The program examples/query_length.p finds LENGTH() functions within the current node which do not provide the second "type" parameter: "CHARACTER," "RAW," or "COLUMN". A query like this might be used as part of your effort to ensure that your application is unicode ready.
Query_length.p is not called directly. It is meant to be called by
a program like proparse/api/testrun.p,
which takes care of the details of
reading through a list of programs to be parsed, parsing each of those, and
then handing off the syntax tree
to query_length.p.
Because the calling program, testrun.p,
has already instantiated the persistent parser
interface, proparse.p,
we don't need to do it again here. We just need the handle to
it, which we call parser.
To access the API, all we need to do is include
proparse/api/proparse.i,
passing it the handle to the parser.
This program uses Proparse's named queries. It creates a query for LENGTH nodes, and then loops through each of the nodes in the query result set. For each node in the result set, it checks to see if the LENGTH function is given the second, optional "type" parameter. If not, then it reports on it.
In practice, you would want to change query_length.p so that it
reports each include file only once. ie: Create a temp-table
storing each filename/line-number combination only once, and then
report on that.