Add a function to get the outline from a PDF file

src/calibre/utils/podofo/__init__.py

@@ -5,6 +5,7 @@ from __future__ import absolute_import, division, print_function, unicode_litera
 
 import os
 import shutil
+import sys
 
 from calibre.constants import plugins, preferred_encoding
 from calibre.ebooks.metadata import authors_to_string
@@ -115,6 +116,17 @@ def get_xmp_metadata(path):
     return p.get_xmp_metadata()
 
 
+def get_outline(path=None):
+    if path is None:
+        path = sys.argv[-1]
+    podofo = get_podofo()
+    p = podofo.PDFDoc()
+    with open(path, 'rb') as f:
+        raw = f.read()
+    p.load(raw)
+    return p.get_outline()['children']
+
+
 def get_image_count(path):
     podofo = get_podofo()
     p = podofo.PDFDoc()
@@ -214,5 +226,4 @@ def test_podofo():
 
 
 if __name__ == '__main__':
-    import sys
     get_xmp_metadata(sys.argv[-1])

src/calibre/utils/podofo/doc.cpp

@@ -752,6 +752,9 @@ static PyMethodDef PDFDoc_methods[] = {
     {"create_outline", (PyCFunction)py_create_outline, METH_VARARGS,
      "create_outline(title, pagenum) -> Create an outline, return the first outline item."
     },
+    {"get_outline", (PyCFunction)py_get_outline, METH_NOARGS,
+     "get_outline() -> Get the outline if any in the PDF file."
+    },
     {"get_xmp_metadata", (PyCFunction)PDFDoc_get_xmp_metadata, METH_VARARGS,
      "get_xmp_metadata(raw) -> Get the XMP metadata as raw bytes"
     },

src/calibre/utils/podofo/global.h

@@ -103,6 +103,7 @@ PyObject* py_dedup_type3_fonts(PDFDoc *self, PyObject *args);
 PyObject* py_impose(PDFDoc *self, PyObject *args);
 PyObject* py_dedup_images(PDFDoc *self, PyObject *args);
 PyObject* py_create_outline(PDFDoc *self, PyObject *args);
+PyObject* py_get_outline(PDFDoc *self, PyObject *args);
 }
 }
 

src/calibre/utils/podofo/outlines.cpp

@@ -9,7 +9,6 @@
 using namespace pdf;
 
 
-// create_outline() {{{
 static PyObject *
 create_outline(PDFDoc *self, PyObject *args) {
     PDFOutlineItem *ans;
@@ -53,6 +52,59 @@ error:
     Py_XDECREF(ans);
     return NULL;
 
-} // }}}
+}
+
+static PyObject*
+create_outline_node() {
+	pyunique_ptr ans(PyDict_New());
+	if (!ans) return NULL;
+	pyunique_ptr children(PyList_New(0));
+	if (!children) return NULL;
+	if (PyDict_SetItemString(ans.get(), "children", children.get()) != 0) return NULL;
+	return ans.release();
+}
+
+static void
+convert_outline(PDFDoc *self, PyObject *parent, PdfOutlineItem *item) {
+	pyunique_ptr title(podofo_convert_pdfstring(item->GetTitle()));
+	if (!title) return;
+	pyunique_ptr node(create_outline_node());
+	if (!node) return;
+	if (PyDict_SetItemString(node.get(), "title", title.get()) != 0) return;
+	PdfDestination* dest = item->GetDestination(self->doc);
+	if (dest) {
+		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;
+		if (PyDict_SetItemString(node.get(), "dest", d.get()) != 0) return;
+	}
+	PyObject *children = PyDict_GetItemString(parent, "children");
+	if (PyList_Append(children, node.get()) != 0) return;
+
+	if (item->First()) {
+		convert_outline(self, node.get(), item->First());
+		if (PyErr_Occurred()) return;
+	}
+
+	if (item->Next()) {
+		convert_outline(self, parent, item->Next());
+		if (PyErr_Occurred()) return;
+	}
+}
+
+static PyObject *
+get_outline(PDFDoc *self, PyObject *args) {
+	PdfOutlines *root = self->doc->GetOutlines(PoDoFo::ePdfDontCreateObject);
+	if (!root || !root->First()) Py_RETURN_NONE;
+	PyObject *ans = create_outline_node();
+	if (!ans) return NULL;
+	convert_outline(self, ans, root->First());
+	if (PyErr_Occurred()) { Py_DECREF(ans); return NULL; }
+	if (!ans) return NULL;
+
+	return ans;
+}
 
 PYWRAP(create_outline)
+PYWRAP(get_outline)