X-Git-Url: https://git.mdrn.pl/redakcja.git/blobdiff_plain/69d9738d6855e38869678a54991d30e5cddb8e67..e977f7187b10b1bc0a30794cd585c6b840568996:/src/documents/management/commands/fixdc.py?ds=sidebyside

diff --git a/src/documents/management/commands/fixdc.py b/src/documents/management/commands/fixdc.py
new file mode 100644
index 00000000..3f4a848e
--- /dev/null
+++ b/src/documents/management/commands/fixdc.py
@@ -0,0 +1,52 @@
+# This file is part of FNP-Redakcja, licensed under GNU Affero GPLv3 or later.
+# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
+#
+from librarian import RDFNS, WLURI, ValidationError
+from librarian.dcparser import BookInfo
+from documents.management import XmlUpdater
+from documents.management.commands import XmlUpdaterCommand
+
+
+class FixDC(XmlUpdater):
+    commit_desc = "auto-fixing DC"
+    retain_publishable = True
+    only_first_chunk = True
+
+    def fix_wluri(elem, change, verbose):
+        try:
+            WLURI.strict(elem.text)
+        except ValidationError:
+            correct_field = str(WLURI.from_slug(
+                                WLURI(elem.text.strip()).slug))
+            try:
+                WLURI.strict(correct_field)
+            except ValidationError:
+                # Can't make a valid WLURI out of it, leave as is.
+                return False
+            if verbose:
+                print("Changing %s from %s to %s" % (
+                        elem.tag, elem.text, correct_field
+                    ))
+            elem.text = correct_field
+            return True
+    for field in BookInfo.FIELDS:
+        if field.validator == WLURI:
+            XmlUpdater.fixes_elements('.//' + field.uri)(fix_wluri)
+
+    @XmlUpdater.fixes_elements(".//" + RDFNS("Description"))
+    def fix_rdfabout(elem, change, verbose):
+        correct_about = change.tree.book.correct_about()
+        attr_name = RDFNS("about")
+        current_about = elem.get(attr_name)
+        if current_about != correct_about:
+            if verbose:
+                print("Changing rdf:about from %s to %s" % (
+                        current_about, correct_about
+                    ))
+            elem.set(attr_name, correct_about)
+            return True
+
+
+class Command(XmlUpdaterCommand):
+    updater = FixDC
+    help = 'Fixes obvious errors in DC: rdf:about and WLURI format.'