After opening a resource you get access to a resource.header object which describes the resource in more detail. This is a list of normalized labels but also provides some additional functionality. Let's take a look:
from frictionless import Resource
with Resource('capital-3.csv') as resource:
print(f'Header: {resource.header}')
print(f'Labels: {resource.header.labels}')
print(f'Fields: {resource.header.fields}')
print(f'Field Names: {resource.header.field_names}')
print(f'Field Numbers: {resource.header.field_numbers}')
print(f'Errors: {resource.header.errors}')
print(f'Valid: {resource.header.valid}')
print(f'As List: {resource.header.to_list()}')
Header: ['id', 'name']
Labels: ['id', 'name']
Fields: [{'name': 'id', 'type': 'integer'}, {'name': 'name', 'type': 'string'}]
Field Names: ['id', 'name']
Field Numbers: [1, 2]
Errors: []
Valid: True
As List: ['id', 'name']The example above shows a case when a header is valid. For a header that contains errors in its tabular structure, this information can be very useful, revealing discrepancies, duplicates or missing cell information:
from pprint import pprint
from frictionless import Resource
with Resource([['name', 'name'], ['value', 'value']]) as resource:
pprint(resource.header.errors)
[{'type': 'duplicate-label',
'title': 'Duplicate Label',
'description': 'Two columns in the header row have the same value. Column '
'names should be unique.',
'message': 'Label "name" in the header at position "2" is duplicated to a '
'label: at position "1"',
'tags': ['#table', '#header', '#label'],
'note': 'at position "1"',
'labels': ['name', 'name'],
'rowNumbers': [1],
'label': 'name',
'fieldName': 'name2',
'fieldNumber': 2}]The extract, resource.read_rows() and other functions return or yield row objects. In Python, this returns a dictionary with the following information. Note: this example uses the Detector object, which tweaks how different aspects of metadata are detected.
from frictionless import Resource, Detector
detector = Detector(schema_patch={'missingValues': ['1']})
with Resource('capital-3.csv', detector=detector) as resource:
for row in resource.row_stream:
print(f'Row: {row}')
print(f'Cells: {row.cells}')
print(f'Fields: {row.fields}')
print(f'Field Names: {row.field_names}')
print(f'Value of field "name": {row["name"]}') # accessed as a dict
print(f'Row Number: {row.row_number}') # counted row number starting from 1
print(f'Blank Cells: {row.blank_cells}')
print(f'Error Cells: {row.error_cells}')
print(f'Errors: {row.errors}')
print(f'Valid: {row.valid}')
print(f'As Dict: {row.to_dict(json=False)}')
print(f'As List: {row.to_list(json=True)}') # JSON compatible data types
break
Row: Unprocessed: Unprocessed: {}
Cells: ['1', 'London']
Fields: [{'name': 'id', 'type': 'integer'}, {'name': 'name', 'type': 'string'}]
Field Names: ['id', 'name']
Value of field "name": London
Row Number: 2
Blank Cells: {'id': '1'}
Error Cells: {}
Errors: []
Valid: True
As Dict: {'id': None, 'name': 'London'}
As List: [None, 'London']As we can see, this output provides a lot of information which is especially useful when a row is not valid. Our row is valid but we demonstrated how it can preserve data about missing values. It also preserves data about all cells that contain errors:
from pprint import pprint
from frictionless import Resource
with Resource([['name'], ['value', 'value']]) as resource:
for row in resource.row_stream:
pprint(row.errors)
[{'type': 'extra-cell',
'title': 'Extra Cell',
'description': 'This row has more values compared to the header row (the '
'first row in the data source). A key concept is that all the '
'rows in tabular data must have the same number of columns.',
'message': 'Row at position "2" has an extra value in field at position "2"',
'tags': ['#table', '#row', '#cell'],
'note': '',
'cells': ['value', 'value'],
'rowNumber': 2,
'cell': 'value',
'fieldName': '',
'fieldNumber': 2}]Header representation Compares the header row read from the data source (the "labels") with the fields declared in the schema, and reports the mismatches as errors. > Constructor of this object is not Public API
(labels: List[str], *, fields: List[Field], row_numbers: List[int], ignore_case: bool = False, fields_match: types.IFieldsMatch = exact)
Returns the fields, in the order expected in the data. Under `exact`, this is just the schema fields unchanged. Under the name-matched modes, fields are reordered to match the labels; labels without a matching field get a fresh `any`-typed field (even where such a label is an error, so that it is reported once rather than once per row), and fields not present in labels are dropped. Duplicate labels are rejected, as they make the mapping ambiguous.
() -> List[Field]
Convert to a list
Row representation > Constructor of this object is not Public API This object is returned by `extract`, `resource.read_rows`, and other functions. ```python rows = extract("data/table.csv") for row in rows: # work with the Row ```
(cells: List[Any], *, handlers: Dict[str, _CellHandler], row_number: int)
(*, csv: bool = False, json: bool = False, types: Optional[List[str]] = None) -> Dict[str, Any]
(*, json: bool = False, types: Optional[List[str]] = None)
(**options: Any)