IGN:Restore PoDoFo based set pdf metadata functionality

src/calibre/ebooks/metadata/pdf.py

@@ -5,8 +5,9 @@ __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
 
 from functools import partial
 
-from calibre import plugins, prints
-from calibre.ebooks.metadata import MetaInformation, string_to_authors#, authors_to_string
+from calibre import prints
+from calibre.constants import plugins
+from calibre.ebooks.metadata import MetaInformation, string_to_authors, authors_to_string
 
 pdfreflow, pdfreflow_error = plugins['pdfreflow']
 
@@ -44,64 +45,13 @@ def get_metadata(stream, cover=True):
 
     return mi
 
-
-
 get_quick_metadata = partial(get_metadata, cover=False)
 
-'''
-import sys, os, cStringIO
+import cStringIO
 from threading import Thread
 
-from calibre import StreamReadWrapper
-from calibre.ptempfile import TemporaryDirectory
-try:
-    from calibre.utils.PythonMagickWand import \
-        NewMagickWand, MagickReadImage, MagickSetImageFormat, \
-        MagickWriteImage, ImageMagick
-    _imagemagick_loaded = True
-except:
-    _imagemagick_loaded = False
-from calibre.ebooks.metadata import MetaInformation, string_to_authors, 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, get_metadata_quick
-from calibre.utils.poppler import get_metadata as get_metadata_poppler, NotAvailable
-
-def get_quick_metadata(stream):
-    try:
-       return get_metadata_poppler(stream, False)
-    except NotAvailable:
-        pass
-
-    return get_metadata_pypdf(stream)
-    raw = stream.read()
-    mi = get_metadata_quick(raw)
-    if mi.title == '_':
-        mi.title = getattr(stream, 'name', _('Unknown'))
-        mi.title = mi.title.rpartition('.')[0]
-    return mi
-
-
-def get_metadata(stream, extract_cover=True):
-    try:
-       return get_metadata_poppler(stream, extract_cover)
-    except NotAvailable:
-        pass
-    try:
-        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)
-    return mi
-
+from calibre.utils.podofo import set_metadata as podofo_set_metadata, Unavailable
 
 def set_metadata(stream, mi):
     stream.seek(0)
@@ -116,25 +66,6 @@ def set_metadata(stream, mi):
     set_metadata_pypdf(stream, mi)
 
 
-def get_metadata_pypdf(stream):
-    """ Return metadata as a L{MetaInfo} object """
-    from pyPdf import PdfFileReader
-    mi = MetaInformation(_('Unknown'), [_('Unknown')])
-    try:
-        with StreamReadWrapper(stream) as stream:
-            info = PdfFileReader(stream).getDocumentInfo()
-            if info.title:
-                mi.title = info.title
-            if info.author:
-                mi.author = info.author
-                mi.authors = string_to_authors(info.author)
-            if info.subject:
-                mi.category = info.subject
-    except Exception, err:
-        msg = u'Couldn\'t read metadata from pdf: %s with error %s'%(mi.title, unicode(err))
-        print >>sys.stderr, msg.encode('utf8')
-    return mi
-
 class MetadataWriter(Thread):
 
     def __init__(self, out_pdf, buf):
@@ -178,13 +109,4 @@ def set_metadata_pypdf(stream, mi):
     stream.write(out_str.read())
     stream.seek(0)
 
-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/ebooks/pdf/main.cpp

@@ -79,6 +79,50 @@ extern "C" {
         return ans;
     }
 
+    static PyObject *
+    pdfreflow_set_metadata(PyObject *self, PyObject *args) {
+        char *pdfdata;
+        Py_ssize_t size;
+        PyObject *info;
+
+        if (!PyArg_ParseTuple(args, "s#O", &pdfdata, &size, &info))
+            return NULL;
+
+        if (!PyDict_Check(info)) {
+            PyErr_SetString(PyExc_ValueError, "Info object must be a dictionary.");
+            return NULL;
+        }
+
+        char Title[10] = "Title", Author[10] = "Author", Keywords[10] = "Keywords";
+        char *keys[3] = { Title, Author, Keywords };
+        map<char *, char *> pinfo;
+        PyObject *val = NULL, *utf8 = NULL;
+
+        for (int i = 0; i < 3; i++) {
+            val = PyDict_GetItemString(info, keys[i]);
+            if (!val || !PyUnicode_Check(val)) continue;
+            utf8 = PyUnicode_AsUTF8String(val);
+            if (!utf8) continue;
+            pinfo[keys[i]] = PyString_AS_STRING(utf8);
+        }
+
+        PyObject *ans = NULL;
+        try {
+            Reflow reflow(pdfdata, static_cast<std::ifstream::pos_type>(size));
+            if (reflow.is_locked()) {
+                PyErr_SetString(PyExc_ValueError, "Setting metadata not possible in encrypeted PDFs");
+                return NULL;
+            }
+            string result = reflow.set_info(pinfo);
+            ans = PyString_FromStringAndSize(result.c_str(), result.size());
+        } catch (std::exception &e) {
+            PyErr_SetString(PyExc_RuntimeError, e.what()); return NULL;
+        } catch (...) {
+            PyErr_SetString(PyExc_RuntimeError,
+                    "Unknown exception raised while getting metadata from PDF"); return NULL;
+        }
+        return ans;
+    }
 
     static 
     PyMethodDef pdfreflow_methods[] = {
@@ -90,6 +134,10 @@ extern "C" {
         "get_metadata(pdf_data, cover)\n\n"
                 "Get metadata and (optionally) cover from the specified PDF."
         },
+        {"set_metadata", pdfreflow_set_metadata, METH_VARARGS,
+        "get_metadata(info_dict)\n\n"
+                "Set metadata in the specified PDF. Currently broken."
+        },
 
         {NULL, NULL, 0, NULL}
     };

src/calibre/ebooks/pdf/reflow.cpp

@@ -680,6 +680,16 @@ void XMLOutputDev::drawImage(GfxState *state, Object *ref, Stream *str,
             colorMap, interpolate, maskColors, inlineImg);
 }
 
+static char stream_pdf[15] = "stream.pdf";
+
+class MemInStream : public MemStream {
+    public:
+        MemInStream(char *buf, size_t st, size_t sz, Object *obj) :
+            MemStream(buf, st, sz, obj) {}
+        ~MemInStream() {}
+        GooString *getFileName() { return new GooString(stream_pdf); }
+};
+
 Reflow::Reflow(char *pdfdata, size_t sz) :
     pdfdata(pdfdata), current_font_size(-1), doc(NULL)
 {
@@ -690,7 +700,7 @@ Reflow::Reflow(char *pdfdata, size_t sz) :
         if (!globalParams)
             throw ReflowException("Failed to allocate Globalparams");
     }
-    MemStream *str = new MemStream(pdfdata, 0, sz, &obj);
+    MemInStream *str = new MemInStream(pdfdata, 0, sz, &obj);
     this->doc = new PDFDoc(str, NULL, NULL);
 
     if (!this->doc->isOk()) {
@@ -909,3 +919,56 @@ char* Reflow::render_first_page(size_t *data_size,
     }
     return buffer;
 }
+
+class MemOutStream : public OutStream {
+    private:
+        ostringstream out;
+
+    public:
+        MemOutStream() :OutStream() {}
+        ~MemOutStream() {}
+        void close() {}
+        int getPos() { return out.tellp(); }
+        void put(char c) { out.put(c); }
+        void printf (const char *format, ...) {
+            vector<char> buf;
+            size_t written = strlen(format)*5;
+            va_list ap;
+            do {
+                buf.reserve(written + 20);
+                va_start(ap, format);
+                written = vsnprintf(&buf[0], buf.capacity(), format, ap);
+                va_end(ap);
+            } while (written >= buf.capacity());
+            out.write(&buf[0], written);            
+        }
+};
+
+string Reflow::set_info(map<char *, char *> sinfo) {
+    XRef *xref = this->doc->getXRef();
+    if (!xref) throw ReflowException("No XRef table");
+    Object *trailer_dict = xref->getTrailerDict();
+    if (!trailer_dict || !trailer_dict->isDict()) throw ReflowException("No trailer dictionary");
+    Object tmp;
+    char INFO[5] = "Info";
+    Object *info = trailer_dict->dictLookup(INFO, &tmp);
+    if (!info) {
+        info = new Object();
+        info->initDict(xref);
+    }
+    if (!info->isDict()) throw ReflowException("Invalid info object");
+
+    for (map<char *, char *>::iterator it = sinfo.begin(); it != sinfo.end(); it++) {
+        Object *tmp = new Object();
+        tmp->initString(new GooString((*it).second));
+        info->dictSet((*it).first, tmp);
+    }
+
+    trailer_dict->dictSet(INFO, info);
+    char out[20] = "/t/out.pdf";
+    this->doc->saveAs(new GooString(out), writeForceRewrite);
+    string ans;
+    return ans;
+}
+
+

src/calibre/ebooks/pdf/reflow.h

@@ -74,6 +74,9 @@ class Reflow {
 
         /* Dump the PDF outline as the file outline.xml in the current directory */
         void dump_outline();
+
+        /* Set the info dictionary. Currently broken. */
+        string set_info(map<char *, char *> info);
 };
 
 class XMLString {