fix #3840: epub_uid should be XML Name

CHANGES

@@ -38,6 +38,7 @@ Bugs fixed
 * #3873: Failure of deprecation warning mechanism of
   ``sphinx.util.compat.Directive``
 * #3874: Bogus warnings for "citation not referenced" for cross-file citations
+* #3840: make checking ``epub_uid`` strict
 
 Testing
 --------

doc/conf.py

@@ -34,6 +34,7 @@ epub_theme = 'epub'
 epub_basename = 'sphinx'
 epub_author = 'Georg Brandl'
 epub_publisher = 'http://sphinx-doc.org/'
+epub_uid = 'web-site'
 epub_scheme = 'url'
 epub_identifier = epub_publisher
 epub_pre_files = [('index.xhtml', 'Welcome')]

doc/config.rst

@@ -1360,7 +1360,10 @@ the `Dublin Core metadata <http://dublincore.org/>`_.
 .. confval:: epub_uid
 
    A unique identifier for the document.  This is put in the Dublin Core
-   metadata.  You may use a random string.  The default value is ``'unknown'``.
+   metadata.  You may use a
+   `XML's Name format <https://www.w3.org/TR/REC-xml/#NT-NameStartChar>`_ string.
+   You can't use hyphen, period, numbers as a first character.
+   The default value is ``'unknown'``.
 
 .. confval:: epub_cover
 

sphinx/builders/epub3.py

@@ -17,7 +17,7 @@ from collections import namedtuple
 from sphinx import package_dir
 from sphinx.config import string_classes, ENUM
 from sphinx.builders import _epub_base
-from sphinx.util import logging
+from sphinx.util import logging, xmlname_checker
 from sphinx.util.fileutil import copy_asset_file
 
 if False:
@@ -89,8 +89,8 @@ class Epub3Builder(_epub_base.EpubBuilder):
                 'conf value "epub_language" (or "language") '
                 'should not be empty for EPUB3')
         # <package> unique-identifier attribute
-        if not self.app.config.epub_uid:
+        if not xmlname_checker().match(self.app.config.epub_uid):
-            self.app.warn('conf value "epub_uid" should not be empty for EPUB3')
+            self.app.warn('conf value "epub_uid" should be XML NAME for EPUB3')
         # dc:title
         if not self.app.config.epub_title:
             self.app.warn(

sphinx/util/__init__.py

@@ -631,3 +631,36 @@ def epoch_to_rfc1123(epoch):
 
 def rfc1123_to_epoch(rfc1123):
     return mktime(strptime(rfc1123, '%a, %d %b %Y %H:%M:%S %Z'))
+
+
+def xmlname_checker():
+    # https://www.w3.org/TR/REC-xml/#NT-Name
+    # Only Python 3.3 or newer support character code in regular expression
+    name_start_chars = [
+        u':', [u'A', u'Z'], u'_',  [u'a', u'z'], [u'\u00C0', u'\u00D6'],
+        [u'\u00D8', u'\u00F6'], [u'\u00F8', u'\u02FF'], [u'\u0370', u'\u037D'],
+        [u'\u037F', u'\u1FFF'], [u'\u200C', u'\u200D'], [u'\u2070', u'\u218F'],
+        [u'\u2C00', u'\u2FEF'], [u'\u3001', u'\uD7FF'], [u'\uF900', u'\uFDCF'],
+        [u'\uFDF0', u'\uFFFD']]
+
+    if sys.version_info.major == 3:
+        name_start_chars.append([u'\U00010000', u'\U000EFFFF'])
+
+    name_chars = [
+        u"\\-", u"\\.", [u'0', u'9'], u'\u00B7', [u'\u0300', u'\u036F'],
+        [u'\u203F', u'\u2040']
+    ]
+
+    def convert(entries, splitter=u'|'):
+        results = []
+        for entry in entries:
+            if isinstance(entry, list):
+                results.append(u'[%s]' % convert(entry, u'-'))
+            else:
+                results.append(entry)
+        return splitter.join(results)
+
+    start_chars_regex = convert(name_start_chars)
+    name_chars_regex = convert(name_chars)
+    return re.compile(u'(%s)(%s|%s)*' % (
+        start_chars_regex, start_chars_regex, name_chars_regex))

tests/test_build_epub.py

@@ -245,3 +245,5 @@ def test_epub_writing_mode(app):
     # vertical / writing-mode (CSS)
     css = (app.outdir / '_static' / 'epub.css').text()
     assert 'writing-mode: vertical-rl;' in css
+
+

tests/test_util.py

@@ -14,7 +14,8 @@ from mock import patch
 
 from sphinx.util import logging
 from sphinx.util import (
-    display_chunk, encode_uri, parselinenos, split_docinfo, status_iterator
+    display_chunk, encode_uri, parselinenos, split_docinfo, status_iterator,
+    xmlname_checker
 )
 
 from sphinx.testing.util import strip_escseq
@@ -115,3 +116,11 @@ def test_parselinenos():
         parselinenos('-', 10)
     with pytest.raises(ValueError):
         parselinenos('3-1', 10)
+
+
+
+def test_xmlname_check():
+    checker = xmlname_checker()
+    assert checker.match('id-pub')
+    assert checker.match('webpage')
+    assert not checker.match('1bfda21')