Cutlery/calibre
1
0
Fork 0
mirror of https://github.com/kovidgoyal/calibre.git synced 2025-07-09 03:04:10 -04:00
Code Issues Packages Projects Releases Wiki Activity

Release GIL while running append

Browse Source
This commit is contained in:
Kovid Goyal 2023-06-17 11:52:50 +05:30
parent 7469c920f2
commit 2acd1ee518
No known key found for this signature in database
GPG Key ID: 06BC317B515ACE7C
1 changed files with 56 additions and 41 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

97
src/calibre/utils/podofo/doc.cpp
View File

@ -333,29 +333,42 @@ PDFDoc_copy_page(PDFDoc *self, PyObject *args) {
// append() {{{ // append() {{{
static void
fix_references(PdfObject &parent, const std::unordered_map<PdfReference, PdfObject*> &ref_map) {
switch(parent.GetDataType()) {
case PdfDataType::Dictionary:
for (auto& pair : parent.GetDictionary()) {
fix_references(pair.second, ref_map);
}
break;
case PdfDataType::Array:
for (auto& child : parent.GetArray()) fix_references(child, ref_map);
break;
case PdfDataType::Reference:
if (auto search = ref_map.find(parent.GetReference()); search != ref_map.end()) {
parent.SetReference(search->second->GetIndirectReference());
}
break;
default:
break;
}
}
static PyObject * static PyObject *
PDFDoc_append(PDFDoc *self, PyObject *args) { PDFDoc_append(PDFDoc *self, PyObject *args) {
class AppendPagesData {
public:
const PdfPage *src_page;
PdfPage *dest_page;
PdfReference dest_page_parent;
AppendPagesData(const PdfPage &src, PdfPage &dest) {
src_page = &src;
dest_page = &dest;
dest_page_parent = dest.GetDictionary().GetKeyAs<PdfReference>("Parent");
}
};
class MapReferences : public std::unordered_map<PdfReference, PdfObject*> {
public:
void apply(PdfObject &parent) {
switch(parent.GetDataType()) {
case PdfDataType::Dictionary:
for (auto& pair : parent.GetDictionary()) {
apply(pair.second );
}
break;
case PdfDataType::Array:
for (auto& child : parent.GetArray()) apply(child);
break;
case PdfDataType::Reference:
if (auto search = find(parent.GetReference()); search != end()) {
parent.SetReference(search->second->GetIndirectReference());
}
break;
default:
break;
}
}
};
static const PdfName inheritableAttributes[] = { static const PdfName inheritableAttributes[] = {
PdfName("Resources"), PdfName("Resources"),
PdfName("MediaBox"), PdfName("MediaBox"),
@ -364,25 +377,25 @@ PDFDoc_append(PDFDoc *self, PyObject *args) {
PdfName::KeyNull PdfName::KeyNull
}; };
PdfMemDocument *dest = self->doc; PdfMemDocument *dest = self->doc;
std::vector<const PdfMemDocument*> docs(PyTuple_GET_SIZE(args));
for (Py_ssize_t i = 0; i < PyTuple_GET_SIZE(args); i++) {
PyObject *doc = PyTuple_GET_ITEM(args, i);
int typ = PyObject_IsInstance(doc, (PyObject*)&PDFDocType);
if (typ == -1) return NULL;
if (typ == 0) { PyErr_SetString(PyExc_TypeError, "You must pass a PDFDoc instance to this method"); return NULL; }
docs[i] = ((PDFDoc*)doc)->doc;
}
PyThreadState *_save; _save = PyEval_SaveThread();
try { try {
std::vector<const PdfMemDocument*> docs(PyTuple_GET_SIZE(args));
for (Py_ssize_t i = 0; i < PyTuple_GET_SIZE(args); i++) {
PyObject *doc = PyTuple_GET_ITEM(args, i);
int typ = PyObject_IsInstance(doc, (PyObject*)&PDFDocType);
if (typ == -1) return NULL;
if (typ == 0) { PyErr_SetString(PyExc_TypeError, "You must pass a PDFDoc instance to this method"); return NULL; }
docs[i] = ((PDFDoc*)doc)->doc;
}
for (auto src : docs) { for (auto src : docs) {
std::unordered_map<PdfReference, PdfObject*> ref_map; MapReferences ref_map;
std::unordered_map<PdfReference, PdfReference> page_parent_map; std::vector<AppendPagesData> pages;
const unsigned initial_page_count = dest->GetPages().GetCount();
// append pages first // append pages first
for (unsigned i = 0; i < src->GetPages().GetCount(); i++) { for (unsigned i = 0; i < src->GetPages().GetCount(); i++) {
const auto& src_page = src->GetPages().GetPageAt(i); const auto& src_page = src->GetPages().GetPageAt(i);
auto& dest_page = dest->GetPages().CreatePage(src_page.GetRect()); auto& dest_page = dest->GetPages().CreatePage(src_page.GetRect());
page_parent_map[dest_page.GetObject().GetIndirectReference()] = dest_page.GetDictionary().GetKeyAs<PdfReference>("Parent"); pages.emplace_back(src_page, dest_page);
dest_page.GetObject() = src_page.GetObject(); dest_page.GetObject() = src_page.GetObject();
dest_page.GetDictionary().RemoveKey("Resource"); dest_page.GetDictionary().RemoveKey("Resource");
dest_page.GetDictionary().RemoveKey("Parent"); dest_page.GetDictionary().RemoveKey("Parent");
@ -396,13 +409,12 @@ PDFDoc_append(PDFDoc *self, PyObject *args) {
} }
} }
// fix references in appended objects // fix references in appended objects
for (auto& elem : ref_map) fix_references(*elem.second, ref_map); for (auto& elem : ref_map) ref_map.apply(*elem.second);
// fixup all pages // fixup all pages
for (unsigned i = 0; i < src->GetPages().GetCount(); i++) { for (auto& x : pages) {
auto& src_page = src->GetPages().GetPageAt(i); auto& src_page = *x.src_page;
auto& dest_page = dest->GetPages().GetPageAt(initial_page_count + i); auto& dest_page = *x.dest_page;
// Reset the parent to the correct value from the stored mapping dest_page.GetDictionary().AddKey("Parent", x.dest_page_parent);
dest_page.GetDictionary().AddKey("Parent", page_parent_map[dest_page.GetObject().GetIndirectReference()]);
// Set the page contents // Set the page contents
if (auto key = src_page.GetDictionary().GetKeyAs<PdfReference>(PdfName::KeyContents); key.IsIndirect()) { if (auto key = src_page.GetDictionary().GetKeyAs<PdfReference>(PdfName::KeyContents); key.IsIndirect()) {
if (auto search = ref_map.find(key); search != ref_map.end()) { if (auto search = ref_map.find(key); search != ref_map.end()) {
@ -416,7 +428,7 @@ PDFDoc_append(PDFDoc *self, PyObject *args) {
if (src_page.GetResources() != nullptr) { if (src_page.GetResources() != nullptr) {
const auto &src_resources = src_page.GetResources()->GetDictionary(); const auto &src_resources = src_page.GetResources()->GetDictionary();
dest_page.GetOrCreateResources().GetDictionary() = src_resources; dest_page.GetOrCreateResources().GetDictionary() = src_resources;
fix_references(dest_page.GetResources()->GetObject(), ref_map); ref_map.apply(dest_page.GetResources()->GetObject());
} else dest_page.GetOrCreateResources(); } else dest_page.GetOrCreateResources();
// Copy inherited properties // Copy inherited properties
@ -425,7 +437,7 @@ PDFDoc_append(PDFDoc *self, PyObject *args) {
auto attribute = src_page.GetDictionary().FindKeyParent(*inherited); auto attribute = src_page.GetDictionary().FindKeyParent(*inherited);
if (attribute != nullptr) { if (attribute != nullptr) {
PdfObject attributeCopy(*attribute); PdfObject attributeCopy(*attribute);
fix_references(attributeCopy, ref_map); ref_map.apply(attributeCopy);
dest_page.GetDictionary().AddKey(*inherited, attributeCopy); dest_page.GetDictionary().AddKey(*inherited, attributeCopy);
} }
inherited++; inherited++;
@ -433,12 +445,15 @@ PDFDoc_append(PDFDoc *self, PyObject *args) {
} }
} }
} catch (const PdfError & err) { } catch (const PdfError & err) {
PyEval_RestoreThread(_save);
podofo_set_exception(err); podofo_set_exception(err);
return NULL; return NULL;
} catch (std::exception & err) { } catch (std::exception & err) {
PyEval_RestoreThread(_save);
PyErr_Format(PyExc_ValueError, "An error occurred while trying to append pages: %s", err.what()); PyErr_Format(PyExc_ValueError, "An error occurred while trying to append pages: %s", err.what());
return NULL; return NULL;
} }
PyEval_RestoreThread(_save);
Py_RETURN_NONE; Py_RETURN_NONE;
} // }}} } // }}}