GetAllTags())
      else:
        for c in child.GetAllTags():
          result.add(f'{c}_{self._name}')
    return result

  def Iterate(self, tag_as=None) -> ('Tag', 'Parent'):
    for child in self._children:
      if not self._append:
        yield (child._name, self._name)
        yield from child.Iterate()
      else:
        child_name = f'{child._name}_{self._name}'
        yield (child_name, self._name)
        yield from child.Iterate(tag_as=child_name)


typetree = TypeTree('everything',
  TypeTree('seafood',
    TypeTree('shellfish',
      TypeTree('processed'),
      TypeTree('wild'),
      TypeTree('frozen'),
      TypeTree('fresh'),
      TypeTree('farmed'),
      TypeTree('valueAdded'), append=True),
    TypeTree('halibut',
      TypeTree('wild'),
      TypeTree('frozen'),
      TypeTree('fresh'),
      TypeTree('valueAdded'), append=True),
    TypeTree('whitefish',
      TypeTree('processed'),
      TypeTree('wild'),
      TypeTree('frozen'),
      TypeTree('fresh'),
      TypeTree('farmed'),
      TypeTree('valueAdded'), append=True),
    TypeTree('finfish',
      TypeTree('processed'),
      TypeTree('wild'),
      TypeTree('frozen'),
      TypeTree('fresh'),
      TypeTree('farmed'),
      TypeTree('valueAdded'), append=True),
    TypeTree('shrimp',
      TypeTree('processed'),
      TypeTree('wild'),
      TypeTree('frozen'),
      TypeTree('fresh'),
      TypeTree('farmed'),
      TypeTree('valueAdded'), append=True),
    TypeTree('cephalopods',
      TypeTree('processed'),
      TypeTree('wild'),
      TypeTree('frozen'),
      TypeTree('valueAdded'), append=True),
    TypeTree('salmon',
      TypeTree('processed'),
      TypeTree('wild'),
      TypeTree('frozen'),
      TypeTree('fresh'),
      TypeTree('farmed'),
      TypeTree('king'),
      TypeTree('valueAdded'), append=True)),
  TypeTree('beef',
    TypeTree('processed'),
    TypeTree('responsible'),
    TypeTree('valueAdded'), append=True),
  TypeTree('pork',
    TypeTree('processed'),
    TypeTree('responsible'),
    TypeTree('valueAdded'), append=True),
  TypeTree('lamb',
    TypeTree('processed'),
    TypeTree('responsible'),
    TypeTree('valueAdded'), append=True),
  TypeTree('mixed',
    TypeTree('processed'),
    TypeTree('responsible'),
    TypeTree('valueAdded'), append=True),
  TypeTree('poultry',
    TypeTree('processed'),
    TypeTree('responsible'),
    TypeTree('valueAdded'), append=True),
  TypeTree('not_meat'),
  TypeTree('Garbage'))


def mfloat(s):
  if s == 'inf':
    return 0.0
  if s == 'nan':
    return 0.0
  return float(s)

def CreateRows(file):
  date_entry = {}
  with open(file) as f:
    for line in f.readlines():
      date, upc, _, A, B, C, tags = line.strip().split(',')
      if date not in date_entry:
        date_entry[date] = {}
      for tag in tags.split('|'):
        if tag not in date_entry[date]:
          date_entry[date][tag] = [
            round(mfloat(A or '0'), 2),
            round(mfloat(B or '0'), 2),
            round(mfloat(C or '0'), 2),
            0, 0, 0]
        else:
          date_entry[date][tag][0] = round(date_entry[date][tag][0] + mfloat(A or '0'), 2)
          date_entry[date][tag][1] = round(date_entry[date][tag][1] + mfloat(B or '0'), 2)
          date_entry[date][tag][2] = round(date_entry[date][tag][2] + mfloat(C or '0'), 2)

  for date, data in date_entry.items():
    for tag, values in data.items():
      values[1] = values[0] / values[2] if values[2] else 0

  tagmap = {tag:parent for tag,parent in typetree.Iterate()}
  for date, data in date_entry.items():
    for tag, values in data.items():
      if tag not in tagmap:
        continue
      parent_values = data[tagmap[tag]]
      for i in range(3):
        if parent_values[i] != 0:
          values[i+3] = round((values[i] / parent_values[i]) * 100.0, 2)
  return date_entry


if __name__ == '__main__':
  tagmap = {tag:parent for tag,parent in typetree.Iterate()}
  csvorder = list(x for x,_ in typetree.Iterate() if x not in ('Garbage', 'not_meat'))

  with open('output.csv', 'w') as f:
    f.write(',everything,,')
    for tag in csvorder:
      f.write(f',{tag},(parent={tagmap[tag]}),,,,')
    f.write('\n')
    f.write('Date,Revenue Total,PPP Total,PoundageTotal')
    for tag in csvorder:
      f.write(',RevenueTotal,PPP Total,Poundage Total,Revenue %Parent,PPP %Parent,Poundage %Parent')
    f.write('\n')

    for date, v in CreateRows('pivot_table_partially_processed.csv').items():
      f.write(date.split(' ')[0])
      f.write(',')
      f.write(','.join(str(v['everything'][i]) for i in range(3)))
      for tag in csvorder:
        try:
          sep = ','.join(str(v[tag][i]) for i in range(6))
          f.write(f',{sep}')
        except:
          f.write(',,,,,,')
      f.write('\n')