diplodocus.org Git - nmh/blob - h/fmt_scan.h

   1
   2 /*
   3  * fmt_scan.h -- definitions for fmt_scan()
   4  */
   5
   6 /*
   7  * This structure describes an "interesting" component.  It holds
   8  * the name & text from the component (if found) and one piece of
   9  * auxiliary info.  The structure for a particular component is located
  10  * by (open) hashing the name and using it as an index into the ptr array
  11  * "wantcomp".  All format entries that reference a particular component
  12  * point to its comp struct (so we only have to do component specific
  13  * processing once.  e.g., parse an address.).
  14  *
  15  * In previous implementations "wantcomp" was made available to other
  16  * functions, but now it's private and is accessed via functions.
  17  */
  18 struct comp {
  19     char        *c_name;        /* component name (in lower case) */
  20     char        *c_text;        /* component text (if found)      */
  21     struct comp *c_next;        /* hash chain linkage             */
  22     short        c_flags;       /* misc. flags (from fmt_scan)    */
  23     short        c_type;        /* type info   (from fmt_compile) */
  24     union {
  25         struct tws *c_u_tws;
  26         struct mailname *c_u_mn;
  27     } c_un;
  28     int          c_refcount;    /* Reference count                */
  29 };
  30
  31 #define c_tws c_un.c_u_tws
  32 #define c_mn  c_un.c_u_mn
  33
  34 /*
  35  * c_type bits
  36  */
  37 #define CT_ADDR       (1<<0)    /* referenced as address    */
  38 #define CT_DATE       (1<<1)    /* referenced as date       */
  39
  40 #define CT_BITS       "\020\01ADDR\02DATE"
  41
  42 /*
  43  * c_flags bits
  44  */
  45 #define CF_TRUE       (1<<0)    /* usually means component is present */
  46 #define CF_PARSED     (1<<1)    /* address/date has been parsed */
  47 #define CF_DATEFAB    (1<<2)    /* datefield fabricated */
  48 #define CF_TRIMMED    (1<<3)    /* Component has been trimmed */
  49
  50 #define CF_BITS       "\020\01TRUE\02PARSED\03CF_DATEFAB\04TRIMMED"
  51
  52 /*
  53  * This structure defines one formatting instruction.
  54  */
  55 struct format {
  56     unsigned char f_type;
  57     char          f_fill;
  58     short         f_width;      /* output field width   */
  59     union {
  60         struct comp *f_u_comp;  /* associated component */
  61         char        *f_u_text;  /* literal text         */
  62         char         f_u_char;  /* literal character    */
  63         int          f_u_value; /* literal value        */
  64     } f_un;
  65     short         f_flags;      /* misc. flags          */
  66 };
  67
  68 #define f_skip f_width          /* instr to skip (false "if") */
  69
  70 #define f_comp  f_un.f_u_comp
  71 #define f_text  f_un.f_u_text
  72 #define f_char  f_un.f_u_char
  73 #define f_value f_un.f_u_value
  74
  75 /*
  76  * f_flags bits
  77  */
  78
  79 #define FF_STRALLOC     (1<<0)  /* String has been allocated */
  80 #define FF_COMPREF      (1<<1)  /* Component reference */
  81
  82 /*
  83  * prototypes used by the format engine
  84  */
  85
  86 /*
  87  * These are the definitions used by the callbacks for fmt_scan()
  88  */
  89
  90 typedef char * (*formataddr_cb)(char *, char *);
  91 typedef char * (*concataddr_cb)(char *, char *);
  92 typedef void (*trace_cb)(void *, struct format *, int, char *, const char *);
  93
  94 struct fmt_callbacks {
  95     formataddr_cb       formataddr;
  96     concataddr_cb       concataddr;
  97     trace_cb            trace_func;
  98     void *              trace_context;
  99 };
 100
 101 /*
 102  * Create a new format string.  Arguments are:
 103  *
 104  * form         - Name of format file.  Will be searched by etcpath(), see that
 105  *                function for details.
 106  * format       - The format string to be used if no format file is given
 107  * default_fs   - The default format string to be used if neither form nor
 108  *                format is given
 109  *
 110  * This function also takes care of processing \ escapes like \n, \t, etc.
 111  *
 112  * Returns an allocated format string.
 113  */
 114
 115 char *new_fs (char *form, char *format, char *default_fs);
 116
 117 /*
 118  * Free memory allocated by new_fs().  It allocates to a static so
 119  * no argument is necessary.
 120  */
 121 void free_fs ();
 122
 123 /*
 124  * Compile a format string into a set of format instructions.  Arguments are:
 125  *
 126  * fstring      - The format string (the "source code").
 127  * fmt          - Returns an allocated array of "struct fmt" elements.  Each
 128  *                struct fmt is one format instruction interpreted by the
 129  *                format engine.
 130  * reset        - If set to true, the format compiler will reset the
 131  *                component hash table.  The component hash table contains
 132  *                all of the references to message components referred to in
 133  *                the format instructions.  If you have multiple format
 134  *                strings that you want to compile and operate on the
 135  *                same message, this should be set to false.
 136  *
 137  * Returns the total number of components referenced by all format instructions
 138  * since the last reset of the hash table.
 139  */
 140
 141 int fmt_compile (char *fstring, struct format **fmt, int reset);
 142
 143 /*
 144  * Interpret a sequence of compiled format instructions.  Arguments are:
 145  *
 146  * format       - Array of format instructions generated by fmt_compile()
 147  * scanl        - Passed-in charstring_t object (created with
 148  *                charstring_create() and later destroyed with
 149  *                charstring_free()) that will contain the output of the
 150  *                format instructions.  Is always terminated with a
 151  *                newline (\n).
 152  * width        - Maximum number of displayed characters.  Does not include
 153  *                characters marked as nonprinting or (depending on the
 154  *                encoding) bytes in a multibyte encoding that exceed the
 155  *                character's column width.
 156  * dat          - An integer array that contains data used by certain format
 157  *                functions.  Currently the following instructions use
 158  *                dat[]:
 159  *
 160  *              dat[0] - %(msg), %(dat)
 161  *              dat[1] - %(cur)
 162  *              dat[2] - %(size)
 163  *              dat[3] - %(width)
 164  *              dat[4] - %(unseen)
 165  *
 166  * callbacks    - A set of a callback functions used by the format engine.
 167  *                Can be NULL.  If structure elements are NULL, a default
 168  *                function will be used.  Callback structure elements are:
 169  *
 170  *              formataddr      - A callback for the %(formataddr) instruction
 171  *              concataddr      - A callback for the %(concataddr) instruction
 172  *              trace           - Called for every format instruction executed
 173  *
 174  * The return value is a pointer to the next format instruction to
 175  * execute, which is currently always NULL.
 176  */
 177
 178 struct format *fmt_scan (struct format *format, charstring_t scanl, int width,
 179                          int *dat, struct fmt_callbacks *callbacks);
 180
 181 /*
 182  * Free a format structure and/or component hash table.  Arguments are:
 183  *
 184  * format       - An array of format structures allocated by fmt_compile,
 185  *                or NULL.
 186  * reset        - If true, reset and remove all references in the component
 187  *                hash table.
 188  */
 189
 190 void fmt_free (struct format *fmt, int reset);
 191
 192 /*
 193  * Free all of the component text structures in the component hash table
 194  */
 195
 196 void fmt_freecomptext(void);
 197
 198 /*
 199  * Search for a component structure in the component hash table.  Arguments are:
 200  *
 201  * component    - The name of the component to search for.  By convention
 202  *                all component names used in format strings are lower case,
 203  *                but for backwards compatibility this search is done in
 204  *                a case-SENSITIVE manner.
 205  *
 206  * This function returns a "struct comp" corresponding to the named component,
 207  * or NULL if the component is not found in the hash table.
 208  */
 209
 210 struct comp *fmt_findcomp(char *component);
 211
 212 /*
 213  * Search for a component structure in the component hash table.
 214  *
 215  * Identical to fmd_findcomp(), but is case-INSENSITIVE.
 216  */
 217
 218 struct comp *fmt_findcasecomp(char *component);
 219
 220 /*
 221  * Add a component entry to the component hash table
 222  *
 223  * component    - The name of the component to add to the hash table.
 224  *
 225  * If the component is already in the hash table, this function will do
 226  * nothing.  Returns 1 if a component was added, 0 if it already existed.
 227  */
 228
 229 int fmt_addcompentry(char *component);
 230
 231 /*
 232  * Add a string to a component hash table entry.  Arguments are:
 233  *
 234  * component    - The name of the component to add text to.  The component
 235  *                is searched for in a case-INSENSITIVE manner (note that
 236  *                this is different than fmt_findcomp()).  If the component
 237  *                is not found in the hash table, this function will silently
 238  *                return.
 239  * text         - The text to add to a component hash table entry.  Note that
 240  *                if the last character of the existing component
 241  *                text is a newline AND it is marked as an address
 242  *                component (the the CT_ADDR flag is set) existing
 243  *                component buffer is a newline, it will be separated
 244  *                from previous text by ",\n\t"; otherwise if the last
 245  *                character of the previous text is a newline it will
 246  *                simply be separated by a "\t".  This unusual processing
 247  *                is designed to handle the case where you have multiple
 248  *                headers with the same name (e.g.: multiple "cc:" headers,
 249  *                even though that isn't technically allowed in the RFCs).
 250  *
 251  * This function is designed to be called when you start processing a new
 252  * component.  The function returns the integer value of the hash table
 253  * bucket corresponding to this component.  If there was no entry found
 254  * in the component hash table, this function will return -1.
 255  */
 256
 257 int fmt_addcomptext(char *component, char *text);
 258
 259 /*
 260  * Append to an existing component.  Arguments are:
 261  *
 262  * bucket       - The hash table bucket corresponding to this component,
 263  *                as returned by fmt_addcomp().  If -1, this function will
 264  *                return with no actions performed.
 265  * component    - The component to append text to.  Like fmt_addcomp, the
 266  *                component is searched case-INSENSITIVELY.
 267  * text         - The text to append to the component.  No special processing
 268  *                is done.
 269  *
 270  * This function is designed to be called when you are processing continuation
 271  * lines on the same header (state == FLDPLUS).
 272  */
 273
 274 void fmt_appendcomp(int bucket, char *component, char *text);
 275
 276 /*
 277  * Iterate over the complete hash table of component structures.
 278  *
 279  * Arguments are:
 280  *
 281  * comp         - Pointer to the current component structure.  The next
 282  *                component in the hash table after this component.  To
 283  *                start (or restart) the iteration of the hash table
 284  *                this argument should be NULL.
 285  * bucket       - Pointer to hash bucket.  Will be managed by this function,
 286  *                the caller should not modify this value.
 287  *
 288  * Returns the next component in the hash table.  This value should be
 289  * passed into the next call to fmt_nextcomp().  Returns NULL at the end
 290  * of the hash table.
 291  */
 292
 293 struct comp *fmt_nextcomp(struct comp *comp, unsigned int *bucket);
 294
 295 /*
 296  * The implementation of the %(formataddr) function.  This is available for
 297  * programs to provide their own local implementation if they wish to do
 298  * special processing (see uip/replsbr.c for an example).  Arguments are:
 299  *
 300  * orig         - Existing list of addresses
 301  * str          - New address(es) to append to list.
 302  *
 303  * This function returns an allocated string containing the new list of
 304  * addresses.
 305  */
 306
 307 char *formataddr(char *orig, char *str);
 308
 309 /*
 310  * The implementation of the %(concataddr) function.  Arguments and behavior
 311  * are the same as %(formataddr).  Again, see uip/replsbr.c to see how you
 312  * can override this behavior.
 313  */
 314
 315 char *concataddr(char *orig, char *str);