TB9.2 Example Applications: examples/PIPS/antiword/src/finddata.c Source File

00001 /*
+00002  * finddata.c
+00003  * Copyright (C) 2000-2002 A.J. van Os; Released under GPL
+00004  *
+00005  * Description:
+00006  * Find the blocks that contain the data of MS Word files
+00007  */
+00008 
+00009 #include <stdio.h>
+00010 #include <stdlib.h>
+00011 #include "antiword.h"
+00012 
+00013 
+00014 /*
+00015  * bAddDataBlocks - Add the blocks to the data block list
+00016  *
+00017  * Returns TRUE when successful, otherwise FALSE
+00018  */
+00019 BOOL
+00020 bAddDataBlocks(ULONG ulDataPosFirst, ULONG ulTotalLength,
+00021         ULONG ulStartBlock, const ULONG *aulBBD, size_t tBBDLen)
+00022 {
+00023         data_block_type tDataBlock;
+00024         ULONG   ulDataPos, ulOffset, ulIndex;
+00025         long    lToGo;
+00026         BOOL    bSuccess;
+00027 
+00028         fail(ulTotalLength > (ULONG)LONG_MAX);
+00029         fail(ulStartBlock > MAX_BLOCKNUMBER && ulStartBlock != END_OF_CHAIN);
+00030         fail(aulBBD == NULL);
+00031 
+00032         NO_DBG_HEX(ulDataPosFirst);
+00033         NO_DBG_DEC(ulTotalLength);
+00034 
+00035         lToGo = (long)ulTotalLength;
+00036 
+00037         ulDataPos = ulDataPosFirst;
+00038         ulOffset = ulDataPosFirst;
+00039         for (ulIndex = ulStartBlock;
+00040              ulIndex != END_OF_CHAIN && lToGo > 0;
+00041              ulIndex = aulBBD[ulIndex]) {
+00042                 if (ulIndex == UNUSED_BLOCK || ulIndex >= (ULONG)tBBDLen) {
+00043                         DBG_DEC(ulIndex);
+00044                         DBG_DEC(tBBDLen);
+00045                         return FALSE;
+00046                 }
+00047                 if (ulOffset >= BIG_BLOCK_SIZE) {
+00048                         ulOffset -= BIG_BLOCK_SIZE;
+00049                         continue;
+00050                 }
+00051                 tDataBlock.ulFileOffset =
+00052                         (ulIndex + 1) * BIG_BLOCK_SIZE + ulOffset;
+00053                 tDataBlock.ulDataPos = ulDataPos;
+00054                 tDataBlock.ulLength = min(BIG_BLOCK_SIZE - ulOffset,
+00055                                                 (ULONG)lToGo);
+00056                 fail(tDataBlock.ulLength > BIG_BLOCK_SIZE);
+00057                 ulOffset = 0;
+00058                 if (!bAdd2DataBlockList(&tDataBlock)) {
+00059                         DBG_HEX(tDataBlock.ulFileOffset);
+00060                         DBG_HEX(tDataBlock.ulDataPos);
+00061                         DBG_DEC(tDataBlock.ulLength);
+00062                         return FALSE;
+00063                 }
+00064                 ulDataPos += tDataBlock.ulLength;
+00065                 lToGo -= (long)tDataBlock.ulLength;
+00066         }
+00067         bSuccess = lToGo == 0 ||
+00068                 (ulTotalLength == (ULONG)LONG_MAX && ulIndex == END_OF_CHAIN);
+00069         DBG_DEC_C(!bSuccess, lToGo);
+00070         DBG_DEC_C(!bSuccess, ulTotalLength);
+00071         DBG_DEC_C(!bSuccess, ulIndex);
+00072         return bSuccess;
+00073 } /* end of bAddDataBlocks */
+00074 
+00075 /*
+00076  * bGet6DocumentData - make a list of the data blocks of Word 6/7 files
+00077  *
+00078  * Code for "fast saved" files.
+00079  *
+00080  * Returns TRUE when successful, otherwise FALSE
+00081  */
+00082 BOOL
+00083 bGet6DocumentData(FILE *pFile, ULONG ulStartBlock,
+00084         const ULONG *aulBBD, size_t tBBDLen, const UCHAR *aucHeader)
+00085 {
+00086         UCHAR   *aucBuffer;
+00087         ULONG   ulBeginTextInfo, ulOffset, ulTotLength;
+00088         size_t  tTextInfoLen;
+00089         int     iIndex, iOff, iType, iLen, iPieces;
+00090 
+00091         DBG_MSG("bGet6DocumentData");
+00092 
+00093         fail(pFile == NULL);
+00094         fail(aulBBD == NULL);
+00095         fail(aucHeader == NULL);
+00096 
+00097         ulBeginTextInfo = ulGetLong(0x160, aucHeader);
+00098         DBG_HEX(ulBeginTextInfo);
+00099         tTextInfoLen = (size_t)ulGetLong(0x164, aucHeader);
+00100         DBG_DEC(tTextInfoLen);
+00101 
+00102         aucBuffer = xmalloc(tTextInfoLen);
+00103         if (!bReadBuffer(pFile, ulStartBlock,
+00104                         aulBBD, tBBDLen, BIG_BLOCK_SIZE,
+00105                         aucBuffer, ulBeginTextInfo, tTextInfoLen)) {
+00106                 aucBuffer = xfree(aucBuffer);
+00107                 return FALSE;
+00108         }
+00109         NO_DBG_PRINT_BLOCK(aucBuffer, tTextInfoLen);
+00110 
+00111         iOff = 0;
+00112         while (iOff < (int)tTextInfoLen) {
+00113                 iType = (int)ucGetByte(iOff, aucBuffer);
+00114                 iOff++;
+00115                 if (iType == 0) {
+00116                         iOff++;
+00117                         continue;
+00118                 }
+00119                 iLen = (int)usGetWord(iOff, aucBuffer);
+00120                 iOff += 2;
+00121                 if (iType == 1) {
+00122                         iOff += iLen;
+00123                         continue;
+00124                 }
+00125                 if (iType != 2) {
+00126                         werr(0, "Unknown type of 'fastsaved' format");
+00127                         aucBuffer = xfree(aucBuffer);
+00128                         return FALSE;
+00129                 }
+00130                 /* Type 2 */
+00131                 NO_DBG_DEC(iLen);
+00132                 iOff += 2;
+00133                 iPieces = (iLen - 4) / 12;
+00134                 DBG_DEC(iPieces);
+00135                 for (iIndex = 0; iIndex < iPieces; iIndex++) {
+00136                         ulOffset = ulGetLong(
+00137                                 iOff + (iPieces + 1) * 4 + iIndex * 8 + 2,
+00138                                 aucBuffer);
+00139                         ulTotLength = ulGetLong(iOff + (iIndex + 1) * 4,
+00140                                                 aucBuffer) -
+00141                                         ulGetLong(iOff + iIndex * 4,
+00142                                                 aucBuffer);
+00143                         if (!bAddDataBlocks(ulOffset, ulTotLength,
+00144                                         ulStartBlock,
+00145                                         aulBBD, tBBDLen)) {
+00146                                 aucBuffer = xfree(aucBuffer);
+00147                                 return FALSE;
+00148                         }
+00149                 }
+00150                 break;
+00151         }
+00152         aucBuffer = xfree(aucBuffer);
+00153         return TRUE;
+00154 } /* end of bGet6DocumentData */
+