Skip to content

Rules for storing data in NeXus

There are several rules which apply for storing single data items in NeXus. There exists a summary in the NeXus documentation outlining most of these rules. However, to guide data providers even further, we have compiled here additional information and explanations.

Namefitting

In general, the names of NeXus group and field items are validated according to the boundaries outlined in the Rules for Storing Data Items in NeXus Files, section "NXDL group and field names": - Recommended names - lower case words separated by underscores and, if needed, with a trailing number

  • Allowed names
  • any combination of upper and lower case letter, numbers, underscores and periods, except that periods cannot be at the start or end of the string
  • This statement is equivalent to matching this regular expression (named validItemName in the nxdl.xsd) XML Schema file: 
    ^[a-zA-Z0-9_]([a-zA-Z0-9_.]*[a-zA-Z0-9_])?$
  • Invalid names:
  • any name not matching this validItemName regular expression

Note that this explicitly also means that it is not allowed to have whitespace (including " ") in NeXus names.

In NeXus base classes and application definitions, there are two options for defining a concept name. If the group or field in the definition is lowercase, that means that any instance must have the exact same (fixed) name. As an example, if there is a field called my_field in an application definition, the only allowed name in a file would be my_field.

Aside from this lower case notation, there is also the option to allow for selectable names. This is achieved by uppercase notation. As an example, if a field in an application definition is called FIELD, the name can be any name as long as it maches the regular expression above. For example, field, field0, any_other_name would be allowed names, while any other name would not be allowed.

There is also the possibility of mixed lowercase and uppercase notation in base classes and application definitions. For example, there might be a userID(NXuser) group. In this case, allowed names include any name that start with user, e.g., user0, user_abcde, as long as the part that replaces the docstring is still valid according to the regex above. Note that here it is also allowed to write user without replacing the uppercase part of the name.

The validation of names is performed by namefitting, i.e., fitting the name that is used by the data provider to the name given in the base class / application definitions.

A python implementation of this process can be found in this function. This function returns twice the length for an exact match, otherwise the number of matching characters (case insensitive) or zero, if name_any is set to True, is returned. This is also the function that is used in the validation implemented in pynxtools.

Special rules for names and namefitting

Aside from these general rules, there are a number of special rules for NeXus names that need to be considered:

  • There is a set of UPPERCASE reserved words (like BLUESKY_, DECTRIS_, IDF_, etc.) that are reserved for certain projects and communities. These are prefixes (typically written as uppercase + undersorce) that cannot be overwritten by namefitting. For the full list, see Rules for Storing Data Items in NeXus Files, section "Reserved prefixes".

  • There is also a set of reserved suffixes that are used to give additional information for a group or field. For the full list, see Rules for Storing Data Items in NeXus Files, section "Reserved suffixes".

  • Additionally to namefitting, data annotation can use further information. For example, in case of NXdata, the axes listed among the @axes shall fit to any instances of AXISNAME and data objects listed in @signal or @auxiliary_signals shall fit to instances of DATA. Such rules are typically given in the base classes (e.g., see here for NXdata). Any tool that makes use of the base classes should implement these special rules in its validation procedure. As an example, pynxtools has a special function for handling NXdata.