When listing fonts check if they are used

src/calibre/utils/podofo/__init__.py

@@ -142,6 +142,18 @@ def get_image_count(path):
     return p.image_count()
 
 
+def list_fonts(pdf_doc):
+    fonts = pdf_doc.list_fonts()
+    ref_map = {f['Reference']: f for f in fonts}
+    for ref in pdf_doc.used_fonts_in_page_range():
+        ref_map[ref]['used'] = True
+    for font in fonts:
+        font['used'] = font.get('used', False)
+        if font['DescendantFont'] and font['used']:
+            ref_map[font['DescendantFont']]['used'] = True
+    return ref_map
+
+
 def test_list_fonts(src):
     podofo = get_podofo()
     p = podofo.PDFDoc()
@@ -149,7 +161,7 @@ def test_list_fonts(src):
         raw = f.read()
     p.load(raw)
     import pprint
-    pprint.pprint(p.list_fonts())
+    pprint.pprint(list_fonts(p))
 
 
 def test_save_to(src, dest):

src/calibre/utils/podofo/doc.cpp

@@ -718,6 +718,9 @@ static PyMethodDef PDFDoc_methods[] = {
     {"list_fonts", (PyCFunction)list_fonts, METH_VARARGS,
      "list_fonts() -> Get list of fonts in document"
     },
+    {"used_fonts_in_page_range", (PyCFunction)used_fonts_in_page_range, METH_VARARGS,
+     "used_fonts_in_page_range() -> Get list of references to fonts used in the specified pages"
+    },
     {"delete_page", (PyCFunction)PDFDoc_delete_page, METH_VARARGS,
      "delete_page(page_num) -> Delete the specified page from the pdf (0 is the first page)."
     },

src/calibre/utils/podofo/fonts.cpp

@@ -11,9 +11,16 @@
 
 using namespace pdf;
 
+static inline PyObject*
+ref_as_tuple(const PdfReference &ref) {
+    unsigned long num = ref.ObjectNumber(), generation = ref.GenerationNumber();
+    return Py_BuildValue("kk", num, generation);
+}
+
+
 static bool
-used_fonts_in_page(const PdfPage *page, PyObject *ans) {
-    PdfContentsTokenizer tokenizer((PdfCanvas*)page);
+used_fonts_in_page(PdfPage *page, PyObject *ans) {
+    PdfContentsTokenizer tokenizer(page);
     bool in_text_block = false;
     const char* token = NULL;
     EPdfContentsType contents_type;
@@ -35,11 +42,9 @@ used_fonts_in_page(const PdfPage *page, PyObject *ans) {
             stack.pop();
             if (stack.size() > 0 && stack.top().IsName()) {
                 const PdfName &reference_name = stack.top().GetName();
-                PdfObject* font = pPage->GetFromResources("Font", reference_name);
+                PdfObject* font = page->GetFromResources("Font", reference_name);
                 if (font) {
-                    const PdfReference &ref = font->Reference();
-                    unsigned long num = ref.ObjectNumber(), generation = ref.GenerationNumber();
-                    pyunique_ptr r(Py_BuildValue("kk", num, generation));
+                    pyunique_ptr r(ref_as_tuple(font->Reference()));
                     if (!r) return false;
                     if (PySet_Add(ans, r.get()) != 0) return false;
                 }
@@ -66,19 +71,33 @@ list_fonts(PDFDoc *self, PyObject *args) {
                     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;
                     if (descriptor) {
                         const PdfObject *ff = descriptor->GetIndirectKey("FontFile");
                         if (!ff) ff = descriptor->GetIndirectKey("FontFile2");
                         if (!ff) ff = descriptor->GetIndirectKey("FontFile3");
+                        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();
                         }
+                    } else if (dict.HasKey("DescendantFonts")) {
+                        const PdfArray &df = dict.GetKey("DescendantFonts")->GetArray();
+                        descendant_font.reset(ref_as_tuple(df[0].GetReference()));
+                        if (!descendant_font) return NULL;
+                    }
+#define V(x) (x ? x.get() : Py_None)
                     pyunique_ptr d(Py_BuildValue(
-                            "{sssss(kk)sL}",
+                            "{ss ss s(kk) sL sO sO}",
                             "BaseFont", name.c_str(),
                             "Subtype", subtype.c_str(),
                             "Reference", num, generation,
-                            "Length", stream_len));
+                            "Length", stream_len,
+                            "DescendantFont", V(descendant_font),
+                            "StreamRef", V(stream_ref)
+                    ));
+#undef V
                     if (!d) { return NULL; }
                     if (PyList_Append(ans.get(), d.get()) != 0) return NULL;
                 }
@@ -99,7 +118,7 @@ used_fonts_in_page_range(PDFDoc *self, PyObject *args) {
     if (!ans) return NULL;
     for (int i = first - 1; i < last; i++) {
         try {
-            const PdfPage *page = self->doc->GetPage(i);
+            PdfPage *page = self->doc->GetPage(i);
             if (!used_fonts_in_page(page, ans.get())) return NULL;
         } catch (const PdfError &err) { continue; }
     }

src/calibre/utils/podofo/global.h

@@ -60,5 +60,6 @@ dictionary_has_key_name(const PdfDictionary &d, T key, const char *name) {
 
 extern "C" {
 PyObject* list_fonts(PDFDoc*, PyObject*);
+PyObject* used_fonts_in_page_range(PDFDoc *self, PyObject *args);
 }
 }