| equal | .eq. | = | not equal | .ne. | != | |
| less than | .lt. | < | greater than | .gt. | > | |
| less than or equal | .le. | <= | greater than or equal | .ge. | >= | |
| and | .and. | and | or | .or. | or | |
| negation | .not. | not | wildcard match | like |
The wildcard match operator, like, has no corresponding Fortran
form. It allows you to do pattern matching using the wildcard
characters * and ?. The * character matches any number of
characters and the ? character matches a single character. The like
operator is asymmetric, the argument on the left must be the column
name and the argument on the right must be a quoted string containing
wildcard characters. Please note that you cannot use wildcard
characters with the equality operator, .eq., they will be interpreted
as ordinary characters.
The columns parameter contains a list of the data base table
columns that are to be retrieved. The items in the list may contain
the usual pattern matching metacharacters (i.e., the wildcard
characters * and ?). If the list is blank (the default
value), all columns will be retrieved. If the first character in the
list is the negation character (~), then only those columns that
are not in the list will be retrieved from the data base.
More than one data base table can be queried simultaneously by passing
more than one name in the first parameter. When more than one data
base table is being queried, all combinations of rows in the tables
are tested and those combinations for which the qualifier is true will
be returned. Column names in the qualifier must be preceded by the
table alias followed by a period. The table alias is a single
character whose value depends on the position of the table name in
list. For example, the second database table in the list of table
names would have the alias b. If the qualifier contained an
expression involving the column version from the second table, it
would be referred to as b.version. Column names referred to in
the second parameter may optionally be preceded by the table alias
followed by a period. If a column name is not preceded by a table
alias, then that column will be retrieved from all tables in which it
is present, or excluded if the negation character precedes the list of
column names. If the second parameter is blank, then every column
from every table will be retrieved.
* and ?), in which case all column names that match the pattern
will be retrieved. If the column name list is preceded by a @, it
is taken to be a file containing the column names. Columns in the file
may be separated by commas or whitespace.
qual is the file name preceded by a @
character.
cdbs_ops.
yes, duplicate rows in the
output will be deleted.
yes, the output table will have a
title consisting of the table column names.
yes, the output will be printed in
upper case. If this parameter is no, output will be printed in
lower case.
yes (the default for command line
mode), the output table will be an ascii file. If it is no (the
default for cgi script mode), the output table will be an html file.
qrysearch.html and
qryresult.html. These contain the html templates to build the
search and result screens, respectively.
The text in the html templates is copied into getref's output except
where variables appear in the template. These variables are replaced
by text generated by getref. Variables in the template file start with
the character string <!-- $ and end with the character string
-->. In between the starting and ending string is the variable
name. The variable thus looks like an html comment to an html browser,
so the template can be checked for style and content without running
the task.
The task supports four variables, output, toplink, txtlink, and
taskurl. The variable names and meanings are the same as those
used by getref, except that query does not support the sumlink
variable. For further information on the meaning of these variables,
read the getref documentation.
task, whcih reads any task
arguments into the proper parameters, calls form_input, which
reads any cgi input, calls query, which takes the task parameters
and uses them to perform a database query, and finally calls either
qryout_html or qryout_text to produce the task output.
The design of form_input is discussed in the documentation of
getref.
The query function fills in a sql query template, executes the
query, and saves the results in a load table. The major portion of the
work is in assempling the strings to be inserted in the template. The
first portion of the work is getting the list of column names, which
is done by get_collist. Getting the list of column names is
non-trivial because the user may use wild cards or the negation
symbol. The column names returned by get_collist are qualified by
the table alias and strip_collist removes them to form unqualified
names. The where clause is converted from Fortran syntax to sql by
ftn_to_sql. After all these pieces of the sql query are built,
they are assembled into the query by build_qry.
The function get_collist loops over all tables that are being
searched. The function read_dbcols returns all the database
columns from a table and match_dbcols finds thos that match the
list of column names the user specified. The function read_dbcols
does a query against the system tables sysobjects and
syscolumns to retrieve all the columns in the table. Columns that
are not convertable to strings (most notably text columns) are not
retrieved. The function match_dbcols loops over all column names
(which may contain wild cards) in the user's column list and tries to
match the pattern against the column names returned by get_collist
by using the pattern matching function patmatch. The result is
stored in an array of flags corresponding to the table columns. If the
negate flag is set, the array is inverted. Finally, the aolumn names
are copied from the list of table columns to the output list when the
corresponding flag is true.
The function ftn_to_sql converts a query qualifier using Fortran
syntax to an equivalent qualifier using sql syntax. The function reads
a token at a time from the qualifier by calling qrylex_next. The
type of the token is returned as a function value. When no more tokens
remain, the function returns. Quoted strings are converted to upper
case and if the string follows a like operator, wild card
characters are converted to their sql equivalents. Numeric constants
are copied as is. Identifiers (either sql keywords or table column
names) are prefixed by the table alias a if they are unqualified
table column names. Symbols and operators are converted from their
Fortran to sql form. The converted (or unconverted) tokens are copied
to the output buffer, each token separated from the previous by a blank.
The function qryout_html uses an html template and a load table to
create an html file. It calls html_template, which prints the
content of the template until it encounters a variable in the
template. At that point, it returns the name of the
variable. Qryout_html prints the text corresponding to the
variable name. It recognizes four variables: output, which
prints the load table as an html table, toplink prints out a link
to the search form, taskurl prints out the bare url of the search
form, and txtlink prints a link that performs the current query,
but outputs the results as an ascii file. All these variables are
fairly trivial, except for txtlink. It calls the function
each_option to retrieve each of the parameters passed to the
task. If the parameter is not in a list of omitted parameters, it is
encoded by the function cgi_encode and appended to the url.
The function (qryout_text) produces ascii output from a load table. The chief complications are determining the width of each column in the output and where rows should be folded so that the width of the output does not exceed 80 characters. The function computes the length of the longest string in each column to determine the column width. It the sums the column widths and when the sum exceeds the maximum value, marks the previous column as the last in the line by setting its width to a negative number. Then when printing the table, the function checks to see if a column has a negative width and if so, follows it with a backslash and a newline.
datafile.h. The filenames use the DATAFILE macro, defined in
system.h to specify the directory name containing the template
files. The url used to invoke the search screen is defined by the
QUERYURL macro defined in datafile.h. The query.cgi shell
script in the util directory, used to run query from a cgi
script script, contains system dependent environment variables and
must be edited and copied to the cgi-bin directory on the system.
This task uses four environment variables to control its access to the
database. CDBS_SERVER sets the database server the task
accesses. CDBS_DB sets the name of the cdbs database that is used
by the task. DADS_DB sets the name of the dads database that is
used by the task. And CDBS_USER sets the user name used to access
the database. If these environment variables are not set, they default
to ZEPPO, cdbs_ops, dads_ops, and a public cdbs account.
In addition to these environment variables that control cdbs access,
the stdb library that this task is linked with requires the two
environment variables SYBASE and ACAREA be set. These
variables define the location of the Sybase and stdb executables and
libraries.
To perform the same query, but also retrieve the pedigree from the row level table, run the following command: