From 929f65ecf2fdb44fa2341ce02ff4513b8dfaa5c5 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Wed, 17 Jul 2019 14:43:56 +0530
Subject: [PATCH] Optionally get font data when listing fonts

---
 src/calibre/utils/podofo/fonts.cpp | 20 ++++++++++++++------
 src/calibre/utils/podofo/global.h  | 26 ++++++++++++++++++++++++++
 2 files changed, 40 insertions(+), 6 deletions(-)

diff --git a/src/calibre/utils/podofo/fonts.cpp b/src/calibre/utils/podofo/fonts.cpp
index 62b125b204..bd322b8e73 100644
--- a/src/calibre/utils/podofo/fonts.cpp
+++ b/src/calibre/utils/podofo/fonts.cpp
@@ -75,6 +75,8 @@ used_fonts_in_page(PdfPage *page, int page_num, PyObject *ans) {
 extern "C" {
 PyObject*
 list_fonts(PDFDoc *self, PyObject *args) {
+    int get_font_data = 0;
+    if (!PyArg_ParseTuple(args, "|i", &get_font_data)) return NULL;
     pyunique_ptr ans(PyList_New(0));
     if (!ans) return NULL;
     try {
@@ -88,15 +90,20 @@ list_fonts(PDFDoc *self, PyObject *args) {
                     const PdfReference &ref = (*it)->Reference();
                     unsigned long num = ref.ObjectNumber(), generation = ref.GenerationNumber();
                     const PdfObject *descriptor = (*it)->GetIndirectKey("FontDescriptor");
-                    long long stream_len = 0;
-                    pyunique_ptr descendant_font, stream_ref;
+                    pyunique_ptr descendant_font, stream_ref, encoding;
+                    PyBytesOutputStream stream_data;
+                    if (dict.HasKey("Encoding") && dict.GetKey("Encoding")->IsName()) {
+                        encoding.reset(PyUnicode_FromString(dict.GetKey("Encoding")->GetName().GetName().c_str()));
+                    }
                     if (descriptor) {
                         const PdfObject *ff = get_font_file(descriptor);
                         if (ff) {
                             stream_ref.reset(ref_as_tuple(ff->Reference()));
                             if (!stream_ref) return NULL;
                             const PdfStream *stream = ff->GetStream();
-                            if (stream) stream_len = stream->GetLength();
+                            if (stream && get_font_data) {
+                                stream->GetCopy(&stream_data);
+                            }
                         }
                     } else if (dict.HasKey("DescendantFonts")) {
                         const PdfArray &df = dict.GetKey("DescendantFonts")->GetArray();
@@ -105,13 +112,14 @@ list_fonts(PDFDoc *self, PyObject *args) {
                     }
 #define V(x) (x ? x.get() : Py_None)
                     pyunique_ptr d(Py_BuildValue(
-                            "{ss ss s(kk) sL sO sO}",
+                            "{ss ss s(kk) sO sO sO sO}",
                             "BaseFont", name.c_str(),
                             "Subtype", subtype.c_str(),
                             "Reference", num, generation,
-                            "Length", stream_len,
+                            "Data", V(stream_data),
                             "DescendantFont", V(descendant_font),
-                            "StreamRef", V(stream_ref)
+                            "StreamRef", V(stream_ref),
+                            "Encoding", V(encoding)
                     ));
 #undef V
                     if (!d) { return NULL; }
diff --git a/src/calibre/utils/podofo/global.h b/src/calibre/utils/podofo/global.h
index ee69492a05..2063d67ed6 100644
--- a/src/calibre/utils/podofo/global.h
+++ b/src/calibre/utils/podofo/global.h
@@ -50,6 +50,32 @@ struct PyObjectDeleter {
 // unique_ptr that uses Py_XDECREF as the destructor function.
 typedef std::unique_ptr<PyObject, PyObjectDeleter> pyunique_ptr;
 
+class PyBytesOutputStream : public PdfOutputStream {
+    private:
+        pyunique_ptr bytes;
+		PyBytesOutputStream( const PyBytesOutputStream & ) ;
+		PyBytesOutputStream & operator=( const PyBytesOutputStream & ) ;
+    public:
+        PyBytesOutputStream() : bytes() {}
+        void Close() {}
+        operator bool() const { return bool(bytes); }
+        PyObject* get() const { return bytes.get(); }
+        pdf_long Write(const char *buf, const pdf_long sz){
+            if (!bytes) {
+                bytes.reset(PyBytes_FromStringAndSize(buf, sz));
+                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(ePdfError_OutOfMemory, __FILE__, __LINE__, NULL);
+                memcpy(PyBytes_AS_STRING(old) + old_sz, buf, sz);
+                bytes.reset(old);
+            }
+            return sz;
+        }
+};
+
+
 template<typename T>
 static inline bool
 dictionary_has_key_name(const PdfDictionary &d, T key, const char *name) {