From 4017e381a0116bf458511dd933ee8f7b4b26c045 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Mon, 28 May 2012 20:30:29 +0530
Subject: [PATCH] Add ability to extract pages from PDF files to calibre podofo
 bindings

---
 src/calibre/utils/podofo/__init__.py | 31 +++++++++++++------------
 src/calibre/utils/podofo/podofo.cpp  | 34 ++++++++++++++++++++++++++++
 2 files changed, 51 insertions(+), 14 deletions(-)

diff --git a/src/calibre/utils/podofo/__init__.py b/src/calibre/utils/podofo/__init__.py
index 78f250dd0e..948962f438 100644
--- a/src/calibre/utils/podofo/__init__.py
+++ b/src/calibre/utils/podofo/__init__.py
@@ -181,20 +181,23 @@ def set_metadata_(path, opath, title, authors, bkp, tags):
         return True
     return False
 
+def delete_all_but(path, pages):
+    ''' Delete all the pages in the pdf except for the specified ones. Negative
+    numbers are counted from the end of the PDF.'''
+    with TemporaryFile('_podofo_in.pdf') as of:
+        shutil.copyfile(path, of)
+
+        p = podofo.PDFDoc()
+        p.open(of)
+        total = p.page_count()
+        pages = { total + x if x < 0 else x for x in pages }
+        for page in xrange(total-1, -1, -1):
+            if page not in pages:
+                p.delete_page(page)
+        os.remove(path)
+        p.save(path)
+
 if __name__ == '__main__':
     f = '/tmp/t.pdf'
-    import StringIO
-    stream = StringIO.StringIO(open(f).read())
-    mi = get_metadata(open(f))
-    print
-    print 'Original metadata:'
-    print mi
-    mi.title = 'Test title'
-    mi.authors = ['Test author', 'author2']
-    mi.book_producer = 'calibre'
-    set_metadata(stream, mi)
-    open('/tmp/x.pdf', 'wb').write(stream.getvalue())
-    print
-    print 'New pdf written to /tmp/x.pdf'
-
+    delete_all_but(f, [0, 1, -2, -1])
 
diff --git a/src/calibre/utils/podofo/podofo.cpp b/src/calibre/utils/podofo/podofo.cpp
index c1f9f84f61..3764fce829 100644
--- a/src/calibre/utils/podofo/podofo.cpp
+++ b/src/calibre/utils/podofo/podofo.cpp
@@ -146,6 +146,33 @@ podofo_PDFDoc_extract_first_page(podofo_PDFDoc *self, PyObject *args, PyObject *
     Py_RETURN_NONE;
 }
 
+static PyObject *
+podofo_PDFDoc_page_count(podofo_PDFDoc *self, PyObject *args, PyObject *kwargs) {
+    int count;
+    try {
+        count = self->doc->GetPageCount();
+    } catch(const PdfError & err) {
+        podofo_set_exception(err);
+        return NULL;
+    }
+    return Py_BuildValue("i", count);
+}
+
+static PyObject *
+podofo_PDFDoc_delete_page(podofo_PDFDoc *self, PyObject *args, PyObject *kwargs) {
+    int num = 0;
+    if (PyArg_ParseTuple(args, "i", &num)) {
+        try {
+            self->doc->DeletePages(num, 1);
+        } catch(const PdfError & err) {
+            podofo_set_exception(err);
+            return NULL;
+        }
+    } else return NULL;
+
+    Py_RETURN_NONE;
+}
+
 static PyObject *
 podofo_convert_pdfstring(const PdfString &s) {
     std::string raw = s.GetStringUtf8();
@@ -321,6 +348,13 @@ static PyMethodDef podofo_PDFDoc_methods[] = {
     {"extract_first_page", (PyCFunction)podofo_PDFDoc_extract_first_page, METH_VARARGS,
      "extract_first_page() -> Remove all but the first page."
     },
+    {"page_count", (PyCFunction)podofo_PDFDoc_page_count, METH_VARARGS,
+     "page_count() -> Number of pages in the PDF."
+    },
+    {"delete_page", (PyCFunction)podofo_PDFDoc_delete_page, METH_VARARGS,
+     "delete_page(page_num) -> Delete the specified page from the pdf (0 is the first page)."
+    },
+
 
     {NULL}  /* Sentinel */
 };