gperf
gperf Utility
gperf
gperf
gperf Utility
gperf
gperf
User's Guide
for the
GNU GPERF Utility
Douglas C. Schmidt
last updated 14 April 1998
for version 2.7 Copyright (C) 1989-1998 Free Software Foundation, Inc.
Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies.
Permission is granted to copy and distribute modified versions of this
manual under the conditions for verbatim copying, provided also that the
section entitled "GNU gperf General Public License" is included exactly as
in the original, and provided that the entire resulting derived work is
distributed under the terms of a permission notice identical to this one.
Permission is granted to copy and distribute translations of this manual
into another language, under the above conditions for modified versions,
except that the section entitled "GNU gperf General Public License" may be
included in a translation approved by the author instead of in the original
English.
Version 1, February 1989
Copyright (C) 1989 Free Software Foundation, Inc. 675 Mass Ave, Cambridge, MA 02139, USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.
The license agreements of most software companies try to keep users at the mercy of those companies. By contrast, our General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. The General Public License applies to the Free Software Foundation's software and to any other program whose authors commit to using it. You can use it for your programs, too.
When we speak of free software, we are referring to freedom, not price. Specifically, the General Public License is designed to make sure that you have the freedom to give away or sell copies of free software, that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things.
To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it.
For example, if you distribute copies of a such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must tell them their rights.
We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software.
Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations.
The precise terms and conditions for copying, distribution and modification follow.
NO WARRANTY
If you develop a new program, and you want it to be of the greatest possible use to humanity, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms.
To do so, attach the following notices to the program. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the "copyright" line and a pointer to where the full notice is found.
one line to give the program's name and a brief idea of what it does. Copyright (C) 19yy name of author 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 1, 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., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
Also add information on how to contact you by electronic and paper mail.
If the program is interactive, make it output a short notice like this when it starts in an interactive mode:
Gnomovision version 69, Copyright (C) 19yy name of author Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. This is free software, and you are welcome to redistribute it under certain conditions; type `show c' for details.
The hypothetical commands `show w' and `show c' should show the appropriate parts of the General Public License. Of course, the commands you use may be called something other than `show w' and `show c'; they could even be mouse-clicks or menu items--whatever suits your program.
You should also get your employer (if you work as a programmer) or your school, if any, to sign a "copyright disclaimer" for the program, if necessary. Here a sample; alter the names:
Yoyodyne, Inc., hereby disclaims all copyright interest in the program `Gnomovision' (a program to direct compilers to make passes at assemblers) written by James Hacker. signature of Ty Coon, 1 April 1989 Ty Coon, President of Vice
That's all there is to it!
gperf Utilitygperf perfect hash function generator utility was
originally written in GNU C++ by Douglas C. Schmidt. It is now also
available in a highly-portable "old-style" C version. The general
idea for the perfect hash function generator was inspired by Keith
Bostic's algorithm written in C, and distributed to net.sources around
1984. The current program is a heavily modified, enhanced, and extended
implementation of Keith's basic idea, created at the University of
California, Irvine. Bugs, patches, and suggestions should be reported
to <[email protected]> and <[email protected]>.
gperf.
gperf is a perfect hash function generator written in C++. It
transforms an n element user-specified keyword set W into
a perfect hash function F. F uniquely maps keywords in
W onto the range 0..k, where k >= n. If
k = n then F is a minimal perfect hash function.
gperf generates a 0..k element static lookup table and a
pair of C functions. These functions determine whether a given
character string s occurs in W, using at most one probe
into the lookup table.
gperf currently generates the reserved keyword recognizer for
lexical analyzers in several production and research compilers and
language processing tools, including GNU C, GNU C++, GNU Pascal, GNU
Modula 3, and GNU indent. Complete C++ source code for gperf is
available via anonymous ftp from ics.uci.edu and
ftp.santafe.edu. gperf was also distributed along with
the GNU libg++ library for several years. A highly portable,
functionally equivalent K&R C version of gperf is archived in
comp.sources.unix, volume 20. Finally, a paper describing
gperf's design and implementation in greater detail is available
in the Second USENIX C++ Conference proceedings.
gperf
A static search structure is an Abstract Data Type with certain
fundamental operations, e.g., initialize, insert,
and retrieve. Conceptually, all insertions occur before any
retrievals. In practice, gperf generates a static array
containing search set keywords and any associated attributes specified
by the user. Thus, there is essentially no execution-time cost for the
insertions. It is a useful data structure for representing static
search sets. Static search sets occur frequently in software system
applications. Typical static search sets include compiler reserved
words, assembler instruction opcodes, and built-in shell interpreter
commands. Search set members, called keywords, are inserted into
the structure only once, usually during program initialization, and are
not generally modified at run-time.
Numerous static search structure implementations exist, e.g., arrays, linked lists, binary search trees, digital search tries, and hash tables. Different approaches offer trade-offs between space utilization and search time efficiency. For example, an n element sorted array is space efficient, though the average-case time complexity for retrieval operations using binary search is proportional to log n. Conversely, hash table implementations often locate a table entry in constant time, but typically impose additional memory overhead and exhibit poor worst case performance.
Minimal perfect hash functions provide an optimal solution for a particular class of static search sets. A minimal perfect hash function is defined by two properties:
For most applications it is far easier to generate perfect hash
functions than minimal perfect hash functions. Moreover,
non-minimal perfect hash functions frequently execute faster than
minimal ones in practice. This phenomena occurs since searching a
sparse keyword table increases the probability of locating a "null"
entry, thereby reducing string comparisons. gperf's default
behavior generates near-minimal perfect hash functions for
keyword sets. However, gperf provides many options that permit
user control over the degree of minimality and perfection.
Static search sets often exhibit relative stability over time. For
example, Ada's 63 reserved words have remained constant for nearly a
decade. It is therefore frequently worthwhile to expend concerted
effort building an optimal search structure once, if it
subsequently receives heavy use multiple times. gperf removes
the drudgery associated with constructing time- and space-efficient
search structures by hand. It has proven a useful and practical tool
for serious programming projects. Output from gperf is currently
used in several production and research compilers, including GNU C, GNU
C++, GNU Pascal, and GNU Modula 3. The latter two compilers are not yet
part of the official GNU distribution. Each compiler utilizes
gperf to automatically generate static search structures that
efficiently identify their respective reserved keywords.
gperf
The perfect hash function generator gperf reads a set of
"keywords" from a keyfile (or from the standard input by
default). It attempts to derive a perfect hashing function that
recognizes a member of the static keyword set with at most a
single probe into the lookup table. If gperf succeeds in
generating such a function it produces a pair of C source code routines
that perform hashing and table lookup recognition. All generated C code
is directed to the standard output. Command-line options described
below allow you to modify the input and output format to gperf.
By default, gperf attempts to produce time-efficient code, with
less emphasis on efficient space utilization. However, several options
exist that permit trading-off execution time for storage space and vice
versa. In particular, expanding the generated table size produces a
sparse search structure, generally yielding faster searches.
Conversely, you can direct gperf to utilize a C switch
statement scheme that minimizes data space storage size. Furthermore,
using a C switch may actually speed up the keyword retrieval time
somewhat. Actual results depend on your C compiler, of course.
In general, gperf assigns values to the characters it is using
for hashing until some set of values gives each keyword a unique value.
A helpful heuristic is that the larger the hash value range, the easier
it is for gperf to find and generate a perfect hash function.
Experimentation is the key to getting the most from gperf.
gperf
You can control the input keyfile format by varying certain command-line
arguments, in particular the `-t' option. The input's appearance
is similar to GNU utilities flex and bison (or UNIX
utilities lex and yacc). Here's an outline of the general
format:
declarations %% keywords %% functions
Unlike flex or bison, all sections of gperf's input
are optional. The following sections describe the input format for each
section.
struct Declarations and C Code Inclusion
The keyword input file optionally contains a section for including
arbitrary C declarations and definitions, as well as provisions for
providing a user-supplied struct. If the `-t' option
is enabled, you must provide a C struct as the last
component in the declaration section from the keyfile file. The first
field in this struct must be a char * identifier called `name',
although it is possible to modify this field's name with the `-K'
option described below.
Here is simple example, using months of the year and their attributes as input:
struct months { char *name; int number; int days; int leap_days; };
%%
january, 1, 31, 31
february, 2, 28, 29
march, 3, 31, 31
april, 4, 30, 30
may, 5, 31, 31
june, 6, 30, 30
july, 7, 31, 31
august, 8, 31, 31
september, 9, 30, 30
october, 10, 31, 31
november, 11, 30, 30
december, 12, 31, 31
Separating the struct declaration from the list of key words and
other fields are a pair of consecutive percent signs, %%,
appearing left justified in the first column, as in the UNIX utility
lex.
Using a syntax similar to GNU utilities flex and bison, it
is possible to directly include C source text and comments verbatim into
the generated output file. This is accomplished by enclosing the region
inside left-justified surrounding %{, %} pairs. Here is
an input fragment based on the previous example that illustrates this
feature:
%{
#include <assert.h>
/* This section of code is inserted directly into the output. */
int return_month_days (struct months *months, int is_leap_year);
%}
struct months { char *name; int number; int days; int leap_days; };
%%
january, 1, 31, 31
february, 2, 28, 29
march, 3, 31, 31
...
It is possible to omit the declaration section entirely. In this case the keyfile begins directly with the first keyword line, e.g.:
january, 1, 31, 31 february, 2, 28, 29 march, 3, 31, 31 april, 4, 30, 30 ...
The second keyfile format section contains lines of keywords and any associated attributes you might supply. A line beginning with `#' in the first column is considered a comment. Everything following the `#' is ignored, up to and including the following newline.
The first field of each non-comment line is always the key itself. It should be given as a simple name, i.e., without surrounding string quotation marks, and be left-justified flush against the first column. In this context, a "field" is considered to extend up to, but not include, the first blank, comma, or newline. Here is a simple example taken from a partial list of C reserved words:
# These are a few C reserved words, see the c.gperf file
# for a complete list of ANSI C reserved words.
unsigned
sizeof
switch
signed
if
default
for
while
return
Note that unlike flex or bison the first %% marker
may be elided if the declaration section is empty.
Additional fields may optionally follow the leading keyword. Fields
should be separated by commas, and terminate at the end of line. What
these fields mean is entirely up to you; they are used to initialize the
elements of the user-defined struct provided by you in the
declaration section. If the `-t' option is not enabled
these fields are simply ignored. All previous examples except the last
one contain keyword attributes.
The optional third section also corresponds closely with conventions
found in flex and bison. All text in this section,
starting at the final %% and extending to the end of the input
file, is included verbatim into the generated output file. Naturally,
it is your responsibility to ensure that the code contained in this
section is valid C.
gperf
Several options control how the generated C code appears on the standard
output. Two C function are generated. They are called hash and
in_word_set, although you may modify the name for
in_word_set with a command-line option. Both functions require
two arguments, a string, char * str, and a length
parameter, int len. Their default function prototypes are
as follows:
static int hash (char *str, int len); int in_word_set (char *str, int len);
By default, the generated hash function returns an integer value
created by adding len to several user-specified str key
positions indexed into an associated values table stored in a
local static array. The associated values table is constructed
internally by gperf and later output as a static local C array called
hash_table; its meaning and properties are described below.
See section Implementation Details of GNU gperf. The relevant key positions are specified via the
`-k' option when running gperf, as detailed in the Options
section below. See section Options to the gperf Utility.
Two options, `-g' (assume you are compiling with GNU C and its
inline feature) and `-a' (assume ANSI C-style function
prototypes), alter the content of both the generated hash and
in_word_set routines. However, function in_word_set may
be modified more extensively, in response to your option settings. The
options that affect the in_word_set structure are:
struct.
switch statement rather than use a large,
(and potentially sparse) static array. Although the exact time and
space savings of this approach vary according to your C compiler's
degree of optimization, this method often results in smaller and faster
code.
If the `-t' and `-S' options are omitted, the
default action is to generate a char * array containing the keys,
together with additional null strings used for padding the array. By
experimenting with the various input and output options, and timing the
resulting C code, you can determine the best option choices for
different keyword set characteristics.
gperf Utility
There are many options to gperf. They were added to make
the program more convenient for use with real applications. "On-line"
help is readily available via the `-h' option. Here is the complete
list of options.
struct type declaration for generated
code. Any text before a pair of consecutive %% is consider part of the
type declaration. Key words and additional fields may follow this, one
group of fields per line. A set of examples for generating perfect hash
tables and functions for Ada, C, and G++, Pascal, and Modula 2 and 3
reserved words are distributed with this release.
gperf to generate code in the language specified by the
option's argument. Languages handled are currently:
#define const to empty
for compilers which don't know about this keyword.
gperf. It does not do anything.
gperf. It does not do anything.
struct.
Perfect_Hash.
isalnum and isgraph do
not guarantee that a character is in this range. Only an explicit
test like `c >= 'A' && c <= 'Z'' guarantees this.) This was the
default in earlier versions of gperf; now the default is to assume
8-bit characters.
strncmp function to perform
string comparisons. The default action is to use strcmp.
<[email protected]>.
<string.h>, at the
beginning of the code. By default, this is not done; the user must
include this header file himself to allow compilation of the code.
switch statement scheme,
rather than an array lookup table. This can lead to a reduction in both
time and space requirements for some keyfiles. The argument to this
option determines how many switch statements are generated. A
value of 1 generates 1 switch containing all the elements, a
value of 2 generates 2 tables with 1/2 the elements in each
switch, etc. This is useful since many C compilers cannot
correctly generate code for large switch statements. This option
was inspired in part by Keith Bostic's original C program.
gperf. It does not do anything.
gperfstrcmp.
However, using `-l' might greatly increase the size of the
generated C code if the lookup table range is large (which implies that
the switch option `-S' is not enabled), since the length table
contains as many elements as there are entries in the lookup table.
gperf does not backtrack it is possible for it to process
all your input keywords without finding a unique mapping for each word.
However, frequently only a very small number of duplicates occur, and
the majority of keys still require one probe into the table.
gperf treats all these keys as part of
an equivalence class and generates a perfect hash function with multiple
comparisons for duplicate keys. It is up to you to completely
disambiguate the keywords by modifying the generated C code. However,
gperf helps you out by organizing the output.
gperf to work on
keyword sets that it otherwise could not handle.
gperf's
running time at the cost of minimizing generated table-size. The
iteration amount represents the number of times to iterate when
resolving a collision. `0' means iterate by the number of keywords.
This option is probably most useful when used in conjunction with options
`-D' and/or `-S' for large keyword sets.
gperf jumps by random amounts.
gperf's execution time, since collisions will begin
earlier and continue throughout the remainder of keyword processing.
See Cichelli's paper from the January 1980 Communications of the ACM for
details.
gperf has difficultly with a certain keyword set try using
`-r' or `-D'.
gperf's runtime, since it must search through a much larger range
of values. Judicious use of the `-f' option helps alleviate this
overhead, however.
gperf is executing. It is useful both for
maintaining the program and for determining whether a given set of
options is actually speeding up the search for a solution. Some useful
information is dumped at the end of the program when the `-d'
option is enabled.
gperf
The following are some limitations with the current release of
gperf:
gperf utility is tuned to execute quickly, and works quickly
for small to medium size data sets (around 1000 keywords). It is
extremely useful for maintaining perfect hash functions for compiler
keyword sets. Several recent enhancements now enable gperf to
work efficiently on much larger keyword sets (over 15,000 keywords).
When processing large keyword sets it helps greatly to have over 8 megs
of RAM.
However, since gperf does not backtrack no guaranteed solution
occurs on every run. On the other hand, it is usually easy to obtain a
solution by varying the option parameters. In particular, try the
`-r' option, and also try changing the default arguments to the
`-s' and `-j' options. To guarantee a solution, use
the `-D' and `-S' options, although the final results are not
likely to be a perfect hash function anymore! Finally, use the
`-f' option if you want gperf to generate the perfect hash
function fast, with less emphasis on making it minimal.
It should be "relatively" easy to replace the current perfect hash function algorithm with a more exhaustive approach; the perfect hash module is essential independent from other program modules. Additional worthwhile improvements include:
switch
option can minimize the data size, at the expense of slightly longer
lookups (note that the gcc compiler generally produces good code for
switch statements, reducing the need for more complex schemes).
gperf
A paper describing the high-level description of the data structures and
algorithms used to implement gperf will soon be available. This
paper is useful not only from a maintenance and enhancement perspective,
but also because they demonstrate several clever and useful programming
techniques, e.g., `Iteration Number' boolean arrays, double
hashing, a "safe" and efficient method for reading arbitrarily long
input from a file, and a provably optimal algorithm for simultaneously
determining both the minimum and maximum elements in a list.
[1] Chang, C.C.: A Scheme for Constructing Ordered Minimal Perfect Hashing Functions Information Sciences 39(1986), 187-195. [2] Cichelli, Richard J. Author's Response to "On Cichelli's Minimal Perfect Hash Functions Method" Communications of the ACM, 23, 12(December 1980), 729. [3] Cichelli, Richard J. Minimal Perfect Hash Functions Made Simple Communications of the ACM, 23, 1(January 1980), 17-19. [4] Cook, C. R. and Oldehoeft, R.R. A Letter Oriented Minimal Perfect Hashing Function SIGPLAN Notices, 17, 9(September 1982), 18-27.
[5] Cormack, G. V. and Horspool, R. N. S. and Kaiserwerth, M. Practical Perfect Hashing Computer Journal, 28, 1(January 1985), 54-58. [6] Jaeschke, G. Reciprocal Hashing: A Method for Generating Minimal Perfect Hashing Functions Communications of the ACM, 24, 12(December 1981), 829-833.
[7] Jaeschke, G. and Osterburg, G. On Cichelli's Minimal Perfect Hash Functions Method Communications of the ACM, 23, 12(December 1980), 728-729.
[8] Sager, Thomas J. A Polynomial Time Generator for Minimal Perfect Hash Functions Communications of the ACM, 28, 5(December 1985), 523-532
[9] Schmidt, Douglas C. GPERF: A Perfect Hash Function Generator Second USENIX C++ Conference Proceedings, April 1990.
[10] Sebesta, R.W. and Taylor, M.A. Minimal Perfect Hash Functions for Reserved Word Lists SIGPLAN Notices, 20, 12(September 1985), 47-53.
[11] Sprugnoli, R. Perfect Hashing Functions: A Single Probe Retrieving Method for Static Sets Communications of the ACM, 20 11(November 1977), 841-850.
[12] Stallman, Richard M. Using and Porting GNU CC Free Software Foundation, 1988.
[13] Stroustrup, Bjarne The C++ Programming Language. Addison-Wesley, 1986.
[14] Tiemann, Michael D. User's Guide to GNU C++ Free Software Foundation, 1989.
This document was generated on 7 November 1998 using the texi2html translator version 1.52.