fix: test file existence using metadata (#8292)

* fix: test file existance using metadata

* fix: use Path more

* fix: don't read the file to see if it exists

* fix: more conservative error handling

* chore: remove unused import

ietf/doc/models.py

@@ -4,7 +4,6 @@
 
 import datetime
 import logging
-import io
 import os
 
 import django.db
@@ -530,16 +529,27 @@ class DocumentInfo(models.Model):
     def replaced_by(self):
         return set([ r.document for r in self.related_that("replaces") ])
 
-    def text(self, size = -1):
+    def _text_path(self):
         path = self.get_file_name()
         root, ext =  os.path.splitext(path)
         txtpath = root+'.txt'
         if ext != '.txt' and os.path.exists(txtpath):
             path = txtpath
+        return path
+    
+    def text_exists(self):
+        path = Path(self._text_path())
+        return path.exists()
+
+    def text(self, size = -1):
+        path = Path(self._text_path())
+        if not path.exists():
+            return None
         try:
-            with io.open(path, 'rb') as file:
+            with path.open('rb') as file:
                 raw = file.read(size)
-        except IOError:
+        except IOError as e:
+            log.log(f"Error reading text for {path}: {e}")
             return None
         text = None
         try:

ietf/doc/tests.py

@@ -3318,3 +3318,18 @@ class InvestigateTests(TestCase):
             self.assertEqual(r.status_code, 200)
             q = PyQuery(r.content)
             self.assertEqual(len(q("#id_name_fragment.is-invalid")), 1)
+
+class LogIOErrorTests(TestCase):
+
+    def test_doc_text_io_error(self):
+
+        d = IndividualDraftFactory()
+
+        with mock.patch("ietf.doc.models.Path") as path_cls_mock:
+            with mock.patch("ietf.doc.models.log.log") as log_mock:
+                path_cls_mock.return_value.exists.return_value = True
+                path_cls_mock.return_value.open.return_value.__enter__.return_value.read.side_effect = IOError("Bad things happened")
+                text = d.text()
+                self.assertIsNone(text)
+                self.assertTrue(log_mock.called)
+                self.assertIn("Bad things happened", log_mock.call_args[0][0])

ietf/doc/utils.py

@@ -1081,7 +1081,7 @@ def build_file_urls(doc: Union[Document, DocHistory]):
             label = "plain text" if t == "txt" else t
             file_urls.append((label, base + doc.name + "-" + doc.rev + "." + t))
 
-        if doc.text():
+        if doc.text_exists():
             file_urls.append(("htmlized", urlreverse('ietf.doc.views_doc.document_html', kwargs=dict(name=doc.name, rev=doc.rev))))
             file_urls.append(("pdfized", urlreverse('ietf.doc.views_doc.document_pdfized', kwargs=dict(name=doc.name, rev=doc.rev))))
         file_urls.append(("bibtex", urlreverse('ietf.doc.views_doc.document_bibtex',kwargs=dict(name=doc.name,rev=doc.rev))))