Version 0.1.1

[m_librarian.git] / m_librarian / inp.py

diff --git a/m_librarian/inp.py b/m_librarian/inp.py
index b3a3d84b43b662bd7870f51d77d0f504d8c18a78..4b772aff703f050d69117e2ef1725eb575d157c8 100644 (file)
--- a/m_librarian/inp.py
+++ b/m_librarian/inp.py
@@ -17,11 +17,18 @@ def split_line(line):
     _l = len(parts)
     if _l < 11:
         raise ValueError('Unknown INP structure: "%s"' % line)
+    archive = None
     if _l == 11:  # Standard structure
         parts.append(None)  # Emulate lang
-    else:  # New structure
+    elif _l == 15:  # New structure
         parts = parts[:12]
-    return parts
+    elif _l == 17:  # Very new structure
+        archive = parts[12]
+        language = parts[13]
+        parts = parts[:11] + [language]
+    else:  # New structure
+        raise ValueError('Unknown INP structure: "%s"' % line)
+    return archive, parts
 
 
 def import_inp_line(archive, parts):
@@ -66,26 +73,43 @@ def import_inp_line(archive, parts):
 
 
 def import_inp(archive, inp):
+    archives = set()
     files = set()
     connection = sqlhub.processConnection
     for file, in connection.queryAll(connection.sqlrepr(
             Select(Book.q.file, Book.q.archive == archive))):
-        files.add(file)
+        files.add((archive, file))
     for line in inp:
         line = line.decode('utf-8')
-        parts = split_line(line)
+        _archive, parts = split_line(line)
+        if _archive and (_archive not in archives):
+            archives.add(_archive)
+            for file, in connection.queryAll(connection.sqlrepr(
+                    Select(Book.q.file, Book.q.archive == _archive))):
+                files.add((_archive, file))
         file = parts[5]
-        if file not in files:
-            files.add(file)
-            import_inp_line(archive, parts)
+        if (_archive or archive, file) not in files:
+            files.add((_archive or archive, file))
+            import_inp_line(_archive or archive, parts)
 
 
-def import_inpx(path):
+def import_inpx(path, pbar_cb=None):
     inpx = ZipFile(path)
+    if pbar_cb:
+        inp_count = 0
+        for name in inpx.namelist():
+            ext = os.path.splitext(name)[1]
+            if ext == '.inp':
+                inp_count += 1
+        pbar_cb.set_max(inp_count)
+    inp_count = 0
     for name in inpx.namelist():
         archive, ext = os.path.splitext(name)
         if ext != '.inp':
             continue
+        if pbar_cb:
+            inp_count += 1
+            pbar_cb.display(inp_count)
         inp = inpx.open(name)
         sqlhub.doInTransaction(import_inp, archive + '.zip', inp)
         inp.close()
@@ -95,3 +119,5 @@ def import_inpx(path):
             connection.query("VACUUM %s" % table.sqlmeta.table)
     elif connection.dbName == 'sqlite':
         connection.query("VACUUM")
+    if pbar_cb:
+        pbar_cb.close()