File calibre-revert-new-podofo.patch of Package calibre
diff --color -Nur calibre-6.20.0.orig/src/calibre/ebooks/pdf/html_writer.py calibre-6.20.0/src/calibre/ebooks/pdf/html_writer.py
--- calibre-6.20.0.orig/src/calibre/ebooks/pdf/html_writer.py 2023-06-11 10:44:12.621384750 -0700
+++ calibre-6.20.0/src/calibre/ebooks/pdf/html_writer.py 2023-06-11 11:20:32.508984678 -0700
@@ -37,7 +37,7 @@
from calibre.utils.logging import default_log
from calibre.utils.monotonic import monotonic
from calibre.utils.podofo import (
- add_image_page, dedup_type3_fonts, get_podofo, remove_unused_fonts,
+ dedup_type3_fonts, get_podofo, remove_unused_fonts,
set_metadata_implementation,
)
from calibre.utils.resources import get_path as P
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/doc.cpp calibre-6.20.0/src/calibre/utils/podofo/doc.cpp
--- calibre-6.20.0.orig/src/calibre/utils/podofo/doc.cpp 2023-06-08 18:36:58.000000000 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/doc.cpp 2023-06-11 11:20:17.614014516 -0700
@@ -7,9 +7,6 @@
#include "global.h"
#include <iostream>
-#include <algorithm>
-#include <new>
-#include <string_view>
using namespace pdf;
@@ -18,7 +15,6 @@
PDFDoc_dealloc(PDFDoc* self)
{
if (self->doc != NULL) delete self->doc;
- Py_CLEAR(self->load_buffer_ref);
Py_TYPE(self)->tp_free((PyObject*)self);
}
@@ -45,10 +41,11 @@
if (!PyArg_ParseTuple(args, "y#", &buffer, &size)) return NULL;
try {
- self->doc->LoadFromBuffer(bufferview(buffer, size));
- Py_CLEAR(self->load_buffer_ref);
- self->load_buffer_ref = args;
- Py_INCREF(args);
+#if PODOFO_VERSION <= 0x000905
+ self->doc->Load(buffer, (long)size);
+#else
+ self->doc->LoadFromBuffer(buffer, (long)size);
+#endif
} catch(const PdfError & err) {
podofo_set_exception(err);
return NULL;
@@ -87,7 +84,7 @@
if (PyArg_ParseTuple(args, "s", &buffer)) {
try {
- self->doc->Save(buffer, save_options);
+ self->doc->Write(buffer);
} catch(const PdfError & err) {
podofo_set_exception(err);
return NULL;
@@ -97,54 +94,16 @@
Py_RETURN_NONE;
}
-class BytesOutputDevice : public OutputStreamDevice {
- private:
- pyunique_ptr bytes;
- size_t written;
- public:
- BytesOutputDevice() : bytes(), written(0) { SetAccess(DeviceAccess::Write); }
- size_t GetLength() const { return written; }
- size_t GetPosition() const { return written; }
- size_t capacity() const { return bytes ? PyBytes_GET_SIZE(bytes.get()) : 0; }
- bool Eof() const { return false; }
-
- void writeBuffer(const char* src, size_t src_sz) {
- if (written + src_sz > capacity()) {
- PyObject* old = bytes.release();
- static const size_t initial_capacity = 1024 * 1024;
- if (old) {
- if (_PyBytes_Resize(&old, std::max(written + src_sz, 2 * std::max(capacity(), initial_capacity))) != 0) {
- throw std::bad_alloc();
- }
- } else {
- old = PyBytes_FromStringAndSize(NULL, std::max(written + src_sz, initial_capacity));
- if (!old) throw std::bad_alloc();
- }
- bytes.reset(old);
- }
- if (bytes) {
- memcpy(PyBytes_AS_STRING(bytes.get()) + written, src, src_sz);
- written += src_sz;
- }
- }
-
- void Flush() { }
- PyObject* Release() {
- auto ans = bytes.release();
- _PyBytes_Resize(&ans, written);
- written = 0;
- return ans;
- }
-};
-
static PyObject *
PDFDoc_write(PDFDoc *self, PyObject *args) {
PyObject *ans;
- BytesOutputDevice d;
try {
- self->doc->Save(d, save_options);
- return d.Release();
+ PdfRefCountedBuffer buffer(1*1024*1024);
+ PdfOutputDevice out(&buffer);
+ self->doc->Write(&out);
+ ans = PyBytes_FromStringAndSize(buffer.GetBuffer(), out.Tell());
+ if (ans == NULL) PyErr_NoMemory();
} catch(const PdfError &err) {
podofo_set_exception(err);
return NULL;
@@ -165,25 +124,11 @@
static PyObject *
PDFDoc_uncompress_pdf(PDFDoc *self, PyObject *args) {
- try {
- auto& objects = self->doc->GetObjects();
- for (auto obj : objects) {
- auto stream = obj->GetStream();
- if (stream == nullptr) continue;
- try {
- try {
- stream->Unwrap();
- } catch (PdfError& e) {
- if (e.GetCode() != PdfErrorCode::Flate) throw e;
- }
- }
- catch (PdfError& e) {
- if (e.GetCode() != PdfErrorCode::UnsupportedFilter) throw e;
- }
+ for (auto &it : self->doc->GetObjects()) {
+ if(it->HasStream()) {
+ PdfMemStream* stream = dynamic_cast<PdfMemStream*>(it->GetStream());
+ stream->Uncompress();
}
- } catch(const PdfError & err) {
- podofo_set_exception(err);
- return NULL;
}
Py_RETURN_NONE;
}
@@ -195,8 +140,7 @@
static PyObject *
PDFDoc_extract_first_page(PDFDoc *self, PyObject *args) {
try {
- auto pages = &self->doc->GetPages();
- while (pages->GetCount() > 1) pages->RemovePageAt(1);
+ while (self->doc->GetPageCount() > 1) self->doc->GetPagesTree()->DeletePage(1);
} catch(const PdfError & err) {
podofo_set_exception(err);
return NULL;
@@ -210,7 +154,7 @@
PDFDoc_page_count(PDFDoc *self, PyObject *args) {
int count;
try {
- count = self->doc->GetPages().GetCount();
+ count = self->doc->GetPageCount();
} catch(const PdfError & err) {
podofo_set_exception(err);
return NULL;
@@ -229,8 +173,8 @@
if( it->IsDictionary() ) {
obj_type = it->GetDictionary().GetKey( PdfName::KeyType );
obj_sub_type = it->GetDictionary().GetKey( PdfName::KeySubtype );
- if( ( obj_type && obj_type->IsName() && ( obj_type->GetName().GetString() == "XObject" ) ) ||
- ( obj_sub_type && obj_sub_type->IsName() && ( obj_sub_type->GetName().GetString() == "Image" ) ) ) count++;
+ if( ( obj_type && obj_type->IsName() && ( obj_type->GetName().GetName() == "XObject" ) ) ||
+ ( obj_sub_type && obj_sub_type->IsName() && ( obj_sub_type->GetName().GetName() == "Image" ) ) ) count++;
}
}
} catch(const PdfError & err) {
@@ -243,11 +187,10 @@
// delete_page() {{{
static PyObject *
PDFDoc_delete_pages(PDFDoc *self, PyObject *args) {
- unsigned int page, count = 1;
- if (PyArg_ParseTuple(args, "I|I", &page, &count)) {
+ int page = 0, count = 1;
+ if (PyArg_ParseTuple(args, "i|i", &page, &count)) {
try {
- auto &pages = self->doc->GetPages();
- while (count-- > 0) pages.RemovePageAt(page - 1);
+ self->doc->DeletePages(page - 1, count);
} catch(const PdfError & err) {
podofo_set_exception(err);
return NULL;
@@ -264,9 +207,10 @@
const char *which;
if (PyArg_ParseTuple(args, "si", &which, &pagenum)) {
try {
- auto page = get_page(self->doc, pagenum-1);
- if (!page) { PyErr_Format(PyExc_ValueError, "page number %d not found in PDF file", pagenum); return NULL; }
- Rect rect;
+ PdfPagesTree* tree = self->doc->GetPagesTree();
+ PdfPage* page = tree->GetPage(pagenum - 1);
+ if (!page) { PyErr_Format(PyExc_ValueError, "page number %d not found in PDF file", pagenum); return NULL; }
+ PdfRect rect;
if (strcmp(which, "MediaBox") == 0) {
rect = page->GetMediaBox();
} else if (strcmp(which, "CropBox") == 0) {
@@ -281,7 +225,7 @@
PyErr_Format(PyExc_KeyError, "%s is not a known box", which);
return NULL;
}
- return Py_BuildValue("dddd", rect.GetLeft(), rect.GetBottom(), rect.Width, rect.Height);
+ return Py_BuildValue("dddd", rect.GetLeft(), rect.GetBottom(), rect.GetWidth(), rect.GetHeight());
} catch(const PdfError & err) {
podofo_set_exception(err);
return NULL;
@@ -299,12 +243,13 @@
const char *which;
if (PyArg_ParseTuple(args, "sidddd", &which, &pagenum, &left, &bottom, &width, &height)) {
try {
- PdfPage* page = get_page(self->doc, pagenum-1);
- if (!page) { PyErr_Format(PyExc_ValueError, "page number %d not found in PDF file", pagenum); return NULL; }
- Rect rect(left, bottom, width, height);
- PdfArray box;
- rect.ToArray(box);
- page->GetObject().GetDictionary().AddKey(PdfName(which), box);
+ PdfPagesTree* tree = self->doc->GetPagesTree();
+ PdfPage* page = tree->GetPage(pagenum - 1);
+ if (!page) { PyErr_Format(PyExc_ValueError, "page number %d not found in PDF file", pagenum); return NULL; }
+ PdfRect rect(left, bottom, width, height);
+ PdfObject box;
+ rect.ToVariant(box);
+ page->GetObject()->GetDictionary().AddKey(PdfName(which), box);
Py_RETURN_NONE;
} catch(const PdfError & err) {
podofo_set_exception(err);
@@ -321,7 +266,9 @@
int from = 0, to = 0;
if (!PyArg_ParseTuple(args, "ii", &from, &to)) return NULL;
try {
- self->doc->GetPages().InsertDocumentPageAt(to - 1, *self->doc, from - 1);
+ PdfPagesTree* tree = self->doc->GetPagesTree();
+ PdfPage* page = tree->GetPage(from - 1);
+ tree->InsertPage(to - 1, page);
} catch(const PdfError & err) {
podofo_set_exception(err);
return NULL;
@@ -340,14 +287,14 @@
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; }
- PDFDoc *pdfdoc = (PDFDoc*)doc;
try {
- self->doc->GetPages().AppendDocumentPages(*pdfdoc->doc);
+ self->doc->Append(*((PDFDoc*)doc)->doc, true);
} catch (const PdfError & err) {
podofo_set_exception(err);
return NULL;
}
+
Py_RETURN_NONE;
} // }}}
@@ -360,7 +307,7 @@
if (!PyArg_ParseTuple(args, "O!|ii", &PDFDocType, &src_doc, &src_page, &at)) return NULL;
try {
- self->doc->GetPages().InsertDocumentPageAt(at, *src_doc->doc, src_page);
+ self->doc->InsertExistingPageAt(*src_doc->doc, src_page, at);
} catch (const PdfError & err) {
podofo_set_exception(err);
return NULL;
@@ -376,11 +323,12 @@
double left, bottom, width, height;
char *box;
if (!PyArg_ParseTuple(args, "isdddd", &num, &box, &left, &bottom, &width, &height)) return NULL;
+
try {
- Rect r(left, bottom, width, height);
- PdfArray o;
- r.ToArray(o);
- self->doc->GetPages().GetPageAt(num).GetObject().GetDictionary().AddKey(PdfName(box), o);
+ PdfRect r(left, bottom, width, height);
+ PdfObject o;
+ r.ToVariant(o);
+ self->doc->GetPage(num)->GetObject()->GetDictionary().AddKey(PdfName(box), o);
} catch(const PdfError & err) {
podofo_set_exception(err);
return NULL;
@@ -388,119 +336,127 @@
PyErr_SetString(PyExc_ValueError, "An unknown error occurred while trying to set the box");
return NULL;
}
+
Py_RETURN_NONE;
} // }}}
// get_xmp_metadata() {{{
static PyObject *
PDFDoc_get_xmp_metadata(PDFDoc *self, PyObject *args) {
+ PoDoFo::PdfObject *metadata = NULL;
+ PoDoFo::PdfStream *str = NULL;
+ PoDoFo::pdf_long len = 0;
+ char *buf = NULL;
+ PyObject *ans = NULL;
+
try {
- auto obj = self->doc->GetCatalog().GetDictionary().FindKey("Metadata");
- if (obj == nullptr) Py_RETURN_NONE;
- auto stream = obj->GetStream();
- if (stream == nullptr) Py_RETURN_NONE;
- std::string s;
- StringStreamDevice ouput(s);
- stream->CopyTo(ouput);
- return PyBytes_FromStringAndSize(s.data(), s.size());
+ if ((metadata = self->doc->GetMetadata()) != NULL) {
+ if ((str = metadata->GetStream()) != NULL) {
+ str->GetFilteredCopy(&buf, &len);
+ if (buf != NULL) {
+ Py_ssize_t psz = len;
+ ans = Py_BuildValue("y#", buf, psz);
+ free(buf); buf = NULL;
+ if (ans == NULL) goto error;
+ }
+ }
+ }
} catch(const PdfError & err) {
- podofo_set_exception(err); return NULL;
+ podofo_set_exception(err); goto error;
} catch (...) {
- PyErr_SetString(PyExc_ValueError, "An unknown error occurred while trying to read the XML metadata"); return NULL;
+ PyErr_SetString(PyExc_ValueError, "An unknown error occurred while trying to read the XML metadata"); goto error;
}
+
+ if (ans != NULL) return ans;
Py_RETURN_NONE;
+error:
+ return NULL;
} // }}}
-// add_image_page() {{{
-static PyObject *
-PDFDoc_add_image_page(PDFDoc *self, PyObject *args) {
- const char *image_data; Py_ssize_t image_data_sz;
- double page_x, page_y, page_width, page_height;
- double image_x, image_y, image_canvas_width, image_canvas_height;
- unsigned int page_num = 1; int preserve_aspect_ratio = 1;
- if (!PyArg_ParseTuple(args, "y#dddddddd|Ip", &image_data, &image_data_sz, &page_x, &page_y, &page_width, &page_height, &image_x, &image_y, &image_canvas_width, &image_canvas_height, &page_num, &preserve_aspect_ratio)) return NULL;
- auto img = self->doc->CreateImage();
- img->LoadFromBuffer(bufferview(image_data, image_data_sz));
- auto &page = self->doc->GetPages().CreatePageAt(page_num-1, Rect(page_x, page_y, page_width, page_height));
- PdfPainter painter;
- painter.SetCanvas(page);
- auto scaling_x = image_canvas_width, scaling_y = image_canvas_height;
- if (preserve_aspect_ratio) {
- auto page_ar = page_width / page_height, img_ar = img->GetRect().Width / img->GetRect().Height;
- if (page_ar > img_ar) {
- scaling_x = img_ar * image_canvas_height;
- image_x = (image_canvas_width - scaling_x) / 2.;
- } else if (page_ar < img_ar) {
- scaling_y = image_canvas_width / img_ar;
- image_y = (image_canvas_height - scaling_y) / 2.;
- }
- }
- painter.DrawImage(*img, image_x, image_y, scaling_x / img->GetRect().Width, scaling_y / img->GetRect().Height);
- painter.FinishDrawing();
- return Py_BuildValue("dd", img->GetRect().Width, img->GetRect().Height);
-}
-// }}}
-
// set_xmp_metadata() {{{
static PyObject *
PDFDoc_set_xmp_metadata(PDFDoc *self, PyObject *args) {
const char *raw = NULL;
Py_ssize_t len = 0;
+ PoDoFo::PdfObject *metadata = NULL, *catalog = NULL;
+ PoDoFo::PdfStream *str = NULL;
+ TVecFilters compressed(1);
+ compressed[0] = ePdfFilter_FlateDecode;
+
if (!PyArg_ParseTuple(args, "y#", &raw, &len)) return NULL;
try {
- auto& metadata = self->doc->GetCatalog().GetOrCreateMetadataObject();
- auto& stream = metadata.GetOrCreateStream();
- stream.SetData(std::string_view(raw, len), true);
- metadata.GetDictionary().RemoveKey(PdfName::KeyFilter);
+ if ((metadata = self->doc->GetMetadata()) != NULL) {
+ if ((str = metadata->GetStream()) == NULL) { PyErr_NoMemory(); goto error; }
+ str->Set(raw, len, compressed);
+ } else {
+ if ((catalog = self->doc->GetCatalog()) == NULL) { PyErr_SetString(PyExc_ValueError, "Cannot set XML metadata as this document has no catalog"); goto error; }
+ if ((metadata = self->doc->GetObjects().CreateObject("Metadata")) == NULL) { PyErr_NoMemory(); goto error; }
+ if ((str = metadata->GetStream()) == NULL) { PyErr_NoMemory(); goto error; }
+ metadata->GetDictionary().AddKey(PoDoFo::PdfName("Subtype"), PoDoFo::PdfName("XML"));
+ str->Set(raw, len, compressed);
+ catalog->GetDictionary().AddKey(PoDoFo::PdfName("Metadata"), metadata->Reference());
+ }
} catch(const PdfError & err) {
- podofo_set_exception(err); return NULL;
+ podofo_set_exception(err); goto error;
} catch (...) {
- PyErr_SetString(PyExc_ValueError, "An unknown error occurred while trying to set the XML metadata"); return NULL;
+ PyErr_SetString(PyExc_ValueError, "An unknown error occurred while trying to set the XML metadata");
+ goto error;
}
Py_RETURN_NONE;
+error:
+ return NULL;
+
} // }}}
// extract_anchors() {{{
static PyObject *
PDFDoc_extract_anchors(PDFDoc *self, PyObject *args) {
+ const PdfObject* catalog = NULL;
PyObject *ans = PyDict_New();
if (ans == NULL) return NULL;
try {
- const PdfObject *dests_ref = self->doc->GetCatalog().GetDictionary().GetKey("Dests");
- auto& pages = self->doc->GetPages();
- if (dests_ref && dests_ref->IsReference()) {
- const PdfObject *dests_obj = self->doc->GetObjects().GetObject(object_as_reference(dests_ref));
- if (dests_obj && dests_obj->IsDictionary()) {
- const PdfDictionary &dests = dests_obj->GetDictionary();
- for (auto itres: dests) {
- if (itres.second.IsArray()) {
- const PdfArray &dest = itres.second.GetArray();
- // see section 8.2 of PDF spec for different types of destination arrays
- // but chromium apparently generates only [page /XYZ left top zoom] type arrays
- if (dest.GetSize() > 4 && dest[1].IsName() && dest[1].GetName().GetString() == "XYZ") {
- const PdfPage *page = get_page(pages, object_as_reference(dest[0]));
- if (page) {
- unsigned int pagenum = page->GetPageNumber();
- double left = dest[2].GetReal(), top = dest[3].GetReal();
- long long zoom = dest[4].GetNumber();
- const std::string &anchor = itres.first.GetString();
- PyObject *key = PyUnicode_DecodeUTF8(anchor.c_str(), anchor.length(), "replace");
- PyObject *tuple = Py_BuildValue("IddL", pagenum, left, top, zoom);
- if (!tuple || !key) { break; }
- int ret = PyDict_SetItem(ans, key, tuple);
- Py_DECREF(key); Py_DECREF(tuple);
- if (ret != 0) break;
- }
- }
- }
- }
- }
- }
+ if ((catalog = self->doc->GetCatalog()) != NULL) {
+ const PdfObject *dests_ref = catalog->GetDictionary().GetKey("Dests");
+ PdfPagesTree *tree = self->doc->GetPagesTree();
+ if (dests_ref && dests_ref->IsReference()) {
+ const PdfObject *dests_obj = self->doc->GetObjects().GetObject(dests_ref->GetReference());
+ if (dests_obj && dests_obj->IsDictionary()) {
+ const PdfDictionary &dests = dests_obj->GetDictionary();
+ const TKeyMap &keys = dests.GetKeys();
+ for (TCIKeyMap itres = keys.begin(); itres != keys.end(); ++itres) {
+ if (itres->second->IsArray()) {
+ const PdfArray &dest = itres->second->GetArray();
+ // see section 8.2 of PDF spec for different types of destination arrays
+ // but chromium apparently generates only [page /XYZ left top zoom] type arrays
+ if (dest.GetSize() > 4 && dest[1].IsName() && dest[1].GetName().GetName() == "XYZ") {
+ const PdfPage *page = tree->GetPage(dest[0].GetReference());
+ if (page) {
+ unsigned int pagenum = page->GetPageNumber();
+ double left = dest[2].GetReal(), top = dest[3].GetReal();
+ long long zoom = dest[4].GetNumber();
+ const std::string &anchor = itres->first.GetName();
+ PyObject *key = PyUnicode_DecodeUTF8(anchor.c_str(), anchor.length(), "replace");
+ PyObject *tuple = Py_BuildValue("IddL", pagenum, left, top, zoom);
+ if (!tuple || !key) { break; }
+ int ret = PyDict_SetItem(ans, key, tuple);
+ Py_DECREF(key); Py_DECREF(tuple);
+ if (ret != 0) break;
+ }
+ }
+ }
+ }
+ }
+ }
+ }
} catch(const PdfError & err) {
podofo_set_exception(err);
+ Py_CLEAR(ans);
+ return NULL;
} catch (...) {
PyErr_SetString(PyExc_ValueError, "An unknown error occurred while trying to set the box");
+ Py_CLEAR(ans);
+ return NULL;
}
if (PyErr_Occurred()) { Py_CLEAR(ans); return NULL; }
return ans;
@@ -516,22 +472,28 @@
}
PdfDictionary &A = link.GetKey("A")->GetDictionary();
PdfObject *uo = A.GetKey("URI");
- const std::string &uri = uo->GetString().GetString();
+ const std::string &uri = uo->GetString().GetStringUtf8();
pyunique_ptr ret(PyObject_CallObject(alter_callback, Py_BuildValue("(N)", PyUnicode_DecodeUTF8(uri.c_str(), uri.length(), "replace"))));
if (!ret) { return; }
if (PyTuple_Check(ret.get()) && PyTuple_GET_SIZE(ret.get()) == 4) {
int pagenum; double left, top, zoom;
if (PyArg_ParseTuple(ret.get(), "iddd", &pagenum, &left, &top, &zoom)) {
- const PdfPage *page = get_page(self->doc, pagenum - 1);
- if (page == NULL) {
- PyErr_Format(PyExc_ValueError, "No page number %d in the PDF file of %d pages", pagenum, self->doc->GetPages().GetCount());
- return;
+ PdfPage *page = NULL;
+ try {
+ page = self->doc->GetPage(pagenum - 1);
+ } catch(const PdfError &err) {
+ (void)err;
+ PyErr_Format(PyExc_ValueError, "No page number %d in the PDF file of %d pages", pagenum, self->doc->GetPageCount());
+ return ;
}
+ if (page) {
+ PdfDestination dest(page, left, top, zoom);
link.RemoveKey("A");
- PdfDestination dest(*page, left, top, zoom);
dest.AddToDictionary(link);
+ }
}
}
+
}
static PyObject *
@@ -542,19 +504,19 @@
bool mark_links = PyObject_IsTrue(py_mark_links);
try {
PdfArray border, link_color;
- border.Add(int64_t(16)); border.Add(int64_t(16)); border.Add(int64_t(1));
- link_color.Add(1.); link_color.Add(0.); link_color.Add(0.);
+ border.push_back((PoDoFo::pdf_int64)16); border.push_back((PoDoFo::pdf_int64)16); border.push_back((PoDoFo::pdf_int64)1);
+ link_color.push_back(1.); link_color.push_back(0.); link_color.push_back(0.);
std::vector<PdfReference> links;
for (auto &it : self->doc->GetObjects()) {
- PdfDictionary *link;
- if(it->TryGetDictionary(link)) {
- if (dictionary_has_key_name(*link, PdfName::KeyType, "Annot") && dictionary_has_key_name(*link, PdfName::KeySubtype, "Link")) {
- PdfObject *akey; PdfDictionary *A;
- if ((akey = link->GetKey("A")) && akey->TryGetDictionary(A)) {
- if (dictionary_has_key_name(*A, PdfName::KeyType, "Action") && dictionary_has_key_name(*A, "S", "URI")) {
- PdfObject *uo = A->GetKey("URI");
+ if(it->IsDictionary()) {
+ PdfDictionary &link = it->GetDictionary();
+ if (dictionary_has_key_name(link, PdfName::KeyType, "Annot") && dictionary_has_key_name(link, PdfName::KeySubtype, "Link")) {
+ if (link.HasKey("A") && link.GetKey("A")->IsDictionary()) {
+ PdfDictionary &A = link.GetKey("A")->GetDictionary();
+ if (dictionary_has_key_name(A, PdfName::KeyType, "Action") && dictionary_has_key_name(A, "S", "URI")) {
+ PdfObject *uo = A.GetKey("URI");
if (uo && uo->IsString()) {
- links.push_back(object_as_reference(it));
+ links.push_back(it->Reference());
}
}
}
@@ -585,134 +547,153 @@
static PyObject *
PDFDoc_pages_getter(PDFDoc *self, void *closure) {
- unsigned long pages = self->doc->GetPages().GetCount();
- PyObject *ans = PyLong_FromUnsignedLong(pages);
+ int pages = self->doc->GetPageCount();
+ PyObject *ans = PyLong_FromLong(static_cast<long>(pages));
if (ans != NULL) Py_INCREF(ans);
return ans;
}
static PyObject *
PDFDoc_version_getter(PDFDoc *self, void *closure) {
- PdfVersion version;
+ int version;
try {
- version = self->doc->GetMetadata().GetPdfVersion();
+ version = self->doc->GetPdfVersion();
} catch(const PdfError & err) {
podofo_set_exception(err);
return NULL;
}
switch(version) {
- case PdfVersion::V1_0:
- return PyUnicode_FromString("1.0");
- case PdfVersion::V1_1:
- return PyUnicode_FromString("1.1");
- case PdfVersion::V1_2:
- return PyUnicode_FromString("1.2");
- case PdfVersion::V1_3:
- return PyUnicode_FromString("1.3");
- case PdfVersion::V1_4:
- return PyUnicode_FromString("1.4");
- case PdfVersion::V1_5:
- return PyUnicode_FromString("1.5");
- case PdfVersion::V1_6:
- return PyUnicode_FromString("1.6");
- case PdfVersion::V1_7:
- return PyUnicode_FromString("1.7");
- case PdfVersion::V2_0:
- return PyUnicode_FromString("2.0");
- case PdfVersion::Unknown:
- return PyUnicode_FromString("");
- }
- return PyUnicode_FromString("");
-}
-
-static PdfDictionary&
-get_or_create_info(PDFDoc *self) {
- PdfObject *info = self->doc->GetTrailer().GetDictionary().FindKey("Info");
- if (info && info->IsDictionary()) return info->GetDictionary();
- auto ninfo = self->doc->GetObjects().CreateDictionaryObject();
- self->doc->GetTrailer().GetDictionary().AddKeyIndirect("Info", ninfo);
- return ninfo.GetDictionary();
-}
-
-static inline PyObject*
-string_metadata_getter(PDFDoc *self, const std::string_view name) {
- auto info = get_or_create_info(self);
- auto obj = info.FindKey(name);
- const PdfString* str;
- return (obj == nullptr || !obj->TryGetString(str)) ? PyUnicode_FromString("") : podofo_convert_pdfstring(*str);
+ case ePdfVersion_1_0:
+ return Py_BuildValue("s", "1.0");
+ case ePdfVersion_1_1:
+ return Py_BuildValue("s", "1.1");
+ case ePdfVersion_1_2:
+ return Py_BuildValue("s", "1.2");
+ case ePdfVersion_1_3:
+ return Py_BuildValue("s", "1.3");
+ case ePdfVersion_1_4:
+ return Py_BuildValue("s", "1.4");
+ case ePdfVersion_1_5:
+ return Py_BuildValue("s", "1.5");
+ case ePdfVersion_1_6:
+ return Py_BuildValue("s", "1.6");
+ case ePdfVersion_1_7:
+ return Py_BuildValue("s", "1.7");
+ default:
+ return Py_BuildValue("");
+ }
+ return Py_BuildValue("");
}
static PyObject *
-PDFDoc_title_getter(PDFDoc *self, void *closure) {
- return string_metadata_getter(self, "Title");
+PDFDoc_getter(PDFDoc *self, int field)
+{
+ PdfString s;
+ PdfInfo *info = self->doc->GetInfo();
+ if (info == NULL) {
+ PyErr_SetString(PyExc_Exception, "You must first load a PDF Document");
+ return NULL;
+ }
+ switch (field) {
+ case 0:
+ s = info->GetTitle(); break;
+ case 1:
+ s = info->GetAuthor(); break;
+ case 2:
+ s = info->GetSubject(); break;
+ case 3:
+ s = info->GetKeywords(); break;
+ case 4:
+ s = info->GetCreator(); break;
+ case 5:
+ s = info->GetProducer(); break;
+ default:
+ PyErr_SetString(PyExc_Exception, "Bad field");
+ return NULL;
+ }
+
+ return podofo_convert_pdfstring(s);
+}
+
+static int
+PDFDoc_setter(PDFDoc *self, PyObject *val, int field) {
+ if (val == NULL || !PyUnicode_Check(val)) {
+ PyErr_SetString(PyExc_ValueError, "Must use unicode objects to set metadata");
+ return -1;
+ }
+ PdfInfo *info = self->doc->GetInfo();
+ if (!info) { PyErr_SetString(Error, "You must first load a PDF Document"); return -1; }
+ const PdfString s = podofo_convert_pystring(val);
+
+ switch (field) {
+ case 0:
+ info->SetTitle(s); break;
+ case 1:
+ info->SetAuthor(s); break;
+ case 2:
+ info->SetSubject(s); break;
+ case 3:
+ info->SetKeywords(s); break;
+ case 4:
+ info->SetCreator(s); break;
+ case 5:
+ info->SetProducer(s); break;
+ default:
+ PyErr_SetString(Error, "Bad field");
+ return -1;
+ }
+
+ return 0;
}
static PyObject *
+PDFDoc_title_getter(PDFDoc *self, void *closure) {
+ return PDFDoc_getter(self, 0);
+}
+static PyObject *
PDFDoc_author_getter(PDFDoc *self, void *closure) {
- return string_metadata_getter(self, "Author");
+ return PDFDoc_getter(self, 1);
}
-
static PyObject *
PDFDoc_subject_getter(PDFDoc *self, void *closure) {
- return string_metadata_getter(self, "Subject");
+ return PDFDoc_getter(self, 2);
}
-
static PyObject *
PDFDoc_keywords_getter(PDFDoc *self, void *closure) {
- return string_metadata_getter(self, "Keywords");
+ return PDFDoc_getter(self, 3);
}
-
static PyObject *
PDFDoc_creator_getter(PDFDoc *self, void *closure) {
- return string_metadata_getter(self, "Creator");
+ return PDFDoc_getter(self, 4);
}
-
static PyObject *
PDFDoc_producer_getter(PDFDoc *self, void *closure) {
- return string_metadata_getter(self, "Producer");
-}
-
-static inline int
-string_metadata_setter(PDFDoc *self, const std::string_view name, PyObject *val) {
- if (!PyUnicode_Check(val)) { PyErr_SetString(PyExc_TypeError, "Must use unicode to set metadata"); return -1; }
- auto& info = get_or_create_info(self);
- const char *raw; Py_ssize_t sz;
- raw = PyUnicode_AsUTF8AndSize(val, &sz);
- if (sz == 0) info.RemoveKey(name);
- else info.AddKey(name, PdfString(std::string_view(raw, sz)));
- return 0;
+ return PDFDoc_getter(self, 5);
}
-
-
static int
PDFDoc_title_setter(PDFDoc *self, PyObject *val, void *closure) {
- return string_metadata_setter(self, "Title", val);
+ return PDFDoc_setter(self, val, 0);
}
-
static int
PDFDoc_author_setter(PDFDoc *self, PyObject *val, void *closure) {
- return string_metadata_setter(self, "Author", val);
+ return PDFDoc_setter(self, val, 1);
}
-
static int
PDFDoc_subject_setter(PDFDoc *self, PyObject *val, void *closure) {
- return string_metadata_setter(self, "Subject", val);
+ return PDFDoc_setter(self, val, 2);
}
-
static int
PDFDoc_keywords_setter(PDFDoc *self, PyObject *val, void *closure) {
- return string_metadata_setter(self, "Keywords", val);
+ return PDFDoc_setter(self, val, 3);
}
-
static int
PDFDoc_creator_setter(PDFDoc *self, PyObject *val, void *closure) {
- return string_metadata_setter(self, "Creator", val);
+ return PDFDoc_setter(self, val, 4);
}
-
static int
PDFDoc_producer_setter(PDFDoc *self, PyObject *val, void *closure) {
- return string_metadata_setter(self, "Producer", val);
+ return PDFDoc_setter(self, val, 5);
}
static PyGetSetDef PDFDoc_getsetters[] = {
@@ -843,10 +824,6 @@
{"set_xmp_metadata", (PyCFunction)PDFDoc_set_xmp_metadata, METH_VARARGS,
"set_xmp_metadata(raw) -> Set the XMP metadata to the raw bytes (which must be a valid XML packet)"
},
- {"add_image_page", (PyCFunction)PDFDoc_add_image_page, METH_VARARGS,
- "add_image_page(image_data, page_idx=0) -> Add the specified image as a full page image, will use the size of the first existing page as page size."
- },
-
{NULL} /* Sentinel */
};
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/fonts.cpp calibre-6.20.0/src/calibre/utils/podofo/fonts.cpp
--- calibre-6.20.0.orig/src/calibre/utils/podofo/fonts.cpp 2023-06-08 18:36:58.000000000 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/fonts.cpp 2023-06-11 11:20:17.615014514 -0700
@@ -7,7 +7,6 @@
#include "global.h"
#include <iostream>
-#include <memory>
#include <stack>
using namespace pdf;
@@ -19,61 +18,47 @@
}
static inline PdfObject*
-get_font_file(PdfObject *descriptor) {
- PdfDictionary *dict;
- PdfObject *ff = NULL;
- if (descriptor->TryGetDictionary(dict)) {
- ff = dict->FindKey("FontFile");
- if (!ff) ff = dict->FindKey("FontFile2");
- if (!ff) ff = dict->FindKey("FontFile3");
- }
- return ff;
-}
-
-static inline const PdfObject*
get_font_file(const PdfObject *descriptor) {
- const PdfDictionary *dict;
- const PdfObject *ff = NULL;
- if (descriptor->TryGetDictionary(dict)) {
- ff = dict->FindKey("FontFile");
- if (!ff) ff = dict->FindKey("FontFile2");
- if (!ff) ff = dict->FindKey("FontFile3");
- }
+ PdfObject *ff = descriptor->GetIndirectKey("FontFile");
+ if (!ff) ff = descriptor->GetIndirectKey("FontFile2");
+ if (!ff) ff = descriptor->GetIndirectKey("FontFile3");
return ff;
}
-
static inline void
-remove_font(PdfIndirectObjectList &objects, PdfObject *font) {
- PdfDictionary *dict;
- if (font->TryGetDictionary(dict)) {
- PdfObject *descriptor = dict->FindKey("FontDescriptor");
- if (descriptor) {
- const PdfObject *ff = get_font_file(descriptor);
- if (ff) objects.RemoveObject(object_as_reference(ff)).reset();
- objects.RemoveObject(object_as_reference(descriptor)).reset();
- }
+remove_font(PdfVecObjects &objects, PdfObject *font) {
+ PdfObject *descriptor = font->GetIndirectKey("FontDescriptor");
+ if (descriptor) {
+ const PdfObject *ff = get_font_file(descriptor);
+ if (ff) delete objects.RemoveObject(ff->Reference());
+ delete objects.RemoveObject(descriptor->Reference());
}
- objects.RemoveObject(object_as_reference(font)).reset();
+ delete objects.RemoveObject(font->Reference());
+}
+
+static inline uint64_t
+ref_as_integer(pdf_objnum num, pdf_gennum gen) {
+ return static_cast<uint64_t>(num) | (static_cast<uint64_t>(gen) << 32);
}
+static inline uint64_t
+ref_as_integer(const PdfReference &ref) { return ref_as_integer(ref.ObjectNumber(), ref.GenerationNumber()); }
+
static void
-used_fonts_in_canvas(const PdfCanvas &canvas, unordered_reference_set &ans) {
- PdfPostScriptTokenizer tokenizer;
- PdfCanvasInputDevice input(canvas);
+used_fonts_in_canvas(PdfCanvas *canvas, unordered_reference_set &ans) {
+ PdfContentsTokenizer tokenizer(canvas);
bool in_text_block = false;
- PdfPostScriptTokenType contents_type;
+ const char* token = NULL;
+ EPdfContentsType contents_type;
PdfVariant var;
std::stack<PdfVariant> stack;
- const PdfDictionary &resources = canvas.GetResources()->GetDictionary();
+ const PdfDictionary &resources = canvas->GetResources()->GetDictionary();
if (!resources.HasKey("Font")) return;
const PdfDictionary &fonts_dict = resources.GetKey("Font")->GetDictionary();
- std::string_view keyword;
- while (tokenizer.TryReadNext(input, contents_type, keyword, var)) {
- if (contents_type == PdfPostScriptTokenType::Variant) stack.push(var);
- if (contents_type != PdfPostScriptTokenType::Keyword) continue;
- const char *token = keyword.data();
+ while (tokenizer.ReadNext(contents_type, token, var)) {
+ if (contents_type == ePdfContentsType_Variant) stack.push(var);
+ if (contents_type != ePdfContentsType_Keyword) continue;
if (strcmp(token, "BT") == 0) {
in_text_block = true;
continue;
@@ -86,8 +71,9 @@
stack.pop();
if (stack.size() > 0 && stack.top().IsName()) {
const PdfName &reference_name = stack.top().GetName();
- const PdfObject *f = fonts_dict.GetKey(reference_name);
- if (f) ans.insert(object_as_reference(f));
+ if (fonts_dict.HasKey(reference_name)) {
+ ans.insert(fonts_dict.GetKey(reference_name)->GetReference());
+ }
}
}
}
@@ -102,10 +88,10 @@
pyunique_ptr item;
if ((*it).IsArray()) {
item.reset(convert_w_array((*it).GetArray()));
- } else if ((*it).IsRealStrict()) {
- item.reset(PyFloat_FromDouble((*it).GetReal()));
} else if ((*it).IsNumber()) {
item.reset(PyLong_FromLongLong((long long)(*it).GetNumber()));
+ } else if ((*it).IsReal()) {
+ item.reset(PyFloat_FromDouble((*it).GetReal()));
} else PyErr_SetString(PyExc_ValueError, "Unknown datatype in w array");
if (!item) return NULL;
if (PyList_Append(ans.get(), item.get()) != 0) return NULL;
@@ -119,16 +105,16 @@
if (!PyArg_ParseTuple(args, "|i", &get_font_data)) return NULL;
pyunique_ptr ans(PyList_New(0));
if (!ans) return NULL;
- const PdfIndirectObjectList &objects = self->doc->GetObjects();
+ const PdfVecObjects &objects = self->doc->GetObjects();
for (auto &it : objects) {
if (it->IsDictionary()) {
const PdfDictionary &dict = it->GetDictionary();
if (dictionary_has_key_name(dict, PdfName::KeyType, "Font") && dict.HasKey("BaseFont")) {
- const std::string &name = dict.GetKey("BaseFont")->GetName().GetString();
- const std::string &subtype = dict.GetKey(PdfName::KeySubtype)->GetName().GetString();
- const PdfReference &ref = object_as_reference(it);
+ const std::string &name = dict.GetKey("BaseFont")->GetName().GetName();
+ const std::string &subtype = dict.GetKey(PdfName::KeySubtype)->GetName().GetName();
+ const PdfReference &ref = it->Reference();
unsigned long num = ref.ObjectNumber(), generation = ref.GenerationNumber();
- const PdfObject *descriptor = dict.FindKey("FontDescriptor");
+ const PdfObject *descriptor = it->GetIndirectKey("FontDescriptor");
pyunique_ptr descendant_font, stream_ref, encoding, w, w2;
PyBytesOutputStream stream_data, to_unicode, cid_gid_map;
if (dict.HasKey("W")) {
@@ -140,33 +126,33 @@
if (!w2) return NULL;
}
if (dict.HasKey("Encoding") && dict.GetKey("Encoding")->IsName()) {
- encoding.reset(PyUnicode_FromString(dict.GetKey("Encoding")->GetName().GetString().c_str()));
+ encoding.reset(PyUnicode_FromString(dict.GetKey("Encoding")->GetName().GetName().c_str()));
if (!encoding) return NULL;
}
- if (dict.HasKey("CIDToGIDMap") && (!dict.GetKey("CIDToGIDMap")->IsName() || strcmp(dict.GetKey("CIDToGIDMap")->GetName().GetString().c_str(), "Identity") != 0)) {
- const PdfObjectStream *stream = dict.GetKey("CIDToGIDMap")->GetStream();
- if (stream) stream->CopyToSafe(cid_gid_map);
+ if (dict.HasKey("CIDToGIDMap") && (!dict.GetKey("CIDToGIDMap")->IsName() || strcmp(dict.GetKey("CIDToGIDMap")->GetName().GetName().c_str(), "Identity") != 0)) {
+ const PdfStream *stream = dict.GetKey("CIDToGIDMap")->GetStream();
+ if (stream) stream->GetFilteredCopy(&cid_gid_map);
}
if (descriptor) {
const PdfObject *ff = get_font_file(descriptor);
if (ff) {
- stream_ref.reset(ref_as_tuple(object_as_reference(ff)));
+ stream_ref.reset(ref_as_tuple(ff->Reference()));
if (!stream_ref) return NULL;
- const PdfObjectStream *stream = ff->GetStream();
+ const PdfStream *stream = ff->GetStream();
if (stream && get_font_data) {
- stream->CopyToSafe(stream_data);
+ stream->GetFilteredCopy(&stream_data);
}
}
} else if (dict.HasKey("DescendantFonts")) {
const PdfArray &df = dict.GetKey("DescendantFonts")->GetArray();
- descendant_font.reset(ref_as_tuple(object_as_reference(df[0])));
+ descendant_font.reset(ref_as_tuple(df[0].GetReference()));
if (!descendant_font) return NULL;
if (get_font_data && dict.HasKey("ToUnicode")) {
- const PdfReference &uref = object_as_reference(dict.GetKey("ToUnicode"));
+ const PdfReference &uref = dict.GetKey("ToUnicode")->GetReference();
PdfObject *t = objects.GetObject(uref);
if (t) {
- PdfObjectStream *stream = t->GetStream();
- if (stream) stream->CopyToSafe(to_unicode);
+ PdfStream *stream = t->GetStream();
+ if (stream) stream->GetFilteredCopy(&to_unicode);
}
}
}
@@ -200,18 +186,18 @@
unsigned long count = 0;
unordered_reference_set used_fonts;
// Look in Pages
- PdfPageCollection *pages = &self->doc->GetPages();
- for (unsigned i = 0; i < pages->GetCount(); i++) {
- used_fonts_in_canvas(self->doc->GetPages().GetPageAt(i), used_fonts);
+ for (int i = 0; i < self->doc->GetPageCount(); i++) {
+ PdfPage *page = self->doc->GetPage(i);
+ if (page) used_fonts_in_canvas(page, used_fonts);
}
// Look in XObjects
- PdfIndirectObjectList &objects = self->doc->GetObjects();
- for (PdfObject *k : objects) {
+ PdfVecObjects &objects = self->doc->GetObjects();
+ for (auto &k : objects) {
if (k->IsDictionary()) {
const PdfDictionary &dict = k->GetDictionary();
if (dictionary_has_key_name(dict, PdfName::KeyType, "XObject") && dictionary_has_key_name(dict, PdfName::KeySubtype, "Form")) {
- std::unique_ptr<PdfXObjectForm> xo;
- if (PdfXObject::TryCreateFromObject<PdfXObjectForm>(*k, xo)) used_fonts_in_canvas(*xo, used_fonts);
+ PdfXObject xo(k);
+ used_fonts_in_canvas(&xo, used_fonts);
}
}
}
@@ -222,14 +208,14 @@
if (k->IsDictionary()) {
const PdfDictionary &dict = k->GetDictionary();
if (dictionary_has_key_name(dict, PdfName::KeyType, "Font")) {
- const std::string &font_type = dict.GetKey(PdfName::KeySubtype)->GetName().GetString();
+ const std::string &font_type = dict.GetKey(PdfName::KeySubtype)->GetName().GetName();
if (font_type == "Type0") {
- all_fonts.insert(object_as_reference(k));
+ all_fonts.insert(k->Reference());
} else if (font_type == "Type3") {
- all_fonts.insert(object_as_reference(k));
- type3_fonts.insert(object_as_reference(k));
- for (auto &x : dict.GetKey("CharProcs")->GetDictionary()) {
- const PdfReference &ref = object_as_reference(x.second);
+ all_fonts.insert(k->Reference());
+ type3_fonts.insert(k->Reference());
+ for (auto &x : dict.GetKey("CharProcs")->GetDictionary().GetKeys()) {
+ const PdfReference &ref = x.second->GetReference();
if (charprocs_usage.find(ref) == charprocs_usage.end()) charprocs_usage[ref] = 1;
else charprocs_usage[ref] += 1;
}
@@ -243,18 +229,16 @@
PdfObject *font = objects.GetObject(ref);
if (font) {
count++;
- PdfDictionary *dict;
- if (font->TryGetDictionary(dict)) {
if (type3_fonts.find(ref) != type3_fonts.end()) {
- for (auto &x : dict->FindKey("CharProcs")->GetDictionary()) {
- charprocs_usage[object_as_reference(x.second)] -= 1;
+ for (auto &x : font->GetIndirectKey("CharProcs")->GetDictionary().GetKeys()) {
+ charprocs_usage[x.second->GetReference()] -= 1;
}
} else {
- for (auto &x : dict->FindKey("DescendantFonts")->GetArray()) {
- PdfObject *dfont = objects.GetObject(object_as_reference(x));
+ for (auto &x : font->GetIndirectKey("DescendantFonts")->GetArray()) {
+ PdfObject *dfont = objects.GetObject(x.GetReference());
if (dfont) remove_font(objects, dfont);
}
- }}
+ }
remove_font(objects, font);
}
}
@@ -262,7 +246,7 @@
for (auto &x : charprocs_usage) {
if (x.second == 0u) {
- objects.RemoveObject(x.first).reset();
+ delete objects.RemoveObject(x.first);
}
}
@@ -274,16 +258,14 @@
const char *data; Py_ssize_t sz;
unsigned long num, gen;
if (!PyArg_ParseTuple(args, "y#kk", &data, &sz, &num, &gen)) return NULL;
- const PdfIndirectObjectList &objects = self->doc->GetObjects();
- PdfObject *font = objects.GetObject(PdfReference(num, static_cast<uint16_t>(gen)));
+ const PdfVecObjects &objects = self->doc->GetObjects();
+ PdfObject *font = objects.GetObject(PdfReference(num, static_cast<pdf_gennum>(gen)));
if (!font) { PyErr_SetString(PyExc_KeyError, "No font with the specified reference found"); return NULL; }
- PdfDictionary *dict;
- if (!font->TryGetDictionary(dict)) { PyErr_SetString(PyExc_ValueError, "Font does not have a descriptor"); return NULL; }
- PdfObject *descriptor = dict->FindKey("FontDescriptor");
+ const PdfObject *descriptor = font->GetIndirectKey("FontDescriptor");
if (!descriptor) { PyErr_SetString(PyExc_ValueError, "Font does not have a descriptor"); return NULL; }
PdfObject *ff = get_font_file(descriptor);
- PdfObjectStream *stream = ff->GetStream();
- stream->SetData(bufferview(data, sz));
+ PdfStream *stream = ff->GetStream();
+ stream->Set(data, sz);
Py_RETURN_NONE;
}
@@ -292,61 +274,60 @@
const char *data; Py_ssize_t sz;
PyObject *references;
if (!PyArg_ParseTuple(args, "y#O!", &data, &sz, &PyTuple_Type, &references)) return NULL;
- PdfIndirectObjectList &objects = self->doc->GetObjects();
+ PdfVecObjects &objects = self->doc->GetObjects();
PdfObject *font_file = NULL;
- PdfDictionary *dict;
for (Py_ssize_t i = 0; i < PyTuple_GET_SIZE(references); i++) {
unsigned long num, gen;
if (!PyArg_ParseTuple(PyTuple_GET_ITEM(references, i), "kk", &num, &gen)) return NULL;
- PdfObject *font = objects.GetObject(PdfReference(num, static_cast<uint16_t>(gen)));
+ PdfObject *font = objects.GetObject(PdfReference(num, static_cast<pdf_gennum>(gen)));
if (!font) { PyErr_SetString(PyExc_KeyError, "No font with the specified reference found"); return NULL; }
-
- PdfObject *dobj = NULL;
- if (font->TryGetDictionary(dict)) { dobj = dict->FindKey("FontDescriptor"); }
+ PdfObject *dobj = font->GetIndirectKey("FontDescriptor");
if (!dobj) { PyErr_SetString(PyExc_ValueError, "Font does not have a descriptor"); return NULL; }
if (!dobj->IsDictionary()) { PyErr_SetString(PyExc_ValueError, "Font does not have a dictionary descriptor"); return NULL; }
PdfDictionary &descriptor = dobj->GetDictionary();
const char *font_file_key = NULL;
- PdfObject *ff = NULL;
- if ((ff = descriptor.FindKey("FontFile"))) { font_file_key = "FontFile"; }
- else if ((ff = descriptor.FindKey("FontFile2"))) { font_file_key = "FontFile2"; }
- else if ((ff = descriptor.FindKey("FontFile3"))) { font_file_key = "FontFile3"; }
- else { PyErr_SetString(PyExc_ValueError, "Font descriptor does not have file data"); return NULL; }
+ if (descriptor.HasKey("FontFile")) font_file_key = "FontFile";
+ else if (descriptor.HasKey("FontFile2")) font_file_key = "FontFile2";
+ else if (descriptor.HasKey("FontFile3")) font_file_key = "FontFile3";
+ else { PyErr_SetString(PyExc_ValueError, "Font descriptor does not have file data"); return NULL; }
+ PdfObject *ff = dobj->GetIndirectKey(font_file_key);
if (i == 0) {
font_file = ff;
- PdfObjectStream *stream = ff->GetStream();
- stream->SetData(bufferview(data, sz));
+ PdfStream *stream = ff->GetStream();
+ stream->Set(data, sz);
} else {
- objects.RemoveObject(object_as_reference(ff)).reset();
- descriptor.AddKey(font_file_key, object_as_reference(font_file));
+ delete objects.RemoveObject(ff->Reference());
+ descriptor.AddKey(font_file_key, font_file->Reference());
}
}
Py_RETURN_NONE;
}
class CharProc {
- charbuff buf;
+ char *buf; pdf_long sz;
PdfReference ref;
CharProc( const CharProc & ) ;
CharProc & operator=( const CharProc & ) ;
public:
- CharProc(const PdfReference &reference, const PdfObject *o) : buf(), ref(reference) {
- const PdfObjectStream *stream = o->GetStream();
- buf = stream->GetCopySafe();
+ CharProc(const PdfReference &reference, const PdfObject *o) : buf(NULL), sz(0), ref(reference) {
+ const PdfStream *stream = o->GetStream();
+ stream->GetFilteredCopy(&buf, &sz);
}
CharProc(CharProc &&other) noexcept :
- buf(std::move(other.buf)), ref(other.ref) {
- other.buf = charbuff();
+ buf(other.buf), sz(other.sz), ref(other.ref) {
+ other.buf = NULL;
}
CharProc& operator=(CharProc &&other) noexcept {
- buf = std::move(other.buf); other.buf = charbuff(); ref = other.ref;
+ if (buf) podofo_free(buf);
+ buf = other.buf; other.buf = NULL; sz = other.sz; ref = other.ref;
return *this;
}
+ ~CharProc() noexcept { if (buf) podofo_free(buf); buf = NULL; }
bool operator==(const CharProc &other) const noexcept {
- return buf.size() == other.buf.size() && memcmp(buf.data(), other.buf.data(), buf.size()) == 0;
+ return other.sz == sz && memcmp(buf, other.buf, sz) == 0;
}
- std::size_t hash() const noexcept { return buf.size(); }
+ std::size_t hash() const noexcept { return sz; }
const PdfReference& reference() const noexcept { return ref; }
};
@@ -363,16 +344,16 @@
unordered_reference_set all_type3_fonts;
char_proc_reference_map cp_map;
- PdfIndirectObjectList &objects = self->doc->GetObjects();
+ PdfVecObjects &objects = self->doc->GetObjects();
for (auto &k : objects) {
if (!k->IsDictionary()) continue;
const PdfDictionary &dict = k->GetDictionary();
if (dictionary_has_key_name(dict, PdfName::KeyType, "Font")) {
- const std::string &font_type = dict.GetKey(PdfName::KeySubtype)->GetName().GetString();
+ const std::string &font_type = dict.GetKey(PdfName::KeySubtype)->GetName().GetName();
if (font_type == "Type3") {
- all_type3_fonts.insert(object_as_reference(k));
- for (auto &x : dict.GetKey("CharProcs")->GetDictionary()) {
- const PdfReference &ref = object_as_reference(x.second);
+ all_type3_fonts.insert(k->Reference());
+ for (auto &x : dict.GetKey("CharProcs")->GetDictionary().GetKeys()) {
+ const PdfReference &ref = x.second->GetReference();
const PdfObject *cpobj = objects.GetObject(ref);
if (!cpobj || !cpobj->HasStream()) continue;
CharProc cp(ref, cpobj);
@@ -392,7 +373,7 @@
for (auto &ref : x.second) {
if (ref != canonical_ref) {
ref_map[ref] = x.first.reference();
- objects.RemoveObject(ref).reset();
+ delete objects.RemoveObject(ref);
count++;
}
}
@@ -401,13 +382,11 @@
if (count > 0) {
for (auto &ref : all_type3_fonts) {
PdfObject *font = objects.GetObject(ref);
- PdfDictionary *d;
- if (!font->TryGetDictionary(d)) continue;
- PdfDictionary dict = d->FindKey("CharProcs")->GetDictionary();
+ PdfDictionary dict = font->GetIndirectKey("CharProcs")->GetDictionary();
PdfDictionary new_dict = PdfDictionary(dict);
bool changed = false;
- for (auto &k : dict) {
- auto it = ref_map.find(object_as_reference(k.second));
+ for (auto &k : dict.GetKeys()) {
+ auto it = ref_map.find(k.second->GetReference());
if (it != ref_map.end()) {
new_dict.AddKey(k.first, (*it).second);
changed = true;
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/global.h calibre-6.20.0/src/calibre/utils/podofo/global.h
--- calibre-6.20.0.orig/src/calibre/utils/podofo/global.h 2023-06-08 18:36:58.000000000 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/global.h 2023-06-11 11:20:17.615014514 -0700
@@ -15,7 +15,6 @@
#include <unordered_set>
#include <unordered_map>
using namespace PoDoFo;
-using namespace std::literals;
namespace pdf {
@@ -26,7 +25,6 @@
PyObject_HEAD
/* Type-specific fields go here. */
PdfMemDocument *doc;
- PyObject *load_buffer_ref;
} PDFDoc;
@@ -54,7 +52,7 @@
// unique_ptr that uses Py_XDECREF as the destructor function.
typedef std::unique_ptr<PyObject, PyObjectDeleter> pyunique_ptr;
-class PyBytesOutputStream : public OutputStream {
+class PyBytesOutputStream : public PdfOutputStream {
private:
pyunique_ptr bytes;
PyBytesOutputStream( const PyBytesOutputStream & ) ;
@@ -64,18 +62,18 @@
void Close() {}
operator bool() const { return bool(bytes); }
PyObject* get() const { return bytes.get(); }
- protected:
- void writeBuffer(const char *buf, size_t sz){
+ pdf_long Write(const char *buf, const pdf_long sz){
if (!bytes) {
bytes.reset(PyBytes_FromStringAndSize(buf, sz));
- if (!bytes) throw PdfError(PdfErrorCode::OutOfMemory, __FILE__, __LINE__, NULL);
+ if (!bytes) throw PdfError(ePdfError_OutOfMemory, __FILE__, __LINE__, NULL);
} else {
size_t old_sz = PyBytes_GET_SIZE(bytes.get());
PyObject *old = bytes.release();
- if (_PyBytes_Resize(&old, old_sz + sz) != 0) throw PdfError(PdfErrorCode::OutOfMemory, __FILE__, __LINE__, NULL);
+ if (_PyBytes_Resize(&old, old_sz + sz) != 0) throw PdfError(ePdfError_OutOfMemory, __FILE__, __LINE__, NULL);
memcpy(PyBytes_AS_STRING(old) + old_sz, buf, sz);
bytes.reset(old);
}
+ return sz;
}
};
@@ -84,56 +82,10 @@
static inline bool
dictionary_has_key_name(const PdfDictionary &d, T key, const char *name) {
const PdfObject *val = d.GetKey(key);
- if (val && val->IsName() && val->GetName().GetString() == name) return true;
+ if (val && val->IsName() && val->GetName().GetName() == name) return true;
return false;
}
-static inline const PdfPage*
-get_page(const PdfPageCollection &pages, const PdfReference &ref) {
- try {
- return &pages.GetPage(ref);
- } catch(PdfError &) { }
- return nullptr;
-}
-
-static inline const PdfPage*
-get_page(const PdfDocument *doc, const PdfReference &ref) {
- try {
- return &doc->GetPages().GetPage(ref);
- } catch(PdfError &) { }
- return nullptr;
-}
-
-static inline const PdfPage*
-get_page(const PdfDocument *doc, const unsigned num) {
- try {
- return &doc->GetPages().GetPageAt(num);
- } catch(PdfError &) { }
- return nullptr;
-}
-
-static inline PdfPage*
-get_page(PdfDocument *doc, const unsigned num) {
- try {
- return &doc->GetPages().GetPageAt(num);
- } catch(PdfError &) { }
- return nullptr;
-}
-
-static inline PdfReference
-object_as_reference(const PdfObject &o) {
- return o.IsReference() ? o.GetReference() : o.GetIndirectReference();
-}
-
-static inline PdfReference
-object_as_reference(const PdfObject *o) {
- return o->IsReference() ? o->GetReference() : o->GetIndirectReference();
-}
-
-// Needed to avoid PoDoFo clobbering the /Info and XMP metadata with its own nonsense
-// rename to NoMetadataUdate after https://github.com/podofo/podofo/commit/96689eb6e45b71eae1577ecb2d4a796c52e9a813
-static const PdfSaveOptions save_options = PdfSaveOptions::NoModifyDateUpdate;
-
class PdfReferenceHasher {
public:
size_t operator()(const PdfReference & obj) const {
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/images.cpp calibre-6.20.0/src/calibre/utils/podofo/images.cpp
--- calibre-6.20.0.orig/src/calibre/utils/podofo/images.cpp 2023-06-08 18:36:58.000000000 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/images.cpp 2023-06-11 11:20:17.616014512 -0700
@@ -10,40 +10,39 @@
using namespace pdf;
class Image {
- charbuff buf;
- int64_t width, height;
+ char *buf; pdf_long sz;
+ pdf_int64 width, height;
PdfReference ref;
Image( const Image & ) ;
Image & operator=( const Image & ) ;
- bool is_valid;
public:
- Image(const PdfReference &reference, const PdfObject *o) : buf(), width(0), height(0), ref(reference) {
- const PdfObjectStream *stream = o->GetStream();
+ Image(const PdfReference &reference, const PdfObject *o) : buf(NULL), sz(0), width(0), height(0), ref(reference) {
+ const PdfStream *stream = o->GetStream();
try {
- buf = stream->GetCopySafe();
- is_valid = true;
+ stream->GetFilteredCopy(&buf, &sz);
} catch(...) {
- buf = charbuff();
- is_valid = false;
+ buf = NULL; sz = -1;
}
const PdfDictionary &dict = o->GetDictionary();
if (dict.HasKey("Width") && dict.GetKey("Width")->IsNumber()) width = dict.GetKey("Width")->GetNumber();
if (dict.HasKey("Height") && dict.GetKey("Height")->IsNumber()) height = dict.GetKey("Height")->GetNumber();
}
Image(Image &&other) noexcept :
- buf(std::move(other.buf)), width(other.width), height(other.height), ref(other.ref) {
- other.buf = charbuff(); is_valid = other.is_valid;
+ buf(other.buf), sz(other.sz), width(other.width), height(other.height), ref(other.ref) {
+ other.buf = NULL;
}
Image& operator=(Image &&other) noexcept {
- buf = std::move(other.buf); other.buf = charbuff(); ref = other.ref;
- width = other.width; height = other.height; is_valid = other.is_valid;
+ if (buf) podofo_free(buf);
+ buf = other.buf; other.buf = NULL; sz = other.sz; ref = other.ref;
+ width = other.width; height = other.height;
return *this;
}
+ ~Image() noexcept { if (buf) podofo_free(buf); buf = NULL; }
bool operator==(const Image &other) const noexcept {
- return other.width == width && is_valid && other.is_valid && other.height == height && other.buf == buf;
+ return other.sz == sz && sz > -1 && other.width == width && other.height == height && memcmp(buf, other.buf, sz) == 0;
}
- std::size_t hash() const noexcept { return buf.size(); }
+ std::size_t hash() const noexcept { return sz; }
const PdfReference& reference() const noexcept { return ref; }
};
@@ -57,14 +56,14 @@
static PyObject*
dedup_images(PDFDoc *self, PyObject *args) {
unsigned long count = 0;
- PdfIndirectObjectList &objects = self->doc->GetObjects();
+ PdfVecObjects &objects = self->doc->GetObjects();
image_reference_map image_map;
for (auto &k : objects) {
if (!k->IsDictionary()) continue;
const PdfDictionary &dict = k->GetDictionary();
if (dictionary_has_key_name(dict, PdfName::KeyType, "XObject") && dictionary_has_key_name(dict, PdfName::KeySubtype, "Image")) {
- Image img(object_as_reference(k), k);
+ Image img(k->Reference(), k);
auto it = image_map.find(img);
if (it == image_map.end()) {
std::vector<PdfReference> vals;
@@ -79,7 +78,7 @@
for (auto &ref : x.second) {
if (ref != canonical_ref) {
ref_map[ref] = x.first.reference();
- objects.RemoveObject(ref).reset();
+ delete objects.RemoveObject(ref);
count++;
}
}
@@ -96,11 +95,11 @@
const PdfDictionary &xobject = resources.GetKey("XObject")->GetDictionary();
PdfDictionary new_xobject = PdfDictionary(xobject);
bool changed = false;
- for (const auto &x : xobject) {
- if (x.second.IsReference()) {
+ for (auto &x : xobject.GetKeys()) {
+ if (x.second->IsReference()) {
try {
- const PdfReference &r = ref_map.at(object_as_reference(x.second));
- new_xobject.AddKey(x.first, r);
+ const PdfReference &r = ref_map.at(x.second->GetReference());
+ new_xobject.AddKey(x.first.GetName(), r);
changed = true;
} catch (const std::out_of_range &err) { (void)err; continue; }
}
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/impose.cpp calibre-6.20.0/src/calibre/utils/podofo/impose.cpp
--- calibre-6.20.0.orig/src/calibre/utils/podofo/impose.cpp 2023-06-08 18:36:58.000000000 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/impose.cpp 2023-06-11 11:20:17.616014512 -0700
@@ -6,25 +6,24 @@
*/
#include "global.h"
-#include <sstream>
-#include <string>
using namespace pdf;
static void
-impose_page(PdfMemDocument *doc, unsigned int dest_page_num, unsigned int src_page_num) {
- auto &src_page = doc->GetPages().GetPageAt(src_page_num);
- auto xobj = doc->CreateXObjectForm(src_page.GetMediaBox(), "HeaderFooter");
- xobj->FillFromPage(src_page);
- auto &dest = doc->GetPages().GetPageAt(dest_page_num);
- dest.GetOrCreateResources().AddResource("XObject", xobj->GetIdentifier(), xobj->GetObject().GetIndirectReference());
- // prepend the header footer xobject to the stream. This means header/footer is drawn first then the contents, which works
- // since chromium does not draw in margin areas. The reverse, i.e. appending, does not work with older WebEngine before Qt 6.5.
- PdfContents *contents = dest.GetContents();
- std::ostringstream s;
- s << "q\n1 0 0 1 0 0 cm\n/" << xobj->GetIdentifier().GetString() << " Do\nQ\n" << contents->GetCopy();
- contents->Reset();
- contents->GetStreamForAppending().SetData(s.str());
+impose_page(PdfMemDocument *doc, unsigned long dest_page_num, unsigned long src_page_num) {
+ PdfXObject *xobj = new PdfXObject(doc, src_page_num, "HeaderFooter");
+ PdfPage *dest = doc->GetPage(dest_page_num);
+ dest->AddResource(xobj->GetIdentifier(), xobj->GetObject()->Reference(), "XObject");
+ PdfStream *stream = dest->GetContents()->GetStream();
+ char *buffer = NULL; pdf_long sz;
+ stream->GetFilteredCopy(&buffer, &sz);
+ stream->BeginAppend();
+ stream->Append("q\n1 0 0 1 0 0 cm\n/");
+ stream->Append(xobj->GetIdentifier().GetName());
+ stream->Append(" Do\nQ\n");
+ stream->Append(buffer, sz);
+ stream->EndAppend();
+ podofo_free(buffer);
}
static PyObject*
@@ -34,8 +33,7 @@
for (unsigned long i = 0; i < count; i++) {
impose_page(self->doc, dest_page_num - 1 + i, src_page_num - 1 + i);
}
- auto& pages = self->doc->GetPages();
- while (count-- && src_page_num <= pages.GetCount()) pages.RemovePageAt(src_page_num - 1);
+ self->doc->DeletePages(src_page_num - 1, count);
Py_RETURN_NONE;
}
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/__init__.py calibre-6.20.0/src/calibre/utils/podofo/__init__.py
--- calibre-6.20.0.orig/src/calibre/utils/podofo/__init__.py 2023-06-11 10:44:12.782384417 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/__init__.py 2023-06-11 11:20:17.616014512 -0700
@@ -70,7 +70,7 @@
touched = True
try:
- tags = prep(', '.join(x.strip() for x in tags if x.strip()))
+ tags = prep(', '.join([x.strip() for x in tags if x.strip()]))
if tags != pdf_doc.keywords:
pdf_doc.keywords = tags
touched = True
@@ -165,23 +165,6 @@
print(f'Modified pdf with {num} glyphs removed saved to:', dest)
-def add_image_page(pdf_doc, image_data, page_size=None, page_num=1, preserve_aspect_ratio=True):
- if page_size is None:
- from qt.core import QPageSize
- p = QPageSize(QPageSize.PageSizeId.A4).rect(QPageSize.Unit.Point)
- page_size = p.left(), p.top(), p.width(), p.height()
- pdf_doc.add_image_page(
- image_data, *page_size, *page_size, page_num, preserve_aspect_ratio)
-
-
-def test_add_image_page(image='/t/t.jpg', dest='/t/t.pdf', **kw):
- image_data = open(image, 'rb').read()
- podofo = get_podofo()
- p = podofo.PDFDoc()
- add_image_page(p, image_data, **kw)
- p.save(dest)
-
-
def test_list_fonts(src):
podofo = get_podofo()
p = podofo.PDFDoc()
@@ -205,40 +188,33 @@
def test_podofo():
import tempfile
+ from io import BytesIO
from calibre.ebooks.metadata.book.base import Metadata
from calibre.ebooks.metadata.xmp import metadata_to_xmp_packet
# {{{
raw = b"%PDF-1.1\n%\xe2\xe3\xcf\xd3\n1 0 obj<</Type/Catalog/Metadata 6 0 R/Pages 2 0 R>>\nendobj\n2 0 obj<</Type/Pages/Count 1/Kids[ 3 0 R]/MediaBox[ 0 0 300 144]>>\nendobj\n3 0 obj<</Type/Page/Contents 4 0 R/Parent 2 0 R/Resources<</Font<</F1<</Type/Font/BaseFont/Times-Roman/Subtype/Type1>>>>>>>>\nendobj\n4 0 obj<</Length 55>>\nstream\n BT\n /F1 18 Tf\n 0 0 Td\n (Hello World) Tj\n ET\nendstream\nendobj\n5 0 obj<</Author(\xfe\xff\x00U\x00n\x00k\x00n\x00o\x00w\x00n)/CreationDate(D:20140919134038+05'00')/Producer(PoDoFo - http://podofo.sf.net)/Title(\xfe\xff\x00n\x00e\x00w\x00t)>>\nendobj\n6 0 obj<</Type/Metadata/Filter/FlateDecode/Length 584/Subtype/XML>>\nstream\nx\x9c\xed\x98\xcd\xb2\x930\x14\xc7\xf7}\n&.\x1d\x1ahoGa\x80\x8e\xb6\xe3x\x17ua\xaf\xe3\xd2\t\xc9i\x1b\x0b\x81&a\xc0\xfbj.|$_\xc1\xd0r\xe9\xb7V\x9d\xbb\x83\x15\x9c\x9c\xff\xff\x97\x8fs\xb2 \x18W9\xa1k\xd0V\x0cK.B\xf4\xf3\xfb\x0fdq\x16\xa2\xcf\xa3\x993\xcb'\xb0\xe2\xef\x1f%\xcc\x1f?<\xd0\xc75\xf5\x18\x1aG\xbd\xa0\xf2\xab4OA\x13\xabJ\x13\xa1\xfc*D\x84e1\xf8\xe6\xbd\x0ec\x14\xf5,+\x90l\xe1\x7f\x9c\xbek\x92\xccW\x88VZ\xe7>\xc6eY\xf6\xcba?\x93K\xecz\x9e\x87\x9d\x01\x1e\x0cl\x93a\xaboB\x93\xca\x16\xea\xc5\xd6\xa3q\x99\x82\xa2\x92\xe7\x9ag\xa2qc\xb45\xcb\x0b\x99l\xad\x18\xc5\x90@\nB+\xec\xf6]\x8c\xacZK\xe2\xac\xd0!j\xec\x8c!\xa3>\xdb\xfb=\x85\x1b\xd2\x9bD\xef#M,\xe15\xd4O\x88X\x86\xa8\xb2\x19,H\x91h\x14\x05x7z`\x81O<\x02|\x99VOBs\x9d\xc0\x7f\xe0\x05\x94\xfa\xd6)\x1c\xb1jx^\xc4\tW+\x90'\x13xK\x96\xf8Hy\x96X\xabU\x11\x7f\x05\xaa\xff\xa4=I\xab\x95T\x02\xd1\xd9)u\x0e\x9b\x0b\xcb\x8e>\x89\xb5\xc8Jqm\x91\x07\xaa-\xee\xc8{\x972=\xdd\xfa+\xe5d\xea\xb9\xad'\xa1\xfa\xdbj\xee\xd3,\xc5\x15\xc9M-9\xa6\x96\xdaD\xce6Wr\xd3\x1c\xdf3S~|\xc1A\xe2MA\x92F{\xb1\x0eM\xba?3\xdd\xc2\x88&S\xa2!\x1a8\xee\x9d\xedx\xb6\xeb=\xb8C\xff\xce\xf1\x87\xaf\xfb\xde\xe0\xd5\xc8\xf3^:#\x7f\xe8\x04\xf8L\xf2\x0fK\xcd%W\xe9\xbey\xea/\xa5\x89`D\xb2m\x17\t\x92\x822\xb7\x02(\x1c\x13\xc5)\x1e\x9c-\x01\xff\x1e\xc0\x16\xd5\xe5\r\xaaG\xcc\x8e\x0c\xff\xca\x8e\x92\x84\xc7\x12&\x93\xd6\xb3\x89\xd8\x10g\xd9\xfai\xe7\xedv\xde6-\x94\xceR\x9bfI\x91\n\x85\x8e}nu9\x91\xcd\xefo\xc6+\x90\x1c\x94\xcd\x05\x83\xea\xca\xd17\x16\xbb\xb6\xfc\xa22\xa9\x9bn\xbe0p\xfd\x88wAs\xc3\x9a+\x19\xb7w\xf2a#=\xdf\xd3A:H\x07\xe9 \x1d\xa4\x83t\x90\x0e\xd2A:H\x07yNH/h\x7f\xd6\x80`!*\xd18\xfa\x05\x94\x80P\xb0\nendstream\nendobj\nxref\n0 7\n0000000000 65535 f \n0000000015 00000 n \n0000000074 00000 n \n0000000148 00000 n \n0000000280 00000 n \n0000000382 00000 n \n0000000522 00000 n \ntrailer\n<</ID[<4D028D512DEBEFD964756764AD8FF726><4D028D512DEBEFD964756764AD8FF726>]/Info 5 0 R/Root 1 0 R/Size 7>>\nstartxref\n1199\n%%EOF\n" # noqa
# }}}
- mi = Metadata('title1', ['xmp_author'])
+ mi = Metadata('title1', ['author1'])
+ xmp_packet = metadata_to_xmp_packet(mi)
podofo = get_podofo()
p = podofo.PDFDoc()
p.load(raw)
- p.title = 'info title'
- p.author = 'info author'
- p.keywords = 'a, b'
- if p.version != '1.1':
- raise ValueError('Incorrect PDF version')
- xmp_packet = metadata_to_xmp_packet(mi)
- # print(p.get_xmp_metadata().decode())
+ p.title = mi.title
+ p.author = mi.authors[0]
p.set_xmp_metadata(xmp_packet)
- # print(p.get_xmp_metadata().decode())
+ buf = BytesIO()
+ p.save_to_fileobj(buf)
+ raw = buf.getvalue()
with tempfile.NamedTemporaryFile(suffix='.pdf', delete=False) as f:
- p.save_to_fileobj(f)
- f.seek(0)
- fraw = f.read()
- wraw = p.write()
- if fraw != wraw:
- raise ValueError("write() and save_to_fileobj() resulted in different output")
+ f.write(raw)
try:
p = podofo.PDFDoc()
p.open(f.name)
- if (p.title, p.author, p.keywords) != ('info title', 'info author', 'a, b'):
+ if (p.title, p.author) != (mi.title, mi.authors[0]):
raise ValueError('podofo failed to set title and author in Info dict {} != {}'.format(
- (p.title, p.author, p.keywords), ('info title', 'info author', 'a, b')))
- xmp = p.get_xmp_metadata().decode()
- if 'xmp_author' not in xmp:
- raise ValueError('Failed to set XML block, received:\n' + xmp)
+ (p.title, p.author), (mi.title, mi.authors[0])))
+ if not p.get_xmp_metadata():
+ raise ValueError('podofo failed to write XMP packet')
del p
finally:
os.remove(f.name)
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/outline.cpp calibre-6.20.0/src/calibre/utils/podofo/outline.cpp
--- calibre-6.20.0.orig/src/calibre/utils/podofo/outline.cpp 2023-06-08 18:36:58.000000000 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/outline.cpp 2023-06-11 11:20:17.617014510 -0700
@@ -6,7 +6,6 @@
*/
#include "global.h"
-#include <memory>
using namespace pdf;
@@ -46,36 +45,43 @@
static PyObject *
create(PDFOutlineItem *self, PyObject *args) {
PyObject *as_child;
- PDFOutlineItem *ans = NULL;
+ PDFOutlineItem *ans;
unsigned int num;
double left = 0, top = 0, zoom = 0;
+ PdfPage *page;
PyObject *title_buf;
if (!PyArg_ParseTuple(args, "UIO|ddd", &title_buf, &num, &as_child, &left, &top, &zoom)) return NULL;
ans = PyObject_New(PDFOutlineItem, &PDFOutlineItemType);
- if (ans == NULL) return NULL;
+ if (ans == NULL) goto error;
ans->doc = self->doc;
- pyunique_ptr decref_ans_on_exit((PyObject*)ans);
try {
PdfString title = podofo_convert_pystring(title_buf);
- const PdfPage *page = get_page(self->doc, num - 1);
- if (!page) { PyErr_Format(PyExc_ValueError, "Invalid page number: %u", num); return NULL; }
- auto dest = std::make_shared<PdfDestination>(*page, left, top, zoom);
+ try {
+ page = self->doc->GetPage(num - 1);
+ } catch(const PdfError &err) { (void)err; page = NULL; }
+ if (page == NULL) { PyErr_Format(PyExc_ValueError, "Invalid page number: %u", num); goto error; }
+ PdfDestination dest(page, left, top, zoom);
if (PyObject_IsTrue(as_child)) {
ans->item = self->item->CreateChild(title, dest);
} else
ans->item = self->item->CreateNext(title, dest);
} catch (const PdfError &err) {
- podofo_set_exception(err); return NULL;
+ podofo_set_exception(err); goto error;
} catch(const std::exception & err) {
- PyErr_Format(PyExc_ValueError, "An error occurred while trying to create the outline: %s", err.what()); return NULL;
+ PyErr_Format(PyExc_ValueError, "An error occurred while trying to create the outline: %s", err.what());
+ goto error;
} catch (...) {
- PyErr_SetString(PyExc_Exception, "An unknown error occurred while trying to create the outline item"); return NULL;
+ PyErr_SetString(PyExc_Exception, "An unknown error occurred while trying to create the outline item");
+ goto error;
}
- return (PyObject*) decref_ans_on_exit.release();
+ return (PyObject*) ans;
+error:
+ Py_XDECREF(ans);
+ return NULL;
}
static PyMethodDef methods[] = {
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/outlines.cpp calibre-6.20.0/src/calibre/utils/podofo/outlines.cpp
--- calibre-6.20.0.orig/src/calibre/utils/podofo/outlines.cpp 2023-06-08 18:36:58.000000000 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/outlines.cpp 2023-06-11 11:20:17.618014508 -0700
@@ -15,36 +15,43 @@
PyObject *title_buf;
unsigned int pagenum;
double left = 0, top = 0, zoom = 0;
+ PdfPage *page;
if (!PyArg_ParseTuple(args, "UI|ddd", &title_buf, &pagenum, &left, &top, &zoom)) return NULL;
ans = PyObject_New(PDFOutlineItem, &PDFOutlineItemType);
- if (ans == NULL) return NULL;
- pyunique_ptr decref_ans_on_exit((PyObject*)ans);
+ if (ans == NULL) goto error;
try {
PdfString title = podofo_convert_pystring(title_buf);
- PdfOutlines &outlines = self->doc->GetOrCreateOutlines();
- ans->item = outlines.CreateRoot(title);
- if (ans->item == NULL) {PyErr_NoMemory(); return NULL;}
+ PdfOutlines *outlines = self->doc->GetOutlines();
+ if (outlines == NULL) {PyErr_NoMemory(); goto error;}
+ ans->item = outlines->CreateRoot(title);
+ if (ans->item == NULL) {PyErr_NoMemory(); goto error;}
ans->doc = self->doc;
- auto page = get_page(self->doc, pagenum -1);
- if (!page) {
- PyErr_Format(PyExc_ValueError, "Invalid page number: %u", pagenum - 1); return NULL;
+ try {
+ page = self->doc->GetPage(pagenum - 1);
+ } catch (const PdfError &err) {
+ (void)err;
+ PyErr_Format(PyExc_ValueError, "Invalid page number: %u", pagenum - 1); goto error;
}
- auto dest = std::make_shared<PdfDestination>(*page, left, top, zoom);
+ PdfDestination dest(page, left, top, zoom);
ans->item->SetDestination(dest);
} catch(const PdfError & err) {
- podofo_set_exception(err); return NULL;
+ podofo_set_exception(err); goto error;
} catch(const std::exception & err) {
PyErr_Format(PyExc_ValueError, "An error occurred while trying to create the outline: %s", err.what());
- return NULL;
+ goto error;
} catch (...) {
PyErr_SetString(PyExc_ValueError, "An unknown error occurred while trying to create the outline");
- return NULL;
+ goto error;
}
- return decref_ans_on_exit.release();
+ return (PyObject*)ans;
+error:
+ Py_XDECREF(ans);
+ return NULL;
+
}
static PyObject*
@@ -64,9 +71,9 @@
pyunique_ptr node(create_outline_node());
if (!node) return;
if (PyDict_SetItemString(node.get(), "title", title.get()) != 0) return;
- auto dest = item->GetDestination();
+ PdfDestination* dest = item->GetDestination(self->doc);
if (dest) {
- PdfPage *page = dest->GetPage();
+ PdfPage *page = dest->GetPage(self->doc);
long pnum = page ? page->GetPageNumber() : -1;
pyunique_ptr d(Py_BuildValue("{sl sd sd sd}", "page", pnum, "top", dest->GetTop(), "left", dest->GetLeft(), "zoom", dest->GetZoom()));
if (!d) return;
@@ -88,7 +95,7 @@
static PyObject *
get_outline(PDFDoc *self, PyObject *args) {
- PdfOutlines *root = self->doc->GetOutlines();
+ PdfOutlines *root = self->doc->GetOutlines(PoDoFo::ePdfDontCreateObject);
if (!root || !root->First()) Py_RETURN_NONE;
PyObject *ans = create_outline_node();
if (!ans) return NULL;
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/output.cpp calibre-6.20.0/src/calibre/utils/podofo/output.cpp
--- calibre-6.20.0.orig/src/calibre/utils/podofo/output.cpp 2023-06-08 18:36:58.000000000 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/output.cpp 2023-06-11 11:20:17.618014508 -0700
@@ -10,10 +10,11 @@
using namespace PoDoFo;
#define NUKE(x) { Py_XDECREF(x); x = NULL; }
-#define PODOFO_RAISE_ERROR(code) throw ::PoDoFo::PdfError(code, __FILE__, __LINE__)
+class pyerr : public std::exception {
+};
-class MyOutputDevice : public OutputStreamDevice {
+class OutputDevice : public PdfOutputDevice {
private:
PyObject *tell_func;
@@ -25,21 +26,20 @@
void update_written() {
size_t pos;
- pos = GetPosition();
+ pos = Tell();
if (pos > written) written = pos;
}
public:
- MyOutputDevice(PyObject *file) : tell_func(0), seek_func(0), read_func(0), write_func(0), flush_func(0), written(0) {
- SetAccess(DeviceAccess::Write);
-#define GA(f, a) { if((f = PyObject_GetAttrString(file, a)) == NULL) throw std::exception(); }
+ OutputDevice(PyObject *file) : tell_func(0), seek_func(0), read_func(0), write_func(0), flush_func(0), written(0) {
+#define GA(f, a) { if((f = PyObject_GetAttrString(file, a)) == NULL) throw pyerr(); }
GA(tell_func, "tell");
GA(seek_func, "seek");
GA(read_func, "read");
GA(write_func, "write");
GA(flush_func, "flush");
}
- ~MyOutputDevice() {
+ ~OutputDevice() {
NUKE(tell_func); NUKE(seek_func); NUKE(read_func); NUKE(write_func); NUKE(flush_func);
}
@@ -47,7 +47,7 @@
long PrintVLen(const char* pszFormat, va_list args) {
- if( !pszFormat ) { PODOFO_RAISE_ERROR(PdfErrorCode::InvalidHandle); }
+ if( !pszFormat ) { PODOFO_RAISE_ERROR( ePdfError_InvalidHandle ); }
#ifdef _MSC_VER
return _vscprintf(pszFormat, args) + 1;
@@ -60,10 +60,10 @@
char *buf;
int res;
- if( !pszFormat ) { PODOFO_RAISE_ERROR(PdfErrorCode::InvalidHandle); }
+ if( !pszFormat ) { PODOFO_RAISE_ERROR( ePdfError_InvalidHandle ); }
buf = new (std::nothrow) char[lBytes+1];
- if (buf == NULL) { PyErr_NoMemory(); throw std::exception(); }
+ if (buf == NULL) { PyErr_NoMemory(); throw pyerr(); }
// Note: PyOS_vsnprintf produces broken output on windows
res = vsnprintf(buf, lBytes, pszFormat, args);
@@ -71,7 +71,7 @@
if (res < 0) {
PyErr_SetString(PyExc_Exception, "Something bad happened while calling vsnprintf");
delete[] buf;
- throw std::exception();
+ throw pyerr();
}
Write(buf, static_cast<size_t>(res));
@@ -97,7 +97,7 @@
char *buf = NULL;
Py_ssize_t len = 0;
- if ((temp = PyLong_FromSize_t(lLen)) == NULL) throw std::exception();
+ if ((temp = PyLong_FromSize_t(lLen)) == NULL) throw pyerr();
ret = PyObject_CallFunctionObjArgs(read_func, temp, NULL);
NUKE(temp);
if (ret != NULL) {
@@ -112,24 +112,24 @@
if (PyErr_Occurred() == NULL)
PyErr_SetString(PyExc_Exception, "Failed to read data from python file object");
- throw std::exception();
+ throw pyerr();
}
void Seek(size_t offset) {
PyObject *ret, *temp;
- if ((temp = PyLong_FromSize_t(offset)) == NULL) throw std::exception();
+ if ((temp = PyLong_FromSize_t(offset)) == NULL) throw pyerr();
ret = PyObject_CallFunctionObjArgs(seek_func, temp, NULL);
NUKE(temp);
if (ret == NULL) {
if (PyErr_Occurred() == NULL)
PyErr_SetString(PyExc_Exception, "Failed to seek in python file object");
- throw std::exception();
+ throw pyerr();
}
Py_DECREF(ret);
}
- size_t GetPosition() const {
+ size_t Tell() const {
PyObject *ret;
unsigned long ans;
@@ -137,27 +137,25 @@
if (ret == NULL) {
if (PyErr_Occurred() == NULL)
PyErr_SetString(PyExc_Exception, "Failed to call tell() on python file object");
- throw std::exception();
+ throw pyerr();
}
if (!PyNumber_Check(ret)) {
Py_DECREF(ret);
PyErr_SetString(PyExc_Exception, "tell() method did not return a number");
- throw std::exception();
+ throw pyerr();
}
ans = PyLong_AsUnsignedLongMask(ret);
Py_DECREF(ret);
- if (PyErr_Occurred() != NULL) throw std::exception();
+ if (PyErr_Occurred() != NULL) throw pyerr();
return static_cast<size_t>(ans);
}
- bool Eof() const { return false; }
-
- void writeBuffer(const char* pBuffer, size_t lLen) {
+ void Write(const char* pBuffer, size_t lLen) {
PyObject *ret, *temp = NULL;
temp = PyBytes_FromStringAndSize(pBuffer, static_cast<Py_ssize_t>(lLen));
- if (temp == NULL) throw std::exception();
+ if (temp == NULL) throw pyerr();
ret = PyObject_CallFunctionObjArgs(write_func, temp, NULL);
NUKE(temp);
@@ -165,7 +163,7 @@
if (ret == NULL) {
if (PyErr_Occurred() == NULL)
PyErr_SetString(PyExc_Exception, "Failed to call write() on python file object");
- throw std::exception();
+ throw pyerr();
}
Py_DECREF(ret);
update_written();
@@ -179,11 +177,10 @@
PyObject* pdf::write_doc(PdfMemDocument *doc, PyObject *f) {
- MyOutputDevice d(f);
+ OutputDevice d(f);
try {
- doc->Save(d, save_options);
- d.Flush();
+ doc->Write(&d);
} catch(const PdfError & err) {
podofo_set_exception(err); return NULL;
} catch (...) {
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/podofo.cpp calibre-6.20.0/src/calibre/utils/podofo/podofo.cpp
--- calibre-6.20.0.orig/src/calibre/utils/podofo/podofo.cpp 2023-06-08 18:36:58.000000000 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/podofo.cpp 2023-06-11 11:20:17.618014508 -0700
@@ -7,19 +7,34 @@
using namespace PoDoFo;
#include "global.h"
-#include <iostream>
PyObject *pdf::Error = NULL;
-static char podofo_doc[] = "Wrapper for the PoDoFo PDF library";
+class PyLogMessage : public PdfError::LogMessageCallback {
-static void
-pdf_log_message(PdfLogSeverity logSeverity, const std::string_view& msg) {
- if (logSeverity == PdfLogSeverity::Error || logSeverity == PdfLogSeverity::Warning) {
- const char *level = logSeverity == PdfLogSeverity::Error ? "ERROR" : "WARNING";
- std::cerr << "PoDoFo" << level << ": " << msg << std::endl;
- }
-}
+ public:
+ ~PyLogMessage() {}
+
+ void LogMessage(ELogSeverity severity, const char* prefix, const char* msg, va_list & args ) {
+ if (severity > eLogSeverity_Warning) return;
+ if (prefix)
+ fprintf(stderr, "%s", prefix);
+
+ vfprintf(stderr, msg, args);
+ }
+
+ void LogMessage(ELogSeverity severity, const wchar_t* prefix, const wchar_t* msg, va_list & args ) {
+ if (severity > eLogSeverity_Warning) return;
+ if (prefix)
+ fwprintf(stderr, prefix);
+
+ vfwprintf(stderr, msg, args);
+ }
+};
+
+PyLogMessage log_message;
+
+static char podofo_doc[] = "Wrapper for the PoDoFo PDF library";
static int
exec_module(PyObject *m) {
@@ -30,10 +45,11 @@
if (pdf::Error == NULL) return -1;
PyModule_AddObject(m, "Error", pdf::Error);
+ PdfError::SetLogMessageCallback((PdfError::LogMessageCallback*)&log_message);
+ PdfError::EnableDebug(false);
+
Py_INCREF(&pdf::PDFDocType);
PyModule_AddObject(m, "PDFDoc", (PyObject *)&pdf::PDFDocType);
-
- PdfCommon::SetLogMessageCallback(pdf_log_message);
return 0;
}
diff --color -Nur calibre-6.20.0.orig/src/calibre/utils/podofo/utils.cpp calibre-6.20.0/src/calibre/utils/podofo/utils.cpp
--- calibre-6.20.0.orig/src/calibre/utils/podofo/utils.cpp 2023-06-08 18:36:58.000000000 -0700
+++ calibre-6.20.0/src/calibre/utils/podofo/utils.cpp 2023-06-11 11:20:17.619014506 -0700
@@ -6,33 +6,29 @@
*/
#include "global.h"
-#include <sstream>
-#include <stdexcept>
-#include <string_view>
using namespace pdf;
void
pdf::podofo_set_exception(const PdfError &err) {
- const char *msg = err.what();
+ const char *msg = PdfError::ErrorMessage(err.GetError());
+ if (msg == NULL) msg = err.what();
std::stringstream stream;
stream << msg << "\n";
- const PdErrorInfoStack &s = err.GetCallStack();
- for (auto info : s) {
- stream << "File: " << info.GetFilePath() << " Line: " << info.GetLine() << " " << info.GetInformation() << "\n";
+ const TDequeErrorInfo &s = err.GetCallstack();
+ for (TDequeErrorInfo::const_iterator it = s.begin(); it != s.end(); it++) {
+ const PdfErrorInfo &info = (*it);
+ stream << "File: " << info.GetFilename() << " Line: " << info.GetLine() << " " << info.GetInformation() << "\n";
}
PyErr_SetString(Error, stream.str().c_str());
}
PyObject *
pdf::podofo_convert_pdfstring(const PdfString &s) {
- return PyUnicode_FromString(s.GetString().c_str());
+ return PyUnicode_FromString(s.GetStringUtf8().c_str());
}
const PdfString
pdf::podofo_convert_pystring(PyObject *val) {
- Py_ssize_t len;
- const char *data = PyUnicode_AsUTF8AndSize(val, &len);
- if (data == NULL) throw std::runtime_error("Failed to convert python string to UTF-8, possibly not a string object");
- return PdfString(std::string_view(data, len));
+ return PdfString(reinterpret_cast<const pdf_utf8*>(PyUnicode_AsUTF8(val)));
}
