NAME¶
Term::Query - Table-driven query routine.
SYNOPSIS¶
- "use Term::Query"
- "qw( query query_table query_table_set_defaults
query_table_process );"
"$result =
query $
prompt, $
flags, [
$
optional_args ];"
"$
ok =
query_table \@
array;"
"
query_table_set_defaults \@
array;"
"$
ok =
query_table_process \@
array, \&flagsub,
\&querysub;"
DESCRIPTION¶
query¶
The
query subroutine fulfills the need for a generalized
question-response subroutine, with programmatic defaulting, validation,
condition and error checking.
Given
$prompt and
$flags, and possibly
additional arguments, depending upon the characters in
$flags,
query issues a prompt to STDOUT and
solicits input from STDIN. The input is validated against a set of test
criteria as configured by the characters in
$flags; if
any of the tests fail, an error message is noted, and the query is
reattempted.
When STDIN is not a tty (not interactive), prompts are not issued, and errors
cause a return rather than attempting to obtain more input. This
non-interactive behaviour can be disabled by setting the variable
$Foce_Interactive as below:
$Term::Query::Force_Interactive = 1;
When $Force_Interactive is a non-null, non-zero value,
query will issue
prompts, error messages, and ask for additional input even when the input is
not interactive.
query_table¶
The
query_table subroutine performs multiple queries, by invoking
query, setting associated variables with the results of each query.
Prompts, flags, and other arguments for each query are given in an array,
called a
query table, which is passed to the
query_table
subroutine by reference.
query_table_set_defaults¶
The
query_table_set_defaults subroutine causes any variables named in the
given
query table array to be assigned their corresponding default
values, if any. This is a non-interactive subroutine.
query_table_process¶
A general interface to processing a
query table is available with the
query_table_process subroutine. It accepts a
query table array,
and two subroutine references, a
&flagsub and a
&querysub. The
&flagsub is invoked on each each
flag character given in the
$flags argument of the
query table (see below). The
&querysub is invoked for each
query in the
query table.
The
query_table and
query_table_set_defaults subroutines both use
query_table_process to perform their functions.
Query Table¶
The format of the
query table array passed to
query_table,
query_table_set_defaults, and
query_table_process subroutines
is:
@array = ( $prompt1, $flags1, [ $arglist1, ... ],
$prompt2, $flags2, [ $arglist2, ... ],
...
$promptN, $flagsN, [ $arglistN, ... ] );
In English, there are three items per query: a
prompt string, a
flags string, and an array of arguments. Note that the syntax used
above uses "[ ... ]" to denote a Perl 5 anonymous array, not an
optional set of arguments. Of course, if there are no arguments for a
particular query, the corresponding anonymous array can be the null string or
zero.
The query table design is such that a query table can be created with a set of
variables, their defaults, value constraints, and help strings, and it can be
used to both initialize the variables' values and to interactively set their
new values. The
query_table_set_defaults subroutine performs the
former, while
query_table does the latter.
Flag Characters¶
With typical usage, given
$prompt and
$flags,
query prints
$prompt
and then waits for input from the user. The handling of the response depends
upon the flag characters given in the
$flags string.
The flag characters indicate the type of input, how to process it, acceptable
values, etc. Some flags simply indicate the type or processing of the input,
and do not require additional arguments. Other flags require that subsequent
arguments to the
query subroutine be given. The arguments must be given
in the same order as their corresponding flag characters.
The ordering of the flags in the
$flags argument is
important -- it determines the ordering of the tests. For example, if both the
a and
m flags are given as "am", then this indicates
that an
after subroutine call should be performed first, followed by a
regular expression
match test.
All tests are applied in the order given in the
$flags
until a particular test fails. When a test fails, an error message is
generated and the input is reattempted, except in the case of the
I
flag.
Flag Characters Without Arguments¶
- i
- The input must be an integer.
- n
- The input must be a number, real or integer.
- Y
- The input is a "yes" or "no", with a default answer of
"yes".
- N
- The input is a "yes" or "no", with a default answer of
"no".
- r
- Some input is required; an empty response will be refused. This
option is only meaningful when there is no default input (see the d
flag character below).
- s
- Strip and squeeze the input. Leading and trailing blanks are eliminated,
and embedded whitespace is "squeezed" to single blank
characters. This flag is implied by the k and K flags.
- H
- Do not treat input of ? as a request for help. This disables
automatic help, unless implemented with the after (a flag)
subroutine.
Flag Characters With Arguments¶
The following flag characters indicate the presence of an argument to
query. The arguments must occur in the same order as their
corresponding flag characters. For example, if both the
V and
h
flags are given as "Vh", then the first argument must be the
variable name, and the next the help string, in that order.
- a \&after
- The next argument is the after subroutine, to be invoked after the
input has been solicited. This feature provides for an "open
ended" input validation, completely at the control of the user of the
Query module. The after subroutine is invoked in this manner:
&$after( \$input );
If the after sub returns an "undef", then query processing
stops with an immediate "undef" return value.
If the after sub returns a null or zero value, then the input is
rejected and resolicted. No error messages are displayed except the
"Please try again." message.
Since the after sub has the reference to the
$input variable, it is free to change the value of
input indirectly; ie:
$$input = $some_new_value;
- b \&before
- The next argument is the before subroutine, to be invoked before
any input is attempted. If the before sub returns a non-null,
non-zero value, the current query will be attempted. If a null or zero
value is returned, the current query will be abandoned, with a null
return.
This feature, used in a query table, allows for selective queries to
be programmed by using before subs on the optional queries. For
example, using the following anonymous sub as the b flag argument:
sub { $> == 0; }
will cause the corresponding query to only be issued for the
"root" user.
The ordering of the b flag in the $flags
argument is unimportant, since, by definition, this test is always
performed before attempting any input.
- d $default
- The next argument is the default input. This string is used instead
of an empty response from the user. The default value can be a scalar
value, a reference to a scalar value, or a reference to a subroutine,
which will be invoked for its result only if a default value is needed (no
input is given).
- h $help_string
- The next argument is the help string, which is printed in response
to an input of " ?". In order to enter a ? as
actual text, it must be prefixed with a backslash: "\".
- k \@array
- The next argument is a reference to an array of allowable keywords. The
input is matched against the array elements in a case-insensitive manner,
with unambiguous abbreviations allowed. This flag implies the s
flag.
The matching can be made case-sensitive by setting the following variable
prior to the invocation of query:
$Query::Case_sensitive = 1;
By default, this variable is null.
- K \@array
- The next argument is a reference to an array of disallowed keywords In
this case, for the input to be unacceptable, it must match exactly,
case-insensitive, one of the array elements. This flag implies the
s flag.
The k option is useful for soliciting new, unique keywords to a
growing list. Adding new fields to a database, for example.
The matching can be made case-sensitive by setting the
$Query::Case_sensitive variable (see above).
- l $maxlen
- The next argument specifies the maximum length of the input.
- m $regular_expression
- The next argument specifies a regular expression pattern against which the
input will be matched.
- I $reference
- The next argument is the input: either a simple scalar value, or a
reference to a value, such as a "SCALAR" variable
reference (eg: "\$somevar"), or a "CODE" reference
(eg: "sub {..}"). In any case, the resulting value is used as
input instead of reading from STDIN.
If the input returned by the reference does not match other constraints,
additional input is not attempted. An error message is noted, and an
"undef" return is taken.
This option is handy for applications which have already acquired the input,
and wish to use the validation features of "query".
It is also useful to embed a query definition in a query table which
does not actually perform a query, but instead does a variable assignment
dynamically, using the I reference value.
- J $reference
- The next argument is the input reference, as with the I
flag, except that if the input fails any of the constraints, additional
input is solicited from the input. In other words, the J flag sets
a one-time only input reference. Think of it as jumping into
the query loop with an initial input.
- V variable_name_or_ref
- The next argument is the variable name or reference to receive the
validated input as its value. This option, and its corresponding variable
name, would normally be present on all entries used with
query_table in order to retain to the values resulting from each
query.
The value can either be a string representing the variable name, or a
reference to a variable, eg: "\$some_var".
Details¶
The query processing proceeds basically in the same order as defined by the
flags argument, with some exceptions. For example, the
before
subroutine is always performed prior to input.
There are implicit precedences in the ordering of some of the
flag tests.
Generally, flags have their corresponding tests performed in the same order as
the given flags. Some flag tests, however, require that other flags' tests be
performed beforehand in order to be effective. For example, when given the
k flag and an
s flag, stripping the input would only be
effective if the strip were done on the input before testing the input against
the keyword table. In other words, the
s flag has precedence over the
k flag. If the user supplies the
flags string as "ks",
the effective ordering would still be "sk".
The table below indicates the precedences of the flag tests:
Given Flag Flags With Higher Precedence
========== ================================
i (int) s (strip), d (default), h (help)
k (key) s (strip), d (default), h (help)
K (nonkey) s (strip), d (default), h (help)
l (maxlen) d (default), h (help)
m (match) d (default), h (help)
n (numeric) s (strip), d (default), h (help)
N (no) s (strip), d (default), h (help)
r (required) d (default), h (help)
s (strip) d (default), h (help)
Y (yes) s (strip), d (default), h (help)
Except for the implied precedence indicated in the table above, the ordering of
the flag tests proceeds in the same order as given in the
flags
argument.
Excepting the precedences above, query processing proceeds generally as
described below.
- •
- If the b flag was given, the "before" subroutine is
invoked as a "pre-input" test. If the sub returns a 0, empty
string, or undef, the query is abandoned. Otherwise, processing
continues.
- •
- If the I or J flags were given, then input is obtained,
without prompting, from the associated reference. If the reference type is
"CODE", then it is invoked and the resulting return value is
used as the input. Otherwise the reference is evaluated in a scalar
context and used as the input. The J flag test is only done once,
on the first entry into the input loop.
- •
- In the absence either the I or J flags, "query"
will issue the given prompt and obtain input from STDIN. If an EOF occurs,
an "undef" value will result.
- •
- The input is examined for "null" input (that is, the empty
string), and processing quits in this case. Since most input is obtained
from STDIN, a null input indicates an end-of-file (EOF). If the input is
not null, a terminating newline is removed, and the input testing
continues. At this point, an empty input string does not indicate an
EOF.
- •
- If the s, k, or K flags were given, the input is
trimmed of leading and trailing blanks, and all whitespace is
"squeezed" to single blanks.
- •
- If the input is an empty response, and there is a default input
(d flag), use it instead.
- •
- Unless the H flag is given, if the input is the character
"?" with nothing else, then print some helpful
information. If the user had supplied a help string, it is printed,
otherwise the message:
You are being asked " $prompt"
is displayed. Also, some information about the expected response, according
to any given flag characters, is displayed. Finally, the user is returned
to the prompt, and given another opportunity to enter a response.
- •
- If input is required (indicated by the r flag), and if the
input is empty, produce an error message, and query again.
- •
- If there was a a flag, the corresponding after subroutine is
invoked with the input reference as its argument. If the subroutine
returns a non-null, non-zero value, the input succeeds, otherwise it
fails. It is up to the after subroutine to display any appropriate
error messages.
- •
- If the query was flagged Y or N, match the input against the
pattern:
/^(y(es?)?|no?)$/i
If the match fails, print an error message, and query again. When the match
succeeds, replace the input with the complete word "yes" or
"no";
- •
- If an integer response is required (i flagged), check for integer
input. If not, print an error, and query again. A successful integer input
is returned.
- •
- If a numeric response is required (n flagged), check for proper
numeric input (either integer or real format). Errors produce a warning,
and another query.
- •
- If the query was given a keyword table (flagged with k), the
input is matched against the allowable keyword list. If an exact match is
found, the keyword is returned as the input. Failing an exact match, an
abbreviation search is performed against the keywords. If a single match
is found, it is returned as the input. If no match is found, an error
message is produced, and the user is returned to the query to try again.
Otherwise, the input was ambiguous, an error noted showing the matches,
and the user is queried again.
The matching is case-insensitive or not, according to the value of the
variable $Query::Case_sensitive, which is nil, by default. The variable
may be set by the user to change the matching from case-insensitive to
case-sensitive.
- •
- If the query was given an unacceptable keyword list (flagged with
K), the input is compared against the unacceptable keywords. If it
matches any keywords exactly, an error is noted, and the query is
performed again.
The matching is case-insensitive by default. Set the variable
$Query::Case_sensitive to a non-null, non-zero value to make the keyword
matching case-sensitive.
- •
- If the query was m flagged with a Perl regular expression pattern,
then the input is matched against the pattern. Failures are noted with an
error message, and the query reattempted.
- •
- If the query was l flagged with a maximum input length, the length
of the input is checked against the maximum. A length violation is noted
with an error message and the user is queried again.
- •
- If the query has a variable defined with the V flag, the variable
is assigned the input string. This is always done last, after and only if
all tests are successful.
If the variable is a string name and not qualified with a package name (ie:
$foo::variable), then the variable is qualified at the level outside of
the Query.pm module.
- •
- Finally, having passed whatever conditions were flagged, the input is
returned to the user.
EXAMPLE¶
The following are typical usage samples:
- •
- To perform a simple "yes" or "no" query, with
"no" as the default answer:
$ans = &query("Do you wish to quit? (yn)",'N');
- •
- An equivalent alternative is:
query "Do you wish to quit? (yn)", 'NV', \$ans;
- •
- To perform the same query, with some supplied helpful information:
$ans = &query("Do you wish to quit? (yn)",'Nh',<<'EOF');
You are being asked if you wish to quit. If you answer "yes",
then all changes will be lost. An answer of "no", will allow
you to return to continue making changes.
EOF
- •
- To solicit an integer input:
$mode = &query("Please enter the file mode:",'idh','644',<<'EOF');
Please enter the 3 digit numeric file mode; if you are unsure
of how the file mode is used, please see the man page for "chmod".
EOF
- •
- To solicit one of several keywords:
@keys = split(' ','SGI DEC IBM Sun HP Apple');
$vendor = &query('Please enter a vendor:','rkd',\@keys,'SGI');
- •
- To solicit a new, unique keyword to be used as a database field name, with
a regexp pattern to check it against:
@fields = split(' ','Index Vendor Title'); # existing fields
$newfield = &query('New field name:','rKm',\@fields,'^\w+$');
ENVIRONMENT¶
- COLUMNS
- This variable is used to control the width of output when listing the
keyword arrays. If not defined, 80 is used by default.
DEPENDENCIES¶
- Carp.pm
- Used to produce usage error messages.
- Array::PrintCols::print_cols
- Used to produce displays of the keyword arrays.
FILES¶
None.
AUTHOR¶
Copyright (C) 1995 Alan K. Stebbens <aks@hub.ucsb.edu>
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation; either version 2 of the License, or (at your option) any later
version.
This program is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc., 675 Mass
Ave, Cambridge, MA 02139, USA.
DIAGNOSTICS¶
- Input is required.
- Issued when an empty response is given, and there is no default
input.
- Please answer with 'yes' or 'no', or enter '?' for help.
- Issued for Y or N flagged queries, and the input is not
reconizeable.
- Please enter an integer number.
- Printed when non-integer input is given for i flagged queries.
- Please enter a number, real or integer.
- Printed when non-numeric input is given for n flagged queries.
- The input '$input' is ambiguous; it matches the
following:
- Issued in response to k flagged queries with input which matches
more than one of the allowed keywords.
- The input '$input' fails to match any of the allowed
keywords:
- Printed when input to a k flagged query does not match any of the
keywords.
- The input '%s' matches a disallowed keyword '%s'.
- Printed when the input matches one of the unacceptable keywords given on a
K flagged query.
- '%s' fails to match '%s'
- This results from input failing to match the regular expression given on a
m flagged query.
- Input is %d characters too long; cannot exceed %d characters.
- The length of the input exceeded the maximum length given with the
l flag argument.
- Please try again, or enter '?' for help.
- query: The k flag needs an array reference.
- The next argument in the argument list to query wasn't an array
reference.
- query: The K flag needs an array reference.
- The next argument in the argumentlist to query wasn't an array
reference.
BUGS¶