Data Exchange Mappings¶
Each Data Exchange mapping is an XML file describing the structure of the JSON, XML or CSV format being mapped. Below you find the elements specific for each of the mapping types. The attributes that you can specify for each element are shared
JSON Mapping elements¶
The following are the elements allowed in a JSON mapping
AimmsJSONMappingelement, the mandatory root of a JSON mapping
ObjectMappingelement, a mapping element used to map a JSON object value (enclosed in curly brackets
ArrayMappingelement, a mapping element used to map a JSON array value (enclosed in square brackets
]). A JSON array mapping can only have a single child mapping, specifying the type of every element in the array.
ValueMappingelement, a mapping element used to map a integer, double or string value in a JSON file
RowMappingelement (underneath an
ObjectMapping), a mapping element used to map all underlying mapping nodes as an array of row arrays (containing heterogeneous value types)
RowMappingelement (underneath a
ColumnMapping), a mapping element used to map the value of a particular element in a
ColumnMappingto a specific identifier in the model.
ColumnMappingelement (underneath an
ObjectMapping), a mapping element used to map all underlying mapping nodes as an array of column arrays (containing homogeneous value types)
ColumnMappingelement (underneath a
RowMapping), a mapping element used to map the value of a particular element in a
RowMappingto a specific identifier in the model.
The represent row-oriented data, the
ColumnMapping will provide the most compact JSON representations and will execute the fastest.
XML Mapping elements¶
The following are the elements allowed in a XML mapping
AimmsXMLMappingelement, the mandatory root of a XML mapping
ElementObjectMappingelement, a mapping element used to map an XML element that holds child elements, but no value
ElementValueMappingelement, a mapping element used to map an XML element that holds a value, but no child elements
AttributeMappingelement, a mapping element used to map the value of an attribute of an XML element
CSV Mapping elements¶
The following are the elements allowed in a CSV mapping
AimmsCSVMappingelement, the mandatory root of a CSV mapping. It should contain a single
RowMappingelement, a mapping element used to map rows of a CSV table
ColumnMappingelement, a mapping element used to map the value of a column in a CSV table
Excel Mapping elements¶
The following are the elements allowed in a CSV mapping
AimmsExcelMappingelement, the mandatory root of an Excel mapping. It can contain multiple
ExcelSheetMappingelement, a mapping element used to map an Excel sheet
RowMappingelement, a mapping element used to map a row in an Excel sheet
ColumnMappingelement, a mapping element used to map the value of a column in an Excel sheet
The attributes of the elements in a Data Exchange mapping are shared among the different types of mappings, although not all attributes are supported by every type of mapping element.
The available mapping attributes are:
The name attribute¶
name attribute specifies the name of the mapped element in a JSON, XML, CSV or Excel format. Not every element needs a name, for instance to root value in a JSON file, or the child mapping of a JSON array.
The binds-to attribute¶
binds-to attribute, which can be added to the mapping of any value-holding element. The
binds-to attribute will also provide an index binding for all sibling mapping elements of mapping element for which it is specified, or for the parent element in case the
binds-to attribute is applied to an
The name-binds-to attribute¶
name-binds-to attribute provides a way of binding the name of an element in a JSON or XML file to an index in your AIMMS model. You would typically use this if a JSON or XML file holds elements with different names but with the same structure. Rather than creating a mapping for each of the elements you can create a mapping where the element names serves as an extra index in the binding of the multi-dimensional identifiers mapped to the values contained in each of the elements.
name-regex attribute should be used in conjuction with a
name-binds-to attribute, to specify a regular expression to restrict the element to which the
name-binds-to attribute should be applied.
name-regex-prefix attribute you can specify a prefix that is used in the JSON, XML, CSV or Excel file, but which should not be included in the element names in the model. Note that the value of the
name-regex-prefix attribute is automatically prepended to the regular expression specified in the
name-regex attribute, and subsequently removed from the match if a match has been found.
The iterative-binds-to attribute¶
iterative-binds-to attribute can be used if the given JSON or XML format does not hold an explicit value which can be bound to an index in your model. The
iterative-binds-to attribute will generate elements using an increasing integer counter.
iterative-prefix attribute can be used alongside the
iterative-binds-to attribute. All elements created in the model will be prefixed with the prefix specified here. If you don’t specify a prefix, the element names will be just increasing integer values.
Assigning a value of 1 to the the
iterative-existing attribute causes the
iterative-binds-to attribute to not generate new elements, but instead to use existing elements of the set bound to the index specified in the
iterative-binds-to attribute, starting at the element with ordinal 1. If a generated element is not present, the reading will stop with an error.
iterative-reset attribute can be specified at the parent element of a mapping element which holds a
iterative-binds-to attribute. It will cause the integer counter of all direct child mappings to be reset to 1.
The implicit-binds-to attribute¶
By default, if a node in a mapping has sibling nodes, any index bound via a
binds-to attribute at such a node n can be used in any attribute of all nodes in the subtree starting at the parent node of n. Via the
implicit-binds-to attribute you can make such an index available for use in subtrees starting at even higher parent nodes. You can use this, for instance, if an id of a JSON/XML data structure, that you intend to use as the index value for all data in such a data structure, is stored deeper in such a data structure. By means of the
implicit-binds-to attribute you can make sure that the Data Exchange library will first read the entire subtree containing the index value, prior to reading the subtrees where this index is referenced in e.g. a
The binds-existing attribute¶
binds-existing attribute can be used in conjunction with the
iterative-binds-to attribute to indicate that, when reading a data file, now new set elements will be created based on node values or names. If a newly read or generated element is not present in the set, any data value underneath the node to which the element is bound will be skipped (cf. to
iterative-reset which will return with an error in such a case). This allows for a filtering mechanism where a data file can only be partially read for all nodes that correspond to existing set elements in the model.
External bindings in mappings¶
Directly underneath the root node of any mapping you can specify one or more
ExternalBinding nodes. An external mapping node has two attributes:
binds-to attribute you can specify the index which should be bound externally to the scalar element parameter specified through the
As a result of an
ExternalBinding, any externally bound index cannot be bound any longer within the document, and any use of an externally bound index in multi-dimensional identifiers used in e.g. a
maps-to attribute will refer to the slice of that identifier associated with the element parameter specified through the
You can use an
ExternalBinding node to read or write a document only for the slice associated with the specified element parameter. Alternatively, you can use it to bind it in an
included-mapping to the current value of an index bound in an outer mapping at the node containing the
The maps-to attribute¶
You can assign the
maps-to attribute to any value-holding mapping element. Its value should be a reference to an identifier in your model, including the indices bound at this location in the mapping tree. Note that this should match the dimension of the identifier exactly, and that the root domain of the identifier should match the root domains of the indices.
write-filter attribute can be specified at any node in the mapping tree, and should be a reference to an identifier in the model including the bound indices at this location as for the
maps-to attribute. For any tuple of bound indices for which the
write-filter attribute does not hold a non-default value, the corresponding part of the generate JSON, XML or CSV file will be skipped.
By default, the Data Exchange library assumes that all string values will hold up to 1024 characters. Through the
max-string-size attribute a maximum string size up to 8KB can be specified.
The range-existing attribute¶
If the identifier associated with a
maps-to attribute is an element parameter, the
range-existing attribute can be used to that any values encountered that do not correspond to an existing element in the range set, should be skipped, rather than creating a new element in the range set for such a value.
The force-dense attribute¶
force-denseattribute should also contain a reference to an identifier plus bound indices as for the
this attribute you can force a specific density pattern by specifying a domain for which nodes should be generated, regardless of whether non-default data is present to fill such nodes, e.g. because the identifier specified in the
maps-toattribute of the node itself, or any of its sub-nodes, holds no non-default data. Note that a density pattern enforced through the
force-denseattribute is still subject to a write filter specified in a
Enforcing a density pattern may be important when the bound indices are generated through the
iterative-binds-to attribute, and not explicitly represented through data-holding node bound to a regular
binds-to attribute. In such cases, not writing nodes that hold no non-default data, may lead to inconsistent numbering of generated elements when reading the generated JSON or XML files back in. When reading a JSON, XML, CSV or Excel file, the library will assign a value of 1 for the identifier specified in the
force-dense attribute to any tuple encountered, such that the same file will be generated when writing back the file using the same mapping based on the data just read in.
None of the
force-dense attributes may contain an identifier slice, but must be bound to indices in the mapping for all dimensions of the given identifier. Thus, for instance, specifying a value of 1 to the
force-dense attribute to enforce full density is not allowed. Instead you should create a full-dimensional parameter holding 1 for every tuple in its domain and assign that to the
To enforce slicing for a particular index, you can specify an
ExternalBinding node directly underneath the root node of the mapping.
The dense-children attribute¶
dense-children you can indicate that when a node will be written, because of the density pattern of all of its children, all direct value-holding child elements with the same bound indices as the parent node, will be written in a dense manner. For example, with this attribute you can cause all columns in a table row to be written to a CSV or Excel file, whenever at least one of the columns holds a non-default value.
With this attribute you cannot cause an array to be written in a dense manner, as the array elements need to bind an additional index. To enforce writing an array in a dense manner, you have to use the
The value attribute¶
value attribute you can specify that, when writing a file, the value of a value-holding mapping element should become the static string value specified through this attribute. When reading a file, a node with a
value attribute will be silently ignored.
Any value-holding mapping element may have only one of the
value attributes specified.
The included-mapping attribute¶
included-mapping attribute, you can indicate that the contents of an object or array element in a given JSON or XML file should be read/written using a mapping, the name of which is contained in the string parameter specified in this attribute. The dimension of the string parameter should match the indices already bound at the given node. With this attribute you can specify a data-driven mapping name for a certain sub-tree of a JSON or XML file, e.g., to specify a table-specific mapping, where the table name is already bound in a parent node of the node at hand.
Alternatively, if the string value of the
included-mapping attribute starts with the
@ character, the remainder of the value will be interpreted as the fixed name of a mapping to be applied for the node at hand, instead of as a string parameter holding mapping names.
Note that when reading the contents of the node associated with the included mapping you cannot refer to the indices already bound at that node in the containing mapping, i.e., the contents of the tree node should be able to be read/written as if read from/written to a completely separate JSON/XML file.
It is possible, however, to externally bind the values of bound indices to indices used in the included mapping by specifying an
ExternalBinding node underneath the node containing the
included-mapping attribute. To this end, the included mapping itself should have an possess an
ExternalBinding for the index you want to bind to. In addition, you should specify an
ExternalBinding node underneath the node with
included-mapping attribute, with the
binds-to attribute set to the externally bound index in the included mapping, and the
binding attribute set to the bound index in the outer mapping you want to bind to.
You can use external bindings in combination with included mappings to break a longer mapping into its constituing components. Note, however, that breaking up mappings this way will carry a performance penalty, especially if there is a lot of repetition in the nodes using an included mapping.
The embedded-mapping attribute¶
embedded-mapping attribute, you can indicate that a value-holding element in the given JSON or XML file should hold a string that can be read or written using the mapping specified in this attribute. Note that the mapping element to which this attribute is attached may not have bound indices. The mapping specified in this attribute may be of any type (e.g. XML, JSON, CSV or Excel) and will be represented as a single (base64 encoded) string.
Assigning a value of 1 to the
base64-encoded attribute indicates whether embedded mapped string is or should be base64 encoded.
How does the mapping work for reading and writing?¶
In this section we will explain how the Data Exchange library uses the mapping to read or write a given format.
When reading a JSON, XML, CSV or Excel file using a specified mapping, the Data Exchange library will iterate over the entire tree.
If reading a particular node in the data file, it will first try to bind any indices specified
at the node itself through the
at direct child nodes through the
at deeper child nodes that make their indices available through
All elements assiocated with indices bound this way will be maintained in a stack of bound indices.
Subsequently the Data Exchange library will examine all other child nodes. If such a node is a structural or iterative node, it will recursively try to read the data associated with the child node. If the examined node is a value-holding node mapped to an multi-dimensional identifier, the value will be assigned to that identifier. Finally, if the node itself is a value-holding node mapped onto an identifier, it will also assign this value.
If a node in the mapping contains an included mapping, all externally bound indices bound to the values of bound indices in the outer mapping, will be carried over to the included mapping, prior to reading the subtree associated with the included mapping.
When generating a JSON, XML, CSV or Excel file for a given mapping, at any given node, the Data Exchange library will examine all multi-dimensional identifiers associated with the node or any of its sub-nodes through either the
force-dense attributes, and will try to find the lowest subtuple associated with all these identifiers, for all indices bound at this level (through the
implicit-binds-to attributes) while fixing the indices already found at a previous level. If such a subtuple can be found, the new indices at this level will be stored, and any mappped value-holding nodes at this level will be written the associated values of any multi-dimensional identifiers matching with the value of the currently bound indices, and the Data Exchange library will iterate over all any structural or iterative child nodes recursively. If no further multi-dimensional data can be found for a particular node, the Data Exchange library will track back to the parent node, and try to progress there.
The message here is that an JSON, XML, CSV or Excel sheet tree is generated solely on the basis of multi-dimensional identifiers in the mapping, and never on the basis of any of the
binds-to attributes. Such nodes will be generated based on indices bound by iterating over multi-dimensional data.
Thus, for instance, to generate a JSON array containing only all element names of a set in your model, you must combine a
binds-to attribute, together with a
force-dense attribute consisting an identifier over the index you want to generate the elements for, holding a value of 1 for every element you want to be contained in the array.
If a node in the mapping contains an included mapping, all externally bound indices bound to the values of bound indices in the outer mapping, will be carried over to the included mapping, resulting in the Data Exchange library to use the identifier slices corresponding to the externally bound indices to generate the node contents.