pith/pattern.h

   1 /*
   2  * $Id: pattern.h 942 2008-03-04 18:21:33Z hubert@u.washington.edu $
   3  *
   4  * ========================================================================
   5  * Copyright 2013-2022 Eduardo Chappa
   6  * Copyright 2006-2008 University of Washington
   7  *
   8  * Licensed under the Apache License, Version 2.0 (the "License");
   9  * you may not use this file except in compliance with the License.
  10  * You may obtain a copy of the License at
  11  *
  12  *     http://www.apache.org/licenses/LICENSE-2.0
  13  *
  14  * ========================================================================
  15  */
  16
  17 #ifndef PITH_PATTERN_INCLUDED
  18 #define PITH_PATTERN_INCLUDED
  19
  20
  21 #include "../pith/msgno.h"
  22 #include "../pith/sorttype.h"
  23 #include "../pith/string.h"
  24 #include "../pith/indxtype.h"
  25
  26
  27 /*
  28  * This structure is used to contain strings which are matched against
  29  * header fields. The match is a simple substring match. The match is
  30  * an OR of all the patterns in the PATTERN_S list. That is,
  31  * substring1_matches OR substring2_matches OR substring3_matches.
  32  * If not is set in the _head_ of the PATTERN_S, it is a NOT of the
  33  * whole pattern, that is,
  34  * NOT (substring1_matches OR substring2_matches OR substring3_matches).
  35  * The not variable is not meaningful except in the head member of the
  36  * PATTERN_S list.
  37  */
  38 typedef struct pattern_s {
  39     int               not;              /* NOT of whole pattern */
  40     char             *substring;
  41     struct pattern_s *next;
  42 } PATTERN_S;
  43
  44 /*
  45  * List of these is a list of arbitrary freetext headers and patterns.
  46  * This may be part of a pattern group.
  47  * The isemptyval bit is to keep track of the difference between an arb
  48  * header with no value set and one with the empty value "" set. For the
  49  * other builtin headers this difference is kept track of by whether or
  50  * not the header is in the config file at all or not. Here we want to
  51  * be able to add a header to the config file without necessarily giving
  52  * it a value.
  53  */
  54 typedef struct arbhdr_s {
  55     char            *field;
  56     PATTERN_S       *p;
  57     int              isemptyval;
  58     struct arbhdr_s *next;
  59 } ARBHDR_S;
  60
  61 /*
  62  * A list of intervals of integers.
  63  */
  64 typedef struct intvl_s {
  65     long            imin, imax;
  66     struct intvl_s *next;
  67 } INTVL_S;
  68
  69 /*
  70  * A Pattern group gives characteristics of an envelope to match against. Any of
  71  * the characteristics (to, from, ...) which is non-null must match for the
  72  * whole thing to be considered a match. That is, it is an AND of all the
  73  * non-null members.
  74  */
  75 typedef struct patgrp_s {
  76     char      *nick;
  77     char      *comment;         /* for user, not used for anything */
  78     PATTERN_S *to,
  79               *from,
  80               *sender,
  81               *cc,
  82               *recip,
  83               *partic,
  84               *news,
  85               *subj,
  86               *alltext,
  87               *bodytext,
  88               *keyword,
  89               *charsets;
  90     STRLIST_S *charsets_list;   /* used for efficiency, computed from charset */
  91     ARBHDR_S  *arbhdr;          /* list of arbitrary hdrnames and patterns */
  92     int        fldr_type;       /* see FLDR_* below                     */
  93     PATTERN_S *folder;          /* folder if type FLDR_SPECIFIC         */
  94     int        inabook;         /* see IAB_* below                      */
  95     PATTERN_S *abooks;
  96     int        do_score;
  97     INTVL_S   *score;
  98     int        do_age;
  99     INTVL_S   *age;             /* ages are in days                     */
 100     int        do_size;
 101     INTVL_S   *size;
 102     int        age_uses_sentdate; /* on or off                          */
 103     int        do_cat;
 104     char     **category_cmd;
 105     INTVL_S   *cat;
 106     long       cat_lim;         /* -1 no limit  0 only headers          */
 107     int        bogus;           /* patgrp contains unknown stuff        */
 108     int        stat_new,        /* msg status is New (Unseen)           */
 109                stat_rec,        /* msg status is Recent                 */
 110                stat_del,        /* msg status is Deleted                */
 111                stat_imp,        /* msg is flagged Important             */
 112                stat_ans,        /* msg is flagged Answered              */
 113                stat_8bitsubj,   /* subject contains 8bit chars          */
 114                stat_bom,        /* this is first pine run of the month  */
 115                stat_boy;        /* this is first pine run of the year   */
 116 } PATGRP_S;
 117
 118 #define FLDR_ANY        0
 119 #define FLDR_NEWS       1
 120 #define FLDR_EMAIL      2
 121 #define FLDR_SPECIFIC   3
 122
 123 #define FLDR_DEFL       FLDR_EMAIL
 124
 125
 126 #define IAB_EITHER              0x0     /* don't care if in or not */
 127
 128 #define IAB_TYPE_MASK           0xf
 129 #define IAB_YES                 0x1     /* addresses in any abook */
 130 #define IAB_NO                  0x2     /*   "      not  " */
 131 #define IAB_SPEC_YES            0x3     /* addresses in specific abooks */
 132 #define IAB_SPEC_NO             0x4
 133
 134 /*
 135  * Warning about reply-to. We're using the c-client envelope reply-to which
 136  * means if there isn't a real reply-to it uses the From!
 137  */
 138 #define IAB_ADDR_MASK           0xff0
 139 #define IAB_FROM                0x10    /* from address included in list */
 140 #define IAB_REPLYTO             0x20    /* reply-to address included in list */
 141 #define IAB_SENDER              0x40    /* sender address included in list */
 142 #define IAB_TO                  0x80    /* to address included in list */
 143 #define IAB_CC                  0x100   /* cc address included in list */
 144
 145 #define IAB_DEFL                IAB_EITHER
 146
 147
 148 #define FILTER_STATE    0
 149 #define FILTER_KILL     1
 150 #define FILTER_FOLDER   2
 151
 152 /*
 153  * For the Status parts of a PATGRP_S. For example, stat_del is Deleted
 154  * status. User sets EITHER means they don't care, it always matches.
 155  * YES means it must be deleted to match. NO means it must not be deleted.
 156  */
 157 #define PAT_STAT_EITHER         0  /* we don't care which, yes or no    */
 158 #define PAT_STAT_YES            1  /* yes, this status is true          */
 159 #define PAT_STAT_NO             2  /* no, this status is not true       */
 160
 161 /*
 162  * For the State setting part of a filter action
 163  */
 164 #define ACT_STAT_LEAVE          0  /* leave msg state alone             */
 165 #define ACT_STAT_SET            1  /* set this part of msg state        */
 166 #define ACT_STAT_CLEAR          2  /* clear this part of msg state      */
 167
 168 typedef struct action_s {
 169     unsigned     is_a_role:1;   /* this is a role action                */
 170     unsigned     is_a_incol:1;  /* this is an index color action        */
 171     unsigned     is_a_score:1;  /* this is a score setting action       */
 172     unsigned     is_a_filter:1; /* this is a filter action              */
 173     unsigned     is_a_other:1;  /* this is a miscellaneous action       */
 174     unsigned     is_a_srch:1;   /* this is for Select cmd, no action    */
 175     unsigned     bogus:1;       /* action contains unknown stuff        */
 176     unsigned     been_here_before:1;  /* inheritance loop prevention    */
 177             /* --- These are for roles --- */
 178     ADDRESS     *from;          /* value to set for From                */
 179     ADDRESS     *replyto;       /* value to set for Reply-To            */
 180     char       **cstm;          /* custom headers                       */
 181     char       **smtp;          /* custom SMTP server for this role     */
 182     char       **nntp;          /* custom NNTP server for this role     */
 183     char        *fcc;           /* value to set for Fcc                 */
 184     char        *litsig;        /* value to set Literal Signature       */
 185     char        *sig;           /* value to set for Sig File            */
 186     char        *template;      /* value to set for Template            */
 187     char        *nick;          /* value to set for Nickname            */
 188     int          repl_type;     /* see ROLE_REPL_* below                */
 189     int          forw_type;     /* see ROLE_FORW_* below                */
 190     int          comp_type;     /* see ROLE_COMP_* below                */
 191     char        *inherit_nick;  /* pattern we inherit actions from      */
 192             /* --- This is for indexcoloring --- */
 193     COLOR_PAIR  *incol;         /* colors for index line                */
 194             /* --- This is for scoring --- */
 195     long         scoreval;
 196     HEADER_TOK_S *scorevalhdrtok;
 197             /* --- These are for filtering --- */
 198     int          kill;
 199     long         state_setting_bits;
 200     PATTERN_S   *keyword_set;   /* set these keywords                   */
 201     PATTERN_S   *keyword_clr;   /* clear these keywords                 */
 202     PATTERN_S   *folder;        /* folders to recv. filtered mail       */
 203     int          move_only_if_not_deleted;      /* on or off            */
 204     int          non_terminating;               /* on or off            */
 205             /* --- These are for other --- */
 206               /* sort order of folder */
 207     unsigned     sort_is_set:1;
 208     SortOrder    sortorder;     /* sorting order                        */
 209     int          revsort;       /* whether or not to reverse sort       */
 210               /* Index format of folder */
 211     char        *index_format;
 212     unsigned     startup_rule;
 213 } ACTION_S;
 214
 215 /* flags for first_pattern..., set_role_from_msg, and confirm_role() */
 216 #define PAT_CLOSED      0x00000000      /* closed                            */
 217 #define PAT_OPENED      0x00000001      /* opened successfully               */
 218 #define PAT_OPEN_FAILED 0x00000002
 219 #define PAT_USE_CURRENT 0x00000010      /* use current_val to set up pattern */
 220 #define PAT_USE_CHANGED 0x00000020      /* use changed_val to set up pattern */
 221 #define PAT_USE_MAIN    0x00000040      /* use main_user_val                 */
 222 #define PAT_USE_POST    0x00000080      /* use post_user_val                 */
 223 #define ROLE_COMPOSE    0x00000100      /* roles with compose value != NO   */
 224 #define ROLE_REPLY      0x00000200      /* roles with reply value != NO     */
 225 #define ROLE_FORWARD    0x00000400      /* roles with forward value != NO   */
 226 #define ROLE_INCOL      0x00000800      /* patterns with non-Normal colors  */
 227 #define ROLE_SCORE      0x00001000      /* patterns with non-zero scorevals */
 228 #define ROLE_DO_ROLES   0x00010000      /* role patterns                    */
 229 #define ROLE_DO_INCOLS  0x00020000      /* index line color patterns        */
 230 #define ROLE_DO_SCORES  0x00040000      /* set score patterns               */
 231 #define ROLE_DO_FILTER  0x00080000      /* filter patterns                  */
 232 #define ROLE_DO_OTHER   0x00100000      /* miscellaneous patterns           */
 233 #define ROLE_DO_SRCH    0x00200000      /* index line color patterns        */
 234 #define ROLE_OLD_PAT    0x00400000      /* old patterns variable            */
 235 #define ROLE_OLD_FILT   0x00800000      /* old patterns-filters variable    */
 236 #define ROLE_OLD_SCORE  0x01000000      /* old patterns-scores variable     */
 237 #define ROLE_CHANGES    0x02000000      /* start editing with changes
 238                                            already registered */
 239
 240 #define PAT_OPEN_MASK   0x0000000f
 241 #define PAT_USE_MASK    0x000000f0
 242 #define ROLE_MASK       0x00ffff00
 243
 244 #define ROLE_REPL_NO            0  /* never use for reply                */
 245 #define ROLE_REPL_YES           1  /* use for reply with confirmation    */
 246 #define ROLE_REPL_NOCONF        2  /* use for reply without confirmation */
 247 #define ROLE_FORW_NO            0  /* ... forward ...                    */
 248 #define ROLE_FORW_YES           1
 249 #define ROLE_FORW_NOCONF        2
 250 #define ROLE_COMP_NO            0  /* ... compose ...                    */
 251 #define ROLE_COMP_YES           1
 252 #define ROLE_COMP_NOCONF        2
 253
 254 #define ROLE_REPL_DEFL          ROLE_REPL_YES   /* default reply value */
 255 #define ROLE_FORW_DEFL          ROLE_FORW_YES   /* default forward value */
 256 #define ROLE_COMP_DEFL          ROLE_COMP_NO    /* default compose value */
 257 #define ROLE_NOTAROLE_DEFL      ROLE_COMP_NO
 258
 259 #define INTVL_INF         (2147483646L)
 260 #define INTVL_UNDEF       (INTVL_INF + 1L)
 261 #define SCORE_UNDEF       INTVL_UNDEF
 262 #define SCORE_MIN         (-100)
 263 #define SCORE_MAX         (100)
 264 #define SCOREUSE_GET      0x000
 265 #define SCOREUSE_INVALID  0x001 /* will recalculate scores_in_use next time */
 266 #define SCOREUSE_ROLES    0x010 /* scores are used for roles                */
 267 #define SCOREUSE_INCOLS   0x020 /* scores are used for index line colors    */
 268 #define SCOREUSE_FILTERS  0x040 /* scores are used for filters              */
 269 #define SCOREUSE_OTHER    0x080 /* scores are used for miscellaneous stuff  */
 270 #define SCOREUSE_INDEX    0x100 /* scores are used in index-format          */
 271 #define SCOREUSE_STATEDEP 0x200 /* scores depend on message state           */
 272
 273 /*
 274  * A message is compared with a pattern group to see if it matches.
 275  * If it does match, then there are actions which are taken.
 276  */
 277 typedef struct pat_s {
 278     PATGRP_S          *patgrp;
 279     ACTION_S          *action;
 280     struct pat_line_s *patline;         /* pat_line that goes with this pat */
 281     char              *raw;
 282     unsigned           inherit:1;
 283     struct pat_s      *next;
 284     struct pat_s      *prev;
 285 } PAT_S;
 286
 287 typedef enum {TypeNotSet = 0, Literal, File, Inherit} PAT_TYPE;
 288
 289 /*
 290  * There's one of these for each line in the pinerc variable.
 291  * Normal type=Literal patterns have a patline with both first and last
 292  * pointing to the pattern. Type File has one patline for the file and first
 293  * and last point to the first and last patterns in the file.
 294  * The patterns aren't linked into one giant list, the patlines are.
 295  * To traverse all the patterns you have to go through the patline list
 296  * and then for each patline go from first to last through the patterns.
 297  * That's what next_pattern and friends do.
 298  */
 299 typedef struct pat_line_s {
 300     PAT_TYPE           type;
 301     PAT_S             *first;   /* 1st pattern in list belonging to this line */
 302     PAT_S             *last;
 303     char              *filename; /* If type File, the filename */
 304     char              *filepath;
 305     unsigned           readonly:1;
 306     unsigned           dirty:1; /* needs to be written back to storage */
 307     struct pat_line_s *next;
 308     struct pat_line_s *prev;
 309 } PAT_LINE_S;
 310
 311 typedef struct pat_handle {
 312     PAT_LINE_S *patlinehead;    /* list of in-core, parsed pat lines */
 313     unsigned    dirtypinerc:1;  /* needs to be written */
 314 } PAT_HANDLE;
 315
 316 typedef struct pat_state {
 317     long        rflags;
 318     int         cur_rflag_num;
 319     PAT_LINE_S *patlinecurrent;
 320     PAT_S      *patcurrent;     /* current pat within patline */
 321 } PAT_STATE;
 322
 323 #define PATTERN_MAGIC     "P#Pats"
 324 #define PATTERN_FILE_VERS "01"
 325
 326
 327 /*
 328  * This is a little dangerous. We're passing flags to match_pattern and
 329  * peeling some of them off for our own use while passing the rest on
 330  * to mail_search_full. So we need to define ours so they don't overlap
 331  * with the c-client flags that can be passed to mail_search_full.
 332  * We could formalize it with mrc.
 333  */
 334 #define MP_IN_CCLIENT_CB        0x10000   /* we're in a c-client callback! */
 335 #define MP_NOT                  0x20000   /* use ! of patgrp for search    */
 336
 337
 338 /* match_pattern_folder_specific flags */
 339 #define FOR_PATTERN             0x01
 340 #define FOR_FILTER              0x02
 341 #define FOR_OPTIONSCREEN        0x04
 342
 343
 344 extern PAT_HANDLE **cur_pat_h;
 345
 346
 347 /* exported prototypes */
 348 void        role_process_filters(void);
 349 int         add_to_pattern(PAT_S *, long);
 350 char       *add_pat_escapes(char *);
 351 char       *remove_pat_escapes(char *);
 352 char       *add_roletake_escapes(char *);
 353 char       *add_comma_escapes(char *);
 354 void        set_pathandle(long);
 355 void        close_every_pattern(void);
 356 void        close_patterns(long);
 357 int         nonempty_patterns(long, PAT_STATE *);
 358 int         any_patterns(long, PAT_STATE *);
 359 int         edit_pattern(PAT_S *, int, long);
 360 int         add_pattern(PAT_S *, long);
 361 int         delete_pattern(int, long);
 362 int         shuffle_pattern(int, int, long);
 363 PAT_LINE_S *parse_pat_file(char *);
 364 INTVL_S    *parse_intvl(char *);
 365 char       *stringform_of_intvl(INTVL_S *);
 366 char       *hdrtok_to_stringform(HEADER_TOK_S *);
 367 HEADER_TOK_S *stringform_to_hdrtok(char *);
 368 char       *hdrtok_to_config(HEADER_TOK_S *);
 369 HEADER_TOK_S *config_to_hdrtok(char *);
 370 int         scores_are_used(int);
 371 int         patgrp_depends_on_state(PATGRP_S *);
 372 int         patgrp_depends_on_active_state(PATGRP_S *);
 373 PATTERN_S  *parse_pattern(char *, char *, int);
 374 PATTERN_S  *string_to_pattern(char *);
 375 char       *pattern_to_string(PATTERN_S *);
 376 char       *pattern_to_config(PATTERN_S *);
 377 PATTERN_S  *config_to_pattern(char *);
 378 PATTERN_S  *editlist_to_pattern(char **);
 379 char      **pattern_to_editlist(PATTERN_S *);
 380 PATGRP_S   *nick_to_patgrp(char *, int);
 381 PAT_S      *first_pattern(PAT_STATE *);
 382 PAT_S      *last_pattern(PAT_STATE *);
 383 PAT_S      *prev_pattern(PAT_STATE *);
 384 PAT_S      *next_pattern(PAT_STATE *);
 385 int         write_patterns(long);
 386 void        convert_statebits_to_vals(long, int *, int *, int *, int *);
 387 int         match_pattern(PATGRP_S *, MAILSTREAM *, SEARCHSET *,char *,
 388                           long (*)(MAILSTREAM *, long), long);
 389 void        find_8bitsubj_in_messages(MAILSTREAM *, SEARCHSET *, int, int);
 390 void        find_charsets_in_messages(MAILSTREAM *, SEARCHSET *, PATGRP_S *, int);
 391 int         compare_strlists_for_match(STRLIST_S *, STRLIST_S *);
 392 int         match_pattern_folder(PATGRP_S *, MAILSTREAM *);
 393 int         match_pattern_folder_specific(PATTERN_S *, MAILSTREAM *, int);
 394 SEARCHPGM  *match_pattern_srchpgm(PATGRP_S   *, MAILSTREAM *, SEARCHSET *);
 395 void        calc_extra_hdrs(void);
 396 char       *get_extra_hdrs(void);
 397 void        free_extra_hdrs(void);
 398 void        free_pat(PAT_S **);
 399 void        free_pattern(PATTERN_S **);
 400 void        free_action(ACTION_S **);
 401 PAT_S      *copy_pat(PAT_S *);
 402 PATGRP_S   *copy_patgrp(PATGRP_S *);
 403 ACTION_S   *copy_action(ACTION_S *);
 404 ACTION_S   *combine_inherited_role(ACTION_S *);
 405 void        mail_expunge_prefilter(MAILSTREAM *, int);
 406 void        process_filter_patterns(MAILSTREAM *, MSGNO_S *, long);
 407 void        reprocess_filter_patterns(MAILSTREAM *, MSGNO_S *, int);
 408 int         trivial_patgrp(PATGRP_S *);
 409 void        free_patgrp(PATGRP_S **);
 410 void        free_patline(PAT_LINE_S **patline);
 411 int         some_filter_depends_on_active_state(void);
 412 void        delete_filtered_msgs(MAILSTREAM *);
 413 void        free_intvl(INTVL_S **);
 414 void        free_arbhdr(ARBHDR_S **);
 415
 416
 417 #endif /* PITH_PATTERN_INCLUDED */