The DataModelDict class is used for handling data models that have equivalent representations in XML, JSON, and Python. Constructing data models in this way is convenient as it supports compatibility across different software tools, such as different types of databases.
The DataModelDict class:
- is a child of OrderedDict,
- has methods for converting to/from XML and JSON,
- has methods for searching through elements, and
- has methods that help with constructing and interacting with compliant data models.
The code has no requirements that limit which systems it can be used on, i.e. it should work on Linux, Mac and Windows computers.
The latest release can be installed using pip:
pip install DataModelDict
The code and all documentation is hosted on GitHub and can be directly downloaded at https://github.com/usnistgov/DataModelDict.
Some considerations need to be taken into account for designing data models that allow for exact reversible transformations between the three formats:
Valid, full XML requires that there is exactly one root element. In other words, the top-level DataModelDict of a data model can have only one key.
Do not use lists of lists for representing data. The XML conversions are only reversible for lists of values or lists of dictionaries. Future updates may allow this.
Avoid using XML attributes if possible. While the XML conversions do reversibly handle attributes, it complicates the Python and JSON representations.
Embedded XML content, i.e. "text with <embed>embedded</embed> content", might not be reversible:
- If this is in a Python/JSON value, converting to XML gives "text with &lt;embed&gt;embedded&lt;/embed&gt; content". This is reversible.
- If this is an XML text field, parsing to Python pulls the embedded elements out of the text, which is not reversible!
XML subelements of the same name within an element should be given consecutively. When parsed, all values of subelements of the same name are collected together in a list. This will alter the original order of subelements if matching names were not originally consecutive.
The Python-JSON conversions use the standard Python JSON library. In converting from Python to JSON, all elements of the DataModelDict must be an instance of a supported data type.
Python | JSON |
---|---|
dict | object |
list, tuple | array |
str | string |
int, float | number |
True | true |
False | false |
None | null |
np.nan | NaN |
np.inf | Infinity |
-np.inf | -Infinity |
As DataModelDict is a child of OrderedDict, it registers as being an instance of dict. Any other objects would first need to be converted to one of these types, e.g. a numpy array would need to be converted to a list.
The Python-XML conversions use the xmltodict Python package. The XML content is constructed based on the Python data types.
Python | XML |
---|---|
dict | subelement |
list, tuple | repeated element |
str | text |
int, float | repr(val) |
True | 'true' |
False | 'false' |
None | '' |
np.nan | 'NaN' |
np.inf | 'Infinity' |
-np.inf | '-Infinity' |
Some characters in the XML text fields will also be converted to avoid conflicts.
- XML limited characters such as <, > and & are converted to their HTML entities.
- \n, \t, \r are converted to \\n, \\t, and \\r
Any dictionary keys starting with '@' will be converted into XML attributes, and the dictionary key '#text' is interpreted as the text value of the element.
The Python-JSON conversions use the standard Python JSON library. In converting from JSON to Python, the conversions of types is straight-forward.
JSON | Python |
---|---|
object | DataModelDict |
array | list |
string | str |
number (int) | int |
number (real) | float |
true | True |
false | False |
null | None |
NaN | np.nan |
Infinity | np.inf |
-Infinity | -np.inf |
The Python-XML conversions use the xmltodict Python package. The text fields will be interpreted based on the following sequential tests:
XML text | Python |
---|---|
text == 'True' or 'true' | True |
text == 'False' or 'false' | False |
text == '' | None |
text == 'NaN' | np.nan |
text == 'Infinity' | np.inf |
text == '-Infinity' | -np.inf |
try int(text) and text == str(int(text)) | int |
try float(text) | float |
otherwise | str |
The int conversion test was updated for version 0.9.8 to check that the values can reversably be changed back into a str. This is necessary to properly handle values, such as journal page numbers, that may contain leading zeroes.
The reverse conversions are done for the special characters mentioned in the Conversion from Python to XML section above.
Any 'attr' attribute fields are converted to elements named '@attr' and corresponding '#text' elements are created if needed.
The DataModelDict class is used for handling data models that have equivalent representations in XML, JSON, and Python. Constructing data models in this way is convenient as it supports compatibility across different software tools, such as different types of databases.
The DataModelDict class:
- is a child of OrderedDict,
- has methods for converting to/from XML and JSON,
- has methods for searching through elements, and
- has methods that help with constructing and interacting with compliant data models.
The code has no requirements that limit which systems it can be used on, i.e. it should work on Linux, Mac and Windows computers.
The latest release can be installed using pip:
pip install DataModelDict
The code and all documentation is hosted on GitHub and can be directly downloaded at https://github.com/usnistgov/DataModelDict.
Some considerations need to be taken into account for designing data models that allow for exact reversible transformations between the three formats:
Valid, full XML requires that there is exactly one root element. In other words, the top-level DataModelDict of a data model can have only one key.
Do not use lists of lists for representing data. The XML conversions are only reversible for lists of values or lists of dictionaries. Future updates may allow this.
Avoid using XML attributes if possible. While the XML conversions do reversibly handle attributes, it complicates the Python and JSON representations.
Embedded XML content, i.e. "text with <embed>embedded</embed> content", might not be reversible:
- If this is in a Python/JSON value, converting to XML gives "text with &lt;embed&gt;embedded&lt;/embed&gt; content". This is reversible.
- If this is an XML text field, parsing to Python pulls the embedded elements out of the text, which is not reversible!
XML subelements of the same name within an element should be given consecutively. When parsed, all values of subelements of the same name are collected together in a list. This will alter the original order of subelements if matching names were not originally consecutive.
The Python-JSON conversions use the standard Python JSON library. In converting from Python to JSON, all elements of the DataModelDict must be an instance of a supported data type.
Python | JSON |
---|---|
dict | object |
list, tuple | array |
str | string |
int, float | number |
True | true |
False | false |
None | null |
np.nan | NaN |
np.inf | Infinity |
-np.inf | -Infinity |
As DataModelDict is a child of OrderedDict, it registers as being an instance of dict. Any other objects would first need to be converted to one of these types, e.g. a numpy array would need to be converted to a list.
The Python-XML conversions use the xmltodict Python package. The XML content is constructed based on the Python data types.
Python | XML |
---|---|
dict | subelement |
list, tuple | repeated element |
str | text |
int, float | repr(val) |
True | 'true' |
False | 'false' |
None | '' |
np.nan | 'NaN' |
np.inf | 'Infinity' |
-np.inf | '-Infinity' |
Some characters in the XML text fields will also be converted to avoid conflicts.
- XML limited characters such as <, > and & are converted to their HTML entities.
- \n, \t, \r are converted to \\n, \\t, and \\r
Any dictionary keys starting with '@' will be converted into XML attributes, and the dictionary key '#text' is interpreted as the text value of the element.
The Python-JSON conversions use the standard Python JSON library. In converting from JSON to Python, the conversions of types is straight-forward.
JSON | Python |
---|---|
object | DataModelDict |
array | list |
string | str |
number (int) | int |
number (real) | float |
true | True |
false | False |
null | None |
NaN | np.nan |
Infinity | np.inf |
-Infinity | -np.inf |
The Python-XML conversions use the xmltodict Python package. The text fields will be interpreted based on the following sequential tests:
XML text | Python |
---|---|
text == 'True' or 'true' | True |
text == 'False' or 'false' | False |
text == '' | None |
text == 'NaN' | np.nan |
text == 'Infinity' | np.inf |
text == '-Infinity' | -np.inf |
try int(text) and text == str(int(text)) | int |
try float(text) | float |
otherwise | str |
The int conversion test was updated for version 0.9.8 to check that the values can reversably be changed back into a str. This is necessary to properly handle values, such as journal page numbers, that may contain leading zeroes.
The reverse conversions are done for the special characters mentioned in the Conversion from Python to XML section above.
Any 'attr' attribute fields are converted to elements named '@attr' and corresponding '#text' elements are created if needed.
class DataModelDict.DataModelDict(*args, kwargs)
Bases:
collections.OrderedDict
Class for handling json/xml equivalent data structures.
append(key, value)
Adds a value for element key by either adding key to the dictionary or appending the value as a list to any current value.
Parameters:
- key (str) – The dictionary key.
- value – The value to add to the dictionary key. If
- key exists, the element is converted to a list if needed and value is appended.
aslist(key)
Gets the value of a dictionary key as a list. Useful for elements whose values may or may not be lists.
Parameters: key (str) – Dictionary key Returns: The dictionary’s element value or [value] depending on if it already is a list. Return type: list find(key, yes={}, no={})
Return the value of a subelement at any level uniquely identified by the specified conditions.
Parameters:
- key (str) – Dictionary key to search for.
- yes (dict) – Key-value terms which the subelement
- must have to be considered a match.
- no (dict) – Key-value terms which the subelement
- must not have to be considered a match.
Returns: The value of the uniquely identified subelement.
Return type: any
Raises: ValueError – If exactly one matching subelement is not identified.
finds(key, yes={}, no={})
Finds the values of all subelements at any level identified by the specified conditions.
Parameters:
- key (str) – Dictionary key to search for.
- yes (dict) – Key-value terms which the subelement
- must have to be considered a match.
- no (dict) – Key-value terms which the subelement
- must not have to be considered a match.
Returns: The values of any matching subelements.
Return type: list
iteraslist(key)
Iterates through the values of a dictionary key. Useful for elements whose values may or may not be lists.
Parameters: key (str) – Dictionary key Yields: any – The dictionary’s value or each element in value if value is a list. iterfinds(key, yes={}, no={})
Iterates over the values of all subelements at any level identified by the specified conditions.
Parameters:
- key (str) – Dictionary key to search for.
- yes (dict) – Key-value terms which the subelement
- must have to be considered a match.
- no (dict) – Key-value terms which the subelement
- must not have to be considered a match.
Yields: any – The values of any matching subelements.
iterpaths(key, yes={}, no={})
Iterates over the path lists to all elements at any level identified by the specified conditions.
Parameters:
- key (str) – Dictionary key to search for.
- yes (dict) – Key-value terms which the subelement
- must have to be considered a match.
- no (dict) – Key-value terms which the subelement
- must not have to be considered a match.
Yields: list of str – The path lists to any matching subelements.
itervaluepaths()
Iterates over path lists to all value elements at any level.
Yields: list – The path lists to all value subelements. json(fp=None, *args, kwargs)
Converts the DataModelDict to JSON content.
Parameters:
- fp (file-like object or None, optional) – An
- open file to write the content to. If None (default), then the content is returned as a str.
- *args (any) – Any other positional arguments
- accepted by json.dump(s)
- **kwargs (any) – Any other keyword arguments
- accepted by json.dump(s)
Returns: The JSON content (only returned if fp is None).
Return type: str, optional
load(model, format=None)
Read in values from a json/xml string or file-like object.
Parameters:
- model (str or file-like object) – The XML or
- JSON content to read. This is allowed to be either a file path, a string representation, or an open file-like object in byte mode.
- format (str or None, optional) – Allows for
- the format of the content to be explicitly stated (‘xml’ or ‘json’). If None (default), will try to determine which format based on if the first character of model is ‘<’ or ‘{‘.
Raises: ValueError – If format is None and unable to identify XML/JON content, or if format is not equal to ‘xml’ or ‘json’.
path(key, yes={}, no={})
Return the path list of a subelement at any level uniquely identified by the specified conditions. Issues an error if either no match, or multiple matches are found.
Parameters:
- key (str) – Dictionary key to search for.
- yes (dict) – Key-value terms which the subelement
- must have to be considered a match.
- no (dict) – Key-value terms which the subelement
- must not have to be considered a match.
Returns: The subelement path list to the uniquely identified subelement.
Return type: list of str
Raises: ValueError – If exactly one matching subelement is not identified.
paths(key, yes={}, no={})
Return a list of all path lists of all elements at any level identified by the specified conditions.
Parameters:
- key (str) – Dictionary key to search for.
- yes (dict) – Key-value terms which the subelement
- must have to be considered a match.
- no (dict) – Key-value terms which the subelement
- must not have to be considered a match.
Returns: The path lists for any matching subelements.
Return type: list
xml(fp=None, indent=None, kwargs)
Return the DataModelDict as XML content.
Parameters:
- fp (file-like object or None, optional) – An
- open file to write the content to. If None (default), then the content is returned as a str.
- indent (int, str or None, optional) – If
- int, number of spaces to indent lines. If str, will use that as the indentation. If None (default), the content will be inline.
- **kwargs (any) – Other keywords supported by
- xmltodict.unparse, except for output which is replaced by fp, and preprocessor, which is controlled.
Returns: The XML content (only returned if fp is None).
Return type: str, optional
DataModelDict.joinpath(path: list, delimiter: str = '.', openbracket: str = '[', closebracket: str = ']') -> str
Takes a path as a list and transforms it into a string.
Parameters:
- path (list) – The path list to join.
- delimiter (str) – The delimiter between subsequent
- element names.
- openbracket (str) – The opening indicator of list
- indices.
- closebracket (str) – The closing indicator of list
- indices.
Return type: The path as a delimited string.
DataModelDict.parsepath(pathstr: str, delimiter: str = '.', openbracket: str = '[', closebracket: str = ']') -> list
Takes a path as a string and parses it into a list of terms.
Parameters:
- pathstr (str) – The path string to parse.
- delimiter (str) – The delimiter between subsequent
- element names.
- openbracket (str) – The opening indicator of list
- indices.
- closebracket (str) – The closing indicator of list
- indices.
Returns: The path as a list.
Return type: list
DataModelDict.uber_open_rmode(data: Union[str, bytes, pathlib.Path, io.IOBase]) -> io.IOBase
Provides a uniform means of reading data from files, file-like objects, and string/bytes content.
Parameters: data (file-like object, file path, or str/bytes file content) – The data that will be opened for reading.
Returns: An open file-like object that is in a bytes read mode. If a file-like object is given, it is passed through after checking that it is for bytes content. If a file path is given, the file is opened in ‘rb’ mode. If bytes or string content is given, the content is returned in a BytesIO object.
Return type: file-like object
Raises:
- ValueError – If a file-like object in text mode is given.
- TypeError – If data is not a file-like object, bytes, str
- or Path.
- FileNotFoundError – If data is a pathlib.Path object and
- is not an existing file.