WordConverter *WordConverterNew(DFDocument *html, const char *abstractPath,
const char *idPrefix, WordPackage *package,
DFBuffer *warnings)
{
WordConverter *converter = (WordConverter *)calloc(1,sizeof(WordConverter));
if ((abstractPath == NULL) || (strlen(abstractPath) == 0))
abstractPath = ".";
converter->html = DFDocumentRetain(html);
converter->abstractPath = strdup(abstractPath);
converter->concretePath = strdup(package->tempPath);
converter->idPrefix = DFStrDup(idPrefix);
converter->package = WordPackageRetain(package);
converter->styles = WordSheetNew(converter->package->styles);
converter->numbering = WordNumberingNew(converter->package);
converter->theme = WordThemeNew(converter->package);
converter->mainSection = WordSectionNew();
converter->objects = WordObjectsNew(converter->package);
converter->footnotes = WordNoteGroupNewFootnotes(converter->package->footnotes);
converter->endnotes = WordNoteGroupNewEndnotes(converter->package->endnotes);
converter->supportedContentTypes = DFHashTableNew((DFCopyFunction)strdup,free);
DFHashTableAdd(converter->supportedContentTypes,"jpg","image/jpeg");
DFHashTableAdd(converter->supportedContentTypes,"jpeg","image/jpeg");
DFHashTableAdd(converter->supportedContentTypes,"tif","image/tiff");
DFHashTableAdd(converter->supportedContentTypes,"tiff","image/tiff");
DFHashTableAdd(converter->supportedContentTypes,"gif","image/gif");
DFHashTableAdd(converter->supportedContentTypes,"bmp","image/bmp");
DFHashTableAdd(converter->supportedContentTypes,"png","image/png");
converter->warnings = DFBufferRetain(warnings);
return converter;
}
static ImageInfo *ImageInfoNew(const char *rId, double widthPts, double heightPts)
{
ImageInfo *info = (ImageInfo *)calloc(1,sizeof(ImageInfo));
info->rId = DFStrDup(rId);
info->widthPts = widthPts;
info->heightPts = heightPts;
return info;
}
static void fixWordLists(DFNode *node, WordConverter *conv)
{
for (DFNode *child = node->first; child != NULL; child = child->next)
fixWordLists(child,conv);
int haveParagraphs = 0;
for (DFNode *child = node->first; child != NULL; child = child->next) {
if (child->tag == WORD_P) {
haveParagraphs = 1;
break;
}
}
if (!haveParagraphs)
return;
int createdHashTables = 0;
DFHashTable *replacementNumIds = NULL;
DFHashTable *itemNoByListKey = NULL;
DFHashTable *lastNumIdByIlvl = NULL;
DFHashTable *itemNoByIlvl = NULL;
int maxIlvl = -1;
for (DFNode *child = node->first; child != NULL; child = child->next) {
if (child->tag != WORD_P)
continue;
DFNode *pPrElem = DFChildWithTag(child,WORD_PPR);
DFNode *numPrElem = DFChildWithTag(pPrElem,WORD_NUMPR);
DFNode *numIdElem = DFChildWithTag(numPrElem,WORD_NUMID);
DFNode *ilvlElem = DFChildWithTag(numPrElem,WORD_ILVL);
const char *numId = DFGetAttribute(numIdElem,WORD_VAL);
const char *ilvl = DFGetAttribute(ilvlElem,WORD_VAL);
if ((numId == NULL) || (atoi(numId) == 0))
continue;
if (!createdHashTables) {
replacementNumIds = DFHashTableNew((DFCopyFunction)xstrdup,free);
itemNoByListKey = DFHashTableNew((DFCopyFunction)xstrdup,free);
lastNumIdByIlvl = DFHashTableNew((DFCopyFunction)xstrdup,free);
itemNoByIlvl = DFHashTableNew((DFCopyFunction)xstrdup,free);
createdHashTables = 1;
}
if (ilvl == NULL)
ilvl = "0";;
WordConcreteNum *concreteNum = WordNumberingConcreteWithId(conv->numbering,numId); // may be NULL
WordNumLevel *numLevel = WordConcreteNumGetLevel(concreteNum,atoi(ilvl)); // may be NULL
const char *levelStart = NULL;
if (numLevel != NULL) {
for (DFNode *lvlChild = numLevel->element->first; lvlChild != NULL; lvlChild = lvlChild->next) {
switch (lvlChild->tag) {
case WORD_START:
levelStart = DFGetAttribute(lvlChild,WORD_VAL);
break;
}
}
}
char *listKey = DFFormatString("%s:%s",numId,ilvl);
char *itemNo = DFStrDup(DFHashTableLookup(itemNoByListKey,listKey));
if (itemNo == NULL) {
itemNo = xstrdup("1");
if ((levelStart != NULL) && (atoi(levelStart) > 1) && (atoi(ilvl) <= maxIlvl)) {
const char *prevNumId = DFHashTableLookup(lastNumIdByIlvl,ilvl);
const char *prevItemNo = DFHashTableLookup(itemNoByIlvl,ilvl);
if ((prevNumId != NULL) && (prevItemNo != NULL)) {
DFHashTableAdd(replacementNumIds,numId,prevNumId);
free(itemNo);
itemNo = DFFormatString("%d",atoi(prevItemNo)+1);
}
}
}
else {
int intValue = atoi(itemNo);
free(itemNo);
itemNo = DFFormatString("%d",intValue+1);
}
DFHashTableAdd(itemNoByListKey,listKey,itemNo);
const char *replNumId = DFHashTableLookup(replacementNumIds,numId);
if (replNumId != NULL) {
numId = replNumId;
DFSetAttribute(numIdElem,WORD_VAL,numId);
}
DFHashTableAdd(lastNumIdByIlvl,ilvl,numId);
DFHashTableAdd(itemNoByIlvl,ilvl,itemNo);
maxIlvl = atoi(ilvl);
free(listKey);
free(itemNo);
}
DFHashTableRelease(replacementNumIds);
DFHashTableRelease(itemNoByListKey);
DFHashTableRelease(lastNumIdByIlvl);
DFHashTableRelease(itemNoByIlvl);
}
char *WordStyleIdForStyle(CSSStyle *style)
{
const char *selector = style->selector;
char *resStyleId = NULL;
if (!strcmp(selector,"table.Normal_Table"))
return strdup("TableNormal");
if (!strcmp(selector,"table.Table_Grid"))
return strdup("TableGrid");
if (!strcmp(selector,"span.Default_Paragraph_Font"))
return strdup("DefaultParagraphFont");
if (!strcmp(selector,"p.List_Paragraph"))
return strdup("ListParagraph");
int headingLevel = CSSSelectorHeadingLevel(selector);
if (headingLevel != 0) {
char *prefix = DFFormatString("heading_%d",headingLevel);
if ((style->className != NULL) && DFStringHasPrefix(style->className,prefix)) {
char *rest = DFSubstring(style->className,strlen(prefix),strlen(style->className));
char *result = DFFormatString("Heading%d%s",headingLevel,rest);
free(rest);
free(prefix);
return result;
}
free(prefix);
}
if (!strcmp(selector,"span.Heading1Char"))
return strdup("Heading1Char");
if (!strcmp(selector,"span.Heading2Char"))
return strdup("Heading2Char");
if (!strcmp(selector,"span.Heading3Char"))
return strdup("Heading3Char");
if (!strcmp(selector,"span.Heading4Char"))
return strdup("Heading4Char");
if (!strcmp(selector,"span.Heading5Char"))
return strdup("Heading5Char");
if (!strcmp(selector,"span.Heading6Char"))
return strdup("Heading6Char");
if (!strcmp(selector,"span.Heading7Char"))
return strdup("Heading7Char");
if (!strcmp(selector,"span.Heading8Char"))
return strdup("Heading8Char");
if (!strcmp(selector,"span.Heading9Char"))
return strdup("Heading9Char");
char *className = CSSSelectorCopyClassName(selector);
switch (CSSSelectorGetTag(selector)) {
case HTML_FIGURE: {
resStyleId = DFStrDup("Figure");
break;
}
case HTML_CAPTION: {
resStyleId = DFStrDup("Caption");
break;
}
case HTML_H1:
case HTML_H2:
case HTML_H3:
case HTML_H4:
case HTML_H5:
case HTML_H6: {
if ((className == NULL) || (strlen(className) == 0)) {
int level = CSSSelectorHeadingLevel(selector);
if ((level >= 1) && (level <= 6)) {
// FIXME: we shouldn't rely on the specific word "Heading" here - instead using the localised name
// FIXME: not covered by tests
resStyleId = DFFormatString("Heading%d",level);
}
}
else {
resStyleId = DFStrDup(className);
}
break;
}
case HTML_P:
resStyleId = DFStrDup(className);
break;
case HTML_SPAN:
resStyleId = DFStrDup(className);
break;
case HTML_TABLE:
resStyleId = DFStrDup(className);
break;
}
free(className);
if (resStyleId == NULL) {
// Note: selector here may start with . (i.e. applies to all elements)
// FIXME: not covered by tests
resStyleId = strdup(selector);
}
return resStyleId;
}
