static int parsePackage(TextPackage *package, const char *string, const char *path, DFError **error)
{
DFBuffer *replaced = DFBufferNew();
if (!strcmp(path,""))
path = ".";
if (!processIncludes(package,string,replaced,path,error)) {
DFBufferRelease(replaced);
return 0;
}
char *currentKey = strdup("");
DFBuffer *currentValue = DFBufferNew();
const char **lines = DFStringSplit(replaced->data,"\n",0);
for (int lineno = 0; lines[lineno]; lineno++) {
const char *line = lines[lineno];
if (!DFStringHasPrefix(line,"#")) {
DFBufferFormat(currentValue,"%s\n",line);
}
else if (DFStringHasPrefix(line,"#item ")) {
package->keys = (char **)realloc(package->keys,(package->nkeys+2)*sizeof(char *));
package->keys[package->nkeys++] = strdup(currentKey);
package->keys[package->nkeys] = NULL;
DFHashTableAdd(package->items,currentKey,currentValue->data);
free(currentKey);
DFBufferRelease(currentValue);
currentKey = DFSubstring(line,6,strlen(line));
currentValue = DFBufferNew();
}
else if (DFStringHasPrefix(line,"##")) {
DFBufferFormat(currentValue,"%s\n",&line[1]);
}
else {
DFErrorFormat(error,"Unknown command: %s on line %d",line,(lineno+1));
return 0;
}
}
package->keys = (char **)realloc(package->keys,(package->nkeys+2)*sizeof(char *));
package->keys[package->nkeys++] = strdup(currentKey);
package->keys[package->nkeys] = NULL;
DFHashTableAdd(package->items,currentKey,currentValue->data);
free(lines);
free(currentKey);
DFBufferRelease(currentValue);
DFBufferRelease(replaced);
return 1;
}
static int processIncludes(TextPackage *package, const char *input, DFBuffer *output, const char *path, DFError **error)
{
int ok = 1;
const char **lines = DFStringSplit(input,"\n",0);
for (int lineno = 0; lines[lineno] && ok; lineno++) {
const char *line = lines[lineno];
if (DFStringHasPrefix(line,"#include \"") && DFStringHasSuffix(line,"\"")) {
char *inclRelPath = DFSubstring(line,10,strlen(line)-1);
char *inclAbsPath = DFAppendPathComponent(path,inclRelPath);
char *inclDirName = DFPathDirName(inclAbsPath);
char *inclContent = DFStringReadFromFile(inclAbsPath,error);
if (inclContent == NULL) {
DFErrorFormat(error,"%s: %s",inclRelPath,DFErrorMessage(error));
ok = 0;
}
else if (!processIncludes(package,inclContent,output,inclDirName,error)) {
ok = 0;
}
free(inclRelPath);
free(inclAbsPath);
free(inclDirName);
free(inclContent);
}
else {
DFBufferFormat(output,"%s\n",line);
}
}
free(lines);
return ok;
}
DFNode *WordConverterGetConcrete(WordPutData *put, DFNode *abstract)
{
// Is the abstract node an element, and does it have an id that matches the prefix used for
// conversion? That is, does it look like it has a corresponding node in the concrete document?
if ((abstract == NULL) || (abstract->tag < MIN_ELEMENT_TAG))
return NULL;
const char *idStr = DFGetAttribute(abstract,HTML_ID);
if ((idStr == NULL) || !DFStringHasPrefix(idStr,put->conv->idPrefix))
return NULL;
// Determine the node sequence number and the document based on the id attribute.
// The format of the attribute is <prefix><seqno>(-<docname>)?, where
//
// <prefix> is the BDT prefix we use to identify nodes that match the original document
// <seqno> is an integer uniquely identifying a node in a given document
// <docname> is the name of the document, either footnotes or endnotes. If absent, it is
// the main content document (that is, document.xml)
//
// Note that the sequence number only makes sense within the context of a specific document. It
// is possible to have two different nodes in different documents that have the same sequence number.
// It is for this reason that the id string identifies both the node and the document.
size_t idLen = strlen(idStr);
size_t prefixLen = strlen(put->conv->idPrefix);
unsigned int seqNo = 0;
size_t pos = prefixLen;
while ((pos < idLen) && (idStr[pos] >= '0') && (idStr[pos] <= '9'))
seqNo = seqNo*10 + (idStr[pos++] - '0');
const char *docName = NULL;
if ((pos < idLen) && (idStr[pos] == '-')) {
pos++;
docName = &idStr[pos];
}
DFDocument *doc = NULL;
if (docName == NULL)
doc = put->conv->package->document;
else if (!strcmp(docName,"footnotes"))
doc = put->conv->package->footnotes;
else if (!strcmp(docName,"endnotes"))
doc = put->conv->package->endnotes;
else
return NULL;
// Check to see if we have a node in the concrete document matching that sequence number
DFNode *node = DFNodeForSeqNo(doc,seqNo);
// Only return the node if it's actually an element
if ((node == NULL) || (node->tag < MIN_ELEMENT_TAG))
return NULL;
return node;
}
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);
}
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;
}
