Code to list fonts in a PDF file

setup/extensions.json

@@ -116,7 +116,7 @@
     },
     {
         "name": "podofo",
-        "sources": "calibre/utils/podofo/utils.cpp calibre/utils/podofo/output.cpp calibre/utils/podofo/doc.cpp calibre/utils/podofo/outline.cpp calibre/utils/podofo/podofo.cpp",
+        "sources": "calibre/utils/podofo/utils.cpp calibre/utils/podofo/output.cpp calibre/utils/podofo/doc.cpp calibre/utils/podofo/outline.cpp calibre/utils/podofo/fonts.cpp calibre/utils/podofo/podofo.cpp",
         "headers": "calibre/utils/podofo/global.h",
         "libraries": "podofo",
         "lib_dirs": "!podofo_lib",

src/calibre/utils/podofo/__init__.py

@@ -142,21 +142,14 @@ def get_image_count(path):
     return p.image_count()
 
 
-def test_outline(src):
+def test_list_fonts(src):
     podofo = get_podofo()
     p = podofo.PDFDoc()
     with open(src, 'rb') as f:
         raw = f.read()
     p.load(raw)
-    total = p.page_count()
-    root = p.create_outline('Table of Contents')
-    for i in range(0, total):
-        root.create('Page %d'%i, i, True)
-    raw = p.write()
-    out = '/tmp/outlined.pdf'
-    with open(out, 'wb') as f:
-        f.write(raw)
-    print('Outlined PDF:', out)
+    import pprint
+    pprint.pprint(p.list_fonts())
 
 
 def test_save_to(src, dest):

src/calibre/utils/podofo/doc.cpp

@@ -429,15 +429,6 @@ PDFDoc_extract_anchors(PDFDoc *self, PyObject *args) {
 
 // alter_links() {{{
 
-template<typename T>
-static inline bool
-dictionary_has_key_name(PdfDictionary &d, T key, const char *name) {
-	const PdfObject *val = d.GetKey(key);
-	if (val && val->IsName() && val->GetName().GetName() == name) return true;
-	return false;
-}
-
-
 static PyObject *
 PDFDoc_alter_links(PDFDoc *self, PyObject *args) {
     int count = 0;
@@ -724,6 +715,9 @@ static PyMethodDef PDFDoc_methods[] = {
     {"alter_links", (PyCFunction)PDFDoc_alter_links, METH_VARARGS,
      "alter_links() -> Change links in the document."
     },
+    {"list_fonts", (PyCFunction)list_fonts, METH_VARARGS,
+     "list_fonts() -> Get list of fonts in document"
+    },
     {"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

@@ -0,0 +1,55 @@
+/*
+ * fonts.cpp
+ * Copyright (C) 2019 Kovid Goyal <kovid at kovidgoyal.net>
+ *
+ * Distributed under terms of the GPL3 license.
+ */
+
+#include "global.h"
+#include <iostream>
+
+using namespace pdf;
+using namespace std;
+
+extern "C" {
+PyObject*
+list_fonts(PDFDoc *self, PyObject *args) {
+    pyunique_ptr ans(PyList_New(0));
+    if (!ans) return NULL;
+    try {
+        const PdfVecObjects &objects = self->doc->GetObjects();
+        for (TCIVecObjects it = objects.begin(); it != objects.end(); it++) {
+            if ((*it)->IsDictionary()) {
+                const PdfDictionary &dict = (*it)->GetDictionary();
+                if (dictionary_has_key_name(dict, PdfName::KeyType, "Font") && dict.HasKey("BaseFont")) {
+                    const string &name = dict.GetKey("BaseFont")->GetName().GetName();
+                    const string &subtype = dict.GetKey(PdfName::KeySubtype)->GetName().GetName();
+                    const PdfReference &ref = (*it)->Reference();
+                    unsigned long num = ref.ObjectNumber(), generation = ref.GenerationNumber();
+                    const PdfObject *descriptor = (*it)->GetIndirectKey("FontDescriptor");
+                    long long stream_len = 0;
+                    if (descriptor) {
+                        const PdfObject *ff = descriptor->GetIndirectKey("FontFile");
+                        if (!ff) ff = descriptor->GetIndirectKey("FontFile2");
+                        if (!ff) ff = descriptor->GetIndirectKey("FontFile3");
+                        const PdfStream *stream = ff->GetStream();
+                        if (stream) stream_len = stream->GetLength();
+                    }
+                    pyunique_ptr d(Py_BuildValue(
+                            "{sssss(kk)sL}",
+                            "BaseFont", name.c_str(),
+                            "Subtype", subtype.c_str(),
+                            "Reference", num, generation,
+                            "Length", stream_len));
+                    if (!d) { return NULL; }
+                    if (PyList_Append(ans.get(), d.get()) != 0) return NULL;
+                }
+            }
+        }
+    } catch (const PdfError &err) {
+        podofo_set_exception(err);
+        return NULL;
+    }
+    return ans.release();
+}
+}

src/calibre/utils/podofo/global.h

@@ -41,4 +41,24 @@ void podofo_set_exception(const PdfError &err);
 PyObject * podofo_convert_pdfstring(const PdfString &s);
 const PdfString podofo_convert_pystring(PyObject *py);
 PyObject* write_doc(PdfMemDocument *doc, PyObject *f);
+
+struct PyObjectDeleter {
+    void operator()(PyObject *obj) {
+        Py_XDECREF(obj);
+    }
+};
+// unique_ptr that uses Py_XDECREF as the destructor function.
+typedef std::unique_ptr<PyObject, PyObjectDeleter> pyunique_ptr;
+
+template<typename T>
+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().GetName() == name) return true;
+	return false;
+}
+
+extern "C" {
+PyObject* list_fonts(PDFDoc*, PyObject*);
+}
 }