1 ==============================
2 CommandLine 2.0 Library Manual
3 ==============================
11 This document describes the CommandLine argument processing library. It will
12 show you how to use it, and what it can do. The CommandLine library uses a
13 declarative approach to specifying the command line options that your program
14 takes. By default, these options declarations implicitly hold the value parsed
15 for the option declared (of course this `can be changed`_).
17 Although there are a **lot** of command line argument parsing libraries out
18 there in many different languages, none of them fit well with what I needed. By
19 looking at the features and problems of other libraries, I designed the
20 CommandLine library to have the following features:
22 #. Speed: The CommandLine library is very quick and uses little resources. The
23 parsing time of the library is directly proportional to the number of
24 arguments parsed, not the number of options recognized. Additionally,
25 command line argument values are captured transparently into user defined
26 global variables, which can be accessed like any other variable (and with the
29 #. Type Safe: As a user of CommandLine, you don't have to worry about
30 remembering the type of arguments that you want (is it an int? a string? a
31 bool? an enum?) and keep casting it around. Not only does this help prevent
32 error prone constructs, it also leads to dramatically cleaner source code.
34 #. No subclasses required: To use CommandLine, you instantiate variables that
35 correspond to the arguments that you would like to capture, you don't
36 subclass a parser. This means that you don't have to write **any**
39 #. Globally accessible: Libraries can specify command line arguments that are
40 automatically enabled in any tool that links to the library. This is
41 possible because the application doesn't have to keep a list of arguments to
42 pass to the parser. This also makes supporting `dynamically loaded options`_
45 #. Cleaner: CommandLine supports enum and other types directly, meaning that
46 there is less error and more security built into the library. You don't have
47 to worry about whether your integral command line argument accidentally got
48 assigned a value that is not valid for your enum type.
50 #. Powerful: The CommandLine library supports many different types of arguments,
51 from simple `boolean flags`_ to `scalars arguments`_ (`strings`_,
52 `integers`_, `enums`_, `doubles`_), to `lists of arguments`_. This is
53 possible because CommandLine is...
55 #. Extensible: It is very simple to add a new argument type to CommandLine.
56 Simply specify the parser that you want to use with the command line option
57 when you declare it. `Custom parsers`_ are no problem.
59 #. Labor Saving: The CommandLine library cuts down on the amount of grunt work
60 that you, the user, have to do. For example, it automatically provides a
61 ``-help`` option that shows the available command line options for your tool.
62 Additionally, it does most of the basic correctness checking for you.
64 #. Capable: The CommandLine library can handle lots of different forms of
65 options often found in real programs. For example, `positional`_ arguments,
66 ``ls`` style `grouping`_ options (to allow processing '``ls -lad``'
67 naturally), ``ld`` style `prefix`_ options (to parse '``-lmalloc
68 -L/usr/lib``'), and interpreter style options.
70 This document will hopefully let you jump in and start using CommandLine in your
71 utility quickly and painlessly. Additionally it should be a simple reference
72 manual to figure out how stuff works.
77 This section of the manual runs through a simple CommandLine'ification of a
78 basic compiler tool. This is intended to show you how to jump into using the
79 CommandLine library in your own program, and show you some of the cool things it
82 To start out, you need to include the CommandLine header file into your program:
86 #include "llvm/Support/CommandLine.h"
88 Additionally, you need to add this as the first line of your main program:
92 int main(int argc, char **argv) {
93 cl::ParseCommandLineOptions(argc, argv);
97 ... which actually parses the arguments and fills in the variable declarations.
99 Now that you are ready to support command line arguments, we need to tell the
100 system which ones we want, and what type of arguments they are. The CommandLine
101 library uses a declarative syntax to model command line arguments with the
102 global variable declarations that capture the parsed values. This means that
103 for every command line option that you would like to support, there should be a
104 global variable declaration to capture the result. For example, in a compiler,
105 we would like to support the Unix-standard '``-o <filename>``' option to specify
106 where to put the output. With the CommandLine library, this is represented like
109 .. _scalars arguments:
114 cl::opt<string> OutputFilename("o", cl::desc("Specify output filename"), cl::value_desc("filename"));
116 This declares a global variable "``OutputFilename``" that is used to capture the
117 result of the "``o``" argument (first parameter). We specify that this is a
118 simple scalar option by using the "``cl::opt``" template (as opposed to the
119 "``cl::list``" template), and tell the CommandLine library that the data
120 type that we are parsing is a string.
122 The second and third parameters (which are optional) are used to specify what to
123 output for the "``-help``" option. In this case, we get a line that looks like
128 USAGE: compiler [options]
131 -help - display available options (-help-hidden for more)
132 -o <filename> - Specify output filename
134 Because we specified that the command line option should parse using the
135 ``string`` data type, the variable declared is automatically usable as a real
136 string in all contexts that a normal C++ string object may be used. For
142 std::ofstream Output(OutputFilename.c_str());
143 if (Output.good()) ...
146 There are many different options that you can use to customize the command line
147 option handling library, but the above example shows the general interface to
148 these options. The options can be specified in any order, and are specified
149 with helper functions like `cl::desc(...)`_, so there are no positional
150 dependencies to remember. The available options are discussed in detail in the
153 Continuing the example, we would like to have our compiler take an input
154 filename as well as an output filename, but we do not want the input filename to
155 be specified with a hyphen (ie, not ``-filename.c``). To support this style of
156 argument, the CommandLine library allows for `positional`_ arguments to be
157 specified for the program. These positional arguments are filled with command
158 line parameters that are not in option form. We use this feature like this:
163 cl::opt<string> InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
165 This declaration indicates that the first positional argument should be treated
166 as the input filename. Here we use the `cl::init`_ option to specify an initial
167 value for the command line option, which is used if the option is not specified
168 (if you do not specify a `cl::init`_ modifier for an option, then the default
169 constructor for the data type is used to initialize the value). Command line
170 options default to being optional, so if we would like to require that the user
171 always specify an input filename, we would add the `cl::Required`_ flag, and we
172 could eliminate the `cl::init`_ modifier, like this:
176 cl::opt<string> InputFilename(cl::Positional, cl::desc("<input file>"), cl::Required);
178 Again, the CommandLine library does not require the options to be specified in
179 any particular order, so the above declaration is equivalent to:
183 cl::opt<string> InputFilename(cl::Positional, cl::Required, cl::desc("<input file>"));
185 By simply adding the `cl::Required`_ flag, the CommandLine library will
186 automatically issue an error if the argument is not specified, which shifts all
187 of the command line option verification code out of your application into the
188 library. This is just one example of how using flags can alter the default
189 behaviour of the library, on a per-option basis. By adding one of the
190 declarations above, the ``-help`` option synopsis is now extended to:
194 USAGE: compiler [options] <input file>
197 -help - display available options (-help-hidden for more)
198 -o <filename> - Specify output filename
200 ... indicating that an input filename is expected.
205 In addition to input and output filenames, we would like the compiler example to
206 support three boolean flags: "``-f``" to force writing binary output to a
207 terminal, "``--quiet``" to enable quiet mode, and "``-q``" for backwards
208 compatibility with some of our users. We can support these by declaring options
209 of boolean type like this:
213 cl::opt<bool> Force ("f", cl::desc("Enable binary output on terminals"));
214 cl::opt<bool> Quiet ("quiet", cl::desc("Don't print informational messages"));
215 cl::opt<bool> Quiet2("q", cl::desc("Don't print informational messages"), cl::Hidden);
217 This does what you would expect: it declares three boolean variables
218 ("``Force``", "``Quiet``", and "``Quiet2``") to recognize these options. Note
219 that the "``-q``" option is specified with the "`cl::Hidden`_" flag. This
220 modifier prevents it from being shown by the standard "``-help``" output (note
221 that it is still shown in the "``-help-hidden``" output).
223 The CommandLine library uses a `different parser`_ for different data types.
224 For example, in the string case, the argument passed to the option is copied
225 literally into the content of the string variable... we obviously cannot do that
226 in the boolean case, however, so we must use a smarter parser. In the case of
227 the boolean parser, it allows no options (in which case it assigns the value of
228 true to the variable), or it allows the values "``true``" or "``false``" to be
229 specified, allowing any of the following inputs:
233 compiler -f # No value, 'Force' == true
234 compiler -f=true # Value specified, 'Force' == true
235 compiler -f=TRUE # Value specified, 'Force' == true
236 compiler -f=FALSE # Value specified, 'Force' == false
238 ... you get the idea. The `bool parser`_ just turns the string values into
239 boolean values, and rejects things like '``compiler -f=foo``'. Similarly, the
240 `float`_, `double`_, and `int`_ parsers work like you would expect, using the
241 '``strtol``' and '``strtod``' C library calls to parse the string value into the
244 With the declarations above, "``compiler -help``" emits this:
248 USAGE: compiler [options] <input file>
251 -f - Enable binary output on terminals
252 -o - Override output filename
253 -quiet - Don't print informational messages
254 -help - display available options (-help-hidden for more)
256 and "``compiler -help-hidden``" prints this:
260 USAGE: compiler [options] <input file>
263 -f - Enable binary output on terminals
264 -o - Override output filename
265 -q - Don't print informational messages
266 -quiet - Don't print informational messages
267 -help - display available options (-help-hidden for more)
269 This brief example has shown you how to use the '`cl::opt`_' class to parse
270 simple scalar command line arguments. In addition to simple scalar arguments,
271 the CommandLine library also provides primitives to support CommandLine option
272 `aliases`_, and `lists`_ of options.
279 So far, the example works well, except for the fact that we need to check the
280 quiet condition like this now:
285 if (!Quiet && !Quiet2) printInformationalMessage(...);
288 ... which is a real pain! Instead of defining two values for the same
289 condition, we can use the "`cl::alias`_" class to make the "``-q``" option an
290 **alias** for the "``-quiet``" option, instead of providing a value itself:
294 cl::opt<bool> Force ("f", cl::desc("Overwrite output files"));
295 cl::opt<bool> Quiet ("quiet", cl::desc("Don't print informational messages"));
296 cl::alias QuietA("q", cl::desc("Alias for -quiet"), cl::aliasopt(Quiet));
298 The third line (which is the only one we modified from above) defines a "``-q``"
299 alias that updates the "``Quiet``" variable (as specified by the `cl::aliasopt`_
300 modifier) whenever it is specified. Because aliases do not hold state, the only
301 thing the program has to query is the ``Quiet`` variable now. Another nice
302 feature of aliases is that they automatically hide themselves from the ``-help``
303 output (although, again, they are still visible in the ``-help-hidden output``).
305 Now the application code can simply use:
310 if (!Quiet) printInformationalMessage(...);
313 ... which is much nicer! The "`cl::alias`_" can be used to specify an
314 alternative name for any variable type, and has many uses.
316 .. _unnamed alternatives using the generic parser:
318 Selecting an alternative from a set of possibilities
319 ----------------------------------------------------
321 So far we have seen how the CommandLine library handles builtin types like
322 ``std::string``, ``bool`` and ``int``, but how does it handle things it doesn't
323 know about, like enums or '``int*``'s?
325 The answer is that it uses a table-driven generic parser (unless you specify
326 your own parser, as described in the `Extension Guide`_). This parser maps
327 literal strings to whatever type is required, and requires you to tell it what
328 this mapping should be.
330 Let's say that we would like to add four optimization levels to our optimizer,
331 using the standard flags "``-g``", "``-O0``", "``-O1``", and "``-O2``". We
332 could easily implement this with boolean options like above, but there are
333 several problems with this strategy:
335 #. A user could specify more than one of the options at a time, for example,
336 "``compiler -O3 -O2``". The CommandLine library would not be able to catch
337 this erroneous input for us.
339 #. We would have to test 4 different variables to see which ones are set.
341 #. This doesn't map to the numeric levels that we want... so we cannot easily
342 see if some level >= "``-O1``" is enabled.
344 To cope with these problems, we can use an enum value, and have the CommandLine
345 library fill it in with the appropriate level directly, which is used like this:
353 cl::opt<OptLevel> OptimizationLevel(cl::desc("Choose optimization level:"),
355 clEnumVal(g , "No optimizations, enable debugging"),
356 clEnumVal(O1, "Enable trivial optimizations"),
357 clEnumVal(O2, "Enable default optimizations"),
358 clEnumVal(O3, "Enable expensive optimizations"),
362 if (OptimizationLevel >= O2) doPartialRedundancyElimination(...);
365 This declaration defines a variable "``OptimizationLevel``" of the
366 "``OptLevel``" enum type. This variable can be assigned any of the values that
367 are listed in the declaration (Note that the declaration list must be terminated
368 with the "``clEnumValEnd``" argument!). The CommandLine library enforces that
369 the user can only specify one of the options, and it ensure that only valid enum
370 values can be specified. The "``clEnumVal``" macros ensure that the command
371 line arguments matched the enum values. With this option added, our help output
376 USAGE: compiler [options] <input file>
379 Choose optimization level:
380 -g - No optimizations, enable debugging
381 -O1 - Enable trivial optimizations
382 -O2 - Enable default optimizations
383 -O3 - Enable expensive optimizations
384 -f - Enable binary output on terminals
385 -help - display available options (-help-hidden for more)
386 -o <filename> - Specify output filename
387 -quiet - Don't print informational messages
389 In this case, it is sort of awkward that flag names correspond directly to enum
390 names, because we probably don't want a enum definition named "``g``" in our
391 program. Because of this, we can alternatively write this example like this:
399 cl::opt<OptLevel> OptimizationLevel(cl::desc("Choose optimization level:"),
401 clEnumValN(Debug, "g", "No optimizations, enable debugging"),
402 clEnumVal(O1 , "Enable trivial optimizations"),
403 clEnumVal(O2 , "Enable default optimizations"),
404 clEnumVal(O3 , "Enable expensive optimizations"),
408 if (OptimizationLevel == Debug) outputDebugInfo(...);
411 By using the "``clEnumValN``" macro instead of "``clEnumVal``", we can directly
412 specify the name that the flag should get. In general a direct mapping is nice,
413 but sometimes you can't or don't want to preserve the mapping, which is when you
419 Another useful argument form is a named alternative style. We shall use this
420 style in our compiler to specify different debug levels that can be used.
421 Instead of each debug level being its own switch, we want to support the
422 following options, of which only one can be specified at a time:
423 "``--debug-level=none``", "``--debug-level=quick``",
424 "``--debug-level=detailed``". To do this, we use the exact same format as our
425 optimization level flags, but we also specify an option name. For this case,
426 the code looks like this:
431 nodebuginfo, quick, detailed
434 // Enable Debug Options to be specified on the command line
435 cl::opt<DebugLev> DebugLevel("debug_level", cl::desc("Set the debugging level:"),
437 clEnumValN(nodebuginfo, "none", "disable debug information"),
438 clEnumVal(quick, "enable quick debug information"),
439 clEnumVal(detailed, "enable detailed debug information"),
442 This definition defines an enumerated command line variable of type "``enum
443 DebugLev``", which works exactly the same way as before. The difference here is
444 just the interface exposed to the user of your program and the help output by
445 the "``-help``" option:
449 USAGE: compiler [options] <input file>
452 Choose optimization level:
453 -g - No optimizations, enable debugging
454 -O1 - Enable trivial optimizations
455 -O2 - Enable default optimizations
456 -O3 - Enable expensive optimizations
457 -debug_level - Set the debugging level:
458 =none - disable debug information
459 =quick - enable quick debug information
460 =detailed - enable detailed debug information
461 -f - Enable binary output on terminals
462 -help - display available options (-help-hidden for more)
463 -o <filename> - Specify output filename
464 -quiet - Don't print informational messages
466 Again, the only structural difference between the debug level declaration and
467 the optimization level declaration is that the debug level declaration includes
468 an option name (``"debug_level"``), which automatically changes how the library
469 processes the argument. The CommandLine library supports both forms so that you
470 can choose the form most appropriate for your application.
474 Parsing a list of options
475 -------------------------
477 Now that we have the standard run-of-the-mill argument types out of the way,
478 lets get a little wild and crazy. Lets say that we want our optimizer to accept
479 a **list** of optimizations to perform, allowing duplicates. For example, we
480 might want to run: "``compiler -dce -constprop -inline -dce -strip``". In this
481 case, the order of the arguments and the number of appearances is very
482 important. This is what the "``cl::list``" template is for. First, start by
483 defining an enum of the optimizations that you would like to perform:
488 // 'inline' is a C++ keyword, so name it 'inlining'
489 dce, constprop, inlining, strip
492 Then define your "``cl::list``" variable:
496 cl::list<Opts> OptimizationList(cl::desc("Available Optimizations:"),
498 clEnumVal(dce , "Dead Code Elimination"),
499 clEnumVal(constprop , "Constant Propagation"),
500 clEnumValN(inlining, "inline", "Procedure Integration"),
501 clEnumVal(strip , "Strip Symbols"),
504 This defines a variable that is conceptually of the type
505 "``std::vector<enum Opts>``". Thus, you can access it with standard vector
510 for (unsigned i = 0; i != OptimizationList.size(); ++i)
511 switch (OptimizationList[i])
514 ... to iterate through the list of options specified.
516 Note that the "``cl::list``" template is completely general and may be used with
517 any data types or other arguments that you can use with the "``cl::opt``"
518 template. One especially useful way to use a list is to capture all of the
519 positional arguments together if there may be more than one specified. In the
520 case of a linker, for example, the linker takes several '``.o``' files, and
521 needs to capture them into a list. This is naturally specified as:
526 cl::list<std::string> InputFilenames(cl::Positional, cl::desc("<Input files>"), cl::OneOrMore);
529 This variable works just like a "``vector<string>``" object. As such, accessing
530 the list is simple, just like above. In this example, we used the
531 `cl::OneOrMore`_ modifier to inform the CommandLine library that it is an error
532 if the user does not specify any ``.o`` files on our command line. Again, this
533 just reduces the amount of checking we have to do.
535 Collecting options as a set of flags
536 ------------------------------------
538 Instead of collecting sets of options in a list, it is also possible to gather
539 information for enum values in a **bit vector**. The representation used by the
540 `cl::bits`_ class is an ``unsigned`` integer. An enum value is represented by a
541 0/1 in the enum's ordinal value bit position. 1 indicating that the enum was
542 specified, 0 otherwise. As each specified value is parsed, the resulting enum's
543 bit is set in the option's bit vector:
547 bits |= 1 << (unsigned)enum;
549 Options that are specified multiple times are redundant. Any instances after
550 the first are discarded.
552 Reworking the above list example, we could replace `cl::list`_ with `cl::bits`_:
556 cl::bits<Opts> OptimizationBits(cl::desc("Available Optimizations:"),
558 clEnumVal(dce , "Dead Code Elimination"),
559 clEnumVal(constprop , "Constant Propagation"),
560 clEnumValN(inlining, "inline", "Procedure Integration"),
561 clEnumVal(strip , "Strip Symbols"),
564 To test to see if ``constprop`` was specified, we can use the ``cl:bits::isSet``
569 if (OptimizationBits.isSet(constprop)) {
573 It's also possible to get the raw bit vector using the ``cl::bits::getBits``
578 unsigned bits = OptimizationBits.getBits();
580 Finally, if external storage is used, then the location specified must be of
581 **type** ``unsigned``. In all other ways a `cl::bits`_ option is equivalent to a
584 .. _additional extra text:
586 Adding freeform text to help output
587 -----------------------------------
589 As our program grows and becomes more mature, we may decide to put summary
590 information about what it does into the help output. The help output is styled
591 to look similar to a Unix ``man`` page, providing concise information about a
592 program. Unix ``man`` pages, however often have a description about what the
593 program does. To add this to your CommandLine program, simply pass a third
594 argument to the `cl::ParseCommandLineOptions`_ call in main. This additional
595 argument is then printed as the overview information for your program, allowing
596 you to include any additional information that you want. For example:
600 int main(int argc, char **argv) {
601 cl::ParseCommandLineOptions(argc, argv, " CommandLine compiler example\n\n"
602 " This program blah blah blah...\n");
606 would yield the help output:
610 **OVERVIEW: CommandLine compiler example
612 This program blah blah blah...**
614 USAGE: compiler [options] <input file>
618 -help - display available options (-help-hidden for more)
619 -o <filename> - Specify output filename
626 Now that you know the basics of how to use the CommandLine library, this section
627 will give you the detailed information you need to tune how command line options
628 work, as well as information on more "advanced" command line option processing
632 .. _positional argument:
633 .. _Positional Arguments:
634 .. _Positional arguments section:
635 .. _positional options:
640 Positional arguments are those arguments that are not named, and are not
641 specified with a hyphen. Positional arguments should be used when an option is
642 specified by its position alone. For example, the standard Unix ``grep`` tool
643 takes a regular expression argument, and an optional filename to search through
644 (which defaults to standard input if a filename is not specified). Using the
645 CommandLine library, this would be specified as:
649 cl::opt<string> Regex (cl::Positional, cl::desc("<regular expression>"), cl::Required);
650 cl::opt<string> Filename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
652 Given these two option declarations, the ``-help`` output for our grep
653 replacement would look like this:
657 USAGE: spiffygrep [options] <regular expression> <input file>
660 -help - display available options (-help-hidden for more)
662 ... and the resultant program could be used just like the standard ``grep``
665 Positional arguments are sorted by their order of construction. This means that
666 command line options will be ordered according to how they are listed in a .cpp
667 file, but will not have an ordering defined if the positional arguments are
668 defined in multiple .cpp files. The fix for this problem is simply to define
669 all of your positional arguments in one .cpp file.
671 Specifying positional options with hyphens
672 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
674 Sometimes you may want to specify a value to your positional argument that
675 starts with a hyphen (for example, searching for '``-foo``' in a file). At
676 first, you will have trouble doing this, because it will try to find an argument
677 named '``-foo``', and will fail (and single quotes will not save you). Note
678 that the system ``grep`` has the same problem:
682 $ spiffygrep '-foo' test.txt
683 Unknown command line argument '-foo'. Try: spiffygrep -help'
685 $ grep '-foo' test.txt
686 grep: illegal option -- f
687 grep: illegal option -- o
688 grep: illegal option -- o
689 Usage: grep -hblcnsviw pattern file . . .
691 The solution for this problem is the same for both your tool and the system
692 version: use the '``--``' marker. When the user specifies '``--``' on the
693 command line, it is telling the program that all options after the '``--``'
694 should be treated as positional arguments, not options. Thus, we can use it
699 $ spiffygrep -- -foo test.txt
702 Determining absolute position with getPosition()
703 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
705 Sometimes an option can affect or modify the meaning of another option. For
706 example, consider ``gcc``'s ``-x LANG`` option. This tells ``gcc`` to ignore the
707 suffix of subsequent positional arguments and force the file to be interpreted
708 as if it contained source code in language ``LANG``. In order to handle this
709 properly, you need to know the absolute position of each argument, especially
710 those in lists, so their interaction(s) can be applied correctly. This is also
711 useful for options like ``-llibname`` which is actually a positional argument
712 that starts with a dash.
714 So, generally, the problem is that you have two ``cl::list`` variables that
715 interact in some way. To ensure the correct interaction, you can use the
716 ``cl::list::getPosition(optnum)`` method. This method returns the absolute
717 position (as found on the command line) of the ``optnum`` item in the
720 The idiom for usage is like this:
724 static cl::list<std::string> Files(cl::Positional, cl::OneOrMore);
725 static cl::list<std::string> Libraries("l", cl::ZeroOrMore);
727 int main(int argc, char**argv) {
729 std::vector<std::string>::iterator fileIt = Files.begin();
730 std::vector<std::string>::iterator libIt = Libraries.begin();
731 unsigned libPos = 0, filePos = 0;
733 if ( libIt != Libraries.end() )
734 libPos = Libraries.getPosition( libIt - Libraries.begin() );
737 if ( fileIt != Files.end() )
738 filePos = Files.getPosition( fileIt - Files.begin() );
742 if ( filePos != 0 && (libPos == 0 || filePos < libPos) ) {
743 // Source File Is next
746 else if ( libPos != 0 && (filePos == 0 || libPos < filePos) ) {
751 break; // we're done with the list
755 Note that, for compatibility reasons, the ``cl::opt`` also supports an
756 ``unsigned getPosition()`` option that will provide the absolute position of
757 that option. You can apply the same approach as above with a ``cl::opt`` and a
758 ``cl::list`` option as you can with two lists.
760 .. _interpreter style options:
761 .. _cl::ConsumeAfter:
762 .. _this section for more information:
764 The ``cl::ConsumeAfter`` modifier
765 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
767 The ``cl::ConsumeAfter`` `formatting option`_ is used to construct programs that
768 use "interpreter style" option processing. With this style of option
769 processing, all arguments specified after the last positional argument are
770 treated as special interpreter arguments that are not interpreted by the command
773 As a concrete example, lets say we are developing a replacement for the standard
774 Unix Bourne shell (``/bin/sh``). To run ``/bin/sh``, first you specify options
775 to the shell itself (like ``-x`` which turns on trace output), then you specify
776 the name of the script to run, then you specify arguments to the script. These
777 arguments to the script are parsed by the Bourne shell command line option
778 processor, but are not interpreted as options to the shell itself. Using the
779 CommandLine library, we would specify this as:
783 cl::opt<string> Script(cl::Positional, cl::desc("<input script>"), cl::init("-"));
784 cl::list<string> Argv(cl::ConsumeAfter, cl::desc("<program arguments>..."));
785 cl::opt<bool> Trace("x", cl::desc("Enable trace output"));
787 which automatically provides the help output:
791 USAGE: spiffysh [options] <input script> <program arguments>...
794 -help - display available options (-help-hidden for more)
795 -x - Enable trace output
797 At runtime, if we run our new shell replacement as ```spiffysh -x test.sh -a -x
798 -y bar``', the ``Trace`` variable will be set to true, the ``Script`` variable
799 will be set to "``test.sh``", and the ``Argv`` list will contain ``["-a", "-x",
800 "-y", "bar"]``, because they were specified after the last positional argument
801 (which is the script name).
803 There are several limitations to when ``cl::ConsumeAfter`` options can be
804 specified. For example, only one ``cl::ConsumeAfter`` can be specified per
805 program, there must be at least one `positional argument`_ specified, there must
806 not be any `cl::list`_ positional arguments, and the ``cl::ConsumeAfter`` option
807 should be a `cl::list`_ option.
810 .. _Internal vs External Storage:
812 Internal vs External Storage
813 ----------------------------
815 By default, all command line options automatically hold the value that they
816 parse from the command line. This is very convenient in the common case,
817 especially when combined with the ability to define command line options in the
818 files that use them. This is called the internal storage model.
820 Sometimes, however, it is nice to separate the command line option processing
821 code from the storage of the value parsed. For example, lets say that we have a
822 '``-debug``' option that we would like to use to enable debug information across
823 the entire body of our program. In this case, the boolean value controlling the
824 debug code should be globally accessible (in a header file, for example) yet the
825 command line option processing code should not be exposed to all of these
826 clients (requiring lots of .cpp files to ``#include CommandLine.h``).
828 To do this, set up your .h file with your option, like this for example:
832 // DebugFlag.h - Get access to the '-debug' command line option
835 // DebugFlag - This boolean is set to true if the '-debug' command line option
836 // is specified. This should probably not be referenced directly, instead, use
837 // the DEBUG macro below.
839 extern bool DebugFlag;
841 // DEBUG macro - This macro should be used by code to emit debug information.
842 // In the '-debug' option is specified on the command line, and if this is a
843 // debug build, then the code specified as the option to the macro will be
844 // executed. Otherwise it will not be.
848 #define DEBUG(X) do { if (DebugFlag) { X; } } while (0)
851 This allows clients to blissfully use the ``DEBUG()`` macro, or the
852 ``DebugFlag`` explicitly if they want to. Now we just need to be able to set
853 the ``DebugFlag`` boolean when the option is set. To do this, we pass an
854 additional argument to our command line argument processor, and we specify where
855 to fill in with the `cl::location`_ attribute:
859 bool DebugFlag; // the actual value
860 static cl::opt<bool, true> // The parser
861 Debug("debug", cl::desc("Enable debug output"), cl::Hidden, cl::location(DebugFlag));
863 In the above example, we specify "``true``" as the second argument to the
864 `cl::opt`_ template, indicating that the template should not maintain a copy of
865 the value itself. In addition to this, we specify the `cl::location`_
866 attribute, so that ``DebugFlag`` is automatically set.
871 This section describes the basic attributes that you can specify on options.
873 * The option name attribute (which is required for all options, except
874 `positional options`_) specifies what the option name is. This option is
875 specified in simple double quotes:
879 cl::opt<**bool**> Quiet("quiet");
883 * The **cl::desc** attribute specifies a description for the option to be
884 shown in the ``-help`` output for the program.
888 * The **cl::value_desc** attribute specifies a string that can be used to
889 fine tune the ``-help`` output for a command line option. Look `here`_ for an
894 * The **cl::init** attribute specifies an initial value for a `scalar`_
895 option. If this attribute is not specified then the command line option value
896 defaults to the value created by the default constructor for the
901 If you specify both **cl::init** and **cl::location** for an option, you
902 must specify **cl::location** first, so that when the command-line parser
903 sees **cl::init**, it knows where to put the initial value. (You will get an
904 error at runtime if you don't put them in the right order.)
908 * The **cl::location** attribute where to store the value for a parsed command
909 line option if using external storage. See the section on `Internal vs
910 External Storage`_ for more information.
914 * The **cl::aliasopt** attribute specifies which option a `cl::alias`_ option is
919 * The **cl::values** attribute specifies the string-to-value mapping to be used
920 by the generic parser. It takes a **clEnumValEnd terminated** list of
921 (option, value, description) triplets that specify the option name, the value
922 mapped to, and the description shown in the ``-help`` for the tool. Because
923 the generic parser is used most frequently with enum values, two macros are
926 #. The **clEnumVal** macro is used as a nice simple way to specify a triplet
927 for an enum. This macro automatically makes the option name be the same as
928 the enum name. The first option to the macro is the enum, the second is
929 the description for the command line option.
931 #. The **clEnumValN** macro is used to specify macro options where the option
932 name doesn't equal the enum name. For this macro, the first argument is
933 the enum value, the second is the flag name, and the second is the
936 You will get a compile time error if you try to use cl::values with a parser
937 that does not support it.
941 * The **cl::multi_val** attribute specifies that this option takes has multiple
942 values (example: ``-sectalign segname sectname sectvalue``). This attribute
943 takes one unsigned argument - the number of values for the option. This
944 attribute is valid only on ``cl::list`` options (and will fail with compile
945 error if you try to use it with other option types). It is allowed to use all
946 of the usual modifiers on multi-valued options (besides
947 ``cl::ValueDisallowed``, obviously).
952 Option modifiers are the flags and expressions that you pass into the
953 constructors for `cl::opt`_ and `cl::list`_. These modifiers give you the
954 ability to tweak how options are parsed and how ``-help`` output is generated to
955 fit your application well.
957 These options fall into five main categories:
959 #. Hiding an option from ``-help`` output
961 #. Controlling the number of occurrences required and allowed
963 #. Controlling whether or not a value must be specified
965 #. Controlling other formatting options
967 #. Miscellaneous option modifiers
969 It is not possible to specify two options from the same category (you'll get a
970 runtime error) to a single option, except for options in the miscellaneous
971 category. The CommandLine library specifies defaults for all of these settings
972 that are the most useful in practice and the most common, which mean that you
973 usually shouldn't have to worry about these.
975 Hiding an option from ``-help`` output
976 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
978 The ``cl::NotHidden``, ``cl::Hidden``, and ``cl::ReallyHidden`` modifiers are
979 used to control whether or not an option appears in the ``-help`` and
980 ``-help-hidden`` output for the compiled program:
984 * The **cl::NotHidden** modifier (which is the default for `cl::opt`_ and
985 `cl::list`_ options) indicates the option is to appear in both help
990 * The **cl::Hidden** modifier (which is the default for `cl::alias`_ options)
991 indicates that the option should not appear in the ``-help`` output, but
992 should appear in the ``-help-hidden`` output.
994 .. _cl::ReallyHidden:
996 * The **cl::ReallyHidden** modifier indicates that the option should not appear
999 Controlling the number of occurrences required and allowed
1000 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1002 This group of options is used to control how many time an option is allowed (or
1003 required) to be specified on the command line of your program. Specifying a
1004 value for this setting allows the CommandLine library to do error checking for
1007 The allowed values for this option group are:
1011 * The **cl::Optional** modifier (which is the default for the `cl::opt`_ and
1012 `cl::alias`_ classes) indicates that your program will allow either zero or
1013 one occurrence of the option to be specified.
1017 * The **cl::ZeroOrMore** modifier (which is the default for the `cl::list`_
1018 class) indicates that your program will allow the option to be specified zero
1023 * The **cl::Required** modifier indicates that the specified option must be
1024 specified exactly one time.
1028 * The **cl::OneOrMore** modifier indicates that the option must be specified at
1031 * The **cl::ConsumeAfter** modifier is described in the `Positional arguments
1034 If an option is not specified, then the value of the option is equal to the
1035 value specified by the `cl::init`_ attribute. If the ``cl::init`` attribute is
1036 not specified, the option value is initialized with the default constructor for
1039 If an option is specified multiple times for an option of the `cl::opt`_ class,
1040 only the last value will be retained.
1042 Controlling whether or not a value must be specified
1043 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1045 This group of options is used to control whether or not the option allows a
1046 value to be present. In the case of the CommandLine library, a value is either
1047 specified with an equal sign (e.g. '``-index-depth=17``') or as a trailing
1048 string (e.g. '``-o a.out``').
1050 The allowed values for this option group are:
1052 .. _cl::ValueOptional:
1054 * The **cl::ValueOptional** modifier (which is the default for ``bool`` typed
1055 options) specifies that it is acceptable to have a value, or not. A boolean
1056 argument can be enabled just by appearing on the command line, or it can have
1057 an explicit '``-foo=true``'. If an option is specified with this mode, it is
1058 illegal for the value to be provided without the equal sign. Therefore
1059 '``-foo true``' is illegal. To get this behavior, you must use
1060 the `cl::ValueRequired`_ modifier.
1062 .. _cl::ValueRequired:
1064 * The **cl::ValueRequired** modifier (which is the default for all other types
1065 except for `unnamed alternatives using the generic parser`_) specifies that a
1066 value must be provided. This mode informs the command line library that if an
1067 option is not provides with an equal sign, that the next argument provided
1068 must be the value. This allows things like '``-o a.out``' to work.
1070 .. _cl::ValueDisallowed:
1072 * The **cl::ValueDisallowed** modifier (which is the default for `unnamed
1073 alternatives using the generic parser`_) indicates that it is a runtime error
1074 for the user to specify a value. This can be provided to disallow users from
1075 providing options to boolean options (like '``-foo=true``').
1077 In general, the default values for this option group work just like you would
1078 want them to. As mentioned above, you can specify the `cl::ValueDisallowed`_
1079 modifier to a boolean argument to restrict your command line parser. These
1080 options are mostly useful when `extending the library`_.
1082 .. _formatting option:
1084 Controlling other formatting options
1085 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1087 The formatting option group is used to specify that the command line option has
1088 special abilities and is otherwise different from other command line arguments.
1089 As usual, you can only specify one of these arguments at most.
1091 .. _cl::NormalFormatting:
1093 * The **cl::NormalFormatting** modifier (which is the default all options)
1094 specifies that this option is "normal".
1098 * The **cl::Positional** modifier specifies that this is a positional argument
1099 that does not have a command line option associated with it. See the
1100 `Positional Arguments`_ section for more information.
1102 * The **cl::ConsumeAfter** modifier specifies that this option is used to
1103 capture "interpreter style" arguments. See `this section for more
1109 * The **cl::Prefix** modifier specifies that this option prefixes its value.
1110 With 'Prefix' options, the equal sign does not separate the value from the
1111 option name specified. Instead, the value is everything after the prefix,
1112 including any equal sign if present. This is useful for processing odd
1113 arguments like ``-lmalloc`` and ``-L/usr/lib`` in a linker tool or
1114 ``-DNAME=value`` in a compiler tool. Here, the '``l``', '``D``' and '``L``'
1115 options are normal string (or list) options, that have the **cl::Prefix**
1116 modifier added to allow the CommandLine library to recognize them. Note that
1117 **cl::Prefix** options must not have the **cl::ValueDisallowed** modifier
1123 * The **cl::Grouping** modifier is used to implement Unix-style tools (like
1124 ``ls``) that have lots of single letter arguments, but only require a single
1125 dash. For example, the '``ls -labF``' command actually enables four different
1126 options, all of which are single letters. Note that **cl::Grouping** options
1129 The CommandLine library does not restrict how you use the **cl::Prefix** or
1130 **cl::Grouping** modifiers, but it is possible to specify ambiguous argument
1131 settings. Thus, it is possible to have multiple letter options that are prefix
1132 or grouping options, and they will still work as designed.
1134 To do this, the CommandLine library uses a greedy algorithm to parse the input
1135 option into (potentially multiple) prefix and grouping options. The strategy
1136 basically looks like this:
1140 parse(string OrigInput) {
1142 1. string input = OrigInput;
1143 2. if (isOption(input)) return getOption(input).parse(); // Normal option
1144 3. while (!isOption(input) && !input.empty()) input.pop_back(); // Remove the last letter
1145 4. if (input.empty()) return error(); // No matching option
1146 5. if (getOption(input).isPrefix())
1147 return getOption(input).parse(input);
1148 6. while (!input.empty()) { // Must be grouping options
1149 getOption(input).parse();
1150 OrigInput.erase(OrigInput.begin(), OrigInput.begin()+input.length());
1152 while (!isOption(input) && !input.empty()) input.pop_back();
1154 7. if (!OrigInput.empty()) error();
1158 Miscellaneous option modifiers
1159 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1161 The miscellaneous option modifiers are the only flags where you can specify more
1162 than one flag from the set: they are not mutually exclusive. These flags
1163 specify boolean properties that modify the option.
1165 .. _cl::CommaSeparated:
1167 * The **cl::CommaSeparated** modifier indicates that any commas specified for an
1168 option's value should be used to split the value up into multiple values for
1169 the option. For example, these two options are equivalent when
1170 ``cl::CommaSeparated`` is specified: "``-foo=a -foo=b -foo=c``" and
1171 "``-foo=a,b,c``". This option only makes sense to be used in a case where the
1172 option is allowed to accept one or more values (i.e. it is a `cl::list`_
1175 .. _cl::PositionalEatsArgs:
1177 * The **cl::PositionalEatsArgs** modifier (which only applies to positional
1178 arguments, and only makes sense for lists) indicates that positional argument
1179 should consume any strings after it (including strings that start with a "-")
1180 up until another recognized positional argument. For example, if you have two
1181 "eating" positional arguments, "``pos1``" and "``pos2``", the string "``-pos1
1182 -foo -bar baz -pos2 -bork``" would cause the "``-foo -bar -baz``" strings to
1183 be applied to the "``-pos1``" option and the "``-bork``" string to be applied
1184 to the "``-pos2``" option.
1188 * The **cl::Sink** modifier is used to handle unknown options. If there is at
1189 least one option with ``cl::Sink`` modifier specified, the parser passes
1190 unrecognized option strings to it as values instead of signaling an error. As
1191 with ``cl::CommaSeparated``, this modifier only makes sense with a `cl::list`_
1194 So far, these are the only three miscellaneous option modifiers.
1201 Some systems, such as certain variants of Microsoft Windows and some older
1202 Unices have a relatively low limit on command-line length. It is therefore
1203 customary to use the so-called 'response files' to circumvent this
1204 restriction. These files are mentioned on the command-line (using the "@file")
1205 syntax. The program reads these files and inserts the contents into argv,
1206 thereby working around the command-line length limits. Response files are
1207 enabled by an optional fourth argument to `cl::ParseEnvironmentOptions`_ and
1208 `cl::ParseCommandLineOptions`_.
1210 Top-Level Classes and Functions
1211 -------------------------------
1213 Despite all of the built-in flexibility, the CommandLine option library really
1214 only consists of one function `cl::ParseCommandLineOptions`_) and three main
1215 classes: `cl::opt`_, `cl::list`_, and `cl::alias`_. This section describes
1216 these three classes in detail.
1218 .. _cl::ParseCommandLineOptions:
1220 The ``cl::ParseCommandLineOptions`` function
1221 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1223 The ``cl::ParseCommandLineOptions`` function is designed to be called directly
1224 from ``main``, and is used to fill in the values of all of the command line
1225 option variables once ``argc`` and ``argv`` are available.
1227 The ``cl::ParseCommandLineOptions`` function requires two parameters (``argc``
1228 and ``argv``), but may also take an optional third parameter which holds
1229 `additional extra text`_ to emit when the ``-help`` option is invoked, and a
1230 fourth boolean parameter that enables `response files`_.
1232 .. _cl::ParseEnvironmentOptions:
1234 The ``cl::ParseEnvironmentOptions`` function
1235 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1237 The ``cl::ParseEnvironmentOptions`` function has mostly the same effects as
1238 `cl::ParseCommandLineOptions`_, except that it is designed to take values for
1239 options from an environment variable, for those cases in which reading the
1240 command line is not convenient or desired. It fills in the values of all the
1241 command line option variables just like `cl::ParseCommandLineOptions`_ does.
1243 It takes four parameters: the name of the program (since ``argv`` may not be
1244 available, it can't just look in ``argv[0]``), the name of the environment
1245 variable to examine, the optional `additional extra text`_ to emit when the
1246 ``-help`` option is invoked, and the boolean switch that controls whether
1247 `response files`_ should be read.
1249 ``cl::ParseEnvironmentOptions`` will break the environment variable's value up
1250 into words and then process them using `cl::ParseCommandLineOptions`_.
1251 **Note:** Currently ``cl::ParseEnvironmentOptions`` does not support quoting, so
1252 an environment variable containing ``-option "foo bar"`` will be parsed as three
1253 words, ``-option``, ``"foo``, and ``bar"``, which is different from what you
1254 would get from the shell with the same input.
1256 The ``cl::SetVersionPrinter`` function
1257 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1259 The ``cl::SetVersionPrinter`` function is designed to be called directly from
1260 ``main`` and *before* ``cl::ParseCommandLineOptions``. Its use is optional. It
1261 simply arranges for a function to be called in response to the ``--version``
1262 option instead of having the ``CommandLine`` library print out the usual version
1263 string for LLVM. This is useful for programs that are not part of LLVM but wish
1264 to use the ``CommandLine`` facilities. Such programs should just define a small
1265 function that takes no arguments and returns ``void`` and that prints out
1266 whatever version information is appropriate for the program. Pass the address of
1267 that function to ``cl::SetVersionPrinter`` to arrange for it to be called when
1268 the ``--version`` option is given by the user.
1273 The ``cl::opt`` class
1274 ^^^^^^^^^^^^^^^^^^^^^
1276 The ``cl::opt`` class is the class used to represent scalar command line
1277 options, and is the one used most of the time. It is a templated class which
1278 can take up to three arguments (all except for the first have default values
1284 template <class DataType, bool ExternalStorage = false,
1285 class ParserClass = parser<DataType> >
1289 The first template argument specifies what underlying data type the command line
1290 argument is, and is used to select a default parser implementation. The second
1291 template argument is used to specify whether the option should contain the
1292 storage for the option (the default) or whether external storage should be used
1293 to contain the value parsed for the option (see `Internal vs External Storage`_
1294 for more information).
1296 The third template argument specifies which parser to use. The default value
1297 selects an instantiation of the ``parser`` class based on the underlying data
1298 type of the option. In general, this default works well for most applications,
1299 so this option is only used when using a `custom parser`_.
1301 .. _lists of arguments:
1304 The ``cl::list`` class
1305 ^^^^^^^^^^^^^^^^^^^^^^
1307 The ``cl::list`` class is the class used to represent a list of command line
1308 options. It too is a templated class which can take up to three arguments:
1313 template <class DataType, class Storage = bool,
1314 class ParserClass = parser<DataType> >
1318 This class works the exact same as the `cl::opt`_ class, except that the second
1319 argument is the **type** of the external storage, not a boolean value. For this
1320 class, the marker type '``bool``' is used to indicate that internal storage
1325 The ``cl::bits`` class
1326 ^^^^^^^^^^^^^^^^^^^^^^
1328 The ``cl::bits`` class is the class used to represent a list of command line
1329 options in the form of a bit vector. It is also a templated class which can
1330 take up to three arguments:
1335 template <class DataType, class Storage = bool,
1336 class ParserClass = parser<DataType> >
1340 This class works the exact same as the `cl::list`_ class, except that the second
1341 argument must be of **type** ``unsigned`` if external storage is used.
1345 The ``cl::alias`` class
1346 ^^^^^^^^^^^^^^^^^^^^^^^
1348 The ``cl::alias`` class is a nontemplated class that is used to form aliases for
1357 The `cl::aliasopt`_ attribute should be used to specify which option this is an
1358 alias for. Alias arguments default to being `cl::Hidden`_, and use the aliased
1359 options parser to do the conversion from string to data.
1363 The ``cl::extrahelp`` class
1364 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
1366 The ``cl::extrahelp`` class is a nontemplated class that allows extra help text
1367 to be printed out for the ``-help`` option.
1375 To use the extrahelp, simply construct one with a ``const char*`` parameter to
1376 the constructor. The text passed to the constructor will be printed at the
1377 bottom of the help message, verbatim. Note that multiple ``cl::extrahelp``
1378 **can** be used, but this practice is discouraged. If your tool needs to print
1379 additional help information, put all that help into a single ``cl::extrahelp``
1386 cl::extrahelp("\nADDITIONAL HELP:\n\n This is the extra help\n");
1388 .. _different parser:
1389 .. _discussed previously:
1394 Parsers control how the string value taken from the command line is translated
1395 into a typed value, suitable for use in a C++ program. By default, the
1396 CommandLine library uses an instance of ``parser<type>`` if the command line
1397 option specifies that it uses values of type '``type``'. Because of this,
1398 custom option processing is specified with specializations of the '``parser``'
1401 The CommandLine library provides the following builtin parser specializations,
1402 which are sufficient for most applications. It can, however, also be extended to
1403 work with new data types and new ways of interpreting the same data. See the
1404 `Writing a Custom Parser`_ for more details on this type of library extension.
1409 * The generic ``parser<t>`` parser can be used to map strings values to any data
1410 type, through the use of the `cl::values`_ property, which specifies the
1411 mapping information. The most common use of this parser is for parsing enum
1412 values, which allows you to use the CommandLine library for all of the error
1413 checking to make sure that only valid enum values are specified (as opposed to
1414 accepting arbitrary strings). Despite this, however, the generic parser class
1415 can be used for any data type.
1420 * The **parser<bool> specialization** is used to convert boolean strings to a
1421 boolean value. Currently accepted strings are "``true``", "``TRUE``",
1422 "``True``", "``1``", "``false``", "``FALSE``", "``False``", and "``0``".
1424 * The **parser<boolOrDefault> specialization** is used for cases where the value
1425 is boolean, but we also need to know whether the option was specified at all.
1426 boolOrDefault is an enum with 3 values, BOU_UNSET, BOU_TRUE and BOU_FALSE.
1427 This parser accepts the same strings as **``parser<bool>``**.
1431 * The **parser<string> specialization** simply stores the parsed string into the
1432 string value specified. No conversion or modification of the data is
1438 * The **parser<int> specialization** uses the C ``strtol`` function to parse the
1439 string input. As such, it will accept a decimal number (with an optional '+'
1440 or '-' prefix) which must start with a non-zero digit. It accepts octal
1441 numbers, which are identified with a '``0``' prefix digit, and hexadecimal
1442 numbers with a prefix of '``0x``' or '``0X``'.
1448 * The **parser<double>** and **parser<float> specializations** use the standard
1449 C ``strtod`` function to convert floating point strings into floating point
1450 values. As such, a broad range of string formats is supported, including
1451 exponential notation (ex: ``1.7e15``) and properly supports locales.
1453 .. _Extension Guide:
1454 .. _extending the library:
1459 Although the CommandLine library has a lot of functionality built into it
1460 already (as discussed previously), one of its true strengths lie in its
1461 extensibility. This section discusses how the CommandLine library works under
1462 the covers and illustrates how to do some simple, common, extensions.
1466 .. _Writing a Custom Parser:
1468 Writing a custom parser
1469 -----------------------
1471 One of the simplest and most common extensions is the use of a custom parser.
1472 As `discussed previously`_, parsers are the portion of the CommandLine library
1473 that turns string input from the user into a particular parsed data type,
1474 validating the input in the process.
1476 There are two ways to use a new parser:
1478 #. Specialize the `cl::parser`_ template for your custom data type.
1480 This approach has the advantage that users of your custom data type will
1481 automatically use your custom parser whenever they define an option with a
1482 value type of your data type. The disadvantage of this approach is that it
1483 doesn't work if your fundamental data type is something that is already
1486 #. Write an independent class, using it explicitly from options that need it.
1488 This approach works well in situations where you would line to parse an
1489 option using special syntax for a not-very-special data-type. The drawback
1490 of this approach is that users of your parser have to be aware that they are
1491 using your parser instead of the builtin ones.
1493 To guide the discussion, we will discuss a custom parser that accepts file
1494 sizes, specified with an optional unit after the numeric size. For example, we
1495 would like to parse "102kb", "41M", "1G" into the appropriate integer value. In
1496 this case, the underlying data type we want to parse into is '``unsigned``'. We
1497 choose approach #2 above because we don't want to make this the default for all
1498 ``unsigned`` options.
1500 To start out, we declare our new ``FileSizeParser`` class:
1504 struct FileSizeParser : public cl::basic_parser<unsigned> {
1505 // parse - Return true on error.
1506 bool parse(cl::Option &O, const char *ArgName, const std::string &ArgValue,
1510 Our new class inherits from the ``cl::basic_parser`` template class to fill in
1511 the default, boiler plate code for us. We give it the data type that we parse
1512 into, the last argument to the ``parse`` method, so that clients of our custom
1513 parser know what object type to pass in to the parse method. (Here we declare
1514 that we parse into '``unsigned``' variables.)
1516 For most purposes, the only method that must be implemented in a custom parser
1517 is the ``parse`` method. The ``parse`` method is called whenever the option is
1518 invoked, passing in the option itself, the option name, the string to parse, and
1519 a reference to a return value. If the string to parse is not well-formed, the
1520 parser should output an error message and return true. Otherwise it should
1521 return false and set '``Val``' to the parsed value. In our example, we
1522 implement ``parse`` as:
1526 bool FileSizeParser::parse(cl::Option &O, const char *ArgName,
1527 const std::string &Arg, unsigned &Val) {
1528 const char *ArgStart = Arg.c_str();
1531 // Parse integer part, leaving 'End' pointing to the first non-integer char
1532 Val = (unsigned)strtol(ArgStart, &End, 0);
1536 case 0: return false; // No error
1537 case 'i': // Ignore the 'i' in KiB if people use that
1538 case 'b': case 'B': // Ignore B suffix
1541 case 'g': case 'G': Val *= 1024*1024*1024; break;
1542 case 'm': case 'M': Val *= 1024*1024; break;
1543 case 'k': case 'K': Val *= 1024; break;
1546 // Print an error message if unrecognized character!
1547 return O.error("'" + Arg + "' value invalid for file size argument!");
1552 This function implements a very simple parser for the kinds of strings we are
1553 interested in. Although it has some holes (it allows "``123KKK``" for example),
1554 it is good enough for this example. Note that we use the option itself to print
1555 out the error message (the ``error`` method always returns true) in order to get
1556 a nice error message (shown below). Now that we have our parser class, we can
1561 static cl::opt<unsigned, false, FileSizeParser>
1562 MFS("max-file-size", cl::desc("Maximum file size to accept"),
1563 cl::value_desc("size"));
1565 Which adds this to the output of our program:
1570 -help - display available options (-help-hidden for more)
1572 -max-file-size=<size> - Maximum file size to accept
1574 And we can test that our parse works correctly now (the test program just prints
1575 out the max-file-size argument value):
1581 $ ./test -max-file-size=123MB
1583 $ ./test -max-file-size=3G
1585 $ ./test -max-file-size=dog
1586 -max-file-size option: 'dog' value invalid for file size argument!
1588 It looks like it works. The error message that we get is nice and helpful, and
1589 we seem to accept reasonable file sizes. This wraps up the "custom parser"
1592 Exploiting external storage
1593 ---------------------------
1595 Several of the LLVM libraries define static ``cl::opt`` instances that will
1596 automatically be included in any program that links with that library. This is
1597 a feature. However, sometimes it is necessary to know the value of the command
1598 line option outside of the library. In these cases the library does or should
1599 provide an external storage location that is accessible to users of the
1600 library. Examples of this include the ``llvm::DebugFlag`` exported by the
1601 ``lib/Support/Debug.cpp`` file and the ``llvm::TimePassesIsEnabled`` flag
1602 exported by the ``lib/VMCore/PassManager.cpp`` file.
1606 TODO: complete this section
1608 .. _dynamically loaded options:
1610 Dynamically adding command line options
1614 TODO: fill in this section