Speed up PDF cover extraction

src/calibre/ebooks/metadata/pdf.py

@@ -18,8 +18,7 @@ except:
 from calibre.ebooks.metadata import MetaInformation, authors_to_string
 from calibre.utils.pdftk import set_metadata as pdftk_set_metadata
 from calibre.utils.podofo import get_metadata as podofo_get_metadata, \
-    set_metadata as podofo_set_metadata, Unavailable, write_first_page, \
-    get_metadata_quick
+    set_metadata as podofo_set_metadata, Unavailable, get_metadata_quick
 
 def get_quick_metadata(stream):
     raw = stream.read()
@@ -32,19 +31,18 @@ def get_quick_metadata(stream):
 
 def get_metadata(stream, extract_cover=True):
     try:
-        mi = podofo_get_metadata(stream)
+        with TemporaryDirectory('_pdfmeta') as tdir:
+            cpath = os.path.join(tdir, 'cover.pdf')
+            if not extract_cover:
+                cpath = None
+            mi = podofo_get_metadata(stream, cpath=cpath)
+            if mi.cover is not None:
+                cdata = get_cover(mi.cover)
+                mi.cover = None
+                if cdata is not None:
+                    mi.cover_data = ('jpg', cdata)
     except Unavailable:
         mi = get_metadata_pypdf(stream)
-    stream.seek(0)
-
-    if extract_cover and _imagemagick_loaded:
-        try:
-            cdata = get_cover(stream)
-            if cdata is not None:
-                mi.cover_data = ('jpg', cdata)
-        except:
-            import traceback
-            traceback.print_exc()
     return mi
 
 
@@ -127,17 +125,13 @@ def set_metadata_pypdf(stream, mi):
     stream.write(out_str.read())
     stream.seek(0)
 
-def get_cover(stream):
-    stream.seek(0)
-    with TemporaryDirectory('_pdfmeta') as tdir:
-        cover_path = os.path.join(tdir, 'cover.pdf')
-        write_first_page(stream, cover_path)
-        with ImageMagick():
-            wand = NewMagickWand()
-            MagickReadImage(wand, cover_path)
-            MagickSetImageFormat(wand, 'JPEG')
-            MagickWriteImage(wand, '%s.jpg' % cover_path)
-        return open('%s.jpg' % cover_path, 'rb').read()
+def get_cover(cover_path):
+    with ImageMagick():
+        wand = NewMagickWand()
+        MagickReadImage(wand, cover_path)
+        MagickSetImageFormat(wand, 'JPEG')
+        MagickWriteImage(wand, '%s.jpg' % cover_path)
+    return open('%s.jpg' % cover_path, 'rb').read()
 
 
 

src/calibre/utils/ipc/worker.py

@@ -39,9 +39,6 @@ PARALLEL_FUNCS = {
       'write_pdf_metadata' :
       ('calibre.utils.podofo.__init__', 'set_metadata_', None),
 
-      'write_pdf_first_page' :
-      ('calibre.utils.podofo.__init__', 'write_first_page_', None),
-
       'save_book' :
       ('calibre.ebooks.metadata.worker', 'save_book', 'notification'),
 }

src/calibre/utils/podofo/__init__.py

@@ -19,39 +19,7 @@ podofo, podofo_err = plugins['podofo']
 
 class Unavailable(Exception): pass
 
-def write_first_page(stream, opath):
-    if not podofo:
-        raise Unavailable(podofo_err)
-    pt = PersistentTemporaryFile('_podofo.pdf')
-    pt.write(stream.read())
-    pt.close()
-    server = Server(pool_size=1)
-    job = ParallelJob('write_pdf_first_page', 'Extract first page of pdf',
-        lambda x,y:x,  args=[pt.name, opath])
-    server.add_job(job)
-    while not job.is_finished:
-        time.sleep(0.1)
-        job.update()
-
-    job.update()
-    server.close()
-    if not job.result:
-        raise ValueError('Failed to extract first page: ' + job.details)
-
-def write_first_page_(inpath, outpath):
-    p = podofo.PDFDoc()
-    p.open(inpath)
-    pages = p.pages
-    if pages < 1:
-        raise ValueError('PDF has no pages')
-    if pages == 1:
-        shutil.copyfile(inpath, outpath)
-        return True
-    p.delete_pages(1, pages-1)
-    p.save(outpath)
-    return True
-
-def get_metadata(stream):
+def get_metadata(stream, cpath=None):
     if not podofo:
         raise Unavailable(podofo_err)
     pt = PersistentTemporaryFile('_podofo.pdf')
@@ -59,7 +27,7 @@ def get_metadata(stream):
     pt.close()
     server = Server(pool_size=1)
     job = ParallelJob('read_pdf_metadata', 'Read pdf metadata',
-        lambda x,y:x,  args=[pt.name])
+        lambda x,y:x,  args=[pt.name, cpath])
     server.add_job(job)
     while not job.is_finished:
         time.sleep(0.1)
@@ -69,7 +37,10 @@ def get_metadata(stream):
     server.close()
     if job.result is None:
         raise ValueError('Failed to read metadata: ' + job.details)
-    title, authors, creator = job.result
+    title, authors, creator, ok = job.result
+    if not ok:
+        print 'Failed to extract cover:'
+        print job.details
     if title == '_':
         title = getattr(stream, 'name', _('Unknown'))
         title = os.path.splitext(title)[0]
@@ -78,6 +49,8 @@ def get_metadata(stream):
     if creator:
         mi.book_producer = creator
     if os.path.exists(pt.name): os.remove(pt.name)
+    if ok:
+        mi.cover = cpath
     return mi
 
 def get_metadata_quick(raw):
@@ -95,7 +68,7 @@ def get_metadata_quick(raw):
     return mi
 
 
-def get_metadata_(path):
+def get_metadata_(path, cpath=None):
     p = podofo.PDFDoc()
     p.open(path)
     title = p.title
@@ -104,7 +77,23 @@ def get_metadata_(path):
     author = p.author
     authors = string_to_authors(author) if author else  [_('Unknown')]
     creator = p.creator
-    return (title, authors, creator)
+    ok = True
+    try:
+        if cpath is not None:
+            pages = p.pages
+            if pages < 1:
+                raise ValueError('PDF has no pages')
+            if True or pages == 1:
+                shutil.copyfile(path, cpath)
+            else:
+                p.extract_first_page()
+                p.save(cpath)
+    except:
+        import traceback
+        traceback.print_exc()
+        ok = False
+
+    return (title, authors, creator, ok)
 
 def prep(val):
     if not val:

src/calibre/utils/podofo/podofo.cpp

@@ -143,18 +143,15 @@ podofo_PDFDoc_version_getter(podofo_PDFDoc *self, void *closure) {
 
  
 static PyObject *
-podofo_PDFDoc_delete_pages(podofo_PDFDoc *self, PyObject *args, PyObject *kwargs) {
-    int first_page, num_pages;
-    if (PyArg_ParseTuple(args, "ii", &first_page, &num_pages)) {
-        try {
-            self->doc->DeletePages(first_page, num_pages);
-        } catch(const PdfError & err) {
-            podofo_set_exception(err);
-            return NULL;
-        }
-    } else return NULL;
-    Py_INCREF(Py_None);
-    return Py_None;
+podofo_PDFDoc_extract_first_page(podofo_PDFDoc *self, PyObject *args, PyObject *kwargs) {
+    int i, num_pages;
+    try {
+        while (self->doc->GetPageCount() > 1) self->doc->GetPagesTree()->DeletePage(1);
+    } catch(const PdfError & err) {
+        podofo_set_exception(err);
+        return NULL;
+    }
+    Py_RETURN_NONE;
 }
 
 static PyObject *
@@ -313,8 +310,8 @@ static PyMethodDef podofo_PDFDoc_methods[] = {
     {"save", (PyCFunction)podofo_PDFDoc_save, METH_VARARGS,
      "Save the PDF document to a path on disk"
     },
-    {"delete_pages", (PyCFunction)podofo_PDFDoc_delete_pages, METH_VARARGS,
-     "delete_pages(start_page, num_pages) -> int, int\nDelete pages from the PDF document."
+    {"extract_first_page", (PyCFunction)podofo_PDFDoc_extract_first_page, METH_VARARGS,
+     "extract_first_page() -> Remove all but the first page."
     },
 
     {NULL}  /* Sentinel */