fhirpath-engine-notes.md

Misc fhirpath-lab notes

Random notes on the various fhirpath engines that are supported by the fhirpath-lab.

Expression Positioning Information

The position information is used in several places, and different engines read it differently. Conversion between the formats is possible, but this is a great place to document what each one is doing to ease debugging if issues arise.

HAPI/Java - fhirpath Parser/Lexer

• The start positioning information is captured in line/column format, and 1 based indexes.
• The end positioning information is captured in line/column format, and 1 based indexes, and refers to the character after the last one in the token (can be at end of line marker).
• to evaluate the length of a node, you would need to subtract the start position from the end position (and consider line breaks).

e.g. position information for the expression showing the boundary characters (note that the single char tokens have the same start and end position)

'start'    . substring(0 , 5)
^------^   |||        |||  || // 'start'
           ^^|        |||  || // .
             ^--------^||  || // substring
                       ^^  || // 0
                           ^^ // 5

• Operators have their position information in the opStart/opEnd properties, and refer to the operator token itself, not the bounds of the operands too.
• delimited identifiers positioning information is similar to a string, as shown above.

Firely .NET parser

The Firely parser uses the sprache Parser combinator library, which captures the position information in a Position object.

• The Position object has a start Line and Column property, both 1 based.
• It also computes the RawPosition, which is a 0 based index into the original string.
• And also has a Length property, which is the length of the token in characters.
• It does not have a stop position information, but the Length can be used to compute the end position as row/col (and would need to consider any newlines in the token).