Skip to content

Commit

Permalink
EPUB: fallback to look for a cover in the first fragment
Browse files Browse the repository at this point in the history 
If no EPUB3 or EPUB2 cover found, look at the first
XHTML fragment from the spine: if it contains a single
image and no text, assume that image can be used as
a cover.
Also relax these conditions if the image has alt"=cover":
it must just be before any text and other images.
  • Loading branch information
@poire-z
poire-z committed Mar 20, 2024
1 parent 738e7eb commit 05b3c3d
Show file tree
Hide file tree
Showing 4 changed files with 201 additions and 8 deletions.
4 changes: 3 additions & 1 deletion crengine/include/epubfmt.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,9 @@
#include "../include/lvtinydom.h"

bool DetectEpubFormat( LVStreamRef stream );
bool ImportEpubDocument( LVStreamRef stream, ldomDocument * doc, LVDocViewCallback * progressCallback, CacheLoadingCallback * formatCallback, bool metadataOnly = false );
bool ImportEpubDocument( LVStreamRef stream, ldomDocument * doc, LVDocViewCallback * progressCallback,
CacheLoadingCallback * formatCallback, bool metadataOnly = false,
const elem_def_t * node_scheme=NULL, const attr_def_t * attr_scheme=NULL, const ns_def_t * ns_scheme=NULL);
lString32 EpubGetRootFilePath( LVContainerRef m_arc );
LVStreamRef GetEpubCoverpage(LVContainerRef arc);

Expand Down
199 changes: 194 additions & 5 deletions crengine/src/epubfmt.cpp
View file
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
#include "../include/epubfmt.h"
#include "../include/fb2def.h"

class EpubItem {
public:
Expand Down Expand Up @@ -336,6 +337,125 @@ void ReadEpubAdobePageMap( ldomDocument * doc, ldomNode * mapRoot, LVPageMap * p
}
}

bool ExtractCoverFilenameFromCoverPageFragment( LVStreamRef stream, lString32 & cover_image_href,
const elem_def_t * node_scheme, const attr_def_t * attr_scheme, const ns_def_t * ns_scheme)
{
// We expect such a cover xhtml file to be small in size
if ( stream.isNull() || stream->GetSize() > 5000 )
return false;
// We needed to get the *_scheme passed from lvdocview.cpp (they are only available there)
// to this ImportEpubDocument() just for this: this HTML parser expects these hardcoded
// elements id to properly handle auto-open-close tags (not passing them would result
// in our parsed HTML being: <RootNode><body><html><body><html><body><div> ...).
ldomDocument * coverDoc = LVParseHTMLStream( stream, node_scheme, attr_scheme, ns_scheme);
if ( !coverDoc )
return false;
// We expect to find a single image, and no text, to consider the image as a possible cover.
lString32 img_href;
bool has_more_images = false;
bool has_text = false;
bool in_body = false;
bool img_has_alt_attribute_equal_cover = false;
ldomNode * coverDocRoot = coverDoc->getRootNode();
/* For debugging:
lString32Collection cssFiles;
lString8 extra;
lString8 html = ldomXPointer(coverDocRoot,0).getHtml(cssFiles, extra, 0);
printf("HTML: %s\n", html.c_str());
*/
// Use our usual non-recursive node walker
ldomNode * n = coverDocRoot;
if (n->isElement() && n->getChildCount() > 0) {
int nextChildIndex = 0;
n = n->getChildNode(nextChildIndex);
while (true) {
// Check only the first time we met a node (nextChildIndex == 0)
// and not when we get back to it from a child to process next sibling
if (nextChildIndex == 0) {
if ( n->isElement() ) {
lUInt16 id = n->getNodeId();
if ( !in_body ) {
// We should ignore text outside <body>, ie. <title> or <style>
if (id == el_body ) {
in_body = true;
}
}
else if (id == el_img ) { // <img src=''>
if ( img_href.empty() ) {
img_href = n->getAttributeValue(attr_src);
lString32 img_alt = n->getAttributeValue(attr_alt);
if ( img_alt.lowercase() == U"cover" ) {
img_has_alt_attribute_equal_cover = true;
}
}
else {
// Too many images for a single cover: give up
has_more_images = true;
break;
}
}
else if (id == el_image ) { // <svg>...<image href=''>
if ( img_href.empty() ) {
img_href = n->getAttributeValue(attr_href);
}
else {
has_more_images = true;
break;
}
}
}
else if ( in_body && n->isText() ) {
lUInt16 id = n->getParentNode()->getNodeId();
if (id == el_stylesheet || id == el_script ) {
// We don't have styles here to know if this element is display:none, but
// skip commmon known ones that should be invisible and may contain text.
}
else {
// This won't ignore \n and &nbsp;, we'll see if we need to do more
lString32 text = n->getText().trim();
if ( text.length() > 0 ) {
// printf("TEXT: #%s#\n", LCSTR(text));
has_text = true;
break;
}
}
}
}
// Process next child
if (n->isElement() && nextChildIndex < n->getChildCount()) {
n = n->getChildNode(nextChildIndex);
nextChildIndex = 0;
continue;
}
// No more child, get back to parent and have it process our sibling
nextChildIndex = n->getNodeIndex() + 1;
n = n->getParentNode();
if (!n) // back to root node
break;
if (n == coverDocRoot && nextChildIndex >= n->getChildCount())
// back to this node, and done with its children
break;
}
}
delete coverDoc;
// printf("EPUB coverpage check: has_text:%d, has_more_images:%d, has_alt=cover:%d img_href:%d (%s)\n",
// has_text, has_more_images, img_has_alt_attribute_equal_cover, !img_href.empty(), LCSTR(img_href));
if ( !img_href.empty() ) {
if ( !has_text && !has_more_images ) {
// Single image with no text in this fragment: it must be a cover image.
cover_image_href = img_href;
return true;
}
if ( img_has_alt_attribute_equal_cover ) {
// If we have met an image with alt="cover" before any text or other images,
// consider it valid.
cover_image_href = img_href;
return true;
}
}
return false;
}

lString32 EpubGetRootFilePath(LVContainerRef m_arc)
{
// check root media type
Expand Down Expand Up @@ -1083,7 +1203,9 @@ class EmbeddedFontStyleParser {
}
};

bool ImportEpubDocument( LVStreamRef stream, ldomDocument * m_doc, LVDocViewCallback * progressCallback, CacheLoadingCallback * formatCallback, bool metadataOnly )
bool ImportEpubDocument( LVStreamRef stream, ldomDocument * m_doc, LVDocViewCallback * progressCallback,
CacheLoadingCallback * formatCallback, bool metadataOnly,
const elem_def_t * node_scheme, const attr_def_t * attr_scheme, const ns_def_t * ns_scheme )
{
LVContainerRef arc = LVOpenArchieve( stream );
if ( arc.isNull() )
Expand Down Expand Up @@ -1282,6 +1404,9 @@ bool ImportEpubDocument( LVStreamRef stream, ldomDocument * m_doc, LVDocViewCall
if (coverId.empty() && name == "cover") {
lString32 content = item->getAttributeValue("content");
coverId = content;
// Note: this is expected to be an id, found in the manifest.
// Some bad EPUBs put there the path to the cover image file, which we then
// won't find in the manifest.
continue;
}
// Has to come before calibre:series_index
Expand Down Expand Up @@ -1380,12 +1505,38 @@ bool ImportEpubDocument( LVStreamRef stream, ldomDocument * m_doc, LVDocViewCall
}
}

// We'll be reusing these later
ldomNode * spine = doc->nodeFromXPath( cs32("package/spine") );
lUInt16 itemref_id = doc->getElementNameIndex(U"itemref");

// If no cover specified among the metadata by the publisher, or if it is
// not found, there may still be one used (and usable by us) in the first
// XHTML fragment, that we may want to look at.
lString32 cover_xhtml_id;
if ( spine && spine->getChildCount() > 1 ) {
// We need at least 2 xhtml fragments to maybe have the 1st hold a cover
int nb_itemrefs = spine->getChildCount();
for (size_t i=0; i<nb_itemrefs; i++) {
ldomNode * item = spine->getChildNode(i);
if ( item->getNodeId() != itemref_id )
continue;
cover_xhtml_id = item->getAttributeValue("idref");
// We'll be looking for it later
break;
}
}

// If no cover to look for, we're done looking for metadata.
// We still prefer finding an ePub3 cover (it may happen that even in an ePub3,
// a ePub2 cover is specified, possibly some remnant metadata, and may not exist).
// We'll have to find out that while iterating the manifest, and exit when found out.
bool look_for_coverid = !coverId.empty();
bool look_for_epub3_cover = isEpub3;
// We'll also have to look for the href of the coverxhtml_id: if none of the previous
// covers is found, we'll look inside the xhtml for a cover image.
bool look_for_coverxhtml_id = !cover_xhtml_id.empty();
lString32 cover_xhtml_href;
bool look_at_cover_xhtml = true;

if ( progressCallback )
progressCallback->OnLoadFileProgress(2);
Expand All @@ -1395,7 +1546,7 @@ bool ImportEpubDocument( LVStreamRef stream, ldomDocument * m_doc, LVDocViewCall
// Iterate all package/manifest/item
lUInt16 item_id = doc->getElementNameIndex(U"item");
for (size_t i=0; i<nb_manifest_items; i++) {
if ( metadataOnly && !look_for_coverid && !look_for_epub3_cover ) {
if ( metadataOnly && !look_for_coverid && !look_for_epub3_cover && !look_for_coverxhtml_id) {
// No more stuff to find, stop iterating
break;
}
Expand All @@ -1417,8 +1568,10 @@ bool ImportEpubDocument( LVStreamRef stream, ldomDocument * m_doc, LVDocViewCall
LVImageSourceRef img = LVCreateStreamImageSource(stream);
if ( !img.isNull() ) {
m_doc_props->setString(DOC_PROP_COVER_FILE, coverFileName);
// We found the epub3 cover, and it is a valid image: stop looking for any epub2 one
// We found the epub3 cover, and it is a valid image: stop looking for any other
look_for_coverid = false;
look_for_coverxhtml_id = false;
look_at_cover_xhtml = false;
}
}
// We found the cover-image item: valid image or not, stop looking for it
Expand All @@ -1434,11 +1587,20 @@ bool ImportEpubDocument( LVStreamRef stream, ldomDocument * m_doc, LVDocViewCall
if ( !img.isNull() ) {
CRLog::info("EPUB coverpage image is correct: %d x %d", img->GetWidth(), img->GetHeight() );
m_doc_props->setString(DOC_PROP_COVER_FILE, coverFileName);
// We found the epub2 cover, and it is a valid image: stop looking for the coverxhtml one.
look_for_coverxhtml_id = false;
look_at_cover_xhtml = false;
}
}
// We found the coverid: valid image or not, stop looking for it
look_for_coverid = false;
}
if ( look_for_coverxhtml_id && id==cover_xhtml_id ) {
// At this point, we just store the href. We will parse it if we end up
// not having found any of the epub3 cover or coverId.
cover_xhtml_href = href;
look_for_coverxhtml_id = false;
}
if (metadataOnly) {
continue;
}
Expand Down Expand Up @@ -1480,14 +1642,42 @@ bool ImportEpubDocument( LVStreamRef stream, ldomDocument * m_doc, LVDocViewCall
}
CRLog::debug("opf: reading items done.");

if ( look_at_cover_xhtml && !cover_xhtml_href.empty() ) {
// No other cover found: look inside the cover xhtml fragment
lString32 cover_xhtml_path = LVCombinePaths(codeBase, cover_xhtml_href);
CRLog::info("EPUB cover xhtml page file: %s", LCSTR(cover_xhtml_path));
LVStreamRef stream = m_arc->OpenStream(cover_xhtml_path.c_str(), LVOM_READ);
lString32 cover_image_href;
if ( ExtractCoverFilenameFromCoverPageFragment(stream, cover_image_href, node_scheme, attr_scheme, ns_scheme) ) {
lString32 codeBase = LVExtractPath( cover_xhtml_path );
if ( codeBase.length()>0 && codeBase.lastChar()!='/' )
codeBase.append(1, U'/');
lString32 cover_image_path = LVCombinePaths(codeBase, cover_image_href);
CRLog::info("EPUB cover image file: %s", LCSTR(cover_image_path));
LVStreamRef stream = m_arc->OpenStream(cover_image_path.c_str(), LVOM_READ);
if ( stream.isNull() ) {
// Try again in case cover_image_path is percent-encoded
cover_image_path = LVCombinePaths(codeBase, DecodeHTMLUrlString(cover_image_href));
CRLog::info("EPUB cover image file pct-decoded: %s", LCSTR(cover_image_path));
stream = m_arc->OpenStream(cover_image_path.c_str(), LVOM_READ);
}
if ( !stream.isNull() ) {
LVImageSourceRef img = LVCreateStreamImageSource(stream);
if ( !img.isNull() ) {
CRLog::info("EPUB coverpage image is correct: %d x %d", img->GetWidth(), img->GetHeight() );
m_doc_props->setString(DOC_PROP_COVER_FILE, cover_image_path);
}
}
}
}

if ( progressCallback )
progressCallback->OnLoadFileProgress(4);

// Gather EpubItems from <package><spine>, which specify which EpubItems,
// and in which order, make out the book content
if ( !metadataOnly && epubItems.length()>0 ) {
CRLog::debug("opf: reading spine");
ldomNode * spine = doc->nodeFromXPath( cs32("package/spine") );
if ( spine ) {
// Some attributes specify that some EpubItems have a specific purpose
// <spine toc="ncx" page-map="page-map">
Expand All @@ -1498,7 +1688,6 @@ bool ImportEpubDocument( LVStreamRef stream, ldomDocument * m_doc, LVDocViewCall
if ( page_map!=NULL )
pageMapHref = LVCombinePaths(codeBase, page_map->href);
// Iterate all package/spine/itemref (each will make a book fragment)
lUInt16 itemref_id = doc->getElementNameIndex(U"itemref");
int nb_itemrefs = spine->getChildCount();
for (size_t i=0; i<nb_itemrefs; i++) {
ldomNode * item = spine->getChildNode(i);
Expand Down
4 changes: 3 additions & 1 deletion crengine/src/lvdocview.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -4439,7 +4439,9 @@ bool LVDocView::LoadDocument(LVStreamRef stream, bool metadataOnly) {
if ( m_callback )
m_callback->OnLoadFileFormatDetected(doc_format_epub);
updateDocStyleSheet();
bool res = ImportEpubDocument( m_stream, m_doc, m_callback, this, metadataOnly );
// See epubfmt.cpp's ExtractCoverFilenameFromCoverPageFragment()
// for why we need to pass fb2_elem_table and such.
bool res = ImportEpubDocument( m_stream, m_doc, m_callback, this, metadataOnly, fb2_elem_table, fb2_attr_table, fb2_ns_table );
if ( !res ) {
setDocFormat( doc_format_none );
createDefaultDocument( cs32("ERROR: Error reading EPUB format"), cs32("Cannot open document") );
Expand Down
2 changes: 1 addition & 1 deletion crengine/src/lvtinydom.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -93,7 +93,7 @@ extern const int gDOMVersionCurrent = DOM_VERSION_CURRENT;

/// change in case of incompatible changes in swap/cache file format to avoid using incompatible swap file
// increment to force complete reload/reparsing of old file
#define CACHE_FILE_FORMAT_VERSION "3.05.73k"
#define CACHE_FILE_FORMAT_VERSION "3.05.74k"
/// increment following value to force re-formatting of old book after load
#define FORMATTING_VERSION_ID 0x0032

Expand Down

0 comments on commit 05b3c3d

Please sign in to comment.