tap_google_sheets/schema.py

   1 import os
   2 import json
   3 from collections import OrderedDict
   4 import singer
   5 from singer import metadata
   6 from tap_google_sheets.streams import STREAMS
   7
   8 LOGGER = singer.get_logger()
   9
  10 # Reference:
  11 # https://github.com/singer-io/getting-started/blob/master/docs/DISCOVERY_MODE.md#Metadata
  12
  13 # Convert column index to column letter
  14 def colnum_string(num):
  15     string = ""
  16     while num > 0:
  17         num, remainder = divmod(num - 1, 26)
  18         string = chr(65 + remainder) + string
  19     return string
  20
  21
  22 # Create sheet_metadata_json with columns from sheet
  23 def get_sheet_schema_columns(sheet):
  24     sheet_title = sheet.get('properties', {}).get('title')
  25     sheet_json_schema = OrderedDict()
  26     data = next(iter(sheet.get('data', [])), {})
  27     row_data = data.get('rowData', [])
  28     # spreadsheet is an OrderedDict, with orderd sheets and rows in the repsonse
  29
  30     headers = row_data[0].get('values', [])
  31     first_values = row_data[1].get('values', [])
  32     # LOGGER.info('first_values = {}'.format(json.dumps(first_values, indent=2, sort_keys=True)))
  33
  34     sheet_json_schema = {
  35         'type': 'object',
  36         'additionalProperties': False,
  37         'properties': {
  38             '__sdc_spreadsheet_id': {
  39                 'type': ['null', 'string']
  40             },
  41             '__sdc_sheet_id': {
  42                 'type': ['null', 'integer']
  43             },
  44             '__sdc_row': {
  45                 'type': ['null', 'integer']
  46             }
  47         }
  48     }
  49
  50     header_list = [] # used for checking uniqueness
  51     columns = []
  52     prior_header = None
  53     i = 0
  54     skipped = 0
  55     # Read column headers until end or 2 consecutive skipped headers
  56     for header in headers:
  57         # LOGGER.info('header = {}'.format(json.dumps(header, indent=2, sort_keys=True)))
  58         column_index = i + 1
  59         column_letter = colnum_string(column_index)
  60         header_value = header.get('formattedValue')
  61         if header_value: # NOT skipped
  62             column_is_skipped = False
  63             skipped = 0
  64             column_name = '{}'.format(header_value)
  65             if column_name in header_list:
  66                 raise Exception('DUPLICATE HEADER ERROR: SHEET: {}, COL: {}, CELL: {}1'.format(
  67                     sheet_title, column_name, column_letter))
  68             header_list.append(column_name)
  69
  70             first_value = None
  71             try:
  72                 first_value = first_values[i]
  73             except IndexError as err:
  74                 raise Exception('NO VALUE IN 2ND ROW FOR HEADER ERROR. SHEET: {}, COL: {}, CELL: {}2. {}'.format(
  75                     sheet_title, column_name, column_letter, err))
  76
  77             column_effective_value = first_value.get('effectiveValue', {})
  78
  79             col_val = None
  80             if column_effective_value == {}:
  81                 column_effective_value_type = 'stringValue'
  82                 LOGGER.info('WARNING: NO VALUE IN 2ND ROW FOR HEADER. SHEET: {}, COL: {}, CELL: {}2.'.format(
  83                     sheet_title, column_name, column_letter))
  84                 LOGGER.info('   Setting column datatype to STRING')
  85             else:
  86                 for key, val in column_effective_value.items():
  87                     if key in ('numberValue', 'stringValue', 'boolValue'):
  88                         column_effective_value_type = key
  89                         col_val = str(val)
  90                     elif key in ('errorType', 'formulaType'):
  91                         col_val = str(val)
  92                         raise Exception('DATA TYPE ERROR 2ND ROW VALUE: SHEET: {}, COL: {}, CELL: {}2, TYPE: {}, VALUE: {}'.format(
  93                             sheet_title, column_name, column_letter, key, col_val))
  94
  95             column_number_format = first_values[i].get('effectiveFormat', {}).get(
  96                 'numberFormat', {})
  97             column_number_format_type = column_number_format.get('type')
  98
  99             # Determine datatype for sheet_json_schema
 100             #
 101             # column_effective_value_type = numberValue, stringValue, boolValue;
 102             #  INVALID: errorType, formulaType
 103             #  https://developers.google.com/sheets/api/reference/rest/v4/spreadsheets/other#ExtendedValue
 104             #
 105             # column_number_format_type = UNEPECIFIED, TEXT, NUMBER, PERCENT, CURRENCY, DATE,
 106             #   TIME, DATE_TIME, SCIENTIFIC
 107             #  https://developers.google.com/sheets/api/reference/rest/v4/spreadsheets/cells#NumberFormatType
 108             #
 109             column_format = None # Default
 110             if column_effective_value == {}:
 111                 col_properties = {'type': ['null', 'string']}
 112                 column_gs_type = 'stringValue'
 113                 LOGGER.info('WARNING: 2ND ROW VALUE IS BLANK: SHEET: {}, COL: {}, CELL: {}2'.format(
 114                         sheet_title, column_name, column_letter))
 115                 LOGGER.info('   Setting column datatype to STRING')
 116             elif column_effective_value_type == 'stringValue':
 117                 col_properties = {'type': ['null', 'string']}
 118                 column_gs_type = 'stringValue'
 119             elif column_effective_value_type == 'boolValue':
 120                 col_properties = {'type': ['null', 'boolean', 'string']}
 121                 column_gs_type = 'boolValue'
 122             elif column_effective_value_type == 'numberValue':
 123                 if column_number_format_type == 'DATE_TIME':
 124                     col_properties = {
 125                         'type': ['null', 'string'],
 126                         'format': 'date-time'
 127                     }
 128                     column_gs_type = 'numberType.DATE_TIME'
 129                 elif column_number_format_type == 'DATE':
 130                     col_properties = {
 131                         'type': ['null', 'string'],
 132                         'format': 'date'
 133                     }
 134                     column_gs_type = 'numberType.DATE'
 135                 elif column_number_format_type == 'TIME':
 136                     col_properties = {
 137                         'type': ['null', 'string'],
 138                         'format': 'time'
 139                     }
 140                     column_gs_type = 'numberType.TIME'
 141                 elif column_number_format_type == 'TEXT':
 142                     col_properties = {'type': ['null', 'string']}
 143                     column_gs_type = 'stringValue'
 144                 else:
 145                     # Interesting - order in the anyOf makes a difference.
 146                     # Number w/ multipleOf must be listed last, otherwise errors occur.
 147                     col_properties =  {
 148                         'anyOf': [
 149                             {
 150                                 'type': 'string'
 151                             },
 152                             {
 153                                 'type': 'null'
 154                             },
 155                             {
 156                                 'type': 'number',
 157                                 'multipleOf': 1e-15
 158                             }
 159                         ]
 160                     }
 161                     column_gs_type = 'numberType'
 162             # Catch-all to deal with other types and set to string
 163             # column_effective_value_type: formulaValue, errorValue, or other
 164             else:
 165                 col_properties = {'type': ['null', 'string']}
 166                 column_gs_type = 'unsupportedValue'
 167                 LOGGER.info('WARNING: UNSUPPORTED 2ND ROW VALUE: SHEET: {}, COL: {}, CELL: {}2, TYPE: {}, VALUE: {}'.format(
 168                         sheet_title, column_name, column_letter, column_effective_value_type, col_val))
 169                 LOGGER.info('Converting to string.')
 170         else: # skipped
 171             column_is_skipped = True
 172             skipped = skipped + 1
 173             column_index_str = str(column_index).zfill(2)
 174             column_name = '__sdc_skip_col_{}'.format(column_index_str)
 175             col_properties = {'type': ['null', 'string']}
 176             column_gs_type = 'stringValue'
 177             LOGGER.info('WARNING: SKIPPED COLUMN; NO COLUMN HEADER. SHEET: {}, COL: {}, CELL: {}1'.format(
 178                 sheet_title, column_name, column_letter))
 179             LOGGER.info('  This column will be skipped during data loading.')
 180
 181         if skipped >= 2:
 182             # skipped = 2 consecutive skipped headers
 183             # Remove prior_header column_name
 184             sheet_json_schema['properties'].pop(prior_header, None)
 185             LOGGER.info('TWO CONSECUTIVE SKIPPED COLUMNS. STOPPING SCAN AT: SHEET: {}, COL: {}, CELL {}1'.format(
 186                 sheet_title, column_name, column_letter))
 187             break
 188
 189         else:
 190             column = {}
 191             column = {
 192                 'columnIndex': column_index,
 193                 'columnLetter': column_letter,
 194                 'columnName': column_name,
 195                 'columnType': column_gs_type,
 196                 'columnSkipped': column_is_skipped
 197             }
 198             columns.append(column)
 199
 200             sheet_json_schema['properties'][column_name] = col_properties
 201
 202         prior_header = column_name
 203         i = i + 1
 204
 205     return sheet_json_schema, columns
 206
 207
 208 # Get Header Row and 1st data row (Rows 1 & 2) from a Sheet on Spreadsheet w/ sheet_metadata query
 209 #   endpoint: spreadsheets/{spreadsheet_id}
 210 #   params: includeGridData = true, ranges = '{sheet_title}'!1:2
 211 # This endpoint includes detailed metadata about each cell - incl. data type, formatting, etc.
 212 def get_sheet_metadata(sheet, spreadsheet_id, client):
 213     sheet_id = sheet.get('properties', {}).get('sheetId')
 214     sheet_title = sheet.get('properties', {}).get('title')
 215     LOGGER.info('sheet_id = {}, sheet_title = {}'.format(sheet_id, sheet_title))
 216
 217     stream_name = 'sheet_metadata'
 218     stream_metadata = STREAMS.get(stream_name)
 219     api = stream_metadata.get('api', 'sheets')
 220     params = stream_metadata.get('params', {})
 221     querystring = '&'.join(['%s=%s' % (key, value) for (key, value) in \
 222         params.items()]).replace('{sheet_title}', sheet_title)
 223     path = '{}?{}'.format(stream_metadata.get('path').replace('{spreadsheet_id}', \
 224         spreadsheet_id), querystring)
 225
 226     sheet_md_results = client.get(path=path, api=api, endpoint=stream_name)
 227     # sheet_metadata: 1st `sheets` node in results
 228     sheet_metadata = sheet_md_results.get('sheets')[0]
 229
 230     # Create sheet_json_schema (for discovery/catalog) and columns (for sheet_metadata results)
 231     sheet_json_schema, columns = get_sheet_schema_columns(sheet_metadata)
 232
 233     return sheet_json_schema, columns
 234
 235
 236 def get_abs_path(path):
 237     return os.path.join(os.path.dirname(os.path.realpath(__file__)), path)
 238
 239 def get_schemas(client, spreadsheet_id):
 240     schemas = {}
 241     field_metadata = {}
 242
 243     for stream_name, stream_metadata in STREAMS.items():
 244         schema_path = get_abs_path('schemas/{}.json'.format(stream_name))
 245         with open(schema_path) as file:
 246             schema = json.load(file)
 247         schemas[stream_name] = schema
 248         mdata = metadata.new()
 249
 250         # Documentation:
 251         # https://github.com/singer-io/getting-started/blob/master/docs/DISCOVERY_MODE.md#singer-python-helper-functions
 252         # Reference:
 253         # https://github.com/singer-io/singer-python/blob/master/singer/metadata.py#L25-L44
 254         mdata = metadata.get_standard_metadata(
 255             schema=schema,
 256             key_properties=stream_metadata.get('key_properties', None),
 257             valid_replication_keys=stream_metadata.get('replication_keys', None),
 258             replication_method=stream_metadata.get('replication_method', None)
 259         )
 260         field_metadata[stream_name] = mdata
 261
 262         if stream_name == 'spreadsheet_metadata':
 263             api = stream_metadata.get('api', 'sheets')
 264             params = stream_metadata.get('params', {})
 265             querystring = '&'.join(['%s=%s' % (key, value) for (key, value) in params.items()])
 266             path = '{}?{}'.format(stream_metadata.get('path').replace('{spreadsheet_id}', \
 267                 spreadsheet_id), querystring)
 268
 269             # GET spreadsheet_metadata, which incl. sheets (basic metadata for each worksheet)
 270             spreadsheet_md_results = client.get(path=path, params=querystring, api=api, \
 271                 endpoint=stream_name)
 272
 273             sheets = spreadsheet_md_results.get('sheets')
 274             if sheets:
 275                 # Loop thru each worksheet in spreadsheet
 276                 for sheet in sheets:
 277                     # GET sheet_json_schema for each worksheet (from function above)
 278                     sheet_json_schema, columns = get_sheet_metadata(sheet, spreadsheet_id, client)
 279                     # LOGGER.info('columns = {}'.format(columns))
 280
 281                     sheet_title = sheet.get('properties', {}).get('title')
 282                     schemas[sheet_title] = sheet_json_schema
 283                     sheet_mdata = metadata.new()
 284                     sheet_mdata = metadata.get_standard_metadata(
 285                         schema=sheet_json_schema,
 286                         key_properties=['__sdc_row'],
 287                         valid_replication_keys=None,
 288                         replication_method='FULL_TABLE'
 289                     )
 290                     field_metadata[sheet_title] = sheet_mdata
 291
 292     return schemas, field_metadata