Moved the code that generates .txt and .html draft files from .xml to a place where potential errors can be caught and displayed better. Related to ticket #2814.

- Legacy-Id: 16862

ietf/submit/forms.py

@@ -11,15 +11,22 @@ import datetime
 import email
 import pytz
 import six
+import sys
 import tempfile
 import xml2rfc
 
 from email.utils import formataddr
 from unidecode import unidecode
 
+if six.PY2:
+    from StringIO import StringIO
+else:
+    from io import StringIO
+
+
 from django import forms
 from django.conf import settings
-from django.utils.html import mark_safe
+from django.utils.html import mark_safe # type:ignore
 from django.urls import reverse as urlreverse
 from django.utils.encoding import force_str
 
@@ -39,6 +46,7 @@ from ietf.submit.parsers.pdf_parser import PDFParser
 from ietf.submit.parsers.plain_parser import PlainParser
 from ietf.submit.parsers.ps_parser import PSParser
 from ietf.submit.parsers.xml_parser import XMLParser
+from ietf.utils import log
 from ietf.utils.draft import Draft
 
 class SubmissionBaseUploadForm(forms.Form):
@@ -128,6 +136,20 @@ class SubmissionBaseUploadForm(forms.Form):
         return self.clean_file("xml", XMLParser)
 
     def clean(self):
+        def format_messages(where, e, log):
+            out = log.write_out.getvalue().splitlines()
+            err = log.write_err.getvalue().splitlines()
+            m = str(e)
+            if m:
+                m = [ m ]
+            else:
+                import traceback
+                typ, val, tb = sys.exc_info()
+                m = traceback.format_exception(typ, val, tb)
+                m = [ l.replace('\n ', ':\n ') for l in m ]
+            msgs = [s for s in (["Error from xml2rfc (%s):" % (where,)] + m + out +  err) if s]
+            return msgs
+
         if self.shutdown and not has_role(self.request.user, "Secretariat"):
             raise forms.ValidationError('The submission tool is currently shut down')
 
@@ -145,6 +167,9 @@ class SubmissionBaseUploadForm(forms.Form):
             xml_file = self.cleaned_data.get('xml')
             name, ext = os.path.splitext(os.path.basename(xml_file.name))
             tfh, tfn = tempfile.mkstemp(prefix=name+'-', suffix='.xml')
+            file_name = {}
+            xml2rfc.log.write_out = StringIO()   # open(os.devnull, "w")
+            xml2rfc.log.write_err = StringIO()   # open(os.devnull, "w")
             try:
                 # We need to write the xml file to disk in order to hand it
                 # over to the xml parser.  XXX FIXME: investigate updating
@@ -154,33 +179,15 @@ class SubmissionBaseUploadForm(forms.Form):
                     for chunk in xml_file.chunks():
                         tf.write(chunk)
                 os.environ["XML_LIBRARY"] = settings.XML_LIBRARY
+                # --- Parse the xml ---
                 try:
                     parser = xml2rfc.XmlRfcParser(str(tfn), quiet=True)
                     self.xmltree = parser.parse(normalize=True)
-                    root = self.xmltree.getroot()
+                    self.xmlroot = self.xmltree.getroot()
-                    ver = root.get('version', '2')
+                    xml_version = self.xmlroot.get('version', '2')
-                    if ver == '2':
+                except Exception as e:
-                        ok, errors = self.xmltree.validate()
+                    raise forms.ValidationError("An exception occurred when trying to [arse the XML file: %s" % e)
-                    else:
+
-                        # XXX TODO: Add v3 validation
-                        ok, errors = True, ''
-                except Exception as exc:
-                    raise forms.ValidationError("An exception occurred when trying to process the XML file: %s" % exc)
-                if not ok:
-                    # Each error has properties:
-                    #
-                    #     message:  the message text
-                    #     domain:   the domain ID (see lxml.etree.ErrorDomains)
-                    #     type:     the message type ID (see lxml.etree.ErrorTypes)
-                    #     level:    the log level ID (see lxml.etree.ErrorLevels)
-                    #     line:     the line at which the message originated (if applicable)
-                    #     column:   the character column at which the message originated (if applicable)
-                    #     filename: the name of the file in which the message originated (if applicable)
-                    raise forms.ValidationError(
-                            [ forms.ValidationError("One or more XML validation errors occurred when processing the XML file:") ] +
-                            [ forms.ValidationError("%s: Line %s: %s" % (xml_file.name, e.line, e.message), code="%s"%e.type) for e in errors ]
-                        )
-                self.xmlroot = self.xmltree.getroot()
                 draftname = self.xmlroot.attrib.get('docName')
                 if draftname is None:
                     raise forms.ValidationError("No docName attribute found in the xml root element")
@@ -212,8 +219,76 @@ class SubmissionBaseUploadForm(forms.Form):
                         if info[item]:
                             info[item] = info[item].strip()
                     self.authors.append(info)
-            except forms.ValidationError:
+
-                raise
+                # --- Prep the xml ---
+                file_name['xml'] = os.path.join(settings.IDSUBMIT_STAGING_PATH, '%s-%s.%s' % (self.filename, self.revision, ext))
+                try:
+                    if xml_version == '3':
+                        prep = xml2rfc.PrepToolWriter(self.xmltree, quiet=True)
+                        self.xmltree.tree = prep.prep()
+                        if self.xmltree.tree == None:
+                            raise forms.ValidationError("Error from xml2rfc (prep): %s" % prep.errors)
+                except Exception as e:
+                        msgs = format_messages('prep', e, xml2rfc.log)
+                        raise forms.ValidationError(msgs)
+
+                # --- Convert to txt ---
+                if not ('txt' in self.cleaned_data and self.cleaned_data['txt']):
+                    file_name['txt'] = os.path.join(settings.IDSUBMIT_STAGING_PATH, '%s-%s.txt' % (self.filename, self.revision))
+                    try:
+                        if xml_version != '3':
+                            pagedwriter = xml2rfc.PaginatedTextRfcWriter(self.xmltree, quiet=True)
+                            pagedwriter.write(file_name['txt'])
+                        else:
+                            writer = xml2rfc.TextWriter(self.xmltree, quiet=True)
+                            writer.write(file_name['txt'])
+                        log.log("In %s: xml2rfc %s generated %s from %s (version %s)" %
+                                (   os.path.dirname(file_name['xml']),
+                                    xml2rfc.__version__,
+                                    os.path.basename(file_name['txt']),
+                                    os.path.basename(file_name['xml']),
+                                    xml_version))
+                    except Exception as e:
+                        msgs = format_messages('txt', e, xml2rfc.log)
+                        raise forms.ValidationError(msgs)
+
+                # --- Convert to xml ---
+                if xml_version == '3':
+                    try:
+                        file_name['html'] = os.path.join(settings.IDSUBMIT_STAGING_PATH, '%s-%s.html' % (self.filename, self.revision))
+                        writer = xml2rfc.HtmlWriter(self.xmltree, quiet=True)
+                        writer.write(file_name['html'])
+                        self.file_types.append('.html')
+                        log.log("In %s: xml2rfc %s generated %s from %s (version %s)" %
+                            (   os.path.dirname(file_name['xml']),
+                                xml2rfc.__version__,
+                                os.path.basename(file_name['html']),
+                                os.path.basename(file_name['xml']),
+                                xml_version))
+                    except Exception as e:
+                        msgs = format_messages('html', e, xml2rfc.log)
+                        raise forms.ValidationError(msgs)
+
+                if xml_version == '2':
+                    ok, errors = self.xmltree.validate()
+                else:
+                    ok, errors = True, ''
+
+                if not ok:
+                    # Each error has properties:
+                    #
+                    #     message:  the message text
+                    #     domain:   the domain ID (see lxml.etree.ErrorDomains)
+                    #     type:     the message type ID (see lxml.etree.ErrorTypes)
+                    #     level:    the log level ID (see lxml.etree.ErrorLevels)
+                    #     line:     the line at which the message originated (if applicable)
+                    #     column:   the character column at which the message originated (if applicable)
+                    #     filename: the name of the file in which the message originated (if applicable)
+                    raise forms.ValidationError(
+                            [ forms.ValidationError("One or more XML validation errors occurred when processing the XML file:") ] +
+                            [ forms.ValidationError("%s: Line %s: %s" % (xml_file.name, r.line, r.message), code="%s"%r.type) for r in errors ] 
+                        )
+
             finally:
                 os.close(tfh)
                 os.unlink(tfn)

ietf/submit/templatetags/submit_tags.py

@@ -8,7 +8,7 @@ import os
 
 from django import template
 from django.conf import settings
-from django.utils.html import mark_safe, escape
+from django.utils.html import mark_safe, escape # type:ignore
 
 register = template.Library()
 

ietf/submit/tests.py

@@ -1774,7 +1774,7 @@ class ApiSubmitTests(TestCase):
         self.assertContains(r, expected, status_code=400)
 
     def test_api_submit_no_title(self):
-        r, author, name = self.post_submission('00', title="")
+        r, author, name = self.post_submission('00', title=" ")
         expected = "Could not extract a valid title from the upload"
         self.assertContains(r, expected, status_code=400)
 

ietf/submit/utils.py

@@ -9,7 +9,6 @@ import io
 import os
 import re
 import six                              # pyflakes:ignore
-import xml2rfc
 if six.PY3:
     from typing import Callable, Optional # pyflakes:ignore
 
@@ -649,57 +648,14 @@ def get_draft_meta(form, saved_files):
     file_name = saved_files
     abstract = None
     file_size = None
-    xml2rfc.log.write_out = open(os.devnull, "w")
-    xml2rfc.log.write_err = open(os.devnull, "w")
     if form.cleaned_data['xml']:
-        try:
-            xmlroot = form.xmltree.getroot()
-            xml_version = xmlroot.get('version', '2')
-            if xml_version == '3':
-                prep = xml2rfc.PrepToolWriter(form.xmltree, quiet=True)
-                form.xmltree.tree = prep.prep()
-                if form.xmltree.tree == None:
-                    raise ValidationError("Error from xml2rfc (prep): %s" % prep.errors)
-        except Exception as e:
-            raise ValidationError("Error from xml2rfc (prep): %s" % e)
-        if not ('txt' in form.cleaned_data and form.cleaned_data['txt']):
-            file_name['txt'] = os.path.join(settings.IDSUBMIT_STAGING_PATH, '%s-%s.txt' % (form.filename, form.revision))
-            try:
-                if xml_version != '3':
-                    pagedwriter = xml2rfc.PaginatedTextRfcWriter(form.xmltree, quiet=True)
-                    pagedwriter.write(file_name['txt'])
-                else:
-                    writer = xml2rfc.TextWriter(form.xmltree, quiet=True)
-                    writer.write(file_name['txt'])
-                log.log("In %s: xml2rfc %s generated %s from %s (version %s)" %
-                        (   os.path.dirname(file_name['xml']),
-                            xml2rfc.__version__,
-                            os.path.basename(file_name['txt']),
-                            os.path.basename(file_name['xml']),
-                            xml_version))
-            except Exception as e:
-                raise ValidationError("Error from xml2rfc (text): %s" % e) 
-            file_size = os.stat(file_name['txt']).st_size
-        if xml_version == '3':
-            try:
-                file_name['html'] = os.path.join(settings.IDSUBMIT_STAGING_PATH, '%s-%s.html' % (form.filename, form.revision))
-                writer = xml2rfc.HtmlWriter(form.xmltree, quiet=True)
-                writer.write(file_name['html'])
-                form.file_types.append('.html')
-                log.log("In %s: xml2rfc %s generated %s from %s (version %s)" %
-                    (   os.path.dirname(file_name['xml']),
-                        xml2rfc.__version__,
-                        os.path.basename(file_name['html']),
-                        os.path.basename(file_name['xml']),
-                        xml_version))
-            except Exception as e:
-                raise ValidationError("Error from xml2rfc (html): %s" % e) 
         # Some meta-information, such as the page-count, can only
         # be retrieved from the generated text file.  Provide a
         # parsed draft object to get at that kind of information.
+        file_name['txt'] = os.path.join(settings.IDSUBMIT_STAGING_PATH, '%s-%s.txt' % (form.filename, form.revision))
+        file_size = os.stat(file_name['txt']).st_size
         with io.open(file_name['txt']) as txt_file:
             form.parsed_draft = Draft(txt_file.read(), txt_file.name)
-
     else:
         file_size = form.cleaned_data['txt'].size