static void WordPutNumPr(DFNode *concrete, CSSProperties *newp)
{
DFNode *children[PREDEFINED_TAG_COUNT];
childrenToArray(concrete,children);
CSSProperties *oldp = CSSPropertiesNew();
WordGetNumPr(concrete,oldp);
if (!DFStringEquals(CSSGet(oldp,"-word-numId"),CSSGet(newp,"-word-numId")) ||
!DFStringEquals(CSSGet(oldp,"-word-ilvl"),CSSGet(newp,"-word-ilvl"))) {
if (CSSGet(newp,"-word-numId") != NULL) {
children[WORD_NUMID] = DFCreateElement(concrete->doc,WORD_NUMID);
DFSetAttribute(children[WORD_NUMID],WORD_VAL,CSSGet(newp,"-word-numId"));
if (CSSGet(newp,"-word-ilvl") != NULL) {
children[WORD_ILVL] = DFCreateElement(concrete->doc,WORD_ILVL);
DFSetAttribute(children[WORD_ILVL],WORD_VAL,CSSGet(newp,"-word-ilvl"));
}
else {
children[WORD_ILVL] = NULL;
}
}
else {
children[WORD_NUMID] = NULL;
children[WORD_ILVL] = NULL;
}
}
replaceChildrenFromArray(concrete,children,WordNumPr_Children);
CSSPropertiesRelease(oldp);
}
int identicalAttributesExcept(DFNode *first, DFNode *second, Tag except)
{
// FIXME: except paremeter is ignored
// This is O(n^2), but it's not really a problem as we generally only have a very small
// number of attributes
for (unsigned int i = 0; i < first->attrsCount; i++) {
Tag tag = first->attrs[i].tag;
if (tag == HTML_ID)
continue;
const char *firstValue = first->attrs[i].value;
const char *secondValue = DFGetAttribute(second,tag);
if (!DFStringEquals(firstValue,secondValue))
return 0;
}
for (unsigned int i = 0; i < second->attrsCount; i++) {
Tag tag = second->attrs[i].tag;
if (tag == HTML_ID)
continue;
const char *secondValue = second->attrs[i].value;
const char *firstValue = DFGetAttribute(first,tag);
if (!DFStringEquals(secondValue,firstValue))
return 0;
}
return 1;
}
static DFNode *WordParagraphContentCreate(WordPutData *put, DFNode *abstract)
{
switch (abstract->tag) {
case HTML_SPAN: {
const char *spanClass = DFGetAttribute(abstract,HTML_CLASS);
if (DFStringEquals(spanClass,DFFieldClass)) {
DFNode *concrete = DFCreateElement(put->contentDoc,WORD_FLDSIMPLE);
char *nodeText = DFNodeTextToString(abstract);
DFSetAttribute(concrete,WORD_INSTR,nodeText);
free(nodeText);
put->conv->haveFields = 1;
return concrete;
}
else if (DFStringEquals(spanClass,DFBookmarkClass)) {
const char *bookmarkId = DFGetAttribute(abstract,WORD_ID);
const char *bookmarkName = DFGetAttribute(abstract,WORD_NAME);
if ((bookmarkId == NULL) || (bookmarkName == NULL))
return NULL;;
WordBookmark *bookmark = WordObjectsBookmarkWithName(put->conv->objects,bookmarkName);
if (bookmark == NULL)
return NULL;;
DFNode *concrete = DFCreateElement(put->contentDoc,WORD_BOOKMARK);
DFSetAttribute(concrete,WORD_ID,bookmarkId);
DFSetAttribute(concrete,WORD_NAME,bookmarkName);
bookmark->element = concrete;
WordParagraphContentPut(put,abstract,concrete);
return concrete;
}
else {
DFNode *concrete = DFCreateElement(put->contentDoc,WORD_R);
WordParagraphContentPut(put,abstract,concrete);
return concrete;
}
}
case HTML_INS:
case HTML_DEL:
return WordChangeLens.create(put,abstract);
case HTML_A: {
const char *href = DFGetAttribute(abstract,HTML_HREF);
if (href != NULL) {
if (HTML_nodeIsHyperlink(abstract))
return WordHyperlinkLens.create(put,abstract);
else
return WordFieldLens.create(put,abstract);
}
return NULL;
}
default:
return NULL;
}
}
static void updateListTypes(WordPutData *put)
{
const char **htmlIds = DFHashTableCopyKeys(put->numIdByHtmlId);
for (int i = 0; htmlIds[i]; i++) {
const char *htmlId = htmlIds[i];
const char *numId = DFHashTableLookup(put->numIdByHtmlId,htmlId);
WordConcreteNum *num = WordNumberingConcreteWithId(put->conv->numbering,numId);
if (num == NULL)
continue; // FIXME: remove entry from both maps so it is re-created
DFNode *listNode = DFNodeForSeqNo(put->conv->html,(unsigned int)atoi(htmlId));
assert(listNode != NULL);
const char *htmlType = DFGetAttribute(listNode,CONV_LISTTYPE);
const char *htmlIlvl = DFGetAttribute(listNode,CONV_ILVL);
WordNumLevel *level = WordConcreteNumGetLevel(num,atoi(htmlIlvl));
if (level == NULL)
continue; // FIXME: remove entry from both maps so it is re-created
const char *wordType = WordNumLevelToListStyleType(level);
if (!DFStringEquals(wordType,htmlType)) {
// Make a copy of numId, as it may be freed during the first call to DFHashTableRemove
char *numIdCopy = strdup(numId);
DFHashTableRemove(put->numIdByHtmlId,htmlId);
DFHashTableRemove(put->htmlIdByNumId,numIdCopy);
free(numIdCopy);
if (num->referenceCount == 1)
WordNumberingRemoveConcrete(put->conv->numbering,num);
}
}
free(htmlIds);
}
// Finds the type of measurement used for cell widths in a table. If the table is empty, or
// the cells use different types of widths, this returns NULL.
static const char *cellWidthTypeForTable(DFNode *tbl)
{
const char *commonType = NULL;
for (DFNode *tr = tbl->first; tr != NULL; tr = tr->next) {
if (tr->tag != WORD_TR)
continue;
for (DFNode *tc = tr->first; tc != NULL; tc = tc->next) {
if (tc->tag != WORD_TC)
continue;
DFNode *tcPr = DFChildWithTag(tc,WORD_TCPR);
DFNode *tcW = DFChildWithTag(tcPr,WORD_TCW);
if (tcW != NULL) {
const char *type = DFGetAttribute(tcW,WORD_TYPE);
if (type == NULL)
type = "dxa";
if (commonType == NULL) // First cell we've encountered
commonType = type;
else if (!DFStringEquals(commonType,type))
return NULL;
}
}
}
return commonType;
}
static void removeCaptionBookmarksRecursive(DFNode *node, int inCaption)
{
if (node->tag == WORD_P) {
DFNode *pPr = DFChildWithTag(node,WORD_PPR);
const char *styleId = DFGetChildAttribute(pPr,WORD_PSTYLE,WORD_VAL);
if (DFStringEquals(styleId,"Caption"))
inCaption = 1;
}
DFNode *next;
for (DFNode *child = node->first; child != NULL; child = next) {
next = child->next;
removeCaptionBookmarksRecursive(child,inCaption);
}
if (inCaption) {
switch (node->tag) {
case WORD_BOOKMARKSTART:
case WORD_BOOKMARKEND:
case WORD_BOOKMARK:
DFRemoveNodeButKeepChildren(node);
break;
}
}
}
static void findTargetAndType(WordBookmark *bookmark, WordSheet *sheet)
{
if (DFStringEquals(bookmark->bookmarkName,"_GoBack")) {
bookmark->type = WordBookmarkCursor;
bookmark->target = NULL;
return;
}
// FIXME: Check if the following line is still relevant with the new bookmarks model
assert(bookmark->element != NULL);
DFNode *p = WordFindContainingParagraph(bookmark->element);
if (p == NULL)
return;
DFNode *pPr = DFChildWithTag(p,WORD_PPR);
if (pPr == NULL)
return;
DFNode *pStyle = DFChildWithTag(pPr,WORD_PSTYLE);
if (pStyle == NULL)
return;
const char *styleId = DFGetAttribute(pStyle,WORD_VAL);
if (styleId == NULL)
return;
WordStyle *style = WordSheetStyleForTypeId(sheet,"paragraph",styleId);
if ((style != NULL) && isHeadingOutlineLvl(style->outlineLvl)) {
bookmark->type = WordBookmarkHeading;
bookmark->target = p;
}
else if (DFStringEquals(styleId,"Caption")) {
DFNode *prev = findPreviousElement(p);
if (prev == NULL)
return;
if (prev->tag == WORD_TBL) {
bookmark->type = WordBookmarkTable;
bookmark->target = prev;
}
else if (Word_isFigureParagraph(prev)) {
bookmark->type = WordBookmarkFigure;
bookmark->target = prev;
}
else if (Word_isEquationParagraph(prev)) {
bookmark->type = WordBookmarkEquation;
bookmark->target = prev;
}
}
}
static void simplifyRecursive(WordSimplification *simp, DFNode *node)
{
switch (node->tag) {
case WORD_FLDCHAR: {
const char *type = DFGetAttribute(node,WORD_FLDCHARTYPE);
if (DFStringEquals(type,"begin")) {
if (simp->depth == 0) {
DFBufferRelease(simp->instrText);
simp->instrText = DFBufferNew();
simp->beginNode = node;
simp->endNode = NULL;
simp->inSeparate = 0;
}
simp->depth++;
}
else if (DFStringEquals(type,"end") && (simp->depth > 0)) {
simp->depth--;
if (simp->depth == 0) {
simp->endNode = node;
replaceField(simp);
}
}
else if (DFStringEquals(type,"separate")) {
if (simp->depth == 1)
simp->inSeparate = 1;
}
break;
}
case WORD_INSTRTEXT: {
if ((simp->depth == 1) && !simp->inSeparate) {
char *value = DFNodeTextToString(node);
DFBufferFormat(simp->instrText,"%s",value);
free(value);
}
break;
}
}
DFNode *next;
for (DFNode *child = node->first; child != NULL; child = next) {
next = child->next;
simplifyRecursive(simp,child);
}
}
static DFNode *WordRunContentGet(WordGetData *get, DFNode *concrete)
{
switch (concrete->tag) {
case WORD_T:
case WORD_DELTEXT: {
DFBuffer *buf = DFBufferNew();
DFNodeTextToBuffer(concrete,buf);
DFNode *abstract = DFCreateTextNode(get->conv->html,buf->data);
DFBufferRelease(buf);
return abstract;
}
case WORD_DRAWING:
case WORD_OBJECT:
case WORD_PICT:
return WordDrawingGet(get,concrete);
case WORD_TAB: {
DFNode *span = WordConverterCreateAbstract(get,HTML_SPAN,concrete);
DFSetAttribute(span,HTML_CLASS,DFTabClass);
return span;
}
case WORD_BR: {
const char *type = DFGetAttribute(concrete,WORD_TYPE);
if (DFStringEquals(type,"column")) {
DFNode *span = WordConverterCreateAbstract(get,HTML_SPAN,concrete);
DFSetAttribute(span,HTML_CLASS,DFPlaceholderClass);
DFCreateChildTextNode(span,"[Column break]");
return span;
}
else if (DFStringEquals(type,"page")) {
DFNode *span = WordConverterCreateAbstract(get,HTML_SPAN,concrete);
DFSetAttribute(span,HTML_CLASS,DFPlaceholderClass);
DFCreateChildTextNode(span,"[Page break]");
return span;
}
else {
return WordConverterCreateAbstract(get,HTML_BR,concrete);
}
}
default:
return NULL;
}
}
static void Word_addContentParts(DFNode *child, const char *content, WordCaption *caption)
{
if (content == NULL)
return;
DFNode *nextSibling = child->first;
DFArray *parts = CSSParseContent(content);
for (int i = 0; i < DFArrayCount(parts); i++) {
ContentPart *part = DFArrayItemAt(parts,i);
switch (part->type) {
case ContentPartString: {
DFNode *text = DFCreateTextNode(child->doc,part->value);
if (strlen(part->value) > 0) {
DFNode *span = DFCreateElement(child->doc,HTML_SPAN);
DFAppendChild(span,text);
DFInsertBefore(child,span,nextSibling);
}
break;
}
case ContentPartCounter: {
if (DFStringEquals(part->value,"figure")) {
DFNode *span = DFCreateElement(child->doc,HTML_SPAN);
DFSetAttribute(span,HTML_CLASS,DFFieldClass);
DFCreateChildTextNode(span," SEQ Figure \\* ARABIC ");
DFInsertBefore(child,span,nextSibling);
caption->number = span;
}
else if (DFStringEquals(part->value,"table")) {
DFNode *span = DFCreateElement(child->doc,HTML_SPAN);
DFSetAttribute(span,HTML_CLASS,DFFieldClass);
DFCreateChildTextNode(span," SEQ Table \\* ARABIC ");
DFInsertBefore(child,span,nextSibling);
caption->number = span;
}
break;
default:
break;
}
}
}
DFArrayRelease(parts);
}
static int WordBookmarkIsVisible2(DFNode *concrete)
{
const char *name = DFGetAttribute(concrete,WORD_NAME);
if (name == NULL)
return 0;
if (DFStringEquals(name,"_GoBack") && (concrete->first == NULL))
return 0;
return 1;
}
static int isSeqField(DFNode *node)
{
if (node->tag != HTML_SPAN)
return 0;
if (!DFStringEquals(DFGetAttribute(node,HTML_CLASS),DFFieldClass))
return 0;
char *instr = DFNodeTextToString(node);
const char **args = Word_parseField(instr);
int result = (args[0] != NULL) && !strcmp(args[0],"SEQ");
free(args);
free(instr);
return result;
}
static const char *bookmarkNameForHtmlId(WordConverter *converter, const char *htmlId, const char *refClass)
{
DFNode *htmlElem = DFElementForIdAttr(converter->html,htmlId);
if (htmlElem == NULL)
return NULL;
switch (htmlElem->tag) {
case HTML_H1:
case HTML_H2:
case HTML_H3:
case HTML_H4:
case HTML_H5:
case HTML_H6: {
DFNode *labelSpan = htmlElem->first;
if ((labelSpan == NULL) || (labelSpan->tag != HTML_SPAN))
return NULL;;
const char *labelClass = DFGetAttribute(labelSpan,HTML_CLASS);
if (!DFStringEquals(labelClass,DFBookmarkClass))
return NULL;
return DFGetAttribute(labelSpan,WORD_NAME);
}
case HTML_FIGURE:
case HTML_TABLE: {
WordCaption *caption = WordObjectsCaptionForTarget(converter->objects,htmlElem);
if (caption == NULL)
return NULL;
if (DFStringEquals(refClass,DFRefTextClass) && (caption->textBookmark != NULL))
return caption->textBookmark->bookmarkName;
else if (DFStringEquals(refClass,DFRefLabelNumClass) && (caption->labelNumBookmark != NULL))
return caption->labelNumBookmark->bookmarkName;
else if (DFStringEquals(refClass,DFRefCaptionTextClass) && (caption->captionTextBookmark != NULL))
return caption->captionTextBookmark->bookmarkName;
else if (caption->textBookmark != NULL)
return caption->textBookmark->bookmarkName; // default is entire caption
}
default:
return NULL;
}
}
static void populateTableStructure(DFTable *structure, DFNode *concrete)
{
// Populate table
int row = 0;
for (DFNode *tblChild = concrete->first; tblChild != NULL; tblChild = tblChild->next) {
if (tblChild->tag != WORD_TR)
continue;
DFTableSetRowElement(structure,tblChild,row);
int col = 0;
for (DFNode *trChild = tblChild->first; trChild != NULL; trChild = trChild->next) {
if (trChild->tag != WORD_TC)
continue;
const char *gridSpan = NULL;
const char *vMerge = NULL;
DFNode *tcPr = DFChildWithTag(trChild,WORD_TCPR);
if (tcPr != NULL) {
DFNode *gridSpanNode = DFChildWithTag(tcPr,WORD_GRIDSPAN);
DFNode *vMergeNode = DFChildWithTag(tcPr,WORD_VMERGE);
if (gridSpanNode != NULL)
gridSpan = DFGetAttribute(gridSpanNode,WORD_VAL);
if (vMergeNode != NULL) {
vMerge = DFGetAttribute(vMergeNode,WORD_VAL);
if (vMerge == NULL)
vMerge = "continue";
}
}
int colSpan = (gridSpan != NULL) ? atoi(gridSpan) : 1;
DFCell *cell;
if ((vMerge != NULL) && !DFStringEquals(vMerge,"restart") && (row > 0)) {
cell = DFCellRetain(DFTableGetCell(structure,row-1,col));
if (cell->rowSpan < row + 1 - cell->row)
cell->rowSpan = row + 1 - cell->row;
}
else {
cell = DFCellNew(trChild,row,col);
cell->colSpan = colSpan;
}
for (int i = 0; i < colSpan; i++)
DFTableSetCell(structure,row,col+i,cell);
col += colSpan;
DFCellRelease(cell);
}
row++;
}
}
static DFNode *WordRunContentCreate(WordPutData *put, DFNode *abstract)
{
switch (abstract->tag) {
case DOM_TEXT: {
DFNode *text = DFCreateTextNode(put->contentDoc,abstract->value);
// Text inside a <w:del> element must be stored in a <w:delText> element
// Text *not* inside a <w:del> element is stored in a <w:t> element
Tag tag = WORD_T;
for (DFNode *a = abstract->parent; a != NULL; a = a->parent) {
if (a->tag == HTML_DEL)
tag = WORD_DELTEXT;
}
DFNode *t = DFCreateElement(put->contentDoc,tag);
DFAppendChild(t,text);
char *trimmed = DFStringTrimWhitespace(abstract->value);
if (!DFStringEquals(trimmed,abstract->value))
DFSetAttribute(t,XML_SPACE,"preserve");
free(trimmed);
return t;
}
case HTML_IMG:
return WordDrawingCreate(put,abstract);
case HTML_BR:
return DFCreateElement(put->contentDoc,WORD_BR);
case HTML_SPAN: {
const char *className = DFGetAttribute(abstract,HTML_CLASS);
if (DFStringEquals(className,DFTabClass))
return DFCreateElement(put->contentDoc,WORD_TAB);
return NULL;
}
default:
return NULL;
}
}
static void WordFieldPut(WordPutData *put, DFNode *abstract, DFNode *concrete)
{
switch (abstract->tag) {
case HTML_SPAN: {
const char *className = DFGetAttribute(abstract,HTML_CLASS);
if (!DFStringEquals(className,DFFieldClass))
return;
char *text = DFNodeTextToString(abstract);
DFSetAttribute(concrete,WORD_INSTR,text);
free(text);
break;
}
case HTML_A: {
const char *href = DFGetAttribute(abstract,HTML_HREF);
if ((href == NULL) || (href[0] != '#'))
return;;
const char *targetId = &href[1];
const char *className = DFGetAttribute(abstract,HTML_CLASS);
if (className == NULL)
className = "";;
const char *bookmarkName = bookmarkNameForHtmlId(put->conv,targetId,className);
if (bookmarkName == NULL)
return;;
DFNode *htmlElem = DFElementForIdAttr(put->conv->html,targetId);
if ((htmlElem != NULL) && ((htmlElem->tag == HTML_TABLE) || (htmlElem->tag == HTML_FIGURE))) {
if (!DFStringEquals(className,DFRefTextClass) &&
!DFStringEquals(className,DFRefLabelNumClass) &&
!DFStringEquals(className,DFRefCaptionTextClass))
className = DFRefTextClass;
}
if (DFStringEquals(className,DFRefTextClass) ||
DFStringEquals(className,DFRefLabelNumClass) ||
DFStringEquals(className,DFRefCaptionTextClass))
DFFormatAttribute(concrete,WORD_INSTR," REF %s \\h ",bookmarkName);
else if (DFStringEquals(className,DFRefDirectionClass))
DFFormatAttribute(concrete,WORD_INSTR," REF %s \\p \\h ",bookmarkName);
else
DFFormatAttribute(concrete,WORD_INSTR," REF %s \\r \\h ",bookmarkName);
break;
}
}
}
static void removeRedundantProperties(CSSSheet *sheet)
{
// Remove any properties set on a style that have the same value as the corresponding property
// on the parent style. This is necessary because CSS doesn't support style inheritance (in
// the sense of Word & ODF's styles), so when we save out a HTML file, every style has all
// properties of its ancestors. After reading in a HTML file for the purposes of updating the
// original Word or ODF style, we don't want these extra property settings to remain, so that
// we can avoid adding spurious extra redundant property settings to the original file.
breakCycles(sheet);
const char **sortedSelectors = reverseTopologicalSortedSelectors(sheet);
for (size_t selIndex = 0; sortedSelectors[selIndex]; selIndex++) {
const char *selector = sortedSelectors[selIndex];
CSSStyle *child = CSSSheetLookupSelector(sheet,selector,0,0);
CSSStyle *parent = CSSSheetGetStyleParent(sheet,child);
if (parent == NULL)
continue;
const char **allSuffixes = CSSStyleCopySuffixes(child);
for (int suffixIndex = 0; allSuffixes[suffixIndex]; suffixIndex++) {
const char *suffix = allSuffixes[suffixIndex];
int isCell = !strcmp(suffix," > * > tr > td");
CSSProperties *childProperties = CSSStyleRuleForSuffix(child,suffix);
CSSProperties *parentProperties = CSSStyleRuleForSuffix(parent,suffix);
const char **allNames = CSSPropertiesCopyNames(childProperties);
for (int nameIndex = 0; allNames[nameIndex]; nameIndex++) {
const char *name = allNames[nameIndex];
// In docx's styles.xml, the tblCellMar values in table styles are not inherited
// (this seems like a bug in word, as isn't inconsistent with all other properties)
// So keep these ones.
if (isCell && DFStringHasPrefix(name,"padding-"))
continue;
const char *childVal = CSSGet(childProperties,name);
const char *parentVal = CSSGet(parentProperties,name);
if ((childVal != NULL) && (parentVal != NULL) && DFStringEquals(childVal,parentVal))
CSSPut(childProperties,name,NULL);
}
free(allNames);
}
free(allSuffixes);
}
free(sortedSelectors);
}
static void Word_preProcessHTML(WordConverter *word, DFNode *node)
{
switch (node->tag) {
case HTML_TABLE:
case HTML_FIGURE: {
DFNode *next;
for (DFNode *child = node->first; child != NULL; child = next) {
next = child->next;
if ((child->tag != HTML_CAPTION) && (child->tag != HTML_FIGCAPTION))
continue;
WordCaption *caption = WordCaptionNew(child);
WordObjectsSetCaption(word->objects,caption,node);
caption->contentStart = child->first;
WordCaptionRelease(caption);
const char *className = DFGetAttribute(child,HTML_CLASS);
CSSStyle *style;
if (child->tag == HTML_CAPTION)
style = CSSSheetLookupElement(word->styleSheet,"caption",className,0,0);
else
style = CSSSheetLookupElement(word->styleSheet,"figcaption",className,0,0);
CSSProperties *before = CSSStyleBefore(style);
if (CSSGet(before,"content") != NULL)
Word_addContentParts(child,CSSGet(before,"content"),caption);
child->tag = HTML_P;
DFSetAttribute(child,HTML_CLASS,"Caption");
DFInsertBefore(node->parent,child,node->next);
Word_preProcessHTML(word,child);
}
// The HTML normalization process ensures that apart from the <figcaption> element,
// all children of a <figure> are paragraphs or containers. Currently the editor only
// lets you create figures that contain a single image, so it's always a single
// paragraph. Since the HTML <figure> element gets mapped to a single <w:p> element
// by WordParagraphLens, we want to make sure it only contains inline children.
for (DFNode *child = node->first; child != NULL; child = next) {
next = child->next;
if (HTML_isParagraphTag(child->tag))
DFRemoveNodeButKeepChildren(child);
}
// FIXME: Handle <div>, <pre>, lists, tables etc which could also theoretically
// exist inside the <figure> element
break;
}
case HTML_NAV: {
const char *className = DFGetAttribute(node,HTML_CLASS);
const char *instr = NULL;
if (DFStringEquals(className,DFTableOfContentsClass))
instr = " TOC \\o \"1-3\" ";
else if (DFStringEquals(className,DFListOfFiguresClass))
instr = " TOC \\c \"Figure\" ";
else if (DFStringEquals(className,DFListOfTablesClass))
instr = " TOC \\c \"Table\" ";
if (instr != NULL) {
DFNode *p = DFCreateElement(word->html,HTML_P);
DFNode *field = DFCreateChildElement(p,HTML_SPAN);
DFSetAttribute(field,HTML_CLASS,DFFieldClass);
DFCreateChildTextNode(field,instr);
DFInsertBefore(node->parent,p,node);
DFRemoveNode(node);
}
break;
}
}
DFNode *next;
for (DFNode *child = node->first; child != NULL; child = next) {
next = child->next;
Word_preProcessHTML(word,child);
}
}
static void WordGetStyle(DFNode *concrete, CSSStyle *style, WordConverter *converter)
{
WordSection *section = converter->mainSection;
for (DFNode *child = concrete->first; child != NULL; child = child->next) {
const char *styleId = NULL;
switch (child->tag) {
case WORD_NAME:
CSSStyleSetDisplayName(style,DFGetAttribute(child,WORD_VAL));
break;
case WORD_NEXT: {
// FIXME: do he need to handle style types other than paragraph here?
const char *nextName = DFGetAttribute(child,WORD_VAL);
if (nextName == NULL)
continue;
WordStyle *nextStyle = WordSheetStyleForTypeId(converter->styles,"paragraph",nextName);
if (nextStyle == NULL)
continue;
CSSStyleSetNext(style,nextStyle->selector);
break;
}
case WORD_RPR:
WordGetRPr(child,CSSStyleRule(style),&styleId,converter->theme);
break;
case WORD_PPR: {
WordGetPPr(child,CSSStyleRule(style),&styleId,section);
if (style->headingLevel > 0) {
DFNode *numPr = DFChildWithTag(child,WORD_NUMPR);
if (numPr != NULL)
WordGetNumPrStyle(numPr,style,converter);
}
break;
}
case WORD_TBLPR:
WordGetTblPr(child,CSSStyleRule(style),CSSStyleCell(style),section,&styleId);
break;
case WORD_TRPR:
// FIXME
break;
case WORD_TCPR:
WordGetTcPr(child,CSSStyleRule(style));
break;
case WORD_TBLSTYLEPR:
WordGetTblStylePr(child,style,section,converter->theme);
break;
}
}
// Special case: The ListParagraph style that word automatically adds specifies an indentation
// of 36pt. We don't actually want this, because HTML automatically indents lists anyway. If
// we see this, clear it, but keep the old value around for when we update the word document.
StyleFamily family = WordStyleFamilyForSelector(style->selector);
const char *name = WordSheetStyleIdForSelector(converter->styles,style->selector);
if ((family == StyleFamilyParagraph) && DFStringEquals(name,"ListParagraph")) {
CSSProperties *properties = CSSStyleRule(style);
const char *wordMarginLeft = CSSGet(properties,"margin-left");
CSSPut(properties,"-word-margin-left",wordMarginLeft);
CSSPut(properties,"margin-left",NULL);
}
DFNode *pPr = DFChildWithTag(concrete,WORD_PPR);
DFNode *numPr = DFChildWithTag(pPr,WORD_NUMPR);
const char *numId = DFGetChildAttribute(numPr,WORD_NUMID,WORD_VAL);
if ((numId != NULL) && (atoi(numId) == 0)) {
switch (style->tag) {
case HTML_H1:
case HTML_H2:
case HTML_H3:
case HTML_H4:
case HTML_H5:
case HTML_H6:
case HTML_FIGURE:
case HTML_TABLE: {
char *counterIncrement = DFFormatString("%s 0",style->elementName);
CSSPut(CSSStyleRule(style),"counter-reset","null");
CSSPut(CSSStyleRule(style),"counter-increment",counterIncrement);
CSSPut(CSSStyleBefore(style),"content","none");
free(counterIncrement);
}
}
}
}
static void collapseRecursive(DFNode *node, DFHashTable *bookmarksById)
{
DFNode *next;
for (DFNode *child = node->first; child != NULL; child = next) {
next = child->next;
switch (child->tag) {
case WORD_BOOKMARKSTART:
case WORD_BOOKMARKEND: {
DFArray *startElements = DFArrayNew(NULL,NULL);
DFArray *endElements = DFArrayNew(NULL,NULL);
DFHashTable *startIds = DFHashTableNew((DFCopyFunction)strdup,(DFFreeFunction)free);
DFHashTable *endIds = DFHashTableNew((DFCopyFunction)strdup,(DFFreeFunction)free);
DFNode *n;
for (n = child;
(n != NULL) && ((n->tag == WORD_BOOKMARKSTART) || (n->tag == WORD_BOOKMARKEND));
n = n->next) {
if (n->tag == WORD_BOOKMARKSTART) {
const char *idValue = DFGetAttribute(n,WORD_ID);
if (idValue == NULL)
idValue = "";
DFHashTableAdd(startIds,idValue,idValue);
DFArrayAppend(startElements,n);
}
else {
const char *idValue = DFGetAttribute(n,WORD_ID);
if (idValue == NULL)
idValue = "";
DFHashTableAdd(endIds,idValue,idValue);
DFArrayAppend(endElements,n);
}
}
next = n;
DFArraySort(startElements,bookmarksById,compareStartElements);
for (size_t endIndex = 0; endIndex < DFArrayCount(endElements); endIndex++) {
DFNode *elem = DFArrayItemAt(endElements,endIndex);
const char *endId = DFGetAttribute(elem,WORD_ID);
int found = 0;
DFNode *ancestor;
for (ancestor = elem->parent; (ancestor != NULL) && !found; ancestor = ancestor->parent) {
if ((ancestor->tag == WORD_BOOKMARK) && DFStringEquals(DFGetAttribute(ancestor,WORD_ID),endId)) {
found = 1;
break;
}
}
if (found) {
DFNode *before = ancestor->next;
DFNode *nnext;
for (DFNode *n = child; n != NULL; n = nnext) {
nnext = n->next;
DFInsertBefore(ancestor->parent,n,before);
}
}
}
size_t x = 0;
while (x < DFArrayCount(startElements)) {
DFNode *element = DFArrayItemAt(startElements,x);
const char *bookmarkId = DFGetAttribute(element,WORD_ID);
if (bookmarkId == NULL)
bookmarkId = "";
if (DFHashTableLookup(endIds,bookmarkId) != NULL) {
element->tag = WORD_BOOKMARK;
DFArrayRemove(startElements,x);
}
else {
x++;
}
}
if (DFArrayCount(startElements) > 0) {
for (size_t i = 1; i < DFArrayCount(startElements); i++) {
DFNode *tempParent = DFArrayItemAt(startElements,i-1);
DFNode *tempChild = DFArrayItemAt(startElements,i);
DFAppendChild(tempParent,tempChild);
}
DFNode *last = DFArrayItemAt(startElements,DFArrayCount(startElements)-1);
while (next != NULL) {
DFNode *tempChild = next;
next = next->next;
DFAppendChild(last,tempChild);
}
}
for (size_t eIndex = 0; eIndex < DFArrayCount(startElements); eIndex++) {
DFNode *e = DFArrayItemAt(startElements,eIndex);
e->tag = WORD_BOOKMARK;
}
for (size_t eIndex = 0; eIndex < DFArrayCount(endElements); eIndex++) {
DFNode *e = DFArrayItemAt(endElements,eIndex);
DFRemoveNode(e);
}
if (DFArrayCount(startElements) > 0) {
DFNode *last = DFArrayItemAt(startElements,DFArrayCount(startElements)-1);
collapseRecursive(last,bookmarksById);
}
DFArrayRelease(startElements);
DFArrayRelease(endElements);
DFHashTableRelease(startIds);
DFHashTableRelease(endIds);
break;
}
default:
collapseRecursive(child,bookmarksById);
break;
}
}
}
int CSSGetDefault(CSSProperties *properties)
{
return DFStringEquals(CSSGet(properties,"-uxwrite-default"),"true");
}
void Word_setupBookmarkLinks(WordPutData *put)
{
DFHashTable *referencesById = findReferences(put->conv->html);
const char **sortedIds = DFHashTableCopyKeys(referencesById);
DFSortStringsCaseSensitive(sortedIds);
for (int idIndex = 0; sortedIds[idIndex]; idIndex++) {
const char *targetId = sortedIds[idIndex];
DFArray *references = DFHashTableLookup(referencesById,targetId);
DFNode *targetElem = DFElementForIdAttr(put->conv->html,targetId);
if (targetElem == NULL)
continue;
// The following is only relevant for figures and tables
int refText = 0;
int refLabelNum = 0;
int refCaptionText = 0;
for (int refIndex = 0; refIndex < DFArrayCount(references); refIndex++) {
DFNode *a = DFArrayItemAt(references,refIndex);
const char *className = DFGetAttribute(a,HTML_CLASS);
if (DFStringEquals(className,DFRefTextClass))
refText = 1;
else if (DFStringEquals(className,DFRefLabelNumClass))
refLabelNum = 1;
else if (DFStringEquals(className,DFRefCaptionTextClass))
refCaptionText = 1;
}
DFNode *concrete = WordConverterGetConcrete(put,targetElem);
switch (targetElem->tag) {
case HTML_H1:
case HTML_H2:
case HTML_H3:
case HTML_H4:
case HTML_H5:
case HTML_H6: {
const char *bookmarkId = NULL;
const char *bookmarkName = NULL;
DFNode *bookmarkElem = NULL;
if ((concrete != NULL) && (concrete->tag == WORD_P)) {
// FIXME: We only want to consider the bookmark to be the headings "correct"
// bookmark in the case where it contains all of the heading's content, though
// excluding other bookmarks that might come before or after it.
// If you have the cursor inside a heading bookmark when you save the document,
// word puts a bookmark called _GoBack there, and we of course don't want to
// confuse that with the actual heading's bookmark (if any).
// For now as a temporary hack we just explicitly filter out _GoBack; but there
// needs to be a more general fix, as there may be other bookmarks that end up
// in the heading.
for (DFNode *child = concrete->first; child != NULL; child = child->next) {
if ((child->tag == WORD_BOOKMARK) &&
!DFStringEquals(DFGetAttribute(child,WORD_NAME),"_GoBack")) {
bookmarkElem = child;
bookmarkId = DFGetAttribute(bookmarkElem,WORD_ID);
bookmarkName = DFGetAttribute(bookmarkElem,WORD_NAME);
break;
}
}
}
if ((bookmarkElem == NULL) || (bookmarkId == NULL) || (bookmarkName == NULL)) {
// New bookmark
WordBookmark *bookmark = WordObjectsAddBookmark(put->conv->objects);
bookmarkId =bookmark->bookmarkId;
bookmarkName = bookmark->bookmarkName;
}
DFNode *bookmarkSpan = DFCreateElement(put->conv->package->document,HTML_SPAN);
DFSetAttribute(bookmarkSpan,HTML_CLASS,DFBookmarkClass);
if (bookmarkElem != NULL) {
// FIXME: Not covered by tests
DFFormatAttribute(bookmarkSpan,HTML_ID,"%s%u",put->conv->idPrefix,bookmarkElem->seqNo);
}
DFSetAttribute(bookmarkSpan,WORD_NAME,bookmarkName);
DFSetAttribute(bookmarkSpan,WORD_ID,bookmarkId);
while (targetElem->first != NULL)
DFAppendChild(bookmarkSpan,targetElem->first);
DFAppendChild(targetElem,bookmarkSpan);
break;
}
case HTML_TABLE:
case HTML_FIGURE: {
WordCaption *caption = WordObjectsCaptionForTarget(put->conv->objects,targetElem);
if (caption == NULL)
break;
assert(caption->element != NULL);
assert((caption->number == NULL) || (caption->number->parent == caption->element));
assert((caption->contentStart == NULL) || (caption->contentStart->parent == caption->element));
// Note: caption.number may be null (i.e. if the caption is unnumbered)
// caption.contentStart may be null (if there is no text in the caption)
WordBookmark *captionTextBookmark = NULL;
WordBookmark *labelNumBookmark = NULL;
WordBookmark *textBookmark = NULL;
if (!refCaptionText && !refLabelNum && !refText)
refText = 1;
if (refCaptionText) {
captionTextBookmark = createBookmark(put->conv);
DFNode *nnext;
for (DFNode *n = caption->contentStart; n != NULL; n = nnext) {
nnext = n->next;
DFAppendChild(captionTextBookmark->element,n);
}
DFAppendChild(caption->element,captionTextBookmark->element);
}
if (refLabelNum && (caption->number != NULL)) {
labelNumBookmark = createBookmark(put->conv);
DFNode *numberNext = caption->number->next;
DFNode *nnext;
for (DFNode *n = caption->element->first; (n != NULL) && (n != numberNext); n = nnext) {
nnext = n->next;
DFAppendChild(labelNumBookmark->element,n);
}
DFInsertBefore(caption->element,labelNumBookmark->element,caption->element->first);
}
if (refText) {
textBookmark = createBookmark(put->conv);
DFNode *nnext;
for (DFNode *n = caption->element->first; n != NULL; n = nnext) {
nnext = n->next;
DFAppendChild(textBookmark->element,n);
}
DFAppendChild(caption->element,textBookmark->element);
}
caption->captionTextBookmark = captionTextBookmark;
caption->labelNumBookmark = labelNumBookmark;
caption->textBookmark = textBookmark;
break;
}
}
}
free(sortedIds);
DFHashTableRelease(referencesById);
}
static void Word_postProcessHTML(WordConverter *conv, DFNode *node)
{
DFNode *next;
for (DFNode *child = node->first; child != NULL; child = next) {
next = child->next;
switch (child->tag) {
case HTML_SPAN: {
const char *className = DFGetAttribute(child,HTML_CLASS);
if (DFStringEquals(className,DFBookmarkClass)) {
if (child->first != NULL)
next = child->first;
DFRemoveNodeButKeepChildren(child);
}
break;
}
case HTML_CAPTION: {
const char *counterName = NULL;
if ((child->prev != NULL) && (child->prev->tag == HTML_FIGURE) &&
(DFChildWithTag(child->prev,HTML_FIGCAPTION) == NULL)) {
child->tag = HTML_FIGCAPTION;
counterName = "figure";
DFAppendChild(child->prev,child);
}
else if ((child->prev != NULL) && (child->prev->tag == HTML_TABLE) &&
(DFChildWithTag(child->prev,HTML_CAPTION) == NULL)) {
counterName = "table";
DFInsertBefore(child->prev,child,child->prev->first);
}
else if ((child->next != NULL) && (child->next->tag == HTML_FIGURE) &&
(DFChildWithTag(child->next,HTML_FIGCAPTION) == NULL)) {
child->tag = HTML_FIGCAPTION;
counterName = "figure";
DFInsertBefore(child->next,child,child->next->first);
}
else if ((child->next != NULL) && (child->next->tag == HTML_TABLE) &&
(DFChildWithTag(child->next,HTML_CAPTION) == NULL)) {
counterName = "table";
DFSetAttribute(child,HTML_STYLE,"caption-side: top");
DFInsertBefore(child->next,child,child->next->first);
}
if (counterName != NULL) {
char *beforeText = extractPrefix(child,counterName);
if (beforeText != NULL) {
CSSStyle *style = CSSSheetLookupElement(conv->styleSheet,DFNodeName(child),NULL,1,0);
if (CSSGet(CSSStyleBefore(style),"content") == NULL) {
CSSPut(CSSStyleRule(style),"counter-increment",counterName);
CSSPut(CSSStyleBefore(style),"content",beforeText);
}
}
free(beforeText);
}
break;
}
case HTML_NAV: {
if (HTML_isParagraphTag(node->tag)) {
if (child->prev != NULL) {
DFNode *beforeP = DFCreateElement(conv->package->document,node->tag);
while (child->prev != NULL)
DFInsertBefore(beforeP,child->prev,beforeP->first);
DFInsertBefore(node->parent,beforeP,node);
}
DFInsertBefore(node->parent,child,node);
if ((node->first == NULL) ||
((node->first->tag == HTML_BR) && (node->first->next == NULL))) {
DFRemoveNode(node);
return;
}
next = NULL;
}
break;
}
}
}
for (DFNode *child = node->first; child != NULL; child = next) {
next = child->next;
Word_postProcessHTML(conv,child);
}
}
int CSSStyleIsNumbered(CSSStyle *style)
{
const char *beforeContent = CSSGet(CSSStyleBefore(style),"content");
return ((beforeContent != NULL) && !DFStringEquals(beforeContent,"\"\""));
}
void WordUpdateStyles(WordConverter *converter, CSSSheet *styleSheet)
{
CSSStyle *paraDefault = CSSSheetDefaultStyleForFamily(styleSheet,StyleFamilyParagraph);
if (CSSGet(CSSStyleRule(paraDefault),"margin-top") == NULL)
CSSPut(CSSStyleRule(paraDefault),"margin-top","-word-auto");
if (CSSGet(CSSStyleRule(paraDefault),"margin-bottom") == NULL)
CSSPut(CSSStyleRule(paraDefault),"margin-bottom","-word-auto");
if (converter->package->styles == NULL) // FIXME: create this document
return;;
DFNode *root = converter->package->styles->root;
if ((root == NULL) || (root->tag != WORD_STYLES))
return;;
DFHashTable *remainingSelectors = DFHashTableNew(NULL,NULL); // Used as a set
const char **allSelectors = CSSSheetCopySelectors(styleSheet);
for (int i = 0; allSelectors[i]; i++) {
const char *selector = allSelectors[i];
DFHashTableAdd(remainingSelectors,selector,"");
}
free(allSelectors);
WordSheet *sheet = converter->styles;
DFHashTable *oldConcreteNumIds = WordSheetFindUsedConcreteNumIds(sheet);
updateNumbering(converter,styleSheet);
// Update or remove existing styles
const char **allIdents = WordSheetCopyIdents(sheet);
for (int i = 0; allIdents[i]; i++) {
WordStyle *wordStyle = WordSheetStyleForIdent(sheet,allIdents[i]);
DFNode *element = wordStyle->element;
if (WordStyleIsProtected(wordStyle)) {
DFHashTableRemove(remainingSelectors,wordStyle->selector);
continue;
}
if (!DFStringEquals(wordStyle->type,"paragraph") &&
!DFStringEquals(wordStyle->type,"character") &&
!DFStringEquals(wordStyle->type,"table"))
continue;
CSSStyle *cssStyle = CSSSheetLookupSelector(styleSheet,wordStyle->selector,0,0);
if (cssStyle == NULL) {
// Remove style
WordSheetRemoveStyle(sheet,wordStyle);
continue;
}
// Update style
WordPutStyle(element,cssStyle,converter);
updateDefault(cssStyle,element,styleSheet,converter);
DFHashTableRemove(remainingSelectors,wordStyle->selector);
}
free(allIdents);
// Sort the list of new styles, so that test output is deterministic
const char **sortedSelectors = DFHashTableCopyKeys(remainingSelectors);
DFSortStringsCaseInsensitive(sortedSelectors);
// Add new styles. We do this in two stages - first creating the styles, and then setting their properties.
// This is because the second stage depends on referenced styles (e.g. based on and next) to be already
// present.
for (int selIndex = 0; sortedSelectors[selIndex]; selIndex++) {
const char *selector = sortedSelectors[selIndex];
CSSStyle *style = CSSSheetLookupSelector(styleSheet,selector,0,0);
const char *familyStr = NULL;
StyleFamily family = WordStyleFamilyForSelector(selector);
if (family == StyleFamilyParagraph)
familyStr = "paragraph";
else if (family == StyleFamilyCharacter)
familyStr = "character";
else if (family == StyleFamilyTable)
familyStr = "table";
else
continue;
char *styleId = WordStyleIdForStyle(style);
char *name = WordStyleNameForStyle(style);
if (name == NULL)
name = xstrdup(styleId);;
WordStyle *wordStyle = WordSheetAddStyle(sheet,familyStr,styleId,name,selector);
DFCreateChildElement(wordStyle->element,WORD_QFORMAT);
free(styleId);
free(name);
}
for (int selIndex = 0; sortedSelectors[selIndex]; selIndex++) {
const char *selector = sortedSelectors[selIndex];
StyleFamily family = WordStyleFamilyForSelector(selector);
if ((family != StyleFamilyParagraph) &&
(family != StyleFamilyCharacter) &&
(family != StyleFamilyTable))
continue;
CSSStyle *style = CSSSheetLookupSelector(styleSheet,selector,0,0);
WordStyle *wordStyle = WordSheetStyleForSelector(converter->styles,selector);
assert(wordStyle != NULL);
CSSStyleAddDefaultHTMLProperties(style);
// FIXME: language
// FIXME: not covered by tests
if ((style->headingLevel >= 1) && (style->headingLevel <= 6))
CSSStyleSetNext(style,"p.Normal");
WordPutStyle(wordStyle->element,style,converter);
updateDefault(style,wordStyle->element,styleSheet,converter);
}
free(sortedSelectors);
// Update body style (document defaults)
updateDefaults(converter,styleSheet);
updateBody(converter,styleSheet);
DFHashTable *newConcreteNumIds = WordSheetFindUsedConcreteNumIds(sheet);
const char **oldKeys = DFHashTableCopyKeys(oldConcreteNumIds);
for (int oldIndex = 0; oldKeys[oldIndex]; oldIndex++) {
const char *numId = oldKeys[oldIndex];
if (DFHashTableLookup(newConcreteNumIds,numId) == NULL) {
WordConcreteNum *concreteNum = WordNumberingConcreteWithId(converter->numbering,numId);
if (concreteNum != NULL)
WordNumberingRemoveConcrete(converter->numbering,concreteNum);
}
}
free(oldKeys);
DFHashTableRelease(remainingSelectors);
DFHashTableRelease(oldConcreteNumIds);
DFHashTableRelease(newConcreteNumIds);
}
static void WordPutStyle(DFNode *concrete, CSSStyle *style, WordConverter *converter)
{
// If we find a style with a counter-increment value of "<elementname> 0", this means that it's
// an explicitly unnumbered paragraph. So we set the numbering id to 0, so word doesn't increment
// the corresponding counter when encountering this style
if (CSSGet(CSSStyleRule(style),"counter-increment") != NULL) {
char *zeroIncrement = DFFormatString("%s 0",style->elementName);
if (DFStringEquals(CSSGet(CSSStyleRule(style),"counter-increment"),zeroIncrement))
CSSPut(CSSStyleRule(style),"-word-numId","0");
free(zeroIncrement);
}
int outlineLvl = -1;
if ((style->headingLevel >= 1) && (style->headingLevel <= 6)) {
outlineLvl = style->headingLevel-1;
char *nextSelector = CSSStyleCopyNext(style);
if (nextSelector == NULL) {
CSSStyle *def = CSSSheetDefaultStyleForFamily(converter->styleSheet,StyleFamilyParagraph);
if (def != NULL)
CSSStyleSetNext(style,def->selector);
}
free(nextSelector);
}
DFNode *children[PREDEFINED_TAG_COUNT];
childrenToArray(concrete,children);
char *parentSelector = CSSStyleCopyParent(style);
if (parentSelector == NULL) {
if (style->className != NULL) {
if ((style->tag != HTML_P) && (style->tag != HTML_SPAN) && (style->tag != HTML_TABLE)) {
CSSStyle *parentStyle = CSSSheetLookupElement(converter->styleSheet,style->elementName,NULL,0,0);
if ((parentStyle != NULL) && !parentStyle->latent)
parentSelector = xstrdup(style->elementName);
}
}
}
// Based on
const char *basedOn = WordSheetStyleIdForSelector(converter->styles,parentSelector);
if (basedOn == NULL) {
children[WORD_BASEDON] = NULL;
}
else {
children[WORD_BASEDON] = DFCreateElement(concrete->doc,WORD_BASEDON);
DFSetAttribute(children[WORD_BASEDON],WORD_VAL,basedOn);
}
// Style for next paragraph
children[WORD_NEXT] = NULL;
char *nextSelector = CSSStyleCopyNext(style);
if (nextSelector != NULL) {
StyleFamily thisFamily = WordStyleFamilyForSelector(style->selector);
StyleFamily nextFamily = WordStyleFamilyForSelector(nextSelector);
if (nextFamily == thisFamily) {
children[WORD_NEXT] = DFCreateElement(concrete->doc,WORD_NEXT);
const char *nextStyleId = WordSheetStyleIdForSelector(converter->styles,nextSelector);
DFSetAttribute(children[WORD_NEXT],WORD_VAL,nextStyleId);
}
}
if (WordStyleFamilyForSelector(style->selector) == StyleFamilyTable) {
// Table properties
if (children[WORD_TBLPR] == NULL)
children[WORD_TBLPR] = DFCreateElement(concrete->doc,WORD_TBLPR);;
const char *oldJc = DFGetChildAttribute(children[WORD_TBLPR],WORD_JC,WORD_VAL);
CSSProperties *wholeTable = CSSStyleRuleForSuffix(style,DFTableSuffixWholeTable);
WordPutTblPr(children[WORD_TBLPR],wholeTable,CSSStyleCell(style),converter->mainSection,NULL);
const char *newJc = DFGetChildAttribute(children[WORD_TBLPR],WORD_JC,WORD_VAL);
if (children[WORD_TBLPR]->first == NULL)
children[WORD_TBLPR] = NULL;
if (children[WORD_TRPR] == NULL)
children[WORD_TRPR] = DFCreateElement(concrete->doc,WORD_TRPR);
WordPutTrPr(children[WORD_TRPR],oldJc,newJc);
if (children[WORD_TRPR]->first == NULL)
children[WORD_TRPR] = NULL;
}
else {
// Paragraph properties
if (children[WORD_PPR] == NULL)
children[WORD_PPR] = DFCreateElement(concrete->doc,WORD_PPR);
WordPutPPr(children[WORD_PPR],CSSStyleRule(style),NULL,converter->mainSection,outlineLvl);
if (children[WORD_PPR]->first == NULL)
children[WORD_PPR] = NULL;
// Run properties
if (children[WORD_RPR] == NULL)
children[WORD_RPR] = DFCreateElement(concrete->doc,WORD_RPR);
WordPutRPr(children[WORD_RPR],CSSStyleRule(style),NULL,converter->theme);
if (children[WORD_RPR]->first == NULL)
children[WORD_RPR] = NULL;
}
replaceChildrenFromArray(concrete,children,WordStyle_Children);
free(parentSelector);
free(nextSelector);
}
static void WordPutSectPr(DFNode *concrete, CSSSheet *styleSheet, WordSection *section)
{
// Note: The sectPr element can potentially contain one or more headerReference or
// footerReference elements at the start, before the elements in WordSectPr_Children.
// So the straight childrenToArray/replaceChildrenFromArray method used elsewhere doesn't
// work here - we need to make sure these other elements are maintained
DFNode *children[PREDEFINED_TAG_COUNT];
childrenToArray(concrete,children);
CSSProperties *oldBody = CSSPropertiesNew();
CSSProperties *oldPage = CSSPropertiesNew();
WordGetSectPr(concrete,oldBody,oldPage,section);
CSSProperties *newBody = CSSSheetBodyProperties(styleSheet);
CSSProperties *newPage = CSSSheetPageProperties(styleSheet);
// Page size
if (children[WORD_PGSZ] == NULL)
children[WORD_PGSZ] = DFCreateElement(concrete->doc,WORD_PGSZ);
int updatePageSize = 0;
const char *widthStr = DFGetAttribute(children[WORD_PGSZ],WORD_W);
const char *heightStr = DFGetAttribute(children[WORD_PGSZ],WORD_H);
int widthTwips;
int heightTwips;
if ((widthStr == NULL) || (atoi(widthStr) <= 0) ||
(heightStr == NULL) || (atoi(heightStr) <= 0)) {
// Invalid or missing page size: Set to A4 portrait
widthTwips = A4_WIDTH_TWIPS;
heightTwips = A4_HEIGHT_TWIPS;
updatePageSize = 1;
}
else {
widthTwips = atoi(widthStr);
heightTwips = atoi(heightStr);
}
if (!DFStringEquals(CSSGet(oldPage,"size"),CSSGet(newPage,"size"))) {
const char *newSize = CSSGet(newPage,"size");
if (DFStringEqualsCI(newSize,"A4 portrait")) {
widthTwips = A4_WIDTH_TWIPS;
heightTwips = A4_HEIGHT_TWIPS;
}
else if (DFStringEqualsCI(newSize,"A4 landscape")) {
widthTwips = A4_HEIGHT_TWIPS;
heightTwips = A4_WIDTH_TWIPS;
}
else if (DFStringEqualsCI(newSize,"letter portrait")) {
widthTwips = LETTER_WIDTH_TWIPS;
heightTwips = LETTER_HEIGHT_TWIPS;
}
else if (DFStringEqualsCI(newSize,"letter landscape")) {
widthTwips = LETTER_HEIGHT_TWIPS;
heightTwips = LETTER_WIDTH_TWIPS;
}
else {
widthTwips = A4_WIDTH_TWIPS;
heightTwips = A4_HEIGHT_TWIPS;
}
updatePageSize = 1;
}
if (updatePageSize) {
DFFormatAttribute(children[WORD_PGSZ],WORD_W,"%d",widthTwips);
DFFormatAttribute(children[WORD_PGSZ],WORD_H,"%d",heightTwips);
if (widthTwips > heightTwips)
DFSetAttribute(children[WORD_PGSZ],WORD_ORIENT,"landscape");
else
DFRemoveAttribute(children[WORD_PGSZ],WORD_ORIENT);
}
if (children[WORD_PGMAR] == NULL)
children[WORD_PGMAR] = DFCreateElement(concrete->doc,WORD_PGMAR);
// Page margins
if (!DFStringEquals(CSSGet(oldBody,"margin-left"),CSSGet(newBody,"margin-left")) || updatePageSize)
updateTwipsFromLength(children[WORD_PGMAR],WORD_LEFT,CSSGet(newBody,"margin-left"),widthTwips);
if (!DFStringEquals(CSSGet(oldBody,"margin-right"),CSSGet(newBody,"margin-right")) || updatePageSize)
updateTwipsFromLength(children[WORD_PGMAR],WORD_RIGHT,CSSGet(newBody,"margin-right"),widthTwips);
if (!DFStringEquals(CSSGet(oldBody,"margin-top"),CSSGet(newBody,"margin-top")) || updatePageSize)
updateTwipsFromLength(children[WORD_PGMAR],WORD_TOP,CSSGet(newBody,"margin-top"),widthTwips);
if (!DFStringEquals(CSSGet(oldBody,"margin-bottom"),CSSGet(newBody,"margin-bottom")) || updatePageSize)
updateTwipsFromLength(children[WORD_PGMAR],WORD_BOTTOM,CSSGet(newBody,"margin-bottom"),widthTwips);
if (children[WORD_PGMAR]->attrsCount == 0)
children[WORD_PGMAR] = NULL;;
DFArray *extra = DFArrayNew(NULL,NULL);
for (DFNode *child = concrete->first; child != NULL; child = child->next) {
switch (child->tag) {
case WORD_HEADERREFERENCE:
case WORD_FOOTERREFERENCE:
DFArrayAppend(extra,child);
break;
}
}
replaceChildrenFromArray(concrete,children,WordSectPr_Children);
for (long i = (long)(DFArrayCount(extra)-1); i >= 0; i--) {
DFNode *child = DFArrayItemAt(extra,i);
DFInsertBefore(concrete,child,concrete->first);
}
DFArrayRelease(extra);
CSSPropertiesRelease(oldBody);
CSSPropertiesRelease(oldPage);
}
static DFNode *WordTblGet(WordGetData *get, DFNode *concrete)
{
if (concrete->tag != WORD_TBL)
return NULL;;
DFNode *table = WordConverterCreateAbstract(get,HTML_TABLE,concrete);
ConcreteInfo *cinfo = getConcreteInfo(get->conv,concrete);
calcTotals(get,cinfo);
const char *cellWidthType = cellWidthTypeForTable(concrete);
int autoWidth = DFStringEquals(cellWidthType,"auto");
if ((CSSGet(cinfo->tableProperties,"width") == NULL) && autoWidth) {
CSSPut(cinfo->tableProperties,"width",NULL);
}
else {
// Determine column widths and table width
if (cinfo->totalWidthPts > 0) {
DFNode *colgroup = HTML_createColgroup(get->conv->html,cinfo->structure);
DFAppendChild(table,colgroup);
double tableWidthPct = 100.0;
if (WordSectionContentWidth(get->conv->mainSection) > 0) {
double contentWidthPts = WordSectionContentWidth(get->conv->mainSection)/20.0;
tableWidthPct = 100.0*cinfo->totalWidthPts/contentWidthPts;
if (CSSGet(cinfo->tableProperties,"width") == NULL) {
char buf[100];
CSSPut(cinfo->tableProperties,"width",DFFormatDoublePct(buf,100,tableWidthPct));
}
}
}
if (CSSGet(cinfo->tableProperties,"width") == NULL)
CSSPut(cinfo->tableProperties,"width","100%");
}
DFHashTable *collapsed = CSSCollapseProperties(cinfo->tableProperties);
char *styleValue = CSSSerializeProperties(collapsed);
DFHashTableRelease(collapsed);
if (strlen(styleValue) > 0)
DFSetAttribute(table,HTML_STYLE,styleValue);
free(styleValue);
if ((cinfo->style != NULL) && (cinfo->style->selector != NULL)) {
char *className = CSSSelectorCopyClassName(cinfo->style->selector);
DFSetAttribute(table,HTML_CLASS,className);
free(className);
}
else {
CSSStyle *defaultStyle = CSSSheetDefaultStyleForFamily(get->conv->styleSheet,StyleFamilyTable);
if (defaultStyle != NULL)
DFSetAttribute(table,HTML_CLASS,defaultStyle->className);
}
// Create rows and cells
int row = 0;
for (DFNode *tblChild = concrete->first; tblChild != NULL; tblChild = tblChild->next) {
if (tblChild->tag != WORD_TR)
continue;
DFNode *tr = WordConverterCreateAbstract(get,HTML_TR,tblChild);
DFAppendChild(table,tr);
unsigned int col = 0;
while (col < cinfo->structure->cols) {
DFCell *cell = DFTableGetCell(cinfo->structure,row,col);
if (cell == NULL) {
DFNode *td = DFCreateElement(get->conv->html,HTML_TD);
DFAppendChild(tr,td);
col++;
continue;
}
if (row == cell->row) {
DFNode *td = WordTcGet(get,cell->element);
DFAppendChild(tr,td);
if (cell->colSpan != 1)
DFFormatAttribute(td,HTML_COLSPAN,"%d",cell->colSpan);
if (cell->rowSpan != 1)
DFFormatAttribute(td,HTML_ROWSPAN,"%d",cell->rowSpan);
}
col += cell->colSpan;
}
row++;
}
ConcreteInfoFree(cinfo);
return table;
}
static void Word_fixListSingle(WordConverter *conv, DFNode *node)
{
ListStack stack;
bzero(&stack,sizeof(ListStack));
DFNode *next;
for (DFNode *child = node->first; child != NULL; child = next) {
next = child->next;
int isListItem = 0;
if (child->tag == HTML_P) {
DFNode *elem = child;
const char *numIdStr = DFGetAttribute(elem,WORD_NUMID);
const char *ilvlStr = DFGetAttribute(elem,WORD_ILVL);
DFRemoveAttribute(elem,WORD_NUMID);
DFRemoveAttribute(elem,WORD_ILVL);
// A numId of 0 means that there is no numbering applied to this paragraph
if ((numIdStr != NULL) && (atoi(numIdStr) == 0)) {
numIdStr = NULL;
ilvlStr = NULL;
}
if ((numIdStr != NULL) && (ilvlStr != NULL)) {
isListItem = 1;
int numId = atoi(numIdStr);
int ilvl = atoi(ilvlStr);
ListDimensions dimensions = listIndent(conv,numIdStr,ilvlStr);
// Find the list at the same ilvl, and check if it has the same numId. If not, we're
// starting a new list.
ListFrame *sameLevelFrame = NULL;
for (ListFrame *frame = stack.top; frame != NULL; frame = frame->parent) {
if (frame->ilvl == ilvl)
sameLevelFrame = frame;
}
if ((sameLevelFrame != NULL) && (sameLevelFrame->numId != numId))
fixTrailingParagraphs(&stack,ilvl);
else
fixTrailingParagraphs(&stack,ilvl+1);
if ((stack.top != NULL) && (stack.top->numId != numId))
ListStackPopToAboveIlvl(&stack,ilvl);
else if ((stack.top != NULL) && (stack.top->ilvl > ilvl))
ListStackPopToAboveIlvl(&stack,ilvl+1);
if ((stack.top == NULL) || (stack.top->numId != numId) || (stack.top->ilvl < ilvl)) {
WordConcreteNum *num = WordNumberingConcreteWithId(conv->numbering,numIdStr); // may be NULL
WordNumLevel *level = WordConcreteNumGetLevel(num,ilvl); // may be NULL
const char *type = WordNumLevelToListStyleType(level); // may be NULL
Tag tag;
if (DFStringEquals(type,"disc") ||
DFStringEquals(type,"circle") ||
DFStringEquals(type,"square"))
tag = HTML_UL;
else
tag = HTML_OL;
DFNode *element = DFCreateElement(conv->html,tag);
if (type != NULL)
DFFormatAttribute(element,HTML_STYLE,"list-style-type: %s",type);
if (stack.top != NULL) {
DFNode *li;
if (stack.top->element->last != NULL)
li = stack.top->element->last;
else
li = DFCreateChildElement(stack.top->element,HTML_LI);
DFAppendChild(li,element);
}
else {
DFInsertBefore(node,element,child);
}
ListStackPushFrame(&stack,element,numId,ilvl,dimensions);
}
}
}
if (stack.top != NULL) {
DFNode *li;
if ((stack.top->element->last != NULL) && !isListItem)
li = stack.top->element->last;
else
li = DFCreateChildElement(stack.top->element,HTML_LI);
DFAppendChild(li,child);
}
}
fixTrailingParagraphs(&stack,-1);
while (stack.top != NULL)
ListStackPop(&stack);
}
void WordPutPPr(DFNode *pPr, CSSProperties *properties, const char *styleId, WordSection *section, int outlineLvl)
{
assert(pPr->tag == WORD_PPR);
// The child elements of pPr have to be in a specific order. So we build up a structure based
// on the existing elements (updated as appropriate from newProperties), and then rebuild
// from that
CSSProperties *oldp = CSSPropertiesNew();
CSSProperties *newp = properties;
const char *oldStyleId = NULL;
const char *newStyleId = styleId;
WordGetPPr(pPr,oldp,&oldStyleId,section);
{
DFNode *children[PREDEFINED_TAG_COUNT];
childrenToArray(pPr,children);
int existingOutlineLvl = -1;
if (children[WORD_OUTLINELVL] != NULL) {
const char *value = DFGetAttribute(children[WORD_OUTLINELVL],WORD_VAL);
if (value != NULL)
existingOutlineLvl = atoi(value);
}
// Style name
if (!DFStringEquals(oldStyleId,newStyleId)) {
if (newStyleId != NULL) {
children[WORD_PSTYLE] = DFCreateElement(pPr->doc,WORD_PSTYLE);
DFSetAttribute(children[WORD_PSTYLE],WORD_VAL,newStyleId);
}
else {
children[WORD_PSTYLE] = NULL;
}
}
// Paragraph border (pBdr)
if (children[WORD_PBDR] == NULL)
children[WORD_PBDR] = DFCreateElement(pPr->doc,WORD_PBDR);
WordPutPBdr(children[WORD_PBDR],oldp,newp);
if (children[WORD_PBDR]->first == NULL)
children[WORD_PBDR] = NULL;
// Numbering and outline level
// Don't change these properties for styles with outline level >= 6, as these can't be
// represented with the standard HTML heading elements, which only go from h1 - h6
// (outline levels 0 - 5)
if (existingOutlineLvl < 6) {
if (children[WORD_NUMPR] == NULL)
children[WORD_NUMPR] = DFCreateElement(pPr->doc,WORD_NUMPR);
WordPutNumPr(children[WORD_NUMPR],newp);
if (children[WORD_NUMPR]->first == NULL)
children[WORD_NUMPR] = NULL;
if ((outlineLvl >= 0) && (outlineLvl <= 5)) {
if (children[WORD_OUTLINELVL] == NULL)
children[WORD_OUTLINELVL] = DFCreateElement(pPr->doc,WORD_OUTLINELVL);
DFFormatAttribute(children[WORD_OUTLINELVL],WORD_VAL,"%d",outlineLvl);
}
else {
children[WORD_OUTLINELVL] = NULL;
}
}
// background-color
char *oldBackgroundColor = CSSHexColor(CSSGet(oldp,"background-color"),0);
char *newBackgroundColor = CSSHexColor(CSSGet(newp,"background-color"),0);
if (!DFStringEquals(oldBackgroundColor,newBackgroundColor))
WordPutShd(pPr->doc,&children[WORD_SHD],newBackgroundColor);
free(oldBackgroundColor);
free(newBackgroundColor);
// text-align
if (!DFStringEquals(CSSGet(oldp,"text-align"),CSSGet(newp,"text-align"))) {
const char *newTextAlign = CSSGet(newp,"text-align");
if (newTextAlign != NULL) {
const char *val = NULL;
if (!strcmp(newTextAlign,"left"))
val = "left";
else if (!strcmp(newTextAlign,"right"))
val = "right";
else if (!strcmp(newTextAlign,"center"))
val = "center";
else if (!strcmp(newTextAlign,"justify"))
val = "both";
if (val != NULL) {
children[WORD_JC] = DFCreateElement(pPr->doc,WORD_JC);
DFSetAttribute(children[WORD_JC],WORD_VAL,val);
}
}
else {
children[WORD_JC] = NULL;
}
}
if ((section != NULL) && (WordSectionContentWidth(section) >= 0)) {
const char *oldMarginLeft = CSSGet(oldp,"margin-left");
const char *oldMarginRight = CSSGet(oldp,"margin-right");
const char *oldTextIndent = CSSGet(oldp,"text-indent");
const char *newMarginLeft = CSSGet(newp,"margin-left");
const char *newMarginRight = CSSGet(newp,"margin-right");
const char *newTextIndent = CSSGet(newp,"text-indent");
// Special case of the List_Paragraph style, which is used by Word to ensure lists are indented. We
// don't set this property for HTML, because it automatically indents lists. However we need to ensure
// that it remains unchanged when updating the word document
const char *newWordMarginLeft = CSSGet(newp,"-word-margin-left");
if (newMarginLeft == NULL)
newMarginLeft = newWordMarginLeft;
if ((newMarginLeft == NULL) && (newMarginRight == NULL) && (newTextIndent == NULL)) {
children[WORD_IND] = NULL;
}
else {
if (children[WORD_IND] == NULL)
children[WORD_IND] = DFCreateElement(pPr->doc,WORD_IND);
if (!DFStringEquals(oldMarginLeft,newMarginLeft)) {
if (newMarginLeft != NULL)
updateTwipsFromLength(children[WORD_IND],WORD_LEFT,newMarginLeft,WordSectionContentWidth(section));
else
DFRemoveAttribute(children[WORD_IND],WORD_LEFT);
DFRemoveAttribute(children[WORD_IND],WORD_START);
}
if (!DFStringEquals(oldMarginRight,newMarginRight)) {
if (newMarginRight != NULL)
updateTwipsFromLength(children[WORD_IND],WORD_RIGHT,newMarginRight,WordSectionContentWidth(section));
else
DFRemoveAttribute(children[WORD_IND],WORD_RIGHT);
DFRemoveAttribute(children[WORD_IND],WORD_END);
}
if (!DFStringEquals(oldTextIndent,newTextIndent)) {
if (newTextIndent != NULL) {
CSSLength length = CSSLengthFromString(newTextIndent);
if (CSSLengthIsValid(length)) {
double pts = CSSLengthToPts(length,WordSectionContentWidthPts(section));
int twips = (int)round(pts*20);
if (twips >= 0) {
DFFormatAttribute(children[WORD_IND],WORD_FIRSTLINE,"%d",twips);
DFRemoveAttribute(children[WORD_IND],WORD_HANGING);
}
else {
DFFormatAttribute(children[WORD_IND],WORD_HANGING,"%d",-twips);
DFRemoveAttribute(children[WORD_IND],WORD_FIRSTLINE);
}
}
}
else {
DFRemoveAttribute(children[WORD_IND],WORD_FIRSTLINE);
DFRemoveAttribute(children[WORD_IND],WORD_HANGING);
}
}
}
}
if (!DFStringEquals(CSSGet(oldp,"margin-top"),CSSGet(newp,"margin-top")) ||
!DFStringEquals(CSSGet(oldp,"margin-bottom"),CSSGet(newp,"margin-bottom")) ||
!DFStringEquals(CSSGet(oldp,"line-height"),CSSGet(newp,"line-height"))) {
if ((CSSGet(newp,"margin-top") == NULL) &&
(CSSGet(newp,"margin-bottom") == NULL) &&
(CSSGet(newp,"line-height") == NULL)) {
children[WORD_SPACING] = NULL;
}
else {
children[WORD_SPACING] = DFCreateElement(pPr->doc,WORD_SPACING);
if (DFStringEquals(CSSGet(newp,"margin-top"),"-word-auto")) {
DFSetAttribute(children[WORD_SPACING],WORD_BEFORE,"100");
DFSetAttribute(children[WORD_SPACING],WORD_BEFOREAUTOSPACING,"1");
}
else {
char *before = twipsFromCSS(CSSGet(newp,"margin-top"),WordSectionContentWidth(section));
DFSetAttribute(children[WORD_SPACING],WORD_BEFORE,before);
DFSetAttribute(children[WORD_SPACING],WORD_BEFOREAUTOSPACING,NULL);
free(before);
}
if (DFStringEquals(CSSGet(newp,"margin-bottom"),"-word-auto")) {
DFSetAttribute(children[WORD_SPACING],WORD_AFTER,"100");
DFSetAttribute(children[WORD_SPACING],WORD_AFTERAUTOSPACING,"1");
}
else {
char *after = twipsFromCSS(CSSGet(newp,"margin-bottom"),WordSectionContentWidth(section));
DFSetAttribute(children[WORD_SPACING],WORD_AFTER,after);
DFSetAttribute(children[WORD_SPACING],WORD_AFTERAUTOSPACING,NULL);
free(after);
}
CSSLength lineHeight = CSSLengthFromString(CSSGet(newp,"line-height"));
if (CSSLengthIsValid(lineHeight) && (lineHeight.units == UnitsPct)) {
int value = (int)round(lineHeight.value*2.4);
DFFormatAttribute(children[WORD_SPACING],WORD_LINE,"%d",value);
}
if (children[WORD_SPACING]->attrsCount == 0)
children[WORD_SPACING] = NULL;
}
}
replaceChildrenFromArray(pPr,children,WordPPR_Children);
}
CSSPropertiesRelease(oldp);
}
