]>
diplodocus.org Git - nmh/blob - uip/picksbr.c
3 * picksbr.c -- routines to help pick along...
5 * This code is Copyright (c) 2002, by the authors of nmh. See the
6 * COPYRIGHT file in the root directory of the nmh distribution for
7 * complete copyright information.
12 #include <h/picksbr.h>
15 #ifdef HAVE_SYS_TIME_H
16 # include <sys/time.h>
20 #define PARSE_SWITCHES \
24 X("lbrace", 0, PRLBR) \
25 X("rbrace", 0, PRRBR) \
26 X("cc pattern", 0, PRCC) \
27 X("date pattern", 0, PRDATE) \
28 X("from pattern", 0, PRFROM) \
29 X("search pattern", 0, PRSRCH) \
30 X("subject pattern", 0, PRSUBJ) \
31 X("to pattern", 0, PRTO) \
32 X("-othercomponent pattern", 15, PROTHR) \
33 X("after date", 0, PRAFTR) \
34 X("before date", 0, PRBEFR) \
35 X("datefield field", 5, PRDATF) \
37 #define X(sw, minchars, id) id,
38 DEFINE_SWITCH_ENUM(PARSE
);
41 #define X(sw, minchars, id) { sw, minchars, id },
42 DEFINE_SWITCH_ARRAY(PARSE
, parswit
);
45 /* DEFINITIONS FOR PATTERN MATCHING */
48 * We really should be using re_comp() and re_exec() here. Unfortunately,
49 * pick advertises that lowercase characters matches characters of both
50 * cases. Since re_exec() doesn't exhibit this behavior, we are stuck
51 * with this version. Furthermore, we need to be able to save and restore
52 * the state of the pattern matcher in order to do things "efficiently".
54 * The matching power of this algorithm isn't as powerful as the re_xxx()
55 * routines (no \(xxx\) and \n constructs). Such is life.
71 static char linebuf
[LBSIZE
+ 1];
72 static char decoded_linebuf
[LBSIZE
+ 1];
74 /* the magic array for case-independence */
75 static unsigned char cc
[] = {
76 0000,0001,0002,0003,0004,0005,0006,0007,
77 0010,0011,0012,0013,0014,0015,0016,0017,
78 0020,0021,0022,0023,0024,0025,0026,0027,
79 0030,0031,0032,0033,0034,0035,0036,0037,
80 0040,0041,0042,0043,0044,0045,0046,0047,
81 0050,0051,0052,0053,0054,0055,0056,0057,
82 0060,0061,0062,0063,0064,0065,0066,0067,
83 0070,0071,0072,0073,0074,0075,0076,0077,
84 0100,0141,0142,0143,0144,0145,0146,0147,
85 0150,0151,0152,0153,0154,0155,0156,0157,
86 0160,0161,0162,0163,0164,0165,0166,0167,
87 0170,0171,0172,0133,0134,0135,0136,0137,
88 0140,0141,0142,0143,0144,0145,0146,0147,
89 0150,0151,0152,0153,0154,0155,0156,0157,
90 0160,0161,0162,0163,0164,0165,0166,0167,
91 0170,0171,0172,0173,0174,0175,0176,0177,
93 0200,0201,0202,0203,0204,0205,0206,0207,
94 0210,0211,0212,0213,0214,0215,0216,0217,
95 0220,0221,0222,0223,0224,0225,0226,0227,
96 0230,0231,0232,0233,0234,0235,0236,0237,
97 0240,0241,0242,0243,0244,0245,0246,0247,
98 0250,0251,0252,0253,0254,0255,0256,0257,
99 0260,0261,0262,0263,0264,0265,0266,0267,
100 0270,0271,0272,0273,0274,0275,0276,0277,
101 0300,0301,0302,0303,0304,0305,0306,0307,
102 0310,0311,0312,0313,0314,0315,0316,0317,
103 0320,0321,0322,0323,0324,0325,0326,0327,
104 0330,0331,0332,0333,0334,0335,0336,0337,
105 0340,0341,0342,0343,0344,0345,0346,0347,
106 0350,0351,0352,0353,0354,0355,0356,0357,
107 0360,0361,0362,0363,0364,0365,0366,0367,
108 0370,0371,0372,0373,0374,0375,0376,0377,
112 * DEFINITIONS FOR NEXUS
115 #define nxtarg() (*argp ? *argp++ : NULL)
116 #define prvarg() argp--
118 #define padvise if (!talked++) advise
124 /* for {OR,AND,NOT}action */
126 struct nexus
*un_L_child
;
127 struct nexus
*un_R_child
;
134 char un_expbuf
[ESIZE
];
147 #define n_L_child un.st1.un_L_child
148 #define n_R_child un.st1.un_R_child
150 #define n_header un.st2.un_header
151 #define n_circf un.st2.un_circf
152 #define n_expbuf un.st2.un_expbuf
153 #define n_patbuf un.st2.un_patbuf
155 #define n_datef un.st3.un_datef
156 #define n_after un.st3.un_after
157 #define n_tws un.st3.un_tws
164 static struct nexus
*head
;
167 * prototypes for date routines
169 static struct tws
*tws_parse(char *, int);
170 static struct tws
*tws_special(char *);
175 static void PRaction(struct nexus
*, int);
176 static int gcompile(struct nexus
*, char *);
177 static int advance(char *, char *);
178 static int cclass(unsigned char *, int, int);
179 static int tcompile(char *, struct tws
*, int);
181 static struct nexus
*parse(void);
182 static struct nexus
*nexp1(void);
183 static struct nexus
*nexp2(void);
184 static struct nexus
*nexp3(void);
185 static struct nexus
*newnexus(int (*)());
187 static int ORaction();
188 static int ANDaction();
189 static int NOTaction();
190 static int GREPaction();
191 static int TWSaction();
195 pcompile (char **vec
, char *date
)
198 if ((datesw
= date
) == NULL
)
202 if ((head
= parse ()) == NULL
)
203 return (talked
? 0 : 1);
206 padvise (NULL
, "%s unexpected", *argp
);
214 static struct nexus
*
220 if ((n
= nexp1 ()) == NULL
|| (cp
= nxtarg ()) == NULL
)
224 padvise (NULL
, "%s unexpected", cp
);
230 switch (smatch (cp
, parswit
)) {
232 ambigsw (cp
, parswit
);
236 fprintf (stderr
, "-%s unknown\n", cp
);
241 o
= newnexus (ORaction
);
243 if ((o
->n_R_child
= parse ()))
245 padvise (NULL
, "missing disjunctive");
256 static struct nexus
*
262 if ((n
= nexp2 ()) == NULL
|| (cp
= nxtarg ()) == NULL
)
266 padvise (NULL
, "%s unexpected", cp
);
273 switch (smatch (cp
, parswit
)) {
275 ambigsw (cp
, parswit
);
280 fprintf (stderr
, "-%s unknown\n", cp
);
286 o
= newnexus (ANDaction
);
288 if ((o
->n_R_child
= nexp1 ()))
290 padvise (NULL
, "missing conjunctive");
302 static struct nexus
*
308 if ((cp
= nxtarg ()) == NULL
)
318 switch (smatch (cp
, parswit
)) {
320 ambigsw (cp
, parswit
);
324 fprintf (stderr
, "-%s unknown\n", cp
);
329 n
= newnexus (NOTaction
);
330 if ((n
->n_L_child
= nexp3 ()))
332 padvise (NULL
, "missing negation");
343 static struct nexus
*
348 char buffer
[BUFSIZ
], temp
[64];
351 if ((cp
= nxtarg ()) == NULL
)
355 padvise (NULL
, "%s unexpected", cp
);
363 switch (i
= smatch (cp
, parswit
)) {
365 ambigsw (cp
, parswit
);
369 fprintf (stderr
, "-%s unknown\n", cp
);
374 if ((n
= parse ()) == NULL
) {
375 padvise (NULL
, "missing group");
378 if ((cp
= nxtarg ()) == NULL
) {
379 padvise (NULL
, "missing -rbrace");
382 if (*cp
++ == '-' && smatch (cp
, parswit
) == PRRBR
)
384 padvise (NULL
, "%s unexpected", --cp
);
396 strncpy(temp
, parswit
[i
].sw
, sizeof(temp
));
397 temp
[sizeof(temp
) - 1] = '\0';
398 dp
= *brkstring (temp
, " ", NULL
);
400 if (!(cp
= nxtarg ())) {/* allow -xyz arguments */
401 padvise (NULL
, "missing argument to %s", argp
[-2]);
404 n
= newnexus (GREPaction
);
406 snprintf (buffer
, sizeof(buffer
), "^%s[ \t]*:.*%s", dp
, cp
);
411 n
= newnexus (GREPaction
);
413 if (!(cp
= nxtarg ())) {/* allow -xyz arguments */
414 padvise (NULL
, "missing argument to %s", argp
[-2]);
420 if (!gcompile (n
, dp
)) {
421 padvise (NULL
, "pattern error in %s %s", argp
[-2], cp
);
425 n
->n_patbuf
= getcpy (dp
);
429 padvise (NULL
, "internal error!");
433 if (!(datesw
= nxtarg ()) || *datesw
== '-') {
434 padvise (NULL
, "missing argument to %s", argp
[-2]);
441 if (!(cp
= nxtarg ())) {/* allow -xyz arguments */
442 padvise (NULL
, "missing argument to %s", argp
[-2]);
445 n
= newnexus (TWSaction
);
447 if (!tcompile (cp
, &n
->n_tws
, n
->n_after
= i
== PRAFTR
)) {
448 padvise (NULL
, "unable to parse %s %s", argp
[-2], cp
);
457 static struct nexus
*
458 newnexus (int (*action
)())
463 p
->n_action
= action
;
468 #define args(a) a, fp, msgnum, start, stop
469 #define params args (n)
478 pmatches (FILE *fp
, int msgnum
, long start
, long stop
, int debug
)
483 if (!talked
++ && debug
)
486 return (*head
->n_action
) (args (head
));
491 PRaction (struct nexus
*n
, int level
)
495 for (i
= 0; i
< level
; i
++)
496 fprintf (stderr
, "| ");
498 if (n
->n_action
== ORaction
) {
499 fprintf (stderr
, "OR\n");
500 PRaction (n
->n_L_child
, level
+ 1);
501 PRaction (n
->n_R_child
, level
+ 1);
504 if (n
->n_action
== ANDaction
) {
505 fprintf (stderr
, "AND\n");
506 PRaction (n
->n_L_child
, level
+ 1);
507 PRaction (n
->n_R_child
, level
+ 1);
510 if (n
->n_action
== NOTaction
) {
511 fprintf (stderr
, "NOT\n");
512 PRaction (n
->n_L_child
, level
+ 1);
515 if (n
->n_action
== GREPaction
) {
516 fprintf (stderr
, "PATTERN(%s) %s\n",
517 n
->n_header
? "header" : "body", n
->n_patbuf
);
520 if (n
->n_action
== TWSaction
) {
521 fprintf (stderr
, "TEMPORAL(%s) %s: %s\n",
522 n
->n_after
? "after" : "before", n
->n_datef
,
523 dasctime (&n
->n_tws
, TW_NULL
));
526 fprintf (stderr
, "UNKNOWN(0x%x)\n",
527 (unsigned int)(unsigned long) (*n
->n_action
));
535 if ((*n
->n_L_child
->n_action
) (args (n
->n_L_child
)))
537 return (*n
->n_R_child
->n_action
) (args (n
->n_R_child
));
545 if (!(*n
->n_L_child
->n_action
) (args (n
->n_L_child
)))
547 return (*n
->n_R_child
->n_action
) (args (n
->n_R_child
));
555 return (!(*n
->n_L_child
->n_action
) (args (n
->n_L_child
)));
560 gcompile (struct nexus
*n
, char *astr
)
564 unsigned char *ep
, *dp
, *sp
, *lastep
= 0;
566 dp
= (ep
= (unsigned char *) n
->n_expbuf
) + sizeof n
->n_expbuf
;
567 sp
= (unsigned char *) astr
;
577 if ((c
= *sp
++) != '*')
604 if ((c
= *sp
++) == '^') {
614 if (c
== '-' && *sp
!= '\0' && *sp
!= ']') {
615 for (c
= ep
[-1]+1; c
< *sp
; c
++) {
618 if (c
== '\0' || ep
>= dp
)
624 if (c
== '\0' || ep
>= dp
)
627 } while ((c
= *sp
++) != ']');
634 if ((c
= *sp
++) == '\0')
654 char *p1
, *p2
, *ebp
, *cbp
;
658 fseek (fp
, start
, SEEK_SET
);
662 if (body
&& n
->n_header
)
669 if (fgets (ibuf
, sizeof ibuf
, fp
) == NULL
670 || (stop
&& pos
>= stop
)) {
675 pos
+= (long) strlen (ibuf
);
677 ebp
= ibuf
+ strlen (ibuf
);
680 if (lf
&& c
!= '\n') {
681 if (c
!= ' ' && c
!= '\t') {
697 /* Unfold by skipping the newline. */
701 if (c
&& p1
< &linebuf
[LBSIZE
- 1])
710 /* Attempt to decode as a MIME header. If it's the last header,
711 body will be 1 and lf will be at least 1. */
712 if ((body
== 0 || lf
> 0) &&
713 decode_rfc2047 (linebuf
, decoded_linebuf
, sizeof decoded_linebuf
)) {
714 p1
= decoded_linebuf
;
718 if (advance (p1
, p2
))
726 if (*p1
== c
|| cc
[(unsigned char)*p1
] == c
)
727 if (advance (p1
, p2
))
734 if (advance (p1
, p2
))
742 advance (char *alp
, char *aep
)
744 unsigned char *lp
, *ep
, *curlp
;
746 lp
= (unsigned char *)alp
;
747 ep
= (unsigned char *)aep
;
751 if (*ep
++ == *lp
++ || ep
[-1] == cc
[lp
[-1]])
769 if (cclass (ep
, *lp
++, 1)) {
776 if (cclass (ep
, *lp
++, 0)) {
790 while (*lp
++ == *ep
|| cc
[lp
[-1]] == *ep
)
798 while (cclass (ep
, *lp
++, ep
[-1] == (CCL
| STAR
)))
806 if (advance ((char *) lp
, (char *) ep
))
808 } while (lp
> curlp
);
812 admonish (NULL
, "advance() botch -- you lose big");
819 cclass (unsigned char *aset
, int ac
, int af
)
822 unsigned char c
, *set
;
830 if (*set
++ == c
|| set
[-1] == cc
[c
])
838 tcompile (char *ap
, struct tws
*tb
, int isafter
)
842 if ((tw
= tws_parse (ap
, isafter
)) == NULL
)
851 tws_parse (char *ap
, int isafter
)
856 if ((tw
= tws_special (ap
)) != NULL
) {
857 tw
->tw_sec
= tw
->tw_min
= isafter
? 59 : 0;
858 tw
->tw_hour
= isafter
? 23 : 0;
861 if ((tw
= dparsetime (ap
)) != NULL
)
864 if ((ts
= dlocaltimenow ()) == NULL
)
867 snprintf (buffer
, sizeof(buffer
), "%s %s", ap
, dtwszone (ts
));
868 if ((tw
= dparsetime (buffer
)) != NULL
)
871 snprintf (buffer
, sizeof(buffer
), "%s %02d:%02d:%02d %s", ap
,
872 ts
->tw_hour
, ts
->tw_min
, ts
->tw_sec
, dtwszone (ts
));
873 if ((tw
= dparsetime (buffer
)) != NULL
)
876 snprintf (buffer
, sizeof(buffer
), "%02d %s %04d %s",
877 ts
->tw_mday
, tw_moty
[ts
->tw_mon
], ts
->tw_year
, ap
);
878 if ((tw
= dparsetime (buffer
)) != NULL
)
881 snprintf (buffer
, sizeof(buffer
), "%02d %s %04d %s %s",
882 ts
->tw_mday
, tw_moty
[ts
->tw_mon
], ts
->tw_year
,
884 if ((tw
= dparsetime (buffer
)) != NULL
)
892 tws_special (char *ap
)
899 if (!strcasecmp (ap
, "today"))
900 return dlocaltime (&clock
);
901 if (!strcasecmp (ap
, "yesterday")) {
902 clock
-= (long) (60 * 60 * 24);
903 return dlocaltime (&clock
);
905 if (!strcasecmp (ap
, "tomorrow")) {
906 clock
+= (long) (60 * 60 * 24);
907 return dlocaltime (&clock
);
910 for (i
= 0; tw_ldotw
[i
]; i
++)
911 if (!strcasecmp (ap
, tw_ldotw
[i
]))
914 if ((tw
= dlocaltime (&clock
)) == NULL
)
916 if ((i
-= tw
->tw_wday
) > 0)
922 else /* -ddd days ago */
923 i
= atoi (ap
); /* we should error check this */
925 clock
+= (long) ((60 * 60 * 24) * i
);
926 return dlocaltime (&clock
);
936 char buf
[BUFSIZ
], name
[NAMESZ
];
938 m_getfld_state_t gstate
= 0;
941 fseek (fp
, start
, SEEK_SET
);
943 int bufsz
= sizeof buf
;
944 switch (state
= m_getfld (&gstate
, name
, buf
, &bufsz
, fp
)) {
948 free (bp
), bp
= NULL
;
949 bp
= add (buf
, NULL
);
950 while (state
== FLDPLUS
) {
952 state
= m_getfld (&gstate
, name
, buf
, &bufsz
, fp
);
955 if (!strcasecmp (name
, n
->n_datef
))
963 if (state
== LENERR
|| state
== FMTERR
)
964 advise (NULL
, "format error in message %d", msgnum
);
970 adios (NULL
, "internal error -- you lose");
974 m_getfld_state_destroy (&gstate
);
976 if ((tw
= dparsetime (bp
)) == NULL
)
977 advise (NULL
, "unable to parse %s field in message %d, matching...",
978 n
->n_datef
, msgnum
), state
= 1;
980 state
= n
->n_after
? (twsort (tw
, &n
->n_tws
) > 0)
981 : (twsort (tw
, &n
->n_tws
) < 0);