Usenet 1994 October

home *** CD-ROM | disk | FTP | other *** search

/ Usenet 1994 October / usenetsourcesnewsgroupsinfomagicoctober1994disk2.iso / unix / volume28 / sortmail / part01 next >

Wrap

Text File | 1994-08-18 | 82.9 KB | 3,118 lines

Newsgroups: comp.sources.unix From: Ed.Falk@Eng.Sun.COM (Ed Falk) Subject: v28i114: sortmail - kill files and more for your mail, Part01/01 Message-id: <1.777230853.7352@gw.home.vix.com> Sender: unix-sources-moderator@gw.home.vix.com Approved: vixie@gw.home.vix.com Submitted-By: Ed.Falk@Eng.Sun.COM (Ed Falk) Posting-Number: Volume 28, Issue 114 Archive-Name: sortmail/part01 Hi; I post this program because I believe it is the best one I've seen (or I wouldn't have written it.) It's written in C rather than shell scripts, and runs quite fast. Overview: sortmail(1) examines incoming mail and sorts it for you based on all or part of the header, or even the entire message. Messages can be filed to your mailbox, to a specific mail folder, junked, forwarded to another address or piped through sh(1). The format of the $HOME/.sortmailrc file is similar to an rn KILL file: set default=+other /MAILER-DAEMON/f:+bounces /falk/t:m /scuba/s:+scuba /marko/f:j /^Precedence: junk/h:+other #! /bin/sh # this is a shell archive, meaning: # 1. Remove everything above the #! /bin/sh line # 2. Save the resulting text in a file. # 3. Execute the file with /bin/sh to create the files: # sortmail.1 # sample.forward # sample.sortmailrc # Makefile # sortmail.h # regex.h # sortmail.c # parse.c # regex.c # This archive created: Wed Aug 17 13:42:30 PDT 1994 by falk # # export PATH; PATH=/bin:$PATH # if test -f sortmail.1 then echo shar: will not over-write existing file sortmail.1 else echo shar: extracting 'sortmail.1', 7787 characters sed 's/^X//' > sortmail.1 <<'SHAR_EOF' X.\" %Z%%M% %I% %E%; Copyright (c) 1990 - Sun Microsystems X.\" X.TH SORTMAIL 1 "21 Apr 1991" X.SH NAME Xsortmail \- classify incoming mail X.SH SYNOPSIS X.B sortmail X[ X.B \-v X] [ X.B \-home X.I path X] [ X.B \-mailbox X.I path X] [ X.B \-mailrc X.I initfile X] [ X.B \-sortmailrc X.I initfile X] X.I username X.SH DESCRIPTION XCreate this \fB.forward\fP file in your home directory: X X "| /\fIpath\fP/sortmail \fIuser\fP" X XWhere "/\fIpath\fP/sortmail" is the full path where you installed Xsortmail, and \fIuser\fP is your own userid. The userid must be Xspecified because when mail arrives \fBsortmail\fP could be run as Xroot, daemon, or any number of other id's. X XOnce your \fB.forward\fP file is set up, X.B sortmail Xwill classify incoming mail according to the patterns in $HOME/.sortmailrc. XYour \fB.sortmailrc\fP file is similar to a news KILL file, but somewhat Xmore powerful. You can discard mail, have it delivered to your mailbox, Xhave it filed into a folder, forward it to another address Xor even pipe it through a shell command. X XWhen \fBsortmail\fP starts up, it first reads your \fB.mailrc\fP file to Xfind the value of mail(1) variables that are also used by \fBsortmail\fP. XIt then reads your \fB.sortmailrc\fP file for additional variable settings, Xif any, and for search patterns. (See below.) X X.SH OPTIONS X.TP 15 X.B \-v XVerbose. A message is printed on /dev/console for every mail message. A Xsecond X.B \-v Xcauses a lot more information to be printed. X.TP 15 X\fB\-home\fP \fIpath\fP XSet user's home directory, overriding the default taken from the user's Xpasswd entry. X.TP 15 X\fB\-mailbox\fP \fIpath\fP XSet the user's system mailbox, overriding the default of X\fB/usr/spool/mail/\fP\fIuser\fP. X.TP 15 X\fB\-mailrc\fP \fIpath\fP XSet the user's .mailrc file, overriding the default of X\fI~user/\fP\fB.mailrc\fP. X.TP 15 X\fB\-sortmailrc\fP \fIpath\fP XSet the user's .sortmailrc file, overriding the default of X\fI~user/\fP\fB.sortmailrc\fP. X X.SH SORTMAILRC FORMAT X XYour .sortmailrc file is a series of lines in the form X X \fBset\fP \fIvariable=value\fP X Xand search patterns take the form X X /\fIregular-expression\fP/\fImodifier\fP:\fIcommand\fP X Xwhere \fIregular-expression\fP is any \fBed(1)\fP-style regular Xexpression, \fImodifier\fP is one of \fBt, f, s, h, a\fP, and X\fIcommand\fP is one of \fBm, j, v, f\fP\ \fIfile\fP, \fB+\fP\fIfile\fP, Xor \fB|\fP\ \fIcommand\fP. X X.SH MODIFIERS X.LP XThese modifiers affect how the regular expression is applied to the Xincoming mail. The default is 's'. X.TP 6 X.B s XTest the "Subject:\ " line of the incoming mail against the regular expression. X.TP 6 X.B t XTest the "To:\ ", "Cc:\ " and "Apparently-To:\ " lines of the incoming Xmail against the regular expression. X.TP 6 X.B f XTest the "From:\ " line of the incoming mail against the regular expression. X.TP 6 X.B h XTest the entire header of the incoming mail against the regular expression. X.TP 6 X.B a XTest the entire incoming mail message against the regular expression. X X.SH COMMANDS X.TP 10 X.B m XSend the message to the user's mailbox. X.TP X\fBm\fP \fIaddress\fP XForward the mail to the specified address. X.TP 10 X.B j XDelete the message ("junk" it.) X.TP 10 X\fBf\fP \fIfolder\fP XSave the message in the given mail folder. \fIfolder\fP may be in Xthe formats ~/\fIpath\fP, /\fIabspath\fP, ~\fIuser/path\fP, or X+\fIname\fP. The latter form expands to ~/\fIfolder/name\fP where X\fIfolder\fP is the value specified for the \fBfolder\fP variable X(default is "\fBfolders\fP".) X.TP 10 X\fB+\fP\fIfolder\fP XShorthand for "\fBf +\fP\fIfolder\fP". X.TP 10 X\fB|\fP \fIcommand\fP XPipe the mail message through the given \fIshell-command\fP. Xsh(1) is used. X X.SH VARIABLES X.LP X\fBsortmail\fP uses the following variables, which may be changed in Xyour \fB.mailrc\fP or \fB.sortmailrc\fP files. Some variables may also Xbe set on the command line. X.TP 10 X.B user XThe user on behalf of whom \fBsortmail\fP is running. This value must Xbe specified on the command line, but may be changed later. It is used Xto determine the user's home directory, among other things. X.TP 10 X.B home XThe user's home directory. Used to find intialization files and the Xuser's \fIfolders\fP directory. Default is ~\fIuser\fP. X.TP 10 X.B mailbox XThe user's mail box. Default is \fB/usr/spool/mail/\fP\fIuser\fP. X.TP 10 X.B mailrc XFull path of the user's \fB.mailrc\fP file. There's no real point in Xchanging this. X.TP 10 X.B sortmailrc XFull path of the user's \fB.sortmailrc\fP file. There's no real point in Xchanging this. X.TP 10 X.B mbox XFull path of the user's \fBmbox\fP file. Not currently used by \fBsortmail\fP. X.TP 10 X.B folder XThe user's \fImail folders\fP directory. Mail folders are identified by the Xleading '+' in their name, and are stored in ~\fIuser/folder\fP/. XDefault is "\fBfolders\fP". X.TP 10 X.B default XName of the file into which unclassifiable mail is sent. The default Xis \fImailbox\fP. Another reasonable value might be "\fB+other\fP". X.TP 10 X.B vacation XIf set, mail to \fIuser\fP will also be piped through /usr/ucb/vacation. X.TP 10 X.B logfilename XMessages printed with -v are sent to the specified log file. Default is X/dev/console, or (if /dev/console cannot be opened) stderr. X X.SH EXAMPLES X.LP XHere is a sample \fB.sortmailrc\fP file: X.LP X.RS X.nf X.ft B Xset default=+other X/MAILER-DAEMON/f:+bounces X/falk/t:m X/bldg8/t:m X/joe/f:m X/scuba@sun/t:+scuba X/scuba/s:+scuba X/mwicks/f:j X/homework/t:m falk@kestrel X/jim@apple/f:| /home/falk/bin/fixjim X/^Precedence: junk/h:+other X.ft X.fi X.RE X.LP XIn this example, the folder directory and other variables have whatever Xvalues were specified in \fB.mailrc\fP. Unclassifiable mail is sent to Xthe folder "+other". Mail from "MAILER-DAEMON" is sent to the folder X"+bounces". Mail to "falk" or "bldg8" or from my friend joe is sent Xdirectly to my mailbox. Mail to the scuba club or with "scuba" in the Xsubject line is sent to the "+scuba" folder. Mail from mwicks is Xthrown away without my ever seeing it. Mail to the "homework" alias Xis forwarded to my account on another machine. XMail from my friend jim, who Xformats his mail in a funny way is passed through a shell script which Xcleans up his messages and appends them to my mailbox. Mail messages Xwith "^Precedence:\ junk" anywhere in the header is filed in +other. X XNote that the patterns are applied in the order given; it is important, Xfor example, that the "MAILER-DAEMON" pattern precede the "falk" pattern Xso that mail from MAILER-DAEMON is filed in +bounces even if directed Xto me personally. On the other hand, mail from mwicks will be placed Xin my mailbox if addressed to me personally, otherwise it will Xbe junked. X X.SH NOTES XRemember that sortmail can be executed under any userid, depending on Xwho sent the mail, and whether or not it came from the local machine. XBecause of this, you cannot depend on any user environment to be available, Xespecially environment variables and path. All filenames and program Xnames should be specified as full paths, except that "~", "~user" and X"+folder" forms are understood. The permissions of sortmail and every Xdirectory along its path should be such that any user can run it. X XIf you pipe incoming mail through a program, that program should not Xgenerate any output to stdout or stderr whatsoever. If it does, that Xoutput will be sent back to the originator of the mail as if the mail Xhad bounced. X X.SH SEE ALSO Xed(1), mail(1) X.SH COPYRIGHT X Copyright (c) 1990 by Sun Microsystems, Inc. X Edward A. Falk (falk@sun.com) X XPermission to use, copy, modify, and distribute this software and its Xdocumentation for any purpose and without fee is hereby granted, Xprovided that the above copyright notice appear in all copies and that Xboth that copyright notice and this permission notice appear in Xsupporting documentation. X Xregex.[ch] is covered by the GNU copyleft. X SHAR_EOF len=`wc -c < sortmail.1` if test $len != 7787 ; then echo error: sortmail.1 was $len bytes long, should have been 7787 fi fi # end of overwriting check if test -f sample.forward then echo shar: will not over-write existing file sample.forward else echo shar: extracting 'sample.forward', 41 characters sed 's/^X//' > sample.forward <<'SHAR_EOF' X"| /home/joeshmoe/bin/sortmail joeshmoe" SHAR_EOF len=`wc -c < sample.forward` if test $len != 41 ; then echo error: sample.forward was $len bytes long, should have been 41 fi fi # end of overwriting check if test -f sample.sortmailrc then echo shar: will not over-write existing file sample.sortmailrc else echo shar: extracting 'sample.sortmailrc', 1067 characters sed 's/^X//' > sample.sortmailrc <<'SHAR_EOF' X# This is a sample .sortmailrc file. X# X# The destination for unclassifiable mail is the +other folder. X# The directory containg the folders is Mail X# vacation mode is not currently enabled X# X# all mail from Mailer-Daemon goes to the +bounces folder, otherwise X# all mail to me (joeshmoe) or to anybody in mail building, or from X# my boss fred goes directly to me. X# all mail from mwicks gets ignored. X# all mail from my friend jim who never formats his mail goes through X# my "fixjim" filter. X# all mail to the scuba club or with "scuba" in the subject goes into X# my +scuba folder. X# all mail to the testing alias is forwarded to my my account on another machine X# all mail with "Precedence: junk" in the header goes to +other. X# all unclassified mail goes to +other. X# Xset default=+other Xset folder='Mail' X#set vacation X/Mailer-Daemon/f:+bounces X/MAILER-DAEMON/f:+bounces X/joeshmoe/t:m X/Joeshmoe/t:m X/bldg18/t:m X/fred/f:m X/mwicks/f:j X/jim@apple/f:| /home/joeshmoe/bin/fixjim X/scuba@sun/t:+scuba X/scuba/s:+scuba X/testing/t:m joeshmoe@lab X/^Precedence: junk/h:+other SHAR_EOF len=`wc -c < sample.sortmailrc` if test $len != 1067 ; then echo error: sample.sortmailrc was $len bytes long, should have been 1067 fi fi # end of overwriting check if test -f Makefile then echo shar: will not over-write existing file Makefile else echo shar: extracting 'Makefile', 460 characters sed 's/^X//' > Makefile <<'SHAR_EOF' X X#DBG = -g -DDEBUG XDBG = -O X X# SunOS 4.x X#CFLAGS = ${DBG} X#LDFLAGS = -n -Bdynamic -s X X# SunOS 5.x XCFLAGS = ${DBG} -DSVr4 XLDFLAGS = X X#CC = gcc X XSRCS = sortmail.c parse.c regex.c X XHDRS = sortmail.h regex.h X XOBJS = $(SRCS:.c=.o) X XDOC = sortmail.1 sample.forward sample.sortmailrc X Xsortmail: $(OBJS) X $(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(OBJS) $(LIB) X Xarchive: Makefile $(HDRS) $(SRCS) $(DOC) X shar $(DOC) Makefile $(HDRS) $(SRCS) > archive X Xclean: X -rm *.o archive SHAR_EOF len=`wc -c < Makefile` if test $len != 460 ; then echo error: Makefile was $len bytes long, should have been 460 fi fi # end of overwriting check if test -f sortmail.h then echo shar: will not over-write existing file sortmail.h else echo shar: extracting 'sortmail.h', 1038 characters sed 's/^X//' > sortmail.h <<'SHAR_EOF' X X X /* sortmail variables */ X Xextern char *user ; /* userid */ Xextern char *home ; /* user's $HOME */ Xextern char *mailbox ; /* $MAIL, default = /var/spool/mail/$USER */ Xextern char *mailrc ; /* $MAILRC, default = $HOME/.mailrc */ Xextern char *sortmailrc ; /* initfile, default = $HOME/.sortmailrc */ Xextern char *mbox ; /* MBOX, default = $HOME/mbox */ Xextern char *deflt ; /* where unclasifiable mail goes */ Xextern char *folder ; /* folder, default = $HOME/folders */ Xextern char *vacation ; /* pipe through vacation(1) */ Xextern int verbose ; Xextern char *logfilename ; Xextern FILE *logfile ; X X#define SPOOLDIR "/usr/spool/mail/" X#define MBOX "mbox" X#define FOLDER "folders" X Xtypedef struct patinfo { X struct patinfo *next ; X char *pattern ; /* pattern to match */ X enum {P_SUBJECT, P_TO, P_FROM, P_HEADER, P_ALL} flags ; X enum {C_MAIL, C_FILE, C_PIPE, C_JUNK} command ; X char *dest ; /* destination, if any */ X } PatInfo ; X Xextern PatInfo *patterns, *lastpat ; X X X#define strmatch(a,b) (!strncmp((a),(b),sizeof(b)-1)) X SHAR_EOF len=`wc -c < sortmail.h` if test $len != 1038 ; then echo error: sortmail.h was $len bytes long, should have been 1038 fi fi # end of overwriting check if test -f regex.h then echo shar: will not over-write existing file regex.h else echo shar: extracting 'regex.h', 8080 characters sed 's/^X//' > regex.h <<'SHAR_EOF' X/* Definitions for data structures callers pass the regex library. X Copyright (C) 1985 Free Software Foundation, Inc. X X This program is free software; you can redistribute it and/or modify X it under the terms of the GNU General Public License as published by X the Free Software Foundation; either version 1, or (at your option) X any later version. X X This program is distributed in the hope that it will be useful, X but WITHOUT ANY WARRANTY; without even the implied warranty of X MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the X GNU General Public License for more details. X X You should have received a copy of the GNU General Public License X along with this program; if not, write to the Free Software X Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. X XIn other words, you are welcome to use, share and improve this program. XYou are forbidden to forbid anyone else to use, share and improve Xwhat you give them. Help stamp out software-hoarding! */ X X X/* Define number of parens for which we record the beginnings and ends. X This affects how much space the `struct re_registers' type takes up. */ X#ifndef RE_NREGS X#define RE_NREGS 10 X#endif X X/* These bits are used in the obscure_syntax variable to choose among X alternative regexp syntaxes. */ X X/* 1 means plain parentheses serve as grouping, and backslash X parentheses are needed for literal searching. X 0 means backslash-parentheses are grouping, and plain parentheses X are for literal searching. */ X#define RE_NO_BK_PARENS 1 X X/* 1 means plain | serves as the "or"-operator, and \| is a literal. X 0 means \| serves as the "or"-operator, and | is a literal. */ X#define RE_NO_BK_VBAR 2 X X/* 0 means plain + or ? serves as an operator, and \+, \? are literals. X 1 means \+, \? are operators and plain +, ? are literals. */ X#define RE_BK_PLUS_QM 4 X X/* 1 means | binds tighter than ^ or $. X 0 means the contrary. */ X#define RE_TIGHT_VBAR 8 X X/* 1 means treat \n as an _OR operator X 0 means treat it as a normal character */ X#define RE_NEWLINE_OR 16 X X/* 0 means that a special characters (such as *, ^, and $) always have X their special meaning regardless of the surrounding context. X 1 means that special characters may act as normal characters in some X contexts. Specifically, this applies to: X ^ - only special at the beginning, or after ( or | X $ - only special at the end, or before ) or | X *, +, ? - only special when not after the beginning, (, or | */ X#define RE_CONTEXT_INDEP_OPS 32 X X/* Now define combinations of bits for the standard possibilities. */ X#define RE_SYNTAX_AWK (RE_NO_BK_PARENS | RE_NO_BK_VBAR | RE_CONTEXT_INDEP_OPS) X#define RE_SYNTAX_EGREP (RE_SYNTAX_AWK | RE_NEWLINE_OR) X#define RE_SYNTAX_GREP (RE_BK_PLUS_QM | RE_NEWLINE_OR) X#define RE_SYNTAX_EMACS 0 X X/* This data structure is used to represent a compiled pattern. */ X Xstruct re_pattern_buffer X { X char *buffer; /* Space holding the compiled pattern commands. */ X int allocated; /* Size of space that buffer points to */ X int used; /* Length of portion of buffer actually occupied */ X char *fastmap; /* Pointer to fastmap, if any, or zero if none. */ X /* re_search uses the fastmap, if there is one, X to skip quickly over totally implausible characters */ X char *translate; /* Translate table to apply to all characters before comparing. X Or zero for no translation. X The translation is applied to a pattern when it is compiled X and to data when it is matched. */ X char fastmap_accurate; X /* Set to zero when a new pattern is stored, X set to one when the fastmap is updated from it. */ X char can_be_null; /* Set to one by compiling fastmap X if this pattern might match the null string. X It does not necessarily match the null string X in that case, but if this is zero, it cannot. X 2 as value means can match null string X but at end of range or before a character X listed in the fastmap. */ X }; X X/* Structure to store "register" contents data in. X X Pass the address of such a structure as an argument to re_match, etc., X if you want this information back. X X start[i] and end[i] record the string matched by $ ... $ grouping i, X for i from 1 to RE_NREGS - 1. X start[0] and end[0] record the entire string matched. */ X Xstruct re_registers X { X int start[RE_NREGS]; X int end[RE_NREGS]; X }; X X/* These are the command codes that appear in compiled regular expressions, one per byte. X Some command codes are followed by argument bytes. X A command code can specify any interpretation whatever for its arguments. X Zero-bytes may appear in the compiled regular expression. */ X Xenum regexpcode X { X unused, X exactn, /* followed by one byte giving n, and then by n literal bytes */ X begline, /* fails unless at beginning of line */ X endline, /* fails unless at end of line */ X jump, /* followed by two bytes giving relative address to jump to */ X on_failure_jump, /* followed by two bytes giving relative address of place X to resume at in case of failure. */ X finalize_jump, /* Throw away latest failure point and then jump to address. */ X maybe_finalize_jump, /* Like jump but finalize if safe to do so. X This is used to jump back to the beginning X of a repeat. If the command that follows X this jump is clearly incompatible with the X one at the beginning of the repeat, such that X we can be sure that there is no use backtracking X out of repetitions already completed, X then we finalize. */ X dummy_failure_jump, /* jump, and push a dummy failure point. X This failure point will be thrown away X if an attempt is made to use it for a failure. X A + construct makes this before the first repeat. */ X anychar, /* matches any one character */ X charset, /* matches any one char belonging to specified set. X First following byte is # bitmap bytes. X Then come bytes for a bit-map saying which chars are in. X Bits in each byte are ordered low-bit-first. X A character is in the set if its bit is 1. X A character too large to have a bit in the map X is automatically not in the set */ X charset_not, /* similar but match any character that is NOT one of those specified */ X start_memory, /* starts remembering the text that is matched X and stores it in a memory register. X followed by one byte containing the register number. X Register numbers must be in the range 0 through NREGS. */ X stop_memory, /* stops remembering the text that is matched X and stores it in a memory register. X followed by one byte containing the register number. X Register numbers must be in the range 0 through NREGS. */ X duplicate, /* match a duplicate of something remembered. X Followed by one byte containing the index of the memory register. */ X before_dot, /* Succeeds if before dot */ X at_dot, /* Succeeds if at dot */ X after_dot, /* Succeeds if after dot */ X begbuf, /* Succeeds if at beginning of buffer */ X endbuf, /* Succeeds if at end of buffer */ X wordchar, /* Matches any word-constituent character */ X notwordchar, /* Matches any char that is not a word-constituent */ X wordbeg, /* Succeeds if at word beginning */ X wordend, /* Succeeds if at word end */ X wordbound, /* Succeeds if at a word boundary */ X notwordbound, /* Succeeds if not at a word boundary */ X syntaxspec, /* Matches any character whose syntax is specified. X followed by a byte which contains a syntax code, Sword or such like */ X notsyntaxspec /* Matches any character whose syntax differs from the specified. */ X }; X Xextern char *re_compile_pattern (); X/* Is this really advertised? */ Xextern void re_compile_fastmap (); Xextern int re_search (), re_search_2 (); Xextern int re_match (), re_match_2 (); X X/* 4.2 bsd compatibility (yuck) */ Xextern char *re_comp (); Xextern int re_exec (); X X#ifdef SYNTAX_TABLE Xextern char *re_syntax_table; X#endif SHAR_EOF len=`wc -c < regex.h` if test $len != 8080 ; then echo error: regex.h was $len bytes long, should have been 8080 fi fi # end of overwriting check if test -f sortmail.c then echo shar: will not over-write existing file sortmail.c else echo shar: extracting 'sortmail.c', 6369 characters sed 's/^X//' > sortmail.c <<'SHAR_EOF' X X X#include <stdio.h> X#include <varargs.h> X#include <errno.h> X#include <alloca.h> X#include <sys/param.h> /* defines MAXPATHLEN */ X X#include "sortmail.h" X#include "regex.h" /* use gnu regex */ X X FILE *logfile ; X Xstatic int XReCompile(pat) X char *pat ; X{ X char *err ; X if( (err = re_comp(pat)) != NULL ) { X fprintf(logfile, "sortmail: pattern %s: %s\n", pat, err) ; X return 1 ; X } X return 0 ; X} X Xextern int sys_nerr ; Xextern char *sys_errlist[] ; X X X X char *user = NULL ; X char *home = NULL ; X char *mailbox = NULL ; X char *mailrc = NULL ; X char *sortmailrc = NULL ; X char *mbox = NULL ; X char *deflt = NULL ; X char *vacation = NULL ; X char *folder = NULL ; X char *logfilename = NULL ; X int verbose = 0 ; X X PatInfo *patterns = NULL, *lastpat = NULL ; X Xstatic char tmpfilename[256] ; X Xstatic void dispose() ; Xstatic void append() ; Xstatic void put_pipe() ; Xstatic void forward() ; X Xstatic int Xargcheck(str, argv, nargs, argc) X char *str, *argv ; X int nargs ; X int *argc ; X{ X if( strcmp(str, argv) != 0 ) X return 0 ; X if( (*argc -= nargs) <= 0 ) X die("sortmail: %s option requires %d arguments\n", str, nargs) ; X return 1 ; X} X X X Xmain(argc,argv) X int argc ; X char **argv ; X{ X FILE *tmpfile ; X char line[1024] ; X int n ; X int header ; X PatInfo *pat ; X char *err ; X int firstline ; X int cont = 0 ; X int foundit = 0 ; X X logfile = fopen("/dev/console","w") ; X if( logfile == NULL ) X logfile = stderr ; X X while( --argc > 0 ) X { X ++argv ; X if( argcheck("-v", *argv, 0, &argc) ) X ++verbose ; X else if( argcheck("-home", *argv, 1, &argc) ) X home = *++argv ; X else if( argcheck("-mailbox", *argv, 1, &argc) ) X mailbox = *++argv ; X else if( argcheck("-mailrc", *argv, 1, &argc) ) X mailrc = *++argv ; X else if( argcheck("-sortmailrc", *argv, 1, &argc) ) X sortmailrc = *++argv ; X else X user = *argv ; X } X X#ifdef DEBUG X verbose = 1 ; X#endif /* DEBUG */ X X read_initfiles() ; X X sprintf(tmpfilename, "/tmp/sortmail%d", getpid()) ; X if( (tmpfile = fopen(tmpfilename, "w+")) == NULL ) X { X fprintf(stderr, "sortmail: can't open %s, mail to %s may be lost\n", X tmpfile, user) ; X fprintf(logfile, "sortmail: can't open %s, %s\n", X tmpfile, sys_errlist[errno]) ; X append(stdin, mailbox) ; /* try to rescue situtation */ X exit(1) ; X } X X firstline = 1 ; X while( fgets(line, sizeof(line), stdin) != NULL ) { X if( !firstline && strmatch(line, "From ") ) X putc('>', tmpfile) ; X fputs(line, tmpfile) ; X firstline = 0 ; X } X#ifdef DEBUG X fprintf(logfile, "got a message in %s\n", tmpfilename) ; X#endif /* DEBUG */ X X for( pat = patterns; pat != NULL && !foundit ; pat = pat->next ) X { X if( !ReCompile(pat->pattern) ) X { X rewind(tmpfile) ; X while( fgets(line, sizeof(line), tmpfile) != NULL ) { X if( line[0] == '\n' && pat->flags != P_ALL ) X break ; X if( pat->flags == P_HEADER || X pat->flags == P_SUBJECT && strmatch(line, "Subject: ") || X pat->flags == P_FROM && strmatch(line, "From: ") || X pat->flags == P_TO && X (strmatch(line, "To: ") || strmatch(line, "Cc: ") || X strmatch(line, "Apparently-To: ") ) || X isspace(line[0]) && cont || X pat->flags == P_ALL ) X { X if( re_exec(line) ) X { X dispose(pat,tmpfile) ; X foundit = 1 ; X break ; X } X else X /* if next line is a continuation line, test it too */ X cont = 1 ; X } X else X cont = 0 ; X } X } X } X X if( vacation != NULL ) X { X rewind(tmpfile) ; X if( !ReCompile(user) ) X { X while( fgets(line, sizeof(line), tmpfile) != NULL ) { X if( line[0] == '\n' ) X break ; X if( strmatch(line, "To: ") || strmatch(line, "Cc: ") || X strmatch(line, "Apparently-To: ") ) X { X if( re_exec(line) ) X { X sprintf(line, "/usr/ucb/vacation %s", user) ; X put_pipe(tmpfile, line) ; X if( verbose ) X fprintf(logfile, "message piped through %s\n", line) ; X break ; X } X } X } X } X } X X unlink(tmpfilename) ; X X if( !foundit ) X append(tmpfile, deflt != NULL ? deflt : mailbox) ; X exit(0) ; X} X X X/*VARARGS0*/ Xdie(va_alist) X va_dcl X{ X va_list args ; X char *fmt ; X X va_start(args) ; X fprintf(logfile, "sortmail: ") ; X fmt = va_arg(args, char *); X vfprintf(logfile, fmt, args); X va_end(args) ; X exit(1) ; X} X X X /* This function deals with the file described by 'file' in X * the manner described by "pat". X */ X Xstatic void Xdispose(pat, file) Xregister PatInfo *pat ; Xregister FILE *file ; X{ X char outfile[MAXPATHLEN] ; X char *ptr ; X struct passwd *passwd ; X X switch( pat->command ) { X case C_MAIL: /* append to user's mail file */ X if( pat->dest == NULL || pat->dest[0] == '\0' ) X append(file, mailbox) ; X else X forward(file, pat->dest) ; X break ; X X case C_FILE: /* append to named file. */ X append(file, pat->dest) ; X break ; X X case C_PIPE: /* pipe through named command */ X put_pipe(file, pat->dest) ; X if( verbose ) X fprintf(logfile, "message piped through %s\n", pat->dest) ; X break ; X X case C_JUNK: /* throw it away */ X if( verbose ) X fprintf(logfile, "message deleted\n") ; X break ; X } X} X X Xstatic void Xappend(file, outfile) X FILE *file ; X char *outfile ; X{ X FILE *ofile ; X char line[1024] ; X int i ; X X filename_expand(line, outfile) ; X if( verbose ) X fprintf(logfile, "message filed to %s\n", line) ; X X rewind(file) ; X X if( (ofile = fopen(line, "a")) == NULL ) X { X fprintf(logfile, "sortmail: can't open %s, %s\n", X line,sys_errlist[errno]); X if( strcmp(outfile, mailbox) == 0 ) X fprintf(stderr, X "sortmail: Cannot open %s, %s. Mail to %s may be lost\n", X mailbox, sys_errlist[errno], user) ; X else X append(file, mailbox) ; X } X else X { X while( fgets(line, sizeof(line), file) != NULL ) X fputs(line, ofile) ; X fputs("\n", ofile) ; X fclose(ofile) ; X } X} X X Xstatic void Xput_pipe(file, dest) X FILE *file ; X char *dest ; X{ X FILE *ofile ; X char line[1024] ; X X if( (ofile = popen(dest, "w")) == NULL ) X { X fprintf(logfile, "sortmail: cannot open pipe to command %s\n", X dest ) ; X append(file, mailbox) ; X } X else X { X rewind(file) ; X while( fgets(line, sizeof(line), file) != NULL ) X fputs(line, ofile) ; X pclose(ofile) ; X } X} X X Xstatic void Xforward(file, dest) X FILE *file ; X char *dest ; X{ X char cmd[1024] ; X X sprintf(cmd, "/usr/lib/sendmail %s", dest) ; X put_pipe(file, cmd) ; X if( verbose ) X fprintf(logfile, "message mailed to %s\n", dest) ; X} SHAR_EOF len=`wc -c < sortmail.c` if test $len != 6369 ; then echo error: sortmail.c was $len bytes long, should have been 6369 fi fi # end of overwriting check if test -f parse.c then echo shar: will not over-write existing file parse.c else echo shar: extracting 'parse.c', 7864 characters sed 's/^X//' > parse.c <<'SHAR_EOF' X X X#include <stdio.h> X#include <stdlib.h> /* defines getenv(3) */ X#include <pwd.h> X#include <ctype.h> X#include <string.h> X#include <sys/param.h> X X#include "sortmail.h" X X#define MAXLINE 256 X X#ifdef DEBUG X#define register X#endif X Xstatic char spooldir[] = SPOOLDIR ; X Xstatic void read_initfile() ; Xstatic void parse_var() ; Xstatic void set_var() ; Xstatic void add_pat() ; Xstatic void get_dest() ; X X#define NN(str) ((str)==NULL ? "-none-" : (str)) X Xread_initfiles() X{ X struct passwd *passwd ; X X /* first, try to set up some of the defaults */ X X if( user == NULL ) { X if( (user = (char *) malloc(L_cuserid)) == NULL ) X die("out of memory") ; X if( cuserid(user) == NULL ) X die("can't determine username") ; X } X X if( home == NULL ) { X if( (passwd = getpwnam(user)) != NULL ) X home = strdup(passwd->pw_dir) ; X else if( (home = getenv("HOME")) == NULL ) X die("can't determine %s's directory",user) ; X } X X if( mailbox == NULL ) { X mailbox = (char *) malloc(sizeof(spooldir) + strlen(user)) ; X strcpy(mailbox, spooldir) ; X strcat(mailbox, user) ; X } X X if( mailrc == NULL ) { X mailrc = (char *) malloc(strlen(home) + sizeof("/.mailrc") ) ; X strcpy(mailrc, home) ; X strcat(mailrc, "/.mailrc") ; X } X X if( sortmailrc == NULL ) { X sortmailrc = (char *) malloc(strlen(home) + sizeof("/.sortmailrc") ) ; X strcpy(sortmailrc, home) ; X strcat(sortmailrc, "/.sortmailrc") ; X } X X if( mbox == NULL ) X mbox = MBOX ; X X if( folder == NULL ) X folder = FOLDER ; X X read_initfile(mailrc) ; X read_initfile(sortmailrc) ; X if( verbose >= 2 ) { X fprintf(logfile,"$USER = %s\n", NN(user)) ; X fprintf(logfile,"$HOME = %s\n", NN(home)) ; X fprintf(logfile,"mailbox = %s\n", NN(mailbox)) ; X fprintf(logfile,"mailrc = %s\n", NN(mailrc)) ; X fprintf(logfile,"sortmailrc = %s\n", NN(sortmailrc)) ; X fprintf(logfile,"mbox = %s\n", NN(mbox)) ; X fprintf(logfile,"deflt = %s\n", NN(deflt)) ; X fprintf(logfile,"vacation = %s\n", NN(vacation)) ; X fprintf(logfile,"folder = %s\n", NN(folder)) ; X } X} X X Xstatic void Xread_initfile(file) X char *file ; X{ Xregister FILE *ifile ; X char line[MAXLINE] ; Xregister char *ptr ; X X if( (ifile = fopen(file, "r")) != NULL ) { X while( fgets(line, MAXLINE, ifile) != NULL ) { X if( line[0] != '#' ) X { X for( ptr = line; isspace(*ptr); ++ptr ); X X if( strmatch(ptr,"set") ) /* variable */ X parse_var(ptr) ; X else if( *ptr == '/' ) /* pattern */ X add_pat(ptr) ; X /* else ignore it */ X } X } X fclose(ifile) ; X } X} X X Xstatic void Xparse_var(ptr) Xregister char *ptr ; X{ X ptr += 3 ; /* skip "set" */ X if( !isspace(*ptr) ) X return ; /* something was wrong */ X X for(; isspace(*ptr); ++ptr ); X X if( strmatch(ptr,"user") ) X set_var(&user, ptr+4) ; X else if( strmatch(ptr,"home") ) X set_var(&home, ptr+4) ; X else if( strmatch(ptr,"mailbox") ) X set_var(&mailbox, ptr+7) ; X else if( strmatch(ptr,"mailrc") ) X set_var(&mailrc, ptr+6) ; X else if( strmatch(ptr,"sortmailrc") ) X set_var(&sortmailrc, ptr+10) ; X else if( strmatch(ptr,"mbox") ) X set_var(&mbox, ptr+4) ; X else if( strmatch(ptr,"folder") ) X set_var(&folder, ptr+6) ; X else if( strmatch(ptr,"default") ) X set_var(&deflt, ptr+7) ; X else if( strmatch(ptr,"vacation") ) X set_var(&vacation, ptr+8) ; X else if( strmatch(ptr,"logfile") ) { X set_var(&logfilename, ptr+7) ; X if( (logfile = fopen(logfilename, "a")) == NULL ) X logfile = stderr ; X } X /* else ignore it */ X} X Xstatic void Xset_var(var, ptr) Xregister char **var ; Xregister char *ptr ; X{ X for(; isspace(*ptr); ++ptr ); /* consume blanks before '=' */ X X if( *ptr != '=' ) X { X *var = "" ; X return ; X } X X for(++ptr; isspace(*ptr); ++ptr ); /* consume blanks after '=' */ X X if( *ptr == '\'' ) { X ++ptr ; X *var = strdup(ptr) ; X for(ptr=*var; *ptr != '\'' && *ptr != '\n' && *ptr != '\0'; ++ptr); X *ptr = '\0' ; X } X X else if( *ptr == '"' ) { X ++ptr ; X *var = strdup(ptr) ; X for(ptr=*var; *ptr != '"' && *ptr != '\n' && *ptr != '\0'; ++ptr); X *ptr = '\0' ; X } X X else { X *var = strdup(ptr) ; X for(ptr=*var; *ptr != '\n' && *ptr != '\0'; ++ptr); X *ptr = '\0' ; X } X} X X X X /* try to make sense out of patterns in the form X /regular expression/modifs:command X */ X Xstatic void Xadd_pat(ptr) Xregister char *ptr ; X{ X char buffer[1024] ; /* KLUDGE! should be unlimited */ Xregister char *optr = buffer ; Xregister PatInfo *pat ; Xregister int done, done2 ; X X ++ptr ; /* skip leading '/' */ X X for(done=0;!done;) X switch( *ptr ) { X case '\0': X fprintf(logfile, "sortmail: missing '/' in pattern\n") ; X return ; X case '\\': /* '\' quotes anything */ X *optr++ = *ptr++ ; X *optr++ = *ptr++ ; X break ; X case '/': /* '/' terminates */ X ++ptr ; X done = 1 ; X break ; X case '[': /* '[' starts a set */ X *optr++ = *ptr++ ; X for(done2=0; !done2;) X switch(*ptr) { X case '\0': /* premature EOS, reject */ X fprintf(logfile, "sortmail: missing ']' in pattern\n") ; X return ; X case ']': /* ']' terminates a set */ X done2=1 ; X *optr++ = *ptr++ ; X break ; X default: X *optr++ = *ptr++ ; X break ; X } X break ; X default: X *optr++ = *ptr++ ; X } X *optr++ = '\0' ; X X pat = (PatInfo *) malloc(sizeof(PatInfo)) ; X pat->next = NULL ; X if( patterns == NULL ) X patterns = pat ; X else X lastpat->next = pat ; X lastpat = pat ; X X pat->pattern = strdup(buffer) ; X pat->flags = P_SUBJECT ; X pat->command = C_MAIL ; X pat->dest = NULL ; X X /* parse modifiers part */ X X for(done=0;!done;) X switch( *ptr ) { X case '\0': done = 1 ; break ; X case ':': ++ptr ; done = 1 ; break ; X case 't': pat->flags = P_TO ; ++ptr ; break ; X case 'f': pat->flags = P_FROM ; ++ptr ; break ; X case 's': pat->flags = P_SUBJECT ; ++ptr ; break ; X case 'h': pat->flags = P_HEADER ; ++ptr ; break ; X case 'a': pat->flags = P_ALL ; ++ptr ; break ; X default: ++ptr ; break ; X } X X while( isspace(*ptr) ) X ++ptr ; X X /* parse command part */ X X switch( *ptr ) { X case '\0': break ; X case 'm': X pat->command = C_MAIL ; X get_dest(pat, ++ptr) ; X break ; X case 'j': pat->command = C_JUNK ; ++ptr ; break ; X case 'f': X pat->command = C_FILE ; X get_dest(pat, ++ptr) ; X break ; X case '+': X pat->command = C_FILE ; X get_dest(pat, ptr) ; X break ; X case '|': X pat->command = C_PIPE ; X get_dest(pat, ++ptr) ; X break ; X default: ++ptr ; break ; X } X} X X Xstatic void Xget_dest(pat,ptr) Xregister PatInfo *pat ; Xregister char *ptr ; X{ X for(; isspace(*ptr); ++ptr); X X pat->dest = strdup(ptr) ; X if( (ptr = strchr(pat->dest, '\n')) != NULL ) X *ptr = '\0' ; X} X X X X Xint Xfilename_expand(dest, src) X char *dest ; X char *src ; /* file to match */ X{ X char *lastsl ; /* last '/' in template */ X char *dirstr ; X struct passwd *passent ; X X if( src[0] == '/' ) { /* absolute */ X strcpy(dest, src) ; /* that was easy */ X return 0 ; X } X X if( src[0] == '+' ) { /* folder name */ X strcpy(dest, home) ; X strcat(dest, "/") ; X strcat(dest, folder) ; X strcat(dest, "/") ; X strcat(dest, src+1) ; X return 0 ; X } X X if( strmatch(src, "$HOME/") ) X { X strcpy(dest, home) ; X strcat(dest, src+5) ; X return 0 ; X } X X if( src[0] != '~' ) { /* relative path, assume relative */ X strcpy(dest, home) ; /* to $HOME */ X strcat(dest, "/") ; X strcat(dest, src) ; X return 0 ; X } X X X X /* cases: X * "~" fail X * "~/ get home directory X * "~xx username matching X */ X X if( src[1] == '/' ) X { X strcpy(dest, home) ; X strcat(dest, src+1) ; X } X else X { X dirstr = strchr(src, '/') ; X if( dirstr != NULL ) X *dirstr = '\0' ; X passent = getpwnam(src+1) ; X if( dirstr != NULL ) X *dirstr = '/' ; X if( passent == NULL ) X return -1 ; X X strcpy(dest, passent->pw_dir) ; X if( dirstr != NULL ) X strcat(dest, dirstr) ; X strcpy(src,dest) ; X } X return 0 ; X} SHAR_EOF len=`wc -c < parse.c` if test $len != 7864 ; then echo error: parse.c was $len bytes long, should have been 7864 fi fi # end of overwriting check if test -f regex.c then echo shar: will not over-write existing file regex.c else echo shar: extracting 'regex.c', 44540 characters sed 's/^X//' > regex.c <<'SHAR_EOF' X/* Extended regular expression matching and search. X Copyright (C) 1985 Free Software Foundation, Inc. X X This program is free software; you can redistribute it and/or modify X it under the terms of the GNU General Public License as published by X the Free Software Foundation; either version 1, or (at your option) X any later version. X X This program is distributed in the hope that it will be useful, X but WITHOUT ANY WARRANTY; without even the implied warranty of X MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the X GNU General Public License for more details. X X You should have received a copy of the GNU General Public License X along with this program; if not, write to the Free Software X Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. X XIn other words, you are welcome to use, share and improve this program. XYou are forbidden to forbid anyone else to use, share and improve Xwhat you give them. Help stamp out software-hoarding! */ X X X/* To test, compile with -Dtest. X This Dtestable feature turns this into a self-contained program X which reads a pattern, describes how it compiles, X then reads a string and searches for it. */ X X X#ifdef SVr4 X#define bcopy(s,d,n) memmove(d,s,n) X#define bzero(d,n) memset(d,0,n) X#define bcmp memcmp X#endif X X#ifdef emacs X X/* The `emacs' switch turns on certain special matching commands X that make sense only in emacs. */ X X#include "config.h" X#include "lisp.h" X#include "buffer.h" X#include "syntax.h" X X#else /* not emacs */ X X/* X * Define the syntax stuff, so we can do the \<...\> things. X */ X X#ifndef Sword /* must be non-zero in some of the tests below... */ X#define Sword 1 X#endif X X#define SYNTAX(c) re_syntax_table[c] X X#ifdef SYNTAX_TABLE X Xchar *re_syntax_table; X X#else X Xstatic char re_syntax_table[256]; X Xstatic void Xinit_syntax_once () X{ X register int c; X static int done = 0; X X if (done) X return; X X bzero (re_syntax_table, sizeof re_syntax_table); X X for (c = 'a'; c <= 'z'; c++) X re_syntax_table[c] = Sword; X X for (c = 'A'; c <= 'Z'; c++) X re_syntax_table[c] = Sword; X X for (c = '0'; c <= '9'; c++) X re_syntax_table[c] = Sword; X X done = 1; X} X X#endif /* SYNTAX_TABLE */ X#endif /* not emacs */ X X#include <alloca.h> X#include "regex.h" X X/* Number of failure points to allocate space for initially, X when matching. If this number is exceeded, more space is allocated, X so it is not a hard limit. */ X X#ifndef NFAILURES X#define NFAILURES 80 X#endif NFAILURES X X/* width of a byte in bits */ X X#define BYTEWIDTH 8 X X#ifndef SIGN_EXTEND_CHAR X#define SIGN_EXTEND_CHAR(x) (x) X#endif X Xstatic int obscure_syntax = 0; X X/* Specify the precise syntax of regexp for compilation. X This provides for compatibility for various utilities X which historically have different, incompatible syntaxes. X X The argument SYNTAX is a bit-mask containing the two bits X RE_NO_BK_PARENS and RE_NO_BK_VBAR. */ X Xint Xre_set_syntax (syntax) X{ X int ret; X X ret = obscure_syntax; X obscure_syntax = syntax; X return ret; X} X X/* re_compile_pattern takes a regular-expression string X and converts it into a buffer full of byte commands for matching. X X PATTERN is the address of the pattern string X SIZE is the length of it. X BUFP is a struct re_pattern_buffer * which points to the info X on where to store the byte commands. X This structure contains a char * which points to the X actual space, which should have been obtained with malloc. X re_compile_pattern may use realloc to grow the buffer space. X X The number of bytes of commands can be found out by looking in X the struct re_pattern_buffer that bufp pointed to, X after re_compile_pattern returns. X*/ X X#define PATPUSH(ch) (*b++ = (char) (ch)) X X#define PATFETCH(c) \ X {if (p == pend) goto end_of_pattern; \ X c = * (unsigned char *) p++; \ X if (translate) c = translate[c]; } X X#define PATFETCH_RAW(c) \ X {if (p == pend) goto end_of_pattern; \ X c = * (unsigned char *) p++; } X X#define PATUNFETCH p-- X X#define EXTEND_BUFFER \ X { char *old_buffer = bufp->buffer; \ X if (bufp->allocated == (1<<16)) goto too_big; \ X bufp->allocated *= 2; \ X if (bufp->allocated > (1<<16)) bufp->allocated = (1<<16); \ X if (!(bufp->buffer = (char *) realloc (bufp->buffer, bufp->allocated))) \ X goto memory_exhausted; \ X c = bufp->buffer - old_buffer; \ X b += c; \ X if (fixup_jump) \ X fixup_jump += c; \ X if (laststart) \ X laststart += c; \ X begalt += c; \ X if (pending_exact) \ X pending_exact += c; \ X } X Xstatic int store_jump (), insert_jump (); X Xchar * Xre_compile_pattern (pattern, size, bufp) X char *pattern; X int size; X struct re_pattern_buffer *bufp; X{ X register char *b = bufp->buffer; X register char *p = pattern; X char *pend = pattern + size; X register unsigned c, c1; X char *p1; X unsigned char *translate = (unsigned char *) bufp->translate; X X /* address of the count-byte of the most recently inserted "exactn" command. X This makes it possible to tell whether a new exact-match character X can be added to that command or requires a new "exactn" command. */ X X char *pending_exact = 0; X X /* address of the place where a forward-jump should go X to the end of the containing expression. X Each alternative of an "or", except the last, ends with a forward-jump X of this sort. */ X X char *fixup_jump = 0; X X /* address of start of the most recently finished expression. X This tells postfix * where to find the start of its operand. */ X X char *laststart = 0; X X /* In processing a repeat, 1 means zero matches is allowed */ X X char zero_times_ok; X X /* In processing a repeat, 1 means many matches is allowed */ X X char many_times_ok; X X /* address of beginning of regexp, or inside of last $ */ X X char *begalt = b; X X /* Stack of information saved by \( and restored by $. X Four stack elements are pushed by each $: X First, the value of b. X Second, the value of fixup_jump. X Third, the value of regnum. X Fourth, the value of begalt. */ X X int stackb[40]; X int *stackp = stackb; X int *stacke = stackb + 40; X int *stackt; X X /* Counts \('s as they are encountered. Remembered for the matching $, X where it becomes the "register number" to put in the stop_memory command */ X X int regnum = 1; X X bufp->fastmap_accurate = 0; X X#ifndef emacs X#ifndef SYNTAX_TABLE X /* X * Initialize the syntax table. X */ X init_syntax_once(); X#endif X#endif X X if (bufp->allocated == 0) X { X bufp->allocated = 28; X if (bufp->buffer) X /* EXTEND_BUFFER loses when bufp->allocated is 0 */ X bufp->buffer = (char *) realloc (bufp->buffer, 28); X else X /* Caller did not allocate a buffer. Do it for him */ X bufp->buffer = (char *) malloc (28); X if (!bufp->buffer) goto memory_exhausted; X begalt = b = bufp->buffer; X } X X while (p != pend) X { X if (b - bufp->buffer > bufp->allocated - 10) X /* Note that EXTEND_BUFFER clobbers c */ X EXTEND_BUFFER; X X PATFETCH (c); X X switch (c) X { X case '$': X if (obscure_syntax & RE_TIGHT_VBAR) X { X if (! (obscure_syntax & RE_CONTEXT_INDEP_OPS) && p != pend) X goto normal_char; X /* Make operand of last vbar end before this `$'. */ X if (fixup_jump) X store_jump (fixup_jump, jump, b); X fixup_jump = 0; X PATPUSH (endline); X break; X } X X /* $ means succeed if at end of line, but only in special contexts. X If randomly in the middle of a pattern, it is a normal character. */ X if (p == pend || *p == '\n' X || (obscure_syntax & RE_CONTEXT_INDEP_OPS) X || (obscure_syntax & RE_NO_BK_PARENS X ? *p == ')' X : *p == '\\' && p[1] == ')') X || (obscure_syntax & RE_NO_BK_VBAR X ? *p == '|' X : *p == '\\' && p[1] == '|')) X { X PATPUSH (endline); X break; X } X goto normal_char; X X case '^': X /* ^ means succeed if at beg of line, but only if no preceding pattern. */ X X if (laststart && p[-2] != '\n' X && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS)) X goto normal_char; X if (obscure_syntax & RE_TIGHT_VBAR) X { X if (p != pattern + 1 X && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS)) X goto normal_char; X PATPUSH (begline); X begalt = b; X } X else X PATPUSH (begline); X break; X X case '+': X case '?': X if (obscure_syntax & RE_BK_PLUS_QM) X goto normal_char; X handle_plus: X case '*': X /* If there is no previous pattern, char not special. */ X if (!laststart && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS)) X goto normal_char; X /* If there is a sequence of repetition chars, X collapse it down to equivalent to just one. */ X zero_times_ok = 0; X many_times_ok = 0; X while (1) X { X zero_times_ok |= c != '+'; X many_times_ok |= c != '?'; X if (p == pend) X break; X PATFETCH (c); X if (c == '*') X ; X else if (!(obscure_syntax & RE_BK_PLUS_QM) X && (c == '+' || c == '?')) X ; X else if ((obscure_syntax & RE_BK_PLUS_QM) X && c == '\\') X { X int c1; X PATFETCH (c1); X if (!(c1 == '+' || c1 == '?')) X { X PATUNFETCH; X PATUNFETCH; X break; X } X c = c1; X } X else X { X PATUNFETCH; X break; X } X } X X /* Star, etc. applied to an empty pattern is equivalent X to an empty pattern. */ X if (!laststart) X break; X X /* Now we know whether 0 matches is allowed, X and whether 2 or more matches is allowed. */ X if (many_times_ok) X { X /* If more than one repetition is allowed, X put in a backward jump at the end. */ X store_jump (b, maybe_finalize_jump, laststart - 3); X b += 3; X } X insert_jump (on_failure_jump, laststart, b + 3, b); X pending_exact = 0; X b += 3; X if (!zero_times_ok) X { X /* At least one repetition required: insert before the loop X a skip over the initial on-failure-jump instruction */ X insert_jump (dummy_failure_jump, laststart, laststart + 6, b); X b += 3; X } X break; X X case '.': X laststart = b; X PATPUSH (anychar); X break; X X case '[': X while (b - bufp->buffer X > bufp->allocated - 3 - (1 << BYTEWIDTH) / BYTEWIDTH) X /* Note that EXTEND_BUFFER clobbers c */ X EXTEND_BUFFER; X X laststart = b; X if (*p == '^') X PATPUSH (charset_not), p++; X else X PATPUSH (charset); X p1 = p; X X PATPUSH ((1 << BYTEWIDTH) / BYTEWIDTH); X /* Clear the whole map */ X bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH); X /* Read in characters and ranges, setting map bits */ X while (1) X { X PATFETCH (c); X if (c == ']' && p != p1 + 1) break; X if (*p == '-' && p[1] != ']') X { X PATFETCH (c1); X PATFETCH (c1); X while (c <= c1) X b[c / BYTEWIDTH] |= 1 << (c % BYTEWIDTH), c++; X } X else X { X b[c / BYTEWIDTH] |= 1 << (c % BYTEWIDTH); X } X } X /* Discard any bitmap bytes that are all 0 at the end of the map. X Decrement the map-length byte too. */ X while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) X b[-1]--; X b += b[-1]; X break; X X case '(': X if (! (obscure_syntax & RE_NO_BK_PARENS)) X goto normal_char; X else X goto handle_open; X X case ')': X if (! (obscure_syntax & RE_NO_BK_PARENS)) X goto normal_char; X else X goto handle_close; X X case '\n': X if (! (obscure_syntax & RE_NEWLINE_OR)) X goto normal_char; X else X goto handle_bar; X X case '|': X if (! (obscure_syntax & RE_NO_BK_VBAR)) X goto normal_char; X else X goto handle_bar; X X case '\\': X if (p == pend) goto invalid_pattern; X PATFETCH_RAW (c); X switch (c) X { X case '(': X if (obscure_syntax & RE_NO_BK_PARENS) X goto normal_backsl; X handle_open: X if (stackp == stacke) goto nesting_too_deep; X if (regnum < RE_NREGS) X { X PATPUSH (start_memory); X PATPUSH (regnum); X } X *stackp++ = b - bufp->buffer; X *stackp++ = fixup_jump ? fixup_jump - bufp->buffer + 1 : 0; X *stackp++ = regnum++; X *stackp++ = begalt - bufp->buffer; X fixup_jump = 0; X laststart = 0; X begalt = b; X break; X X case ')': X if (obscure_syntax & RE_NO_BK_PARENS) X goto normal_backsl; X handle_close: X if (stackp == stackb) goto unmatched_close; X begalt = *--stackp + bufp->buffer; X if (fixup_jump) X store_jump (fixup_jump, jump, b); X if (stackp[-1] < RE_NREGS) X { X PATPUSH (stop_memory); X PATPUSH (stackp[-1]); X } X stackp -= 2; X fixup_jump = 0; X if (*stackp) X fixup_jump = *stackp + bufp->buffer - 1; X laststart = *--stackp + bufp->buffer; X break; X X case '|': X if (obscure_syntax & RE_NO_BK_VBAR) X goto normal_backsl; X handle_bar: X insert_jump (on_failure_jump, begalt, b + 6, b); X pending_exact = 0; X b += 3; X if (fixup_jump) X store_jump (fixup_jump, jump, b); X fixup_jump = b; X b += 3; X laststart = 0; X begalt = b; X break; X X#ifdef emacs X case '=': X PATPUSH (at_dot); X break; X X case 's': X laststart = b; X PATPUSH (syntaxspec); X PATFETCH (c); X PATPUSH (syntax_spec_code[c]); X break; X X case 'S': X laststart = b; X PATPUSH (notsyntaxspec); X PATFETCH (c); X PATPUSH (syntax_spec_code[c]); X break; X#endif emacs X X case 'w': X laststart = b; X PATPUSH (wordchar); X break; X X case 'W': X laststart = b; X PATPUSH (notwordchar); X break; X X case '<': X PATPUSH (wordbeg); X break; X X case '>': X PATPUSH (wordend); X break; X X case 'b': X PATPUSH (wordbound); X break; X X case 'B': X PATPUSH (notwordbound); X break; X X case '`': X PATPUSH (begbuf); X break; X X case '\'': X PATPUSH (endbuf); X break; X X case '1': X case '2': X case '3': X case '4': X case '5': X case '6': X case '7': X case '8': X case '9': X c1 = c - '0'; X if (c1 >= regnum) X goto normal_char; X for (stackt = stackp - 2; stackt > stackb; stackt -= 4) X if (*stackt == c1) X goto normal_char; X laststart = b; X PATPUSH (duplicate); X PATPUSH (c1); X break; X X case '+': X case '?': X if (obscure_syntax & RE_BK_PLUS_QM) X goto handle_plus; X X default: X normal_backsl: X /* You might think it would be useful for \ to mean X not to translate; but if we don't translate it X it will never match anything. */ X if (translate) c = translate[c]; X goto normal_char; X } X break; X X default: X normal_char: X if (!pending_exact || pending_exact + *pending_exact + 1 != b X || *pending_exact == 0177 || *p == '*' || *p == '^' X || ((obscure_syntax & RE_BK_PLUS_QM) X ? *p == '\\' && (p[1] == '+' || p[1] == '?') X : (*p == '+' || *p == '?'))) X { X laststart = b; X PATPUSH (exactn); X pending_exact = b; X PATPUSH (0); X } X PATPUSH (c); X (*pending_exact)++; X } X } X X if (fixup_jump) X store_jump (fixup_jump, jump, b); X X if (stackp != stackb) goto unmatched_open; X X bufp->used = b - bufp->buffer; X return 0; X X invalid_pattern: X return "Invalid regular expression"; X X unmatched_open: X return "Unmatched \$"; X X unmatched_close: X return "Unmatched \$"; X X end_of_pattern: X return "Premature end of regular expression"; X X nesting_too_deep: X return "Nesting too deep"; X X too_big: X return "Regular expression too big"; X X memory_exhausted: X return "Memory exhausted"; X} X X/* Store where `from' points a jump operation to jump to where `to' points. X `opcode' is the opcode to store. */ X Xstatic int Xstore_jump (from, opcode, to) X char *from, *to; X char opcode; X{ X from[0] = opcode; X from[1] = (to - (from + 3)) & 0377; X from[2] = (to - (from + 3)) >> 8; X} X X/* Open up space at char FROM, and insert there a jump to TO. X CURRENT_END gives te end of the storage no in use, X so we know how much data to copy up. X OP is the opcode of the jump to insert. X X If you call this function, you must zero out pending_exact. */ X Xstatic int Xinsert_jump (op, from, to, current_end) X char op; X char *from, *to, *current_end; X{ X register char *pto = current_end + 3; X register char *pfrom = current_end; X while (pfrom != from) X *--pto = *--pfrom; X store_jump (from, op, to); X} X X/* Given a pattern, compute a fastmap from it. X The fastmap records which of the (1 << BYTEWIDTH) possible characters X can start a string that matches the pattern. X This fastmap is used by re_search to skip quickly over totally implausible text. X X The caller must supply the address of a (1 << BYTEWIDTH)-byte data area X as bufp->fastmap. X The other components of bufp describe the pattern to be used. */ X Xvoid Xre_compile_fastmap (bufp) X struct re_pattern_buffer *bufp; X{ X unsigned char *pattern = (unsigned char *) bufp->buffer; X int size = bufp->used; X register char *fastmap = bufp->fastmap; X register unsigned char *p = pattern; X register unsigned char *pend = pattern + size; X register int j, k; X unsigned char *translate = (unsigned char *) bufp->translate; X X unsigned char *stackb[NFAILURES]; X unsigned char **stackp = stackb; X X bzero (fastmap, (1 << BYTEWIDTH)); X bufp->fastmap_accurate = 1; X bufp->can_be_null = 0; X X while (p) X { X if (p == pend) X { X bufp->can_be_null = 1; X break; X } X#ifdef SWITCH_ENUM_BUG X switch ((int) ((enum regexpcode) *p++)) X#else X switch ((enum regexpcode) *p++) X#endif X { X case exactn: X if (translate) X fastmap[translate[p[1]]] = 1; X else X fastmap[p[1]] = 1; X break; X X case begline: X case before_dot: X case at_dot: X case after_dot: X case begbuf: X case endbuf: X case wordbound: X case notwordbound: X case wordbeg: X case wordend: X continue; X X case endline: X if (translate) X fastmap[translate['\n']] = 1; X else X fastmap['\n'] = 1; X if (bufp->can_be_null != 1) X bufp->can_be_null = 2; X break; X X case finalize_jump: X case maybe_finalize_jump: X case jump: X case dummy_failure_jump: X bufp->can_be_null = 1; X j = *p++ & 0377; X j += SIGN_EXTEND_CHAR (*(char *)p) << 8; X p += j + 1; /* The 1 compensates for missing ++ above */ X if (j > 0) X continue; X /* Jump backward reached implies we just went through X the body of a loop and matched nothing. X Opcode jumped to should be an on_failure_jump. X Just treat it like an ordinary jump. X For a * loop, it has pushed its failure point already; X if so, discard that as redundant. */ X if ((enum regexpcode) *p != on_failure_jump) X continue; X p++; X j = *p++ & 0377; X j += SIGN_EXTEND_CHAR (*(char *)p) << 8; X p += j + 1; /* The 1 compensates for missing ++ above */ X if (stackp != stackb && *stackp == p) X stackp--; X continue; X X case on_failure_jump: X j = *p++ & 0377; X j += SIGN_EXTEND_CHAR (*(char *)p) << 8; X p++; X *++stackp = p + j; X continue; X X case start_memory: X case stop_memory: X p++; X continue; X X case duplicate: X bufp->can_be_null = 1; X fastmap['\n'] = 1; X case anychar: X for (j = 0; j < (1 << BYTEWIDTH); j++) X if (j != '\n') X fastmap[j] = 1; X if (bufp->can_be_null) X return; X /* Don't return; check the alternative paths X so we can set can_be_null if appropriate. */ X break; X X case wordchar: X for (j = 0; j < (1 << BYTEWIDTH); j++) X if (SYNTAX (j) == Sword) X fastmap[j] = 1; X break; X X case notwordchar: X for (j = 0; j < (1 << BYTEWIDTH); j++) X if (SYNTAX (j) != Sword) X fastmap[j] = 1; X break; X X#ifdef emacs X case syntaxspec: X k = *p++; X for (j = 0; j < (1 << BYTEWIDTH); j++) X if (SYNTAX (j) == (enum syntaxcode) k) X fastmap[j] = 1; X break; X X case notsyntaxspec: X k = *p++; X for (j = 0; j < (1 << BYTEWIDTH); j++) X if (SYNTAX (j) != (enum syntaxcode) k) X fastmap[j] = 1; X break; X#endif emacs X X case charset: X for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) X if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) X { X if (translate) X fastmap[translate[j]] = 1; X else X fastmap[j] = 1; X } X break; X X case charset_not: X /* Chars beyond end of map must be allowed */ X for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) X if (translate) X fastmap[translate[j]] = 1; X else X fastmap[j] = 1; X X for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) X if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) X { X if (translate) X fastmap[translate[j]] = 1; X else X fastmap[j] = 1; X } X break; X } X X /* Get here means we have successfully found the possible starting characters X of one path of the pattern. We need not follow this path any farther. X Instead, look at the next alternative remembered in the stack. */ X if (stackp != stackb) X p = *stackp--; X else X break; X } X} X X/* Like re_search_2, below, but only one string is specified. */ X Xint Xre_search (pbufp, string, size, startpos, range, regs) X struct re_pattern_buffer *pbufp; X char *string; X int size, startpos, range; X struct re_registers *regs; X{ X return re_search_2 (pbufp, 0, 0, string, size, startpos, range, regs, size); X} X X/* Like re_match_2 but tries first a match starting at index STARTPOS, X then at STARTPOS + 1, and so on. X RANGE is the number of places to try before giving up. X If RANGE is negative, the starting positions tried are X STARTPOS, STARTPOS - 1, etc. X It is up to the caller to make sure that range is not so large X as to take the starting position outside of the input strings. X XThe value returned is the position at which the match was found, X or -1 if no match was found, X or -2 if error (such as failure stack overflow). */ X Xint Xre_search_2 (pbufp, string1, size1, string2, size2, startpos, range, regs, mstop) X struct re_pattern_buffer *pbufp; X char *string1, *string2; X int size1, size2; X int startpos; X register int range; X struct re_registers *regs; X int mstop; X{ X register char *fastmap = pbufp->fastmap; X register unsigned char *translate = (unsigned char *) pbufp->translate; X int total = size1 + size2; X int val; X X /* Update the fastmap now if not correct already */ X if (fastmap && !pbufp->fastmap_accurate) X re_compile_fastmap (pbufp); X X /* Don't waste time in a long search for a pattern X that says it is anchored. */ X if (pbufp->used > 0 && (enum regexpcode) pbufp->buffer[0] == begbuf X && range > 0) X { X if (startpos > 0) X return -1; X else X range = 1; X } X X while (1) X { X /* If a fastmap is supplied, skip quickly over characters X that cannot possibly be the start of a match. X Note, however, that if the pattern can possibly match X the null string, we must test it at each starting point X so that we take the first null string we get. */ X X if (fastmap && startpos < total && pbufp->can_be_null != 1) X { X if (range > 0) X { X register int lim = 0; X register unsigned char *p; X int irange = range; X if (startpos < size1 && startpos + range >= size1) X lim = range - (size1 - startpos); X X p = ((unsigned char *) X &(startpos >= size1 ? string2 - size1 : string1)[startpos]); X X if (translate) X { X while (range > lim && !fastmap[translate[*p++]]) X range--; X } X else X { X while (range > lim && !fastmap[*p++]) X range--; X } X startpos += irange - range; X } X else X { X register unsigned char c; X if (startpos >= size1) X c = string2[startpos - size1]; X else X c = string1[startpos]; X c &= 0xff; X if (translate ? !fastmap[translate[c]] : !fastmap[c]) X goto advance; X } X } X X if (range >= 0 && startpos == total X && fastmap && pbufp->can_be_null == 0) X return -1; X X val = re_match_2 (pbufp, string1, size1, string2, size2, startpos, regs, X mstop); X /* Propagate error indication if worse than mere failure. */ X if (val == -2) X return -2; X /* Return position on success. */ X if (0 <= val) X return startpos; X X#ifdef C_ALLOCA X alloca (0); X#endif /* C_ALLOCA */ X X advance: X if (!range) break; X if (range > 0) range--, startpos++; else range++, startpos--; X } X return -1; X} X X#ifndef emacs /* emacs never uses this */ Xint Xre_match (pbufp, string, size, pos, regs) X struct re_pattern_buffer *pbufp; X char *string; X int size, pos; X struct re_registers *regs; X{ X return re_match_2 (pbufp, 0, 0, string, size, pos, regs, size); X} X#endif /* emacs */ X X/* Maximum size of failure stack. Beyond this, overflow is an error. */ X Xint re_max_failures = 2000; X Xstatic int bcmp_translate(); X/* Match the pattern described by PBUFP X against data which is the virtual concatenation of STRING1 and STRING2. X SIZE1 and SIZE2 are the sizes of the two data strings. X Start the match at position POS. X Do not consider matching past the position MSTOP. X X If pbufp->fastmap is nonzero, then it had better be up to date. X X The reason that the data to match are specified as two components X which are to be regarded as concatenated X is so this function can be used directly on the contents of an Emacs buffer. X X -1 is returned if there is no match. -2 is returned if there is X an error (such as match stack overflow). Otherwise the value is the length X of the substring which was matched. */ X Xint Xre_match_2 (pbufp, string1, size1, string2, size2, pos, regs, mstop) X struct re_pattern_buffer *pbufp; X unsigned char *string1, *string2; X int size1, size2; X int pos; X struct re_registers *regs; X int mstop; X{ X register unsigned char *p = (unsigned char *) pbufp->buffer; X register unsigned char *pend = p + pbufp->used; X /* End of first string */ X unsigned char *end1; X /* End of second string */ X unsigned char *end2; X /* Pointer just past last char to consider matching */ X unsigned char *end_match_1, *end_match_2; X register unsigned char *d, *dend; X register int mcnt; X unsigned char *translate = (unsigned char *) pbufp->translate; X X /* Failure point stack. Each place that can handle a failure further down the line X pushes a failure point on this stack. It consists of two char *'s. X The first one pushed is where to resume scanning the pattern; X the second pushed is where to resume scanning the strings. X If the latter is zero, the failure point is a "dummy". X If a failure happens and the innermost failure point is dormant, X it discards that failure point and tries the next one. */ X X unsigned char *initial_stack[2 * NFAILURES]; X unsigned char **stackb = initial_stack; X unsigned char **stackp = stackb, **stacke = &stackb[2 * NFAILURES]; X X /* Information on the "contents" of registers. X These are pointers into the input strings; they record X just what was matched (on this attempt) by some part of the pattern. X The start_memory command stores the start of a register's contents X and the stop_memory command stores the end. X X At that point, regstart[regnum] points to the first character in the register, X regend[regnum] points to the first character beyond the end of the register, X regstart_seg1[regnum] is true iff regstart[regnum] points into string1, X and regend_seg1[regnum] is true iff regend[regnum] points into string1. */ X X unsigned char *regstart[RE_NREGS]; X unsigned char *regend[RE_NREGS]; X unsigned char regstart_seg1[RE_NREGS], regend_seg1[RE_NREGS]; X X /* Set up pointers to ends of strings. X Don't allow the second string to be empty unless both are empty. */ X if (!size2) X { X string2 = string1; X size2 = size1; X string1 = 0; X size1 = 0; X } X end1 = string1 + size1; X end2 = string2 + size2; X X /* Compute where to stop matching, within the two strings */ X if (mstop <= size1) X { X end_match_1 = string1 + mstop; X end_match_2 = string2; X } X else X { X end_match_1 = end1; X end_match_2 = string2 + mstop - size1; X } X X /* Initialize \) text positions to -1 X to mark ones that no $ or $ has been seen for. */ X X for (mcnt = 0; mcnt < sizeof (regend) / sizeof (*regend); mcnt++) X regend[mcnt] = (unsigned char *) -1; X X /* `p' scans through the pattern as `d' scans through the data. X `dend' is the end of the input string that `d' points within. X `d' is advanced into the following input string whenever necessary, X but this happens before fetching; X therefore, at the beginning of the loop, X `d' can be pointing at the end of a string, X but it cannot equal string2. */ X X if (pos <= size1) X d = string1 + pos, dend = end_match_1; X else X d = string2 + pos - size1, dend = end_match_2; X X/* Write PREFETCH; just before fetching a character with *d. */ X#define PREFETCH \ X while (d == dend) \ X { if (dend == end_match_2) goto fail; /* end of string2 => failure */ \ X d = string2; /* end of string1 => advance to string2. */ \ X dend = end_match_2; } X X /* This loop loops over pattern commands. X It exits by returning from the function if match is complete, X or it drops through if match fails at this starting point in the input data. */ X X while (1) X { X if (p == pend) X /* End of pattern means we have succeeded! */ X { X /* If caller wants register contents data back, convert it to indices */ X if (regs) X { X regs->start[0] = pos; X if (dend == end_match_1) X regs->end[0] = d - string1; X else X regs->end[0] = d - string2 + size1; X for (mcnt = 1; mcnt < RE_NREGS; mcnt++) X { X if (regend[mcnt] == (unsigned char *) -1) X { X regs->start[mcnt] = -1; X regs->end[mcnt] = -1; X continue; X } X if (regstart_seg1[mcnt]) X regs->start[mcnt] = regstart[mcnt] - string1; X else X regs->start[mcnt] = regstart[mcnt] - string2 + size1; X if (regend_seg1[mcnt]) X regs->end[mcnt] = regend[mcnt] - string1; X else X regs->end[mcnt] = regend[mcnt] - string2 + size1; X } X } X if (dend == end_match_1) X return (d - string1 - pos); X else X return d - string2 + size1 - pos; X } X X /* Otherwise match next pattern command */ X#ifdef SWITCH_ENUM_BUG X switch ((int) ((enum regexpcode) *p++)) X#else X switch ((enum regexpcode) *p++) X#endif X { X X /* $ is represented by a start_memory, $ by a stop_memory. X Both of those commands contain a "register number" argument. X The text matched within the $ and $ is recorded under that number. X Then, \<digit> turns into a `duplicate' command which X is followed by the numeric value of <digit> as the register number. */ X X case start_memory: X regstart[*p] = d; X regstart_seg1[*p++] = (dend == end_match_1); X break; X X case stop_memory: X regend[*p] = d; X regend_seg1[*p++] = (dend == end_match_1); X break; X X case duplicate: X { X int regno = *p++; /* Get which register to match against */ X register unsigned char *d2, *dend2; X X /* Don't allow matching a register that hasn't been used. X This isn't fully reliable in the current version, X but it is better than crashing. */ X if ((int) regend[regno] <= -1) X goto fail; X X d2 = regstart[regno]; X dend2 = ((regstart_seg1[regno] == regend_seg1[regno]) X ? regend[regno] : end_match_1); X while (1) X { X /* Advance to next segment in register contents, if necessary */ X while (d2 == dend2) X { X if (dend2 == end_match_2) break; X if (dend2 == regend[regno]) break; X d2 = string2, dend2 = regend[regno]; /* end of string1 => advance to string2. */ X } X /* At end of register contents => success */ X if (d2 == dend2) break; X X /* Advance to next segment in data being matched, if necessary */ X PREFETCH; X X /* mcnt gets # consecutive chars to compare */ X mcnt = dend - d; X if (mcnt > dend2 - d2) X mcnt = dend2 - d2; X /* Compare that many; failure if mismatch, else skip them. */ X if (translate ? bcmp_translate (d, d2, mcnt, translate) : bcmp (d, d2, mcnt)) X goto fail; X d += mcnt, d2 += mcnt; X } X } X break; X X case anychar: X /* fetch a data character */ X PREFETCH; X /* Match anything but a newline. */ X if ((translate ? translate[*d++] : *d++) == '\n') X goto fail; X break; X X case charset: X case charset_not: X { X /* Nonzero for charset_not */ X int not = 0; X register int c; X if (*(p - 1) == (unsigned char) charset_not) X not = 1; X X /* fetch a data character */ X PREFETCH; X X if (translate) X c = translate [*d]; X else X c = *d; X X if (c < *p * BYTEWIDTH X && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) X not = !not; X X p += 1 + *p; X X if (!not) goto fail; X d++; X break; X } X X case begline: X if (d == string1 || d[-1] == '\n') X break; X goto fail; X X case endline: X if (d == end2 X || (d == end1 ? (size2 == 0 || *string2 == '\n') : *d == '\n')) X break; X goto fail; X X /* "or" constructs ("|") are handled by starting each alternative X with an on_failure_jump that points to the start of the next alternative. X Each alternative except the last ends with a jump to the joining point. X (Actually, each jump except for the last one really jumps X to the following jump, because tensioning the jumps is a hassle.) */ X X /* The start of a stupid repeat has an on_failure_jump that points X past the end of the repeat text. X This makes a failure point so that, on failure to match a repetition, X matching restarts past as many repetitions have been found X with no way to fail and look for another one. */ X X /* A smart repeat is similar but loops back to the on_failure_jump X so that each repetition makes another failure point. */ X X case on_failure_jump: X if (stackp == stacke) X { X unsigned char **stackx; X if (stacke - stackb > re_max_failures) X return -2; X stackx = (unsigned char **) alloca (2 * (stacke - stackb) X * sizeof (char *)); X bcopy (stackb, stackx, (stacke - stackb) * sizeof (char *)); X stackp = stackx + (stackp - stackb); X stacke = stackx + 2 * (stacke - stackb); X stackb = stackx; X } X mcnt = *p++ & 0377; X mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8; X p++; X *stackp++ = mcnt + p; X *stackp++ = d; X break; X X /* The end of a smart repeat has an maybe_finalize_jump back. X Change it either to a finalize_jump or an ordinary jump. */ X X case maybe_finalize_jump: X mcnt = *p++ & 0377; X mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8; X p++; X /* Compare what follows with the begining of the repeat. X If we can establish that there is nothing that they would X both match, we can change to finalize_jump */ X if (p == pend) X p[-3] = (unsigned char) finalize_jump; X else if (*p == (unsigned char) exactn X || *p == (unsigned char) endline) X { X register int c = *p == (unsigned char) endline ? '\n' : p[2]; X register unsigned char *p1 = p + mcnt; X /* p1[0] ... p1[2] are an on_failure_jump. X Examine what follows that */ X if (p1[3] == (unsigned char) exactn && p1[5] != c) X p[-3] = (unsigned char) finalize_jump; X else if (p1[3] == (unsigned char) charset X || p1[3] == (unsigned char) charset_not) X { X int not = p1[3] == (unsigned char) charset_not; X if (c < p1[4] * BYTEWIDTH X && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) X not = !not; X /* not is 1 if c would match */ X /* That means it is not safe to finalize */ X if (!not) X p[-3] = (unsigned char) finalize_jump; X } X } X p -= 2; X if (p[-1] != (unsigned char) finalize_jump) X { X p[-1] = (unsigned char) jump; X goto nofinalize; X } X X /* The end of a stupid repeat has a finalize-jump X back to the start, where another failure point will be made X which will point after all the repetitions found so far. */ X X case finalize_jump: X stackp -= 2; X X case jump: X nofinalize: X mcnt = *p++ & 0377; X mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8; X p += mcnt + 1; /* The 1 compensates for missing ++ above */ X break; X X case dummy_failure_jump: X if (stackp == stacke) X { X unsigned char **stackx X = (unsigned char **) alloca (2 * (stacke - stackb) X * sizeof (char *)); X bcopy (stackb, stackx, (stacke - stackb) * sizeof (char *)); X stackp = stackx + (stackp - stackb); X stacke = stackx + 2 * (stacke - stackb); X stackb = stackx; X } X *stackp++ = 0; X *stackp++ = 0; X goto nofinalize; X X case wordbound: X if (d == string1 /* Points to first char */ X || d == end2 /* Points to end */ X || (d == end1 && size2 == 0)) /* Points to end */ X break; X if ((SYNTAX (d[-1]) == Sword) X != (SYNTAX (d == end1 ? *string2 : *d) == Sword)) X break; X goto fail; X X case notwordbound: X if (d == string1 /* Points to first char */ X || d == end2 /* Points to end */ X || (d == end1 && size2 == 0)) /* Points to end */ X goto fail; X if ((SYNTAX (d[-1]) == Sword) X != (SYNTAX (d == end1 ? *string2 : *d) == Sword)) X goto fail; X break; X X case wordbeg: X if (d == end2 /* Points to end */ X || (d == end1 && size2 == 0) /* Points to end */ X || SYNTAX (* (d == end1 ? string2 : d)) != Sword) /* Next char not a letter */ X goto fail; X if (d == string1 /* Points to first char */ X || SYNTAX (d[-1]) != Sword) /* prev char not letter */ X break; X goto fail; X X case wordend: X if (d == string1 /* Points to first char */ X || SYNTAX (d[-1]) != Sword) /* prev char not letter */ X goto fail; X if (d == end2 /* Points to end */ X || (d == end1 && size2 == 0) /* Points to end */ X || SYNTAX (d == end1 ? *string2 : *d) != Sword) /* Next char not a letter */ X break; X goto fail; X X#ifdef emacs X case before_dot: X if (PTR_CHAR_POS (d) + 1 >= point) X goto fail; X break; X X case at_dot: X if (PTR_CHAR_POS (d) + 1 != point) X goto fail; X break; X X case after_dot: X if (PTR_CHAR_POS (d) + 1 <= point) X goto fail; X break; X X case wordchar: X mcnt = (int) Sword; X goto matchsyntax; X X case syntaxspec: X mcnt = *p++; X matchsyntax: X PREFETCH; X if (SYNTAX (*d++) != (enum syntaxcode) mcnt) goto fail; X break; X X case notwordchar: X mcnt = (int) Sword; X goto matchnotsyntax; X X case notsyntaxspec: X mcnt = *p++; X matchnotsyntax: X PREFETCH; X if (SYNTAX (*d++) == (enum syntaxcode) mcnt) goto fail; X break; X#else X case wordchar: X PREFETCH; X if (SYNTAX (*d++) == 0) goto fail; X break; X X case notwordchar: X PREFETCH; X if (SYNTAX (*d++) != 0) goto fail; X break; X#endif not emacs X X case begbuf: X if (d == string1) /* Note, d cannot equal string2 */ X break; /* unless string1 == string2. */ X goto fail; X X case endbuf: X if (d == end2 || (d == end1 && size2 == 0)) X break; X goto fail; X X case exactn: X /* Match the next few pattern characters exactly. X mcnt is how many characters to match. */ X mcnt = *p++; X if (translate) X { X do X { X PREFETCH; X if (translate[*d++] != *p++) goto fail; X } X while (--mcnt); X } X else X { X do X { X PREFETCH; X if (*d++ != *p++) goto fail; X } X while (--mcnt); X } X break; X } X continue; /* Successfully matched one pattern command; keep matching */ X X /* Jump here if any matching operation fails. */ X fail: X if (stackp != stackb) X /* A restart point is known. Restart there and pop it. */ X { X if (!stackp[-2]) X { /* If innermost failure point is dormant, flush it and keep looking */ X stackp -= 2; X goto fail; X } X d = *--stackp; X p = *--stackp; X if (d >= string1 && d <= end1) X dend = end_match_1; X } X else break; /* Matching at this starting point really fails! */ X } X return -1; /* Failure to match */ X} X Xstatic int Xbcmp_translate (s1, s2, len, translate) X unsigned char *s1, *s2; X register int len; X unsigned char *translate; X{ X register unsigned char *p1 = s1, *p2 = s2; X while (len) X { X if (translate [*p1++] != translate [*p2++]) return 1; X len--; X } X return 0; X} X X/* Entry points compatible with bsd4.2 regex library */ X X#ifndef emacs X Xstatic struct re_pattern_buffer re_comp_buf; X Xchar * Xre_comp (s) X char *s; X{ X if (!s) X { X if (!re_comp_buf.buffer) X return "No previous regular expression"; X return 0; X } X X if (!re_comp_buf.buffer) X { X if (!(re_comp_buf.buffer = (char *) malloc (200))) X return "Memory exhausted"; X re_comp_buf.allocated = 200; X if (!(re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH))) X return "Memory exhausted"; X } X return re_compile_pattern (s, strlen (s), &re_comp_buf); X} X Xint Xre_exec (s) X char *s; X{ X int len = strlen (s); X return 0 <= re_search (&re_comp_buf, s, len, 0, len, 0); X} X X#endif /* emacs */ X X#ifdef test X X#include <stdio.h> X X/* Indexed by a character, gives the upper case equivalent of the character */ X Xstatic char upcase[0400] = X { 000, 001, 002, 003, 004, 005, 006, 007, X 010, 011, 012, 013, 014, 015, 016, 017, X 020, 021, 022, 023, 024, 025, 026, 027, X 030, 031, 032, 033, 034, 035, 036, 037, X 040, 041, 042, 043, 044, 045, 046, 047, X 050, 051, 052, 053, 054, 055, 056, 057, X 060, 061, 062, 063, 064, 065, 066, 067, X 070, 071, 072, 073, 074, 075, 076, 077, X 0100, 0101, 0102, 0103, 0104, 0105, 0106, 0107, X 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117, X 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127, X 0130, 0131, 0132, 0133, 0134, 0135, 0136, 0137, X 0140, 0101, 0102, 0103, 0104, 0105, 0106, 0107, X 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117, X 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127, X 0130, 0131, 0132, 0173, 0174, 0175, 0176, 0177, X 0200, 0201, 0202, 0203, 0204, 0205, 0206, 0207, X 0210, 0211, 0212, 0213, 0214, 0215, 0216, 0217, X 0220, 0221, 0222, 0223, 0224, 0225, 0226, 0227, X 0230, 0231, 0232, 0233, 0234, 0235, 0236, 0237, X 0240, 0241, 0242, 0243, 0244, 0245, 0246, 0247, X 0250, 0251, 0252, 0253, 0254, 0255, 0256, 0257, X 0260, 0261, 0262, 0263, 0264, 0265, 0266, 0267, X 0270, 0271, 0272, 0273, 0274, 0275, 0276, 0277, X 0300, 0301, 0302, 0303, 0304, 0305, 0306, 0307, X 0310, 0311, 0312, 0313, 0314, 0315, 0316, 0317, X 0320, 0321, 0322, 0323, 0324, 0325, 0326, 0327, X 0330, 0331, 0332, 0333, 0334, 0335, 0336, 0337, X 0340, 0341, 0342, 0343, 0344, 0345, 0346, 0347, X 0350, 0351, 0352, 0353, 0354, 0355, 0356, 0357, X 0360, 0361, 0362, 0363, 0364, 0365, 0366, 0367, X 0370, 0371, 0372, 0373, 0374, 0375, 0376, 0377 X }; X Xmain (argc, argv) X int argc; X char **argv; X{ X char pat[80]; X struct re_pattern_buffer buf; X int i; X char c; X char fastmap[(1 << BYTEWIDTH)]; X X /* Allow a command argument to specify the style of syntax. */ X if (argc > 1) X obscure_syntax = atoi (argv[1]); X X buf.allocated = 40; X buf.buffer = (char *) malloc (buf.allocated); X buf.fastmap = fastmap; X buf.translate = upcase; X X while (1) X { X gets (pat); X X if (*pat) X { X re_compile_pattern (pat, strlen(pat), &buf); X X for (i = 0; i < buf.used; i++) X printchar (buf.buffer[i]); X X putchar ('\n'); X X printf ("%d allocated, %d used.\n", buf.allocated, buf.used); X X re_compile_fastmap (&buf); X printf ("Allowed by fastmap: "); X for (i = 0; i < (1 << BYTEWIDTH); i++) X if (fastmap[i]) printchar (i); X putchar ('\n'); X } X X gets (pat); /* Now read the string to match against */ X X i = re_match (&buf, pat, strlen (pat), 0, 0); X printf ("Match value %d.\n", i); X } X} X X#ifdef NOTDEF Xprint_buf (bufp) X struct re_pattern_buffer *bufp; X{ X int i; X X printf ("buf is :\n----------------\n"); X for (i = 0; i < bufp->used; i++) X printchar (bufp->buffer[i]); X X printf ("\n%d allocated, %d used.\n", bufp->allocated, bufp->used); X X printf ("Allowed by fastmap: "); X for (i = 0; i < (1 << BYTEWIDTH); i++) X if (bufp->fastmap[i]) X printchar (i); X printf ("\nAllowed by translate: "); X if (bufp->translate) X for (i = 0; i < (1 << BYTEWIDTH); i++) X if (bufp->translate[i]) X printchar (i); X printf ("\nfastmap is%s accurate\n", bufp->fastmap_accurate ? "" : "n't"); X printf ("can %s be null\n----------", bufp->can_be_null ? "" : "not"); X} X#endif X Xprintchar (c) X char c; X{ X if (c < 041 || c >= 0177) X { X putchar ('\\'); X putchar (((c >> 6) & 3) + '0'); X putchar (((c >> 3) & 7) + '0'); X putchar ((c & 7) + '0'); X } X else X putchar (c); X} X Xerror (string) X char *string; X{ X puts (string); X exit (1); X} X X#endif test SHAR_EOF len=`wc -c < regex.c` if test $len != 44540 ; then echo error: regex.c was $len bytes long, should have been 44540 fi fi # end of overwriting check exit 0 -- -ed falk, sun microsystems sun!falk, falk@sun.com He who dies with the most friends, wins.