datalist.c

   1 /*
   2  * datalist.c
   3  * Copyright (C) 2000-2002 A.J. van Os; Released under GPL
   4  *
   5  * Description:
   6  * Build, read and destroy a list of Word data blocks
   7  */
   8
   9 #include <stdlib.h>
  10 #include <errno.h>
  11 #include "antiword.h"
  12
  13 #if defined(__riscos)
  14 #define EIO             42
  15 #endif /* __riscos */
  16
  17
  18 /*
  19  * Private structure to hide the way the information
  20  * is stored from the rest of the program
  21  */
  22 typedef struct data_mem_tag {
  23         data_block_type         tInfo;
  24         struct data_mem_tag     *pNext;
  25 } data_mem_type;
  26
  27 /* Variable to describe the start of the data block list */
  28 static data_mem_type    *pAnchor = NULL;
  29 /* Variable needed to read the data block list */
  30 static data_mem_type    *pBlockLast = NULL;
  31 /* Variable needed to read the data block list */
  32 static data_mem_type    *pBlockCurrent = NULL;
  33 static ULONG    ulBlockOffset = 0;
  34 static size_t   tByteNext = 0;
  35 /* Last block read */
  36 static UCHAR    aucBlock[BIG_BLOCK_SIZE];
  37
  38
  39 /*
  40  * vDestroyDataBlockList - destroy the data block list
  41  */
  42 void
  43 vDestroyDataBlockList(void)
  44 {
  45         data_mem_type   *pCurr, *pNext;
  46
  47         DBG_MSG("vDestroyDataBlockList");
  48
  49         pCurr = pAnchor;
  50         while (pCurr != NULL) {
  51                 pNext = pCurr->pNext;
  52                 pCurr = xfree(pCurr);
  53                 pCurr = pNext;
  54         }
  55         pAnchor = NULL;
  56         /* Reset all the control variables */
  57         pBlockLast = NULL;
  58         pBlockCurrent = NULL;
  59         ulBlockOffset = 0;
  60         tByteNext = 0;
  61 } /* end of vDestroyDataBlockList */
  62
  63 /*
  64  * bAdd2DataBlockList - add an element to the data block list
  65  *
  66  * Returns TRUE when successful, otherwise FALSE
  67  */
  68 BOOL
  69 bAdd2DataBlockList(const data_block_type *pDataBlock)
  70 {
  71         data_mem_type   *pListMember;
  72
  73         fail(pDataBlock == NULL);
  74         fail(pDataBlock->ulFileOffset == FC_INVALID);
  75         fail(pDataBlock->ulDataPos == CP_INVALID);
  76         fail(pDataBlock->ulLength == 0);
  77
  78         NO_DBG_MSG("bAdd2DataBlockList");
  79         NO_DBG_HEX(pDataBlock->ulFileOffset);
  80         NO_DBG_HEX(pDataBlock->ulDataPos);
  81         NO_DBG_HEX(pDataBlock->ulLength);
  82
  83         if (pDataBlock->ulFileOffset == FC_INVALID ||
  84             pDataBlock->ulDataPos == CP_INVALID ||
  85             pDataBlock->ulLength == 0) {
  86                 werr(0, "Software (datablock) error");
  87                 return FALSE;
  88         }
  89         /* Check for continuous blocks */
  90         if (pBlockLast != NULL &&
  91             pBlockLast->tInfo.ulFileOffset +
  92              pBlockLast->tInfo.ulLength == pDataBlock->ulFileOffset &&
  93             pBlockLast->tInfo.ulDataPos +
  94              pBlockLast->tInfo.ulLength == pDataBlock->ulDataPos) {
  95                 /* These are continous blocks */
  96                 pBlockLast->tInfo.ulLength += pDataBlock->ulLength;
  97                 return TRUE;
  98         }
  99         /* Make a new block */
 100         pListMember = xmalloc(sizeof(data_mem_type));
 101         /* Add the block to the data list */
 102         pListMember->tInfo = *pDataBlock;
 103         pListMember->pNext = NULL;
 104         if (pAnchor == NULL) {
 105                 pAnchor = pListMember;
 106         } else {
 107                 fail(pBlockLast == NULL);
 108                 pBlockLast->pNext = pListMember;
 109         }
 110         pBlockLast = pListMember;
 111         return TRUE;
 112 } /* end of bAdd2DataBlockList */
 113
 114 /*
 115  * ulGetDataOffset - get the offset in the data block list
 116  *
 117  * Get the fileoffset the current position in the data block list
 118  */
 119 ULONG
 120 ulGetDataOffset(FILE *pFile)
 121 {
 122         return pBlockCurrent->tInfo.ulFileOffset + ulBlockOffset + tByteNext;
 123 } /* end of ulGetDataOffset */
 124
 125 /*
 126  * bSetDataOffset - set the offset in the data block list
 127  *
 128  * Make the given fileoffset the current position in the data block list
 129  */
 130 BOOL
 131 bSetDataOffset(FILE *pFile, ULONG ulFileOffset)
 132 {
 133         data_mem_type   *pCurr;
 134         size_t  tReadLen;
 135
 136         DBG_HEX(ulFileOffset);
 137
 138         for (pCurr = pAnchor; pCurr != NULL; pCurr = pCurr->pNext) {
 139                 if (ulFileOffset < pCurr->tInfo.ulFileOffset ||
 140                     ulFileOffset >= pCurr->tInfo.ulFileOffset +
 141                      pCurr->tInfo.ulLength) {
 142                         /* The file offset is not in this block */
 143                         continue;
 144                 }
 145                 /* Compute the maximum number of bytes to read */
 146                 tReadLen = (size_t)(pCurr->tInfo.ulFileOffset +
 147                                 pCurr->tInfo.ulLength -
 148                                 ulFileOffset);
 149                 /* Compute the real number of bytes to read */
 150                 if (tReadLen > sizeof(aucBlock)) {
 151                         tReadLen = sizeof(aucBlock);
 152                 }
 153                 /* Read the bytes */
 154                 if (!bReadBytes(aucBlock, tReadLen, ulFileOffset, pFile)) {
 155                         return FALSE;
 156                 }
 157                 /* Set the control variables */
 158                 pBlockCurrent = pCurr;
 159                 ulBlockOffset = ulFileOffset - pCurr->tInfo.ulFileOffset;
 160                 tByteNext = 0;
 161                 return TRUE;
 162         }
 163         return FALSE;
 164 } /* end of bSetDataOffset */
 165
 166 /*
 167  * iNextByte - get the next byte from the data block list
 168  */
 169 int
 170 iNextByte(FILE *pFile)
 171 {
 172         ULONG   ulReadOff;
 173         size_t  tReadLen;
 174
 175         fail(pBlockCurrent == NULL);
 176
 177         if (tByteNext >= sizeof(aucBlock) ||
 178             ulBlockOffset + tByteNext >= pBlockCurrent->tInfo.ulLength) {
 179                 if (ulBlockOffset + sizeof(aucBlock) <
 180                                         pBlockCurrent->tInfo.ulLength) {
 181                         /* Same block, next part */
 182                         ulBlockOffset += sizeof(aucBlock);
 183                 } else {
 184                         /* Next block, first part */
 185                         pBlockCurrent = pBlockCurrent->pNext;
 186                         ulBlockOffset = 0;
 187                 }
 188                 if (pBlockCurrent == NULL) {
 189                         /* Past the last part of the last block */
 190                         errno = EIO;
 191                         return EOF;
 192                 }
 193                 tReadLen = (size_t)
 194                                 (pBlockCurrent->tInfo.ulLength - ulBlockOffset);
 195                 if (tReadLen > sizeof(aucBlock)) {
 196                         tReadLen = sizeof(aucBlock);
 197                 }
 198                 ulReadOff = pBlockCurrent->tInfo.ulFileOffset + ulBlockOffset;
 199                 if (!bReadBytes(aucBlock, tReadLen, ulReadOff, pFile)) {
 200                         errno = EIO;
 201                         return EOF;
 202                 }
 203                 tByteNext = 0;
 204         }
 205         return (int)aucBlock[tByteNext++];
 206 } /* end of iNextByte */
 207
 208 /*
 209  * usNextWord - get the next word from the data block list
 210  *
 211  * Read a two byte value in Little Endian order, that means MSB last
 212  *
 213  * All return values can be valid so errno is set in case of error
 214  */
 215 USHORT
 216 usNextWord(FILE *pFile)
 217 {
 218         USHORT  usLSB, usMSB;
 219
 220         usLSB = (USHORT)iNextByte(pFile);
 221         if (usLSB == (USHORT)EOF) {
 222                 errno = EIO;
 223                 return (USHORT)EOF;
 224         }
 225         usMSB = (USHORT)iNextByte(pFile);
 226         if (usMSB == (USHORT)EOF) {
 227                 DBG_MSG("usNextWord: Unexpected EOF");
 228                 errno = EIO;
 229                 return (USHORT)EOF;
 230         }
 231         return (usMSB << 8) | usLSB;
 232 } /* end of usNextWord */
 233
 234 /*
 235  * ulNextLong - get the next long from the data block list
 236  *
 237  * Read a four byte value in Little Endian order, that means MSW last
 238  *
 239  * All return values can be valid so errno is set in case of error
 240  */
 241 ULONG
 242 ulNextLong(FILE *pFile)
 243 {
 244         ULONG   ulLSW, ulMSW;
 245
 246         ulLSW = (ULONG)usNextWord(pFile);
 247         if (ulLSW == (ULONG)EOF) {
 248                 errno = EIO;
 249                 return (ULONG)EOF;
 250         }
 251         ulMSW = (ULONG)usNextWord(pFile);
 252         if (ulMSW == (ULONG)EOF) {
 253                 DBG_MSG("ulNextLong: Unexpected EOF");
 254                 errno = EIO;
 255                 return (ULONG)EOF;
 256         }
 257         return (ulMSW << 16) | ulLSW;
 258 } /* end of ulNextLong */
 259
 260 /*
 261  * usNextWordBE - get the next two byte value
 262  *
 263  * Read a two byte value in Big Endian order, that means MSB first
 264  *
 265  * All return values can be valid so errno is set in case of error
 266  */
 267 USHORT
 268 usNextWordBE(FILE *pFile)
 269 {
 270         USHORT usLSB, usMSB;
 271
 272         usMSB = (USHORT)iNextByte(pFile);
 273         if (usMSB == (USHORT)EOF) {
 274                 errno = EIO;
 275                 return (USHORT)EOF;
 276         }
 277         usLSB = (USHORT)iNextByte(pFile);
 278         if (usLSB == (USHORT)EOF) {
 279                 DBG_MSG("usNextWordBE: Unexpected EOF");
 280                 errno = EIO;
 281                 return (USHORT)EOF;
 282         }
 283         return (usMSB << 8) | usLSB;
 284 } /* end of usNextWordBE */
 285
 286 /*
 287  * ulNextLongBE - get the next four byte value
 288  *
 289  * Read a four byte value in Big Endian order, that means MSW first
 290  *
 291  * All return values can be valid so errno is set in case of error
 292  */
 293 ULONG
 294 ulNextLongBE(FILE *pFile)
 295 {
 296         ULONG   ulLSW, ulMSW;
 297
 298         ulMSW = (ULONG)usNextWordBE(pFile);
 299         if (ulMSW == (ULONG)EOF) {
 300                 errno = EIO;
 301                 return (ULONG)EOF;
 302         }
 303         ulLSW = (ULONG)usNextWordBE(pFile);
 304         if (ulLSW == (ULONG)EOF) {
 305                 DBG_MSG("ulNextLongBE: Unexpected EOF");
 306                 errno = EIO;
 307                 return (ULONG)EOF;
 308         }
 309         return (ulMSW << 16) | ulLSW;
 310 } /* end of ulNextLongBE */
 311
 312 /*
 313  * tSkipBytes - skip over the given number of bytes
 314  *
 315  * Returns the number of skipped bytes
 316  */
 317 size_t
 318 tSkipBytes(FILE *pFile, size_t tToSkip)
 319 {
 320         size_t  tToGo, tMaxMove, tMove;
 321
 322         fail(pFile == NULL);
 323         fail(pBlockCurrent == NULL);
 324
 325         tToGo = tToSkip;
 326         while (tToGo != 0) {
 327                 /* Goto the end of the current block */
 328                 tMaxMove = min(sizeof(aucBlock) - tByteNext,
 329                                 (size_t)(pBlockCurrent->tInfo.ulLength -
 330                                 ulBlockOffset - tByteNext));
 331                 tMove = min(tMaxMove, tToGo);
 332                 tByteNext += tMove;
 333                 tToGo -= tMove;
 334                 if (tToGo != 0) {
 335                         /* Goto the next block */
 336                         if (iNextByte(pFile) == EOF) {
 337                                 return tToSkip - tToGo;
 338                         }
 339                         tToGo--;
 340                 }
 341         }
 342         return tToSkip;
 343 } /* end of tSkipBytes */
 344
 345 /*
 346  * Translate  a data position to an offset in the file.
 347  * Logical to physical offset.
 348  *
 349  * Returns:     FC_INVALID: in case of error
 350  *              otherwise: the computed file offset
 351  */
 352 ULONG
 353 ulDataPos2FileOffset(ULONG ulDataPos)
 354 {
 355         data_mem_type   *pCurr;
 356
 357         fail(ulDataPos == CP_INVALID);
 358
 359         for (pCurr = pAnchor; pCurr != NULL; pCurr = pCurr->pNext) {
 360                 if (ulDataPos < pCurr->tInfo.ulDataPos ||
 361                     ulDataPos >= pCurr->tInfo.ulDataPos +
 362                      pCurr->tInfo.ulLength) {
 363                         /* The data offset is not in this block, try the next */
 364                         continue;
 365                 }
 366                 /* The data offset is in the current block */
 367                 return pCurr->tInfo.ulFileOffset +
 368                                 ulDataPos -
 369                                 pCurr->tInfo.ulDataPos;
 370         }
 371         /* Passed beyond the end of the list */
 372         DBG_HEX_C(ulDataPos != 0, ulDataPos);
 373         return FC_INVALID;
 374 } /* end of ulDataPos2FileOffset */