Add checksum to media objects

svn: r22870

gramps/gen/db/upgrade.py

@@ -24,8 +24,10 @@
 from __future__ import with_statement, unicode_literals
 
 import sys
+import os
 from ..lib.markertype import MarkerType
 from ..lib.tag import Tag
+from ..utils.file import create_checksum
 import time
 import logging
 LOG = logging.getLogger(".citation")
@@ -54,10 +56,11 @@ def gramps_upgrade_17(self):
        1. This upgrade adds tags to event, place, repository, source and 
           citation objects.
        2. Data of Source becomes SourceAttributes Secondary Object
+       3. Add checksum field to media objects.
     """
     length = (len(self.event_map) + len(self.place_map) +
               len(self.repository_map) + len(self.source_map) +
-              len(self.citation_map))
+              len(self.citation_map) + len(self.media_map))
     self.set_total(length)
 
     # ---------------------------------
@@ -166,6 +169,27 @@ def gramps_upgrade_17(self):
             txn.put(handle, new_citation)
         self.update()
 
+    # ---------------------------------
+    # Modify Media
+    # ---------------------------------
+    # Add new checksum field.
+    base_path = self.metadata[b'mediapath']
+    for handle in self.media_map.keys():
+        media = self.media_map[handle]
+        new_media = list(media)
+        if os.path.isabs(new_media[2]):
+            full_path = new_media[2]
+        else:
+            full_path = os.path.join(base_path, new_media[2])
+        checksum = create_checksum(full_path)
+        new_media = new_media[:5] + [checksum] + new_media[5:]
+        new_media = tuple(new_media)
+        with BSDDBTxn(self.env, self.media_map) as txn:
+            if isinstance(handle, UNITYPE):
+                handle = handle.encode('utf-8')
+            txn.put(handle, new_media)
+        self.update()
+
     # Bump up database version. Separate transaction to save metadata.
     with BSDDBTxn(self.env, self.metadata) as txn:
         txn.put(b'version', 17)

gramps/gen/lib/mediaobj.py

@@ -86,11 +86,13 @@ class MediaObject(CitationBase, NoteBase, DateBase, AttributeBase,
             self.path = source.path
             self.mime = source.mime
             self.desc = source.desc
+            self.checksum = source.checksum
             self.thumb = source.thumb
         else:
             self.path = ""
             self.mime = ""
             self.desc = ""
+            self.checksum = ""
             self.thumb = None
 
     def serialize(self, no_text_date = False):
@@ -112,6 +114,7 @@ class MediaObject(CitationBase, NoteBase, DateBase, AttributeBase,
         :rtype: tuple
         """
         return (self.handle, self.gramps_id, self.path, self.mime, self.desc,
+                self.checksum,
                 AttributeBase.serialize(self),
                 CitationBase.serialize(self),
                 NoteBase.serialize(self),
@@ -145,6 +148,7 @@ class MediaObject(CitationBase, NoteBase, DateBase, AttributeBase,
                 "path": self.path, 
                 "mime": self.mime, 
                 "desc": self.desc,
+                "checksum": self.checksum,
                 "attribute_list": AttributeBase.to_struct(self),
                 "citation_list": CitationBase.to_struct(self),
                 "note_list": NoteBase.to_struct(self),
@@ -162,7 +166,7 @@ class MediaObject(CitationBase, NoteBase, DateBase, AttributeBase,
         :type data: tuple
         """
         (self.handle, self.gramps_id, self.path, self.mime, self.desc,
-         attribute_list, citation_list, note_list, self.change,
+         self.checksum, attribute_list, citation_list, note_list, self.change,
          date, tag_list, self.private) = data
 
         AttributeBase.unserialize(self, attribute_list)
@@ -287,3 +291,12 @@ class MediaObject(CitationBase, NoteBase, DateBase, AttributeBase,
     def get_description(self):
         """Return the description of the image."""
         return self.desc
+
+    def set_checksum(self, text):
+        """Set the checksum of the image."""
+        self.checksum = text
+
+    def get_checksum(self):
+        """Return the checksum of the image."""
+        return self.checksum
+

gramps/gen/utils/file.py

@@ -34,6 +34,8 @@ File and folder related utility functions
 import os
 import sys
 import shutil
+import io
+import hashlib
 import logging
 LOG = logging.getLogger(".gen.utils.file")
 
@@ -294,3 +296,15 @@ def fix_encoding(value, errors='strict'):
                 return value.decode(encoding=codeset, errors=errors)
     else:
         return value
+
+def create_checksum(full_path):
+    """
+    Create a md5 hash for the given file.
+    """
+    full_path = os.path.normpath(full_path)
+    try:
+        with io.open(full_path, 'rb') as media_file:
+            md5sum = hashlib.md5(media_file.read()).hexdigest()
+    except IOError:
+            md5sum = ''
+    return md5sum

gramps/gui/editors/displaytabs/gallerytab.py

@@ -64,7 +64,7 @@ from gramps.gen.constfunc import cuni
 from gramps.gen.lib import MediaObject, MediaRef
 from gramps.gen.db import DbTxn
 from gramps.gen.utils.file import (media_path_full, media_path, relative_path,
-                            fix_encoding)
+                                   fix_encoding, create_checksum)
 from ...thumbnails import get_thumbnail_image
 from gramps.gen.errors import WindowActiveError
 from gramps.gen.mime import get_type, is_valid_type
@@ -525,6 +525,9 @@ class GalleryTab(ButtonTab, DbGUIElement):
                         if not is_valid_type(mime):
                             return
                         photo = MediaObject()
+                        self.uistate.set_busy_cursor(True)
+                        photo.set_checksum(create_checksum(name))
+                        self.uistate.set_busy_cursor(False)
                         base_dir = cuni(media_path(self.dbstate.db))
                         if os.path.exists(base_dir):
                             name = relative_path(name, base_dir)

gramps/gui/editors/editmedia.py

@@ -49,8 +49,8 @@ from gramps.gen.lib import MediaObject, NoteType
 from gramps.gen.db import DbTxn
 from gramps.gen.mime import get_description, get_type
 from ..thumbnails import get_thumbnail_image, find_mime_type_pixbuf
-from gramps.gen.utils.file import (media_path_full, find_file, 
+from gramps.gen.utils.file import (media_path_full, find_file, create_checksum,
-                            get_unicode_path_from_file_chooser)
+                                   get_unicode_path_from_file_chooser)
 from .editprimary import EditPrimary
 from ..widgets import (MonitoredDate, MonitoredEntry, PrivacyButton,
                          MonitoredTagList)
@@ -263,8 +263,15 @@ class EditMedia(EditPrimary):
         fname = self.obj.get_path()
         self.file_path.set_text(fname)
         self.determine_mime()
+        self.update_checksum()
         self.draw_preview()
 
+    def update_checksum(self):
+        self.uistate.set_busy_cursor(True)
+        media_path = media_path_full(self.dbstate.db, self.obj.get_path())
+        self.obj.set_checksum(create_checksum(os.path.normpath(media_path)))
+        self.uistate.set_busy_cursor(False)
+
     def save(self, *obj):
         self.ok_button.set_sensitive(False)
 

gramps/gui/editors/editmediaref.py

@@ -49,8 +49,8 @@ from ..utils import open_file_with_default_application
 from gramps.gen.const import THUMBSCALE
 from gramps.gen.mime import get_description, get_type
 from ..thumbnails import get_thumbnail_image, find_mime_type_pixbuf
-from gramps.gen.utils.file import (media_path_full, find_file, 
+from gramps.gen.utils.file import (media_path_full, find_file, create_checksum,
-                            get_unicode_path_from_file_chooser)
+                                   get_unicode_path_from_file_chooser)
 from gramps.gen.lib import NoteType
 from gramps.gen.db import DbTxn
 from ..glade import Glade
@@ -553,7 +553,14 @@ class EditMediaRef(EditReference):
         for obj in (self.descr_window, self.path_obj):
             obj.update()
         self.determine_mime()
+        self.update_checksum()
         self.draw_preview()
+        
+    def update_checksum(self):
+        self.uistate.set_busy_cursor(True)
+        media_path = media_path_full(self.dbstate.db, self.source.get_path())
+        self.source.set_checksum(create_checksum(os.path.normpath(media_path)))
+        self.uistate.set_busy_cursor(False)
 
     def select_file(self, val):
         self.determine_mime()

gramps/gui/views/treemodels/mediamodel.py

@@ -137,9 +137,9 @@ class MediaModel(FlatBaseModel):
         return cuni(data[1])
 
     def column_date(self,data):
-        if data[9]:
+        if data[10]:
             date = Date()
-            date.unserialize(data[9])
+            date.unserialize(data[10])
             return cuni(displayer.display(date))
         return ''
 
@@ -156,17 +156,17 @@ class MediaModel(FlatBaseModel):
         return cuni(data[0])
 
     def column_private(self, data):
-        if data[11]:
+        if data[12]:
             return 'gramps-lock'
         else:
             # There is a problem returning None here.
             return ''
 
     def sort_change(self,data):
-        return "%012x" % data[8]
+        return "%012x" % data[9]
 
     def column_change(self,data):
-        return format_time(data[8])
+        return format_time(data[9])
 
     def column_tooltip(self,data):
         return cuni('Media tooltip')
@@ -183,7 +183,7 @@ class MediaModel(FlatBaseModel):
         """
         tag_color = "#000000000000"
         tag_priority = None
-        for handle in data[10]:
+        for handle in data[11]:
             tag = self.db.get_tag_from_handle(handle)
             this_priority = tag.get_priority()
             if tag_priority is None or this_priority < tag_priority:
@@ -195,5 +195,5 @@ class MediaModel(FlatBaseModel):
         """
         Return the sorted list of tags.
         """
-        tag_list = list(map(self.get_tag_name, data[10]))
+        tag_list = list(map(self.get_tag_name, data[11]))
         return ', '.join(sorted(tag_list, key=glocale.sort_key))

gramps/plugins/export/exportxml.py

@@ -1216,6 +1216,7 @@ class GrampsXmlWriter(UpdateCallback):
         mime_type = obj.get_mime_type()
         path = obj.get_path()
         desc = obj.get_description()
+        checksum = obj.get_checksum()
         if desc:
             desc_text = ' description="%s"' % self.fix(desc)
         else:
@@ -1229,9 +1230,9 @@ class GrampsXmlWriter(UpdateCallback):
             # Always export path with \ replaced with /. Otherwise import
             # from Windows to Linux of gpkg's path to images does not work. 
             path = path.replace('\\','/')
-        self.g.write('%s<file src="%s" mime="%s"%s/>\n'
+        self.g.write('%s<file src="%s" mime="%s" checksum="%s"%s/>\n'
                      % ("  "*(index+1), self.fix(path), self.fix(mime_type), 
-                        desc_text))
+                        checksum, desc_text))
         self.write_attribute_list(obj.get_attribute_list())
         self.write_note_list(obj.get_note_list(), index+1)
         dval = obj.get_date_object()

gramps/plugins/importer/importxml.py

@@ -62,6 +62,7 @@ from gramps.gen.errors import GrampsImportError
 from gramps.gen.utils.id import create_id
 from gramps.gen.utils.db import family_name
 from gramps.gen.utils.unknown import make_unknown, create_explanation_note
+from gramps.gen.utils.file import create_checksum
 from gramps.gen.datehandler import parser, set_date
 from gramps.gen.display.name import displayer as name_displayer
 from gramps.gen.db.dbconst import (PERSON_KEY, FAMILY_KEY, SOURCE_KEY, 
@@ -1603,6 +1604,14 @@ class GrampsParser(UpdateCallback):
             if self.all_abs and not os.path.isabs(src):
                 self.all_abs = False
                 self.info.add('relative-path', None, None)
+        if 'checksum' in attrs:
+            self.object.checksum = attrs['checksum']
+        else:
+            if os.path.isabs(src):
+                full_path = src
+            else:
+                full_path = os.path.join(self.mediapath, src)
+            self.object.checksum = create_checksum(full_path)
 
     def start_childof(self, attrs):
         """

gramps/plugins/view/mediaview.py

@@ -63,7 +63,7 @@ from gramps.gui.views.treemodels import MediaModel
 from gramps.gen.constfunc import win, cuni
 from gramps.gen.config import config
 from gramps.gen.utils.file import (media_path, relative_path, media_path_full, 
-                            fix_encoding)
+                                   fix_encoding, create_checksum)
 from gramps.gen.utils.db import get_media_referents
 from gramps.gui.views.bookmarks import MediaBookmarks
 from gramps.gen.mime import get_type, is_valid_type
@@ -199,6 +199,9 @@ class MediaView(ListView):
                 if not is_valid_type(mime):
                     return
                 photo = MediaObject()
+                self.uistate.set_busy_cursor(True)
+                photo.set_checksum(create_checksum(name))
+                self.uistate.set_busy_cursor(False)
                 base_dir = cuni(media_path(self.dbstate.db))
                 if os.path.exists(base_dir):
                     name = relative_path(name, base_dir)