stylelist.c

   1 /*
   2  * stylelist.c
   3  * Copyright (C) 1998-2002 A.J. van Os; Released under GPL
   4  *
   5  * Description:
   6  * Build, read and destroy a list of Word style information
   7  */
   8
   9 #include <stdlib.h>
  10 #include <stddef.h>
  11 #include <ctype.h>
  12 #include "antiword.h"
  13
  14
  15 /*
  16  * Private structure to hide the way the information
  17  * is stored from the rest of the program
  18  */
  19 typedef struct style_mem_tag {
  20         style_block_type        tInfo;
  21         ULONG                   ulSequenceNumber;
  22         struct style_mem_tag    *pNext;
  23 } style_mem_type;
  24
  25 /* Variables needed to write the Style Information List */
  26 static style_mem_type   *pAnchor = NULL;
  27 static style_mem_type   *pStyleLast = NULL;
  28 /* Values for efficiency reasons */
  29 static const style_mem_type     *pMidPtr = NULL;
  30 static BOOL             bMoveMidPtr = FALSE;
  31 static BOOL             bInSequence = TRUE;
  32
  33
  34 /*
  35  * vDestroyStyleInfoList - destroy the Style Information List
  36  */
  37 void
  38 vDestroyStyleInfoList(void)
  39 {
  40         style_mem_type  *pCurr, *pNext;
  41
  42         DBG_MSG("vDestroyStyleInfoList");
  43
  44         /* Free the Style Information List */
  45         pCurr = pAnchor;
  46         while (pCurr != NULL) {
  47                 pNext = pCurr->pNext;
  48                 pCurr = xfree(pCurr);
  49                 pCurr = pNext;
  50         }
  51         pAnchor = NULL;
  52         /* Reset all control variables */
  53         pStyleLast = NULL;
  54         pMidPtr = NULL;
  55         bMoveMidPtr = FALSE;
  56         bInSequence = TRUE;
  57 } /* end of vDestroyStyleInfoList */
  58
  59 /*
  60  * ucChooseListCharacter - choose our list character
  61  */
  62 static UCHAR
  63 ucChooseListCharacter(UCHAR ucNFC, UCHAR ucListCharacter)
  64 {
  65         if (isprint(ucListCharacter)) {
  66                 return ucListCharacter;
  67         }
  68         if (ucNFC == LIST_BULLETS) {
  69                 switch (ucListCharacter) {
  70                 case 0x00:
  71                 case 0xb7:
  72                         return (UCHAR)OUR_BULLET;
  73                 case 0xa8:
  74                         return (UCHAR)OUR_DIAMOND;
  75                 case 0xde:
  76                         return (UCHAR)'=';
  77                 case 0xe0:
  78                         return (UCHAR)'o';
  79                 case 0xe1:
  80                         return (UCHAR)'(';
  81                 case 0xfe:
  82                         return (UCHAR)' ';
  83                 default:
  84                         DBG_HEX(ucListCharacter);
  85                         return (UCHAR)OUR_BULLET;
  86                 }
  87         }
  88         return (UCHAR)'.';
  89 } /* end of ucChooseListCharacter */
  90
  91 /*
  92  * eGetNumType - get level type from level number
  93  */
  94 level_type_enum
  95 eGetNumType(UCHAR ucLevelNumber)
  96 {
  97         switch (ucLevelNumber) {
  98         case 10:
  99                 return level_type_numbering;
 100         case 11:
 101                 return level_type_sequence;
 102         case 12:
 103                 return level_type_pause;
 104         default:
 105                 if ((int)ucLevelNumber >= 1 && (int)ucLevelNumber <= 9) {
 106                         return level_type_outline;
 107                 }
 108                 return level_type_none;
 109         }
 110 } /* end of eGetNumType */
 111
 112 /*
 113  * vCorrectStyleValues - correct style values that Antiword can't use
 114  */
 115 void
 116 vCorrectStyleValues(style_block_type *pStyleBlock)
 117 {
 118         if (pStyleBlock->usBeforeIndent > 0x7fff) {
 119                 pStyleBlock->usBeforeIndent = 0;
 120         } else if (pStyleBlock->usBeforeIndent > 2160) {
 121                 /* 2160 twips = 1.5 inches or 38.1 mm */
 122                 DBG_DEC(pStyleBlock->usBeforeIndent);
 123                 pStyleBlock->usBeforeIndent = 2160;
 124         }
 125         if (pStyleBlock->usIstd >= 1 &&
 126             pStyleBlock->usIstd <= 9 &&
 127             pStyleBlock->usBeforeIndent < HEADING_GAP) {
 128                 NO_DBG_DEC(pStyleBlock->usBeforeIndent);
 129                 pStyleBlock->usBeforeIndent = HEADING_GAP;
 130         }
 131         if (pStyleBlock->usAfterIndent > 0x7fff) {
 132                 pStyleBlock->usAfterIndent = 0;
 133         } else if (pStyleBlock->usAfterIndent > 2160) {
 134                 /* 2160 twips = 1.5 inches or 38.1 mm */
 135                 DBG_DEC(pStyleBlock->usAfterIndent);
 136                 pStyleBlock->usAfterIndent = 2160;
 137         }
 138         if (pStyleBlock->usIstd >= 1 &&
 139             pStyleBlock->usIstd <= 9 &&
 140             pStyleBlock->usAfterIndent < HEADING_GAP) {
 141                 NO_DBG_DEC(pStyleBlock->usAfterIndent);
 142                 pStyleBlock->usAfterIndent = HEADING_GAP;
 143         }
 144         if (pStyleBlock->sLeftIndent < 0) {
 145                 pStyleBlock->sLeftIndent = 0;
 146         }
 147         if (pStyleBlock->sRightIndent > 0) {
 148                 pStyleBlock->sRightIndent = 0;
 149         }
 150         pStyleBlock->ucListCharacter =
 151                         ucChooseListCharacter(
 152                                 pStyleBlock->ucNFC,
 153                                 pStyleBlock->ucListCharacter);
 154 } /* end of vCorrectStyleValues */
 155
 156 /*
 157  * vAdd2StyleInfoList - Add an element to the Style Information List
 158  */
 159 void
 160 vAdd2StyleInfoList(const style_block_type *pStyleBlock)
 161 {
 162         style_mem_type  *pListMember;
 163
 164         fail(pStyleBlock == NULL);
 165
 166         NO_DBG_MSG("bAdd2StyleInfoList");
 167
 168         if (pStyleBlock->ulFileOffset == FC_INVALID) {
 169                 NO_DBG_DEC(pStyleBlock->usIstd);
 170                 return;
 171         }
 172
 173         NO_DBG_HEX(pStyleBlock->ulFileOffset);
 174         NO_DBG_DEC_C(pStyleBlock->sLeftIndent != 0,
 175                                         pStyleBlock->sLeftIndent);
 176         NO_DBG_DEC_C(pStyleBlock->sRightIndent != 0,
 177                                         pStyleBlock->sRightIndent);
 178         NO_DBG_DEC_C(pStyleBlock->bInList, pStyleBlock->bInList);
 179         NO_DBG_DEC_C(pStyleBlock->bNumPause, pStyleBlock->bNumPause);
 180         NO_DBG_DEC_C(pStyleBlock->usIstd != 0, pStyleBlock->usIstd);
 181         NO_DBG_DEC_C(pStyleBlock->usStartAt != 0, pStyleBlock->usStartAt);
 182         NO_DBG_DEC_C(pStyleBlock->usAfterIndent != 0,
 183                                         pStyleBlock->usAfterIndent);
 184         NO_DBG_DEC_C(pStyleBlock->ucAlignment != 0, pStyleBlock->ucAlignment);
 185         NO_DBG_DEC_C(pStyleBlock->bInList, pStyleBlock->ucNFC);
 186         NO_DBG_HEX_C(pStyleBlock->bInList, pStyleBlock->ucListCharacter);
 187         NO_DBG_DEC_C(pStyleBlock->ucNFC != LIST_BULLETS,
 188                                         pStyleBlock->usStartAt);
 189
 190         if (pStyleLast != NULL &&
 191             pStyleLast->tInfo.ulFileOffset == pStyleBlock->ulFileOffset) {
 192                 /*
 193                  * If two consecutive styles share the same
 194                  * offset, remember only the last style
 195                  */
 196                 fail(pStyleLast->pNext != NULL);
 197                 pStyleLast->tInfo = *pStyleBlock;
 198                 return;
 199         }
 200
 201         /* Create list member */
 202         pListMember = xmalloc(sizeof(style_mem_type));
 203         /* Fill the list member */
 204         pListMember->tInfo = *pStyleBlock;
 205         pListMember->pNext = NULL;
 206         /* Add the sequence number */
 207         pListMember->ulSequenceNumber =
 208                         ulGetSeqNumber(pListMember->tInfo.ulFileOffset);
 209         /* Correct the values where needed */
 210         vCorrectStyleValues(&pListMember->tInfo);
 211         /* Add the new member to the list */
 212         if (pAnchor == NULL) {
 213                 pAnchor = pListMember;
 214                 /* For efficiency */
 215                 pMidPtr = pAnchor;
 216                 bMoveMidPtr = FALSE;
 217                 bInSequence = TRUE;
 218         } else {
 219                 fail(pStyleLast == NULL);
 220                 pStyleLast->pNext = pListMember;
 221                 /* For efficiency */
 222                 if (bMoveMidPtr) {
 223                         pMidPtr = pMidPtr->pNext;
 224                         bMoveMidPtr = FALSE;
 225                 } else {
 226                         bMoveMidPtr = TRUE;
 227                 }
 228                 if (bInSequence) {
 229                         bInSequence = pListMember->ulSequenceNumber >
 230                                         pStyleLast->ulSequenceNumber;
 231                 }
 232         }
 233         pStyleLast = pListMember;
 234 } /* end of vAdd2StyleInfoList */
 235
 236 /*
 237  * Get the record that follows the given recored in the Style Information List
 238  */
 239 const style_block_type *
 240 pGetNextStyleInfoListItem(const style_block_type *pCurr)
 241 {
 242         const style_mem_type    *pRecord;
 243         size_t  tOffset;
 244
 245         if (pCurr == NULL) {
 246                 if (pAnchor == NULL) {
 247                         /* There are no records */
 248                         return NULL;
 249                 }
 250                 /* The first record is the only one without a predecessor */
 251                 return &pAnchor->tInfo;
 252         }
 253         tOffset = offsetof(style_mem_type, tInfo);
 254         /* Many casts to prevent alignment warnings */
 255         pRecord = (style_mem_type *)(void *)((char *)pCurr - tOffset);
 256         fail(pCurr != &pRecord->tInfo);
 257         if (pRecord->pNext == NULL) {
 258                 /* The last record has no successor */
 259                 return NULL;
 260         }
 261         return &pRecord->pNext->tInfo;
 262 } /* end of pGetNextStyleInfoListItem */
 263
 264 /*
 265  * usGetIstd - get the istd that belongs to the given file offset
 266  */
 267 USHORT
 268 usGetIstd(ULONG ulFileOffset)
 269 {
 270         const style_mem_type    *pCurr, *pBest, *pStart;
 271         ULONG   ulSeq, ulBest;
 272
 273         ulSeq = ulGetSeqNumber(ulFileOffset);
 274         if (ulSeq == FC_INVALID) {
 275                 return ISTD_NORMAL;
 276         }
 277         NO_DBG_HEX(ulFileOffset);
 278         NO_DBG_DEC(ulSeq);
 279
 280         if (bInSequence &&
 281             pMidPtr != NULL &&
 282             ulSeq > pMidPtr->ulSequenceNumber) {
 283                 /* The istd is in the second half of the chained list */
 284                 pStart = pMidPtr;
 285         } else {
 286                 pStart = pAnchor;
 287         }
 288
 289         pBest = NULL;
 290         ulBest = 0;
 291         for (pCurr = pStart; pCurr != NULL; pCurr = pCurr->pNext) {
 292                 if (pCurr->ulSequenceNumber != FC_INVALID &&
 293                     (pBest == NULL || pCurr->ulSequenceNumber > ulBest) &&
 294                     pCurr->ulSequenceNumber <= ulSeq) {
 295                         pBest = pCurr;
 296                         ulBest = pCurr->ulSequenceNumber;
 297                 }
 298                 if (bInSequence && pCurr->ulSequenceNumber > ulSeq) {
 299                         break;
 300                 }
 301         }
 302         NO_DBG_DEC(ulBest);
 303
 304         if (pBest == NULL) {
 305                 return ISTD_NORMAL;
 306         }
 307
 308         NO_DBG_DEC(pBest->tInfo.usIstd);
 309         return pBest->tInfo.usIstd;
 310 } /* end of usGetIstd */