Excel Comment ORM. Declare ORM Spec descriptively right on the excel file.
The package allows you to declare orm mapping right in the excel file in the comments then use it to extract data from other similar files.
An example of a template is shown below.
Dynamic Location, Validation, Assumptions, custom Parser are also supported.
pip install exco
import exco
processor = exco.from_excel('./quickstart_template.xlsx')
result = processor.process_excel('./quickstart_data_file.xlsx')
print(result.to_dict())
See Also Quick Start Notebook
Exco relies on yml blocks inside excel comment to build a spec. Each comment can contain multiple blocks. There are three types of echo block: a cell block, a table block, and a column block.
A cellblock is a block of yml that is surrounded by {{--
and --}}
{{--
key: some_int
parser: int
--}}
This means that parse this cell as int
and put it in the result with key some_int
.
Other advance features like, fallback, locator, assumptions, validator are optional. The full specification is shown below.
{{--
key: some_value
# Optional default at_comment_cell
locator: {name: at_comment_cell}
assumptions: # Optional
- {key: right_of, name: right_of, label: marker}
parser: int
# Optional
fallback: 7
# Optional. Dict[name, value].
params: {}
validators: # Optional
- {key: between, name: between, low: 5, high: 10 }
metadata: {unit: km}
--}}
The cell processing flow is
- Locating the Cell
- Check Assumption
- Parse
- Validation
Key where the result will be put. This is required.
This is the most useful one where if the cell is failed locating, testing assumption, parsing.
The fallback value is assumed. If it is not specified, the value None is used. (yml's none is null
).
Locator is a dictionary. The key name
is the class name to be used to locate the cell. Other parameters
can be flattened and put in the same dictionary. These parameters will be pass through the constructor
of such class.
If the locator is left out, it is assumed to be at_comment_cell
which locate the cell with exactly
the same coordinate as the commented cell in the template.
List of dictionaries. The key
is the name of the check. name
is the name of the class to be used in
testing an assumption. Other parameters can be flattened and included in the same dictionary.
Required. Name of the parser class.
Optional. Dictionary of the parameter name to its value. The spread values are passed through the parser constructor.
Optional. List of Dictionary. Each dictionary must have key
for the validation key and name
,
validator's class name, specified. Other parameters to be passed to the validator constructor can be flattened and specified in this dictionary as well.
Optional. Metadata. Dictionary of any object can be put here. The object will be parsed on to the result.
A table block is for specifying the properties of the table. The table block is surrounded by
{{--table
and --}}
. The simplest table block is
{{--table
key: some_table
--}}
This is typically what you need. A more complicated example is shown below:
{{--table
key: some_table
# optional. default at_comment_cell
locator: {name: at_comment_cell}
# optional. default downward
item_direction: downward
end_conditions:
- {name: all_blank}
- {name: max_row, n: 10, inclusive: true}
--}}
where the table will be put in the result. The output for the table is a list of dictionaries where the key in the dictionaries is the column name specified in column blocks.
Optional. Default at_comment_cell. Locator for the table anchor cell. Same as a cell block.
Optional Direction for items in the table. Default downward. (rightward or downward)
A list of dictionary specifying each termination condition. The dictionary contains name
and flattened
parameters.
If any of the end condition is met then the parsing for the table terminates. Depending on the end condition's class, it may or may not include the matching row.
Column block specifications are very similar to the cell block. The simplest example is shown below.
{{--col
table_key: some_table
key: some_col
parser: int
}}
The column block is very similar to the cell block. It is the cell block with table_key
and without
locator
. The table_key
tells which table it belongs to. It must match one of the table's key
defined in the table block.
The position of the column cell when extracting value is computed from the relative position to the table block in the template.
All other features like fallback, assumptions, validations are also supported in the same fashion as the cell block.
Locate the cell right at the comment cell's coordinate in the template.
Locate the cell to the right of the cell or merged cell with the given label. In the case of a merged cell, right_of will pick the rightmost column and topmost row of the merged cell togo to the right of.
Parameters:
label
label to match.n
Optional. Default value is 1. Indicates the number of columns to move from label cell to located cell.
Searches for non-empty cell right of the cell with the given value.
Parameters:
label
label to match.max_empty_col_search
Indicates the maximum number of columns to search for non-empty cell right of label.
Locate the cell to the right of the cell with a regex match. Similarly, in the case of a merged cell, right_of_regex will pick the rightmost column and topmost row of the merged cell togo to the right of.
Parameters:
regex
string for the regular expression.n
Optional. Default value is 1. Indicates the number of columns to move from regex matched cell to located cell.
Locate the cell below of the cell with the given label. In the case of a merged cell, below_of will pick the bottommost row and leftmost column of the merged cell togo to the bottom of.
Parameters:
label
label to match.n
Optional. Default value is 1. Indicates the number of rows to skip from label cell to located cell.
Searches for non-empty cell below of the cell with the given label.
Parameters:
label
label to match.max_empty_row_search
Indicates the maximum number of rows to search for non-empty cell below of label.
Locate the cell below of the cell with a regex match. Similarly, in the case of a merged cell, below_of_regex will pick the bottommost row and leftmost row of the merged cell togo to the bottom of.
Parameters:
regex
string for the regular expression.n
Optional. Default value is 1. Indicates the number of rows to move from regex matched cell to located cell.
Locate the cell between a cell with the given label. This function searches to the right of or below of the cell with the given label. If it is to the right of, the function searches row by row while staying in between the column ranges of the cell with the given label. If it is below of, similarly, we search row by row while staying in between the row ranges of the cell with the given label. On top of this you have the option on how to pick your data (to the right of or beneath of) once you have found your desired key within the cell range.
Parameters:
label
label of cell you want to search in betweenfind
the name of the cell you want to find in between the cell with that label's rangedirection
direction you want to search for your value, choices are:right_of
this will search the right of the cell between the columns row wisebelow_of
this will search the beneath the cell between the rows row wise
perform
the action you want to perform to fetch your data, your options are:right_of
fetch the value to the right of the cell labeled infind
below_of
fetch the value to beneath of the cell labeled infind
n
pick the value n values away in the specified perform direction (right_of or below_of)
This is useful in batch processing allowing us to check that we have the correct label to the left.
Parameters:
label
string to check that it matches.
parse value as string.
parse the value as an integer.
parse the value as float
parser the value as a date.
valid if the value is between(inclusively) high
and low
high
upper bound value.low
lower bound value
Evaluate to true if all columns are blank.
Evaluate to true if the row number(start from 1) including the parsing row is greater than or
equal to n
.
n
number of rowinclusive
Optional. Default True. Whether to include the row in which it evaluates to true.
Evaluate to true if any column in row contains matching cell_value
. Table terminates the row before.
cell_value
cell value to match.
There are two types of dereferencing
- Spec Creation time dereferencing. The string similar to
<<A1>>
will be resolved to the value at A1 of the template file. - Extraction time dereferencing. This string similar to
==A1==
will be resolved to the value at A1 of the extracted file.
Here is an example
{{--
key: ==D2==
assumptions: # Optional
- {key: right_of, name: right_of, label: <<A1>>}
parser: int
fallback: ==D3==
--}}
When an excel file to be extracted has a different sheet name than the one in the template file,
use sheet_name_checkers
parameter in exco.from_excel()
to create ExcelProcessor
with sheet name alias checker.
For example, when the sheet name in the template file is 'Test 1/1/2021', but sheet names can be varied in the extracting files, i.e., 'Test 2/2/2022, 'Test 3/3/2023', etc.
import exco
SheetName = str
SheetNameAliasChecker = Callable[[SheetName], bool]
test_sheet_checker: SheetNameAliasChecker = lambda sheetname: 'Test' in sheetname
checkers: Dict[SheetName, SheetNameAliasChecker] = {'Test 1/1/2021': test_sheet_checker}
processor = exco.from_excel(template_excel_path, sheet_name_checkers=checkers)
In the case where there are hidden sheets that might have information that you dont want to extract, make sure to set the accept_only_visible_sheets parameter to True
processor = exco.from_excel(template_excel_path, sheet_name_checkers=checkers, accept_only_visible_sheets=True)
See Advance Features Notebook. But, in essence,
processor = exco.from_excel('./custom_locator/custom_locator_template.xlsx',
extra_locators={'diagonal_of': DiagonalOfLocator})
Exco will not read Excel 97-2003 workbook (.xls) files. Use XLS2XLSX to convert .xls files to the supported .xlsx format.