Need lxml>3.0 for handling Unicode in PySolr nicely.

[edumed.git] / catalogue / management / commands / importlessons.py

diff --git a/catalogue/management/commands/importlessons.py b/catalogue/management/commands/importlessons.py
index 6252fb5..4e7a38c 100755 (executable)
--- a/catalogue/management/commands/importlessons.py
+++ b/catalogue/management/commands/importlessons.py
@@ -12,7 +12,7 @@ from django.core.management.color import color_style
 from django.core.files import File
 
 from librarian import IOFile
-from catalogue.models import Lesson
+from catalogue.models import Lesson, Section
 
 #from search import Index
 
@@ -21,26 +21,44 @@ class Command(BaseCommand):
     option_list = BaseCommand.option_list + (
         make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
             help='Verbosity level; 0=minimal output, 1=normal output, 2=all output'),
+        make_option('-a', '--attachments', dest='attachments', metavar="PATH", default='materialy',
+            help='Attachments dir path.'),
     )
     help = 'Imports lessons from the specified directories.'
     args = 'directory [directory ...]'
 
-    def import_book(self, file_path, options):
+    def import_book(self, file_path, options, attachments):
         verbose = options.get('verbose')
-        iofile = IOFile.from_filename(file_path)
-        basename, ext = file_path.rsplit('.', 1)
-        if os.path.isdir(basename):
-            for att_name in os.listdir(basename):
-                iofile.attachments[att_name] = IOFile.from_filename(
-                    os.path.join(basename, att_name))
+        iofile = IOFile.from_filename(os.path.join(self.curdir, file_path))
+        iofile.attachments = attachments
         lesson = Lesson.publish(iofile)
 
+    @staticmethod
+    def all_attachments(path):
+        files = {}
+
+        def read_dir(path):
+            for name in os.listdir(path):
+                fullname = os.path.join(path, name)
+                if os.path.isdir(fullname):
+                    read_dir(fullname)
+                else:
+                    f = IOFile.from_filename(fullname)
+                    files[name] = f
+                    files.setdefault(name.replace(" ", ""), f)
+
+        read_dir(path)
+        return files
+
+
     def handle(self, *directories, **options):
         from django.db import transaction
 
         self.style = color_style()
         
         verbose = options.get('verbose')
+        self.curdir = os.path.abspath(os.curdir)
+
 
         # Start transaction management.
         transaction.commit_unless_managed()
@@ -51,15 +69,19 @@ class Command(BaseCommand):
         files_skipped = 0
 
         for dir_name in directories:
-            if not os.path.isdir(dir_name):
-                print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
+            abs_dir = os.path.join(self.curdir, dir_name)
+            if not os.path.isdir(abs_dir):
+                print self.style.ERROR("%s: Not a directory. Skipping." % abs_dir)
             else:
+                att_dir = os.path.join(abs_dir, options['attachments'])
+                attachments = self.all_attachments(att_dir)
+
                 # files queue
-                files = sorted(os.listdir(dir_name))
+                files = sorted(os.listdir(abs_dir))
                 postponed = {}
                 while files:
                     file_name = files.pop(0)
-                    file_path = os.path.join(dir_name, file_name)
+                    file_path = os.path.join(abs_dir, file_name)
                     file_base, ext = os.path.splitext(file_path)
 
                     # Skip files that are not XML files
@@ -73,9 +95,24 @@ class Command(BaseCommand):
                         sys.stdout.flush()
 
                     # Import book files
-                    self.import_book(file_path, options)
-                    files_imported += 1
-                    transaction.commit()
+                    try:
+                        self.import_book(file_path, options, attachments)
+                        files_imported += 1
+                        transaction.commit()
+                    except Section.IncompleteError, e:
+                        if file_name not in postponed or postponed[file_name] < files_imported:
+                            # Push it back into the queue, maybe the missing lessons will show up.
+                            if verbose > 0:
+                                print self.style.NOTICE('Waiting for missing lessons.')
+                            files.append(file_name)
+                            postponed[file_name] = files_imported
+                        else:
+                            # We're in a loop, nothing's being imported - some lesson is really missing.
+                            raise e
+                    except BaseException, e:
+                        import traceback
+                        traceback.print_exc()
+                        files_skipped += 1
 
         # Print results
         print