GdsCollector: proof-of-concept to catch validation errors including filename

Author: kba

ocrd_models/ocrd_models/generatedscollector.py

@@ -1,7 +1,7 @@
 class OcrdGdsCollector(object):
 
-    def __init__(self, messages=None):
-        print('GdsCollector.__init__', self)
+    def __init__(self, filename=None, messages=None):
+        self.filename = filename
         if messages is None:
             self.messages = []
         else:

ocrd_models/ocrd_models/ocrd_page_generateds.py

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Thu Jul 30 13:44:35 2020 by generateDS.py version 2.35.24.
+# Generated Mon Aug 24 15:23:10 2020 by generateDS.py version 2.35.26.
 # Python 3.6.9 (default, Jul 17 2020, 12:50:27)  [GCC 8.4.0]
 #
 # Command line options:
@@ -14950,7 +14950,9 @@ def main():
 ]
 def parse(inFileName, silence=False, print_warnings=True):
     global CapturedNsmap_
-    gds_collector = GdsCollector_()
+    if not filename:
+        filename=inFilename
+    gds_collector = GdsCollector_(filename=filenmae)
     parser = None
     doc = parsexml_(inFileName, parser)
     rootNode = doc.getroot()
@@ -14979,7 +14981,7 @@ def parse(inFileName, silence=False, print_warnings=True):
         sys.stderr.write(separator)
     return rootObj
 
-def parseString(inString, silence=False, print_warnings=True):
+def parseString(inString, filename=None, silence=False, print_warnings=True):
     '''Parse a string, create the object tree, and export it.
 
     Arguments:
@@ -14990,7 +14992,7 @@ def parseString(inString, silence=False, print_warnings=True):
     '''
     parser = None
     rootNode= parsexmlstring_(inString, parser)
-    gds_collector = GdsCollector_()
+    gds_collector = GdsCollector_(filename=filename)
     rootTag, rootClass = get_root_tag(rootNode)
     if rootClass is None:
         rootTag = 'PcGts'

ocrd_models/ocrd_page_parse.py

@@ -1,6 +1,8 @@
 def parse(inFileName, silence=False, print_warnings=True):
     global CapturedNsmap_
-    gds_collector = GdsCollector_()
+    if not filename:
+        filename=inFilename
+    gds_collector = GdsCollector_(filename=filenmae)
     parser = None
     doc = parsexml_(inFileName, parser)
     rootNode = doc.getroot()
@@ -29,7 +31,7 @@ def parse(inFileName, silence=False, print_warnings=True):
         sys.stderr.write(separator)
     return rootObj
 
-def parseString(inString, silence=False, print_warnings=True):
+def parseString(inString, filename=None, silence=False, print_warnings=True):
     '''Parse a string, create the object tree, and export it.
 
     Arguments:
@@ -40,7 +42,7 @@ def parseString(inString, silence=False, print_warnings=True):
     '''
     parser = None
     rootNode= parsexmlstring_(inString, parser)
-    gds_collector = GdsCollector_()
+    gds_collector = GdsCollector_(filename=filename)
     rootTag, rootClass = get_root_tag(rootNode)
     if rootClass is None:
         rootTag = 'PcGts'

requirements_test.txt

@@ -1,5 +1,5 @@
 pytest >= 4.0.0
-generateDS == 2.35.24
+generateDS == 2.35.26
 coverage >= 4.5.2
 sphinx
 codecov >= 2.0.15

tests/model/test_ocrd_page.py

@@ -297,5 +297,18 @@ def test_gdscollector_override(self):
             gdc = pcgts.gds_collector_
             self.assertTrue(isinstance(gdc, OcrdGdsCollector))
 
+    def test_gdscollector_info(self):
+        filename = assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml')
+        with open(filename, 'r') as f:
+            s = f.read()
+            s = s.replace('pc:Page', 'pc:Foo')
+            s = s.encode('utf-8')
+            pcgts = parseString(s, silence=True, filename=filename)
+            gdsc = pcgts.gds_collector_
+            self.assertEqual(gdsc.messages, [])
+            self.assertEqual(gdsc.filename, filename)
+            pcgts.validate_(gdsc, True)
+            self.assertEqual(gdsc.messages, ['Number of values for Page near line 2 is below the minimum allowed, expected at least 1, found 0'])
+
 if __name__ == '__main__':
     main(__file__)