| 1 | /* -*-c-*- |
| 2 | * |
| 3 | * $Id: lbuf.c,v 1.6 2002/01/13 13:32:52 mdw Exp $ |
| 4 | * |
| 5 | * Block-to-line buffering |
| 6 | * |
| 7 | * (c) 1999 Straylight/Edgeware |
| 8 | */ |
| 9 | |
| 10 | /*----- Licensing notice --------------------------------------------------* |
| 11 | * |
| 12 | * This file is part of the mLib utilities library. |
| 13 | * |
| 14 | * mLib is free software; you can redistribute it and/or modify |
| 15 | * it under the terms of the GNU Library General Public License as |
| 16 | * published by the Free Software Foundation; either version 2 of the |
| 17 | * License, or (at your option) any later version. |
| 18 | * |
| 19 | * mLib is distributed in the hope that it will be useful, |
| 20 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 21 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 22 | * GNU Library General Public License for more details. |
| 23 | * |
| 24 | * You should have received a copy of the GNU Library General Public |
| 25 | * License along with mLib; if not, write to the Free |
| 26 | * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, |
| 27 | * MA 02111-1307, USA. |
| 28 | */ |
| 29 | |
| 30 | /*----- Revision history --------------------------------------------------* |
| 31 | * |
| 32 | * $Log: lbuf.c,v $ |
| 33 | * Revision 1.6 2002/01/13 13:32:52 mdw |
| 34 | * Pass line length to line handler function. Provide a @typedef@ for |
| 35 | * handler functions. Allow run-time configuration of line delimiters. |
| 36 | * |
| 37 | * Revision 1.5 2001/02/03 16:23:33 mdw |
| 38 | * Bug fix: handle a disable during a close-induced flush without dumping |
| 39 | * core. |
| 40 | * |
| 41 | * Revision 1.4 2000/06/17 10:38:14 mdw |
| 42 | * Add support for variable buffer sizes. |
| 43 | * |
| 44 | * Revision 1.3 1999/05/22 13:38:50 mdw |
| 45 | * Fix bug which discarded initial portions of incomplete lines. |
| 46 | * |
| 47 | * Revision 1.2 1999/05/17 20:36:08 mdw |
| 48 | * Make the magical constants for the buffer flags uppercase. |
| 49 | * |
| 50 | * Revision 1.1 1999/05/14 21:01:14 mdw |
| 51 | * Integrated `select' handling bits from the background resolver project. |
| 52 | * |
| 53 | */ |
| 54 | |
| 55 | /*----- Header files ------------------------------------------------------*/ |
| 56 | |
| 57 | #include <assert.h> |
| 58 | #include <stdio.h> |
| 59 | #include <stdlib.h> |
| 60 | #include <string.h> |
| 61 | |
| 62 | #include "alloc.h" |
| 63 | #include "arena.h" |
| 64 | #include "lbuf.h" |
| 65 | |
| 66 | /*----- Main code ---------------------------------------------------------*/ |
| 67 | |
| 68 | /* --- @lbuf_flush@ --- * |
| 69 | * |
| 70 | * Arguments: @lbuf *b@ = pointer to buffer block |
| 71 | * @char *p@ = pointer to where to start searching |
| 72 | * @size_t len@ = length of new material added |
| 73 | * |
| 74 | * Returns: --- |
| 75 | * |
| 76 | * Use: Flushes any complete lines in a line buffer. New material |
| 77 | * is assumed to have been added starting at @p@. If @p@ is |
| 78 | * null, then the scan starts at the beginning of the buffer, |
| 79 | * and the size of data already in the buffer is used in place |
| 80 | * of @len@. |
| 81 | * |
| 82 | * It is assumed that the buffer is initially enabled. You |
| 83 | * shouldn't be contributing data to a disabled buffer anyway. |
| 84 | * However, the buffer handler may at some point disable itself, |
| 85 | * and @lbuf_flush@ can cope with this eventuality. Any pending |
| 86 | * data is left at the start of the buffer and can be flushed |
| 87 | * out by calling @lbuf_flush(b, 0, 0)@ if the buffer is ever |
| 88 | * re-enabled. |
| 89 | */ |
| 90 | |
| 91 | void lbuf_flush(lbuf *b, char *p, size_t len) |
| 92 | { |
| 93 | char *l; /* Limit of data in buffer */ |
| 94 | char *q; /* Roving pointer through string */ |
| 95 | char *base; /* Base address of current line */ |
| 96 | int cr; /* Carriage return state */ |
| 97 | |
| 98 | if (b->f & LBUF_CLOSE) { |
| 99 | b->func(0, 0, b->p); |
| 100 | return; |
| 101 | } |
| 102 | |
| 103 | /* --- Initialize variables as necessary --- */ |
| 104 | |
| 105 | if (!p) { |
| 106 | p = b->buf; |
| 107 | cr = 0; |
| 108 | len = b->len; |
| 109 | } else |
| 110 | cr = b->f & LBUF_CR; |
| 111 | |
| 112 | l = p + len; |
| 113 | |
| 114 | /* --- Clear @base@ if I'm discarding an overlong line --- */ |
| 115 | |
| 116 | if (b->len == b->sz) |
| 117 | base = 0; |
| 118 | else |
| 119 | base = b->buf; |
| 120 | |
| 121 | /* --- Now I march through the string --- */ |
| 122 | |
| 123 | for (q = p; q < l; q++) { |
| 124 | |
| 125 | /* --- Quickly discard uninteresting characters --- */ |
| 126 | |
| 127 | switch (b->delim) { |
| 128 | case LBUF_CRLF: |
| 129 | case LBUF_STRICTCRLF: |
| 130 | if (*q != '\r' && *q != '\n') { |
| 131 | cr = 0; |
| 132 | continue; |
| 133 | } |
| 134 | if (*q == '\r') { |
| 135 | cr = 1; |
| 136 | continue; |
| 137 | } |
| 138 | if (!cr && b->delim == LBUF_STRICTCRLF) |
| 139 | continue; |
| 140 | break; |
| 141 | default: |
| 142 | if (*q != b->delim) |
| 143 | continue; |
| 144 | } |
| 145 | |
| 146 | /* --- I have a positive ID on a delimiter --- * |
| 147 | * |
| 148 | * If I'm interested in this string, report it to my owner. |
| 149 | */ |
| 150 | |
| 151 | if (base) { |
| 152 | len = q - base; |
| 153 | if (cr) |
| 154 | len--; /* Exercise: why is this safe? */ |
| 155 | base[len] = 0; |
| 156 | b->func(base, len, b->p); |
| 157 | if (!(b->f & LBUF_ENABLE)) { |
| 158 | base = q + 1; |
| 159 | break; |
| 160 | } |
| 161 | } |
| 162 | base = q + 1; |
| 163 | cr = 0; |
| 164 | } |
| 165 | |
| 166 | /* --- Sift through the aftermath --- */ |
| 167 | |
| 168 | if (base) { |
| 169 | len = l - base; |
| 170 | if (len == b->sz) { |
| 171 | b->buf[len - 1] = 0; |
| 172 | b->func(base, len - 1, b->p); |
| 173 | } else if (base != b->buf) |
| 174 | memmove(b->buf, base, len); |
| 175 | b->len = len; |
| 176 | if (cr) |
| 177 | b->f |= LBUF_CR; |
| 178 | else |
| 179 | b->f &= ~LBUF_CR; |
| 180 | } |
| 181 | } |
| 182 | |
| 183 | /* --- @lbuf_close@ --- * |
| 184 | * |
| 185 | * Arguments: @lbuf *b@ = pointer to buffer block |
| 186 | * |
| 187 | * Returns: --- |
| 188 | * |
| 189 | * Use: Empties the buffer of any data currently lurking in it, and |
| 190 | * informs the client that this has happened. It's assumed that |
| 191 | * the buffer is enabled: you shouldn't be reading close events |
| 192 | * on disabled buffers. The buffer, if allocated, is freed. |
| 193 | */ |
| 194 | |
| 195 | void lbuf_close(lbuf *b) |
| 196 | { |
| 197 | if (b->len && b->len != b->sz) { |
| 198 | b->buf[b->len] = 0; |
| 199 | b->func(b->buf, b->len, b->p); |
| 200 | } |
| 201 | if (b->buf) { |
| 202 | x_free(b->a, b->buf); |
| 203 | b->buf = 0; |
| 204 | } |
| 205 | b->f |= LBUF_CLOSE; |
| 206 | if (b->f & LBUF_ENABLE) |
| 207 | b->func(0, 0, b->p); |
| 208 | } |
| 209 | |
| 210 | /* --- @lbuf_free@ --- * |
| 211 | * |
| 212 | * Arguments: @lbuf *b@ = pointer to buffer block |
| 213 | * @char **p@ = output pointer to free space |
| 214 | * |
| 215 | * Returns: Free buffer size. |
| 216 | * |
| 217 | * Use: Returns the free portion of a line buffer. Data can then be |
| 218 | * written to this portion, and split out into lines by calling |
| 219 | * @lbuf_flush@. A buffer is allocated if none currently |
| 220 | * exists. |
| 221 | */ |
| 222 | |
| 223 | size_t lbuf_free(lbuf *b, char **p) |
| 224 | { |
| 225 | /* --- There's a special case to consider --- * |
| 226 | * |
| 227 | * If a line from the file wouldn't fit in the buffer, I truncate it and |
| 228 | * return what would fit. The rest of the line ought to be discarded. |
| 229 | * This condition is signalled by @len = b->sz@, and means that the entire |
| 230 | * buffer is OK to be trashed. In other cases, @len@ is the amount of |
| 231 | * space currently occupied in the buffer. This special case is the reason |
| 232 | * this routine exists. |
| 233 | */ |
| 234 | |
| 235 | if (b->len != 0 && b->len != b->sz) { |
| 236 | *p = b->buf + b->len; |
| 237 | return (b->sz - b->len); |
| 238 | } else { |
| 239 | if (!b->buf) |
| 240 | b->buf = x_alloc(b->a, b->sz); |
| 241 | *p = b->buf; |
| 242 | return (b->sz); |
| 243 | } |
| 244 | } |
| 245 | |
| 246 | /* --- @lbuf_snarf@ --- * |
| 247 | * |
| 248 | * Arguments: @lbuf *b@ = pointer to buffer block |
| 249 | * @const void *p@ = pointer to input data buffer |
| 250 | * @size_t sz@ = size of data in input buffer |
| 251 | * |
| 252 | * Returns: --- |
| 253 | * |
| 254 | * Use: Snarfs the data from the input buffer and spits it out as |
| 255 | * lines. This interface ignores the complexities of dealing |
| 256 | * with disablement: you should be using @lbuf_free@ to |
| 257 | * contribute data if you want to cope with that. |
| 258 | */ |
| 259 | |
| 260 | void lbuf_snarf(lbuf *b, const void *p, size_t sz) |
| 261 | { |
| 262 | const char *pp = p; |
| 263 | while (sz && (b->f & LBUF_ENABLE)) { |
| 264 | size_t bsz; |
| 265 | char *bp; |
| 266 | |
| 267 | bsz = lbuf_free(b, &bp); |
| 268 | if (bsz > sz) |
| 269 | bsz = sz; |
| 270 | memcpy(bp, pp, bsz); |
| 271 | lbuf_flush(b, bp, bsz); |
| 272 | pp += bsz; |
| 273 | sz -= bsz; |
| 274 | } |
| 275 | } |
| 276 | |
| 277 | /* --- @lbuf_setsize@ --- * |
| 278 | * |
| 279 | * Arguments: @lbuf *b@ = pointer to buffer block |
| 280 | * @size_t sz@ = requested maximum line size |
| 281 | * |
| 282 | * Returns: --- |
| 283 | * |
| 284 | * Use: Modifies the size of the buffer associated with the block. |
| 285 | * It is an error to resize a buffer while it contains data. |
| 286 | */ |
| 287 | |
| 288 | void lbuf_setsize(lbuf *b, size_t sz) |
| 289 | { |
| 290 | if (b->buf) |
| 291 | assert(((void)"Buffer in use in lbuf_setsize", |
| 292 | b->len == 0 || b->len == b->sz)); |
| 293 | if (b->buf) |
| 294 | x_free(b->a, b->buf); |
| 295 | b->sz = sz; |
| 296 | b->buf = 0; |
| 297 | } |
| 298 | |
| 299 | /* --- @lbuf_init@ --- * |
| 300 | * |
| 301 | * Arguments: @lbuf *b@ = pointer to buffer block |
| 302 | * @lbuf_func *func@ = handler function |
| 303 | * @void *p@ = argument pointer for @func@ |
| 304 | * |
| 305 | * Returns: --- |
| 306 | * |
| 307 | * Use: Initializes a line buffer block. Any recognized lines are |
| 308 | * passed to @func@ for processing. No buffer is initially |
| 309 | * allocated; this is done when the buffer is actually required |
| 310 | * for the first time. |
| 311 | */ |
| 312 | |
| 313 | void lbuf_init(lbuf *b, lbuf_func *func, void *p) |
| 314 | { |
| 315 | b->func = func; |
| 316 | b->p = p; |
| 317 | b->len = 0; |
| 318 | b->f = LBUF_ENABLE; |
| 319 | b->delim = LBUF_CRLF; |
| 320 | b->buf = 0; |
| 321 | b->a = arena_global; |
| 322 | lbuf_setsize(b, 256); |
| 323 | } |
| 324 | |
| 325 | /* --- @lbuf_destroy@ --- * |
| 326 | * |
| 327 | * Arguments: @lbuf *b@ = pointer to buffer block |
| 328 | * |
| 329 | * Returns: --- |
| 330 | * |
| 331 | * Use: Deallocates a line buffer and frees any resources it owned. |
| 332 | */ |
| 333 | |
| 334 | void lbuf_destroy(lbuf *b) |
| 335 | { |
| 336 | if (b->buf) { |
| 337 | x_free(b->a, b->buf); |
| 338 | b->buf = 0; |
| 339 | } |
| 340 | } |
| 341 | |
| 342 | /*----- That's all, folks -------------------------------------------------*/ |