Merge pull request #5400 from tk0miya/5399_find_catalog_crashes

Fix #5399: Sphinx crashes if unknown po file exists

CHANGES

@@ -17,6 +17,8 @@ Deprecated
 
 * ``sphinx.io.SphinxI18nReader.set_lineno_for_reporter()`` is deprecated
 * ``sphinx.io.SphinxI18nReader.line`` is deprecated
+* ``sphinx.util.i18n.find_catalog_source_file()`` has changed; the
+  *gettext_compact* argument has been deprecated
 
 Features added
 --------------
@@ -41,6 +43,7 @@ Bugs fixed
 * #2720, #4034: Incorrect links with ``:download:``, duplicate names, and
   parallel builds
 * #5290: autodoc: failed to analyze source code in egg package
+* #5399: Sphinx crashes if unknown po file exists
 
 Testing
 --------

doc/extdev/index.rst

@@ -136,6 +136,12 @@ The following is a list of deprecated interface.
      - 4.0
      - :confval:`autodoc_default_options`
 
+   * - ``gettext_compact`` arguments of
+       ``sphinx.util.i18n.find_catalog_source_files()``
+     - 1.8
+     - 3.0
+     - N/A
+
    * - ``sphinx.io.SphinxI18nReader.set_lineno_for_reporter()``
      - 1.8
      - 3.0

sphinx/builders/__init__.py

@@ -261,7 +261,6 @@ class Builder(object):
             [path.join(self.srcdir, x) for x in self.config.locale_dirs],
             self.config.language,
             charset=self.config.source_encoding,
-            gettext_compact=self.config.gettext_compact,
             force_all=True,
             excluded=Matcher(['**/.?**']))
         message = __('all of %d po files') % len(catalogs)
@@ -284,7 +283,6 @@ class Builder(object):
             self.config.language,
             domains=list(specified_domains),
             charset=self.config.source_encoding,
-            gettext_compact=self.config.gettext_compact,
             excluded=Matcher(['**/.?**']))
         message = __('targets for %d po files that are specified') % len(catalogs)
         self.compile_catalogs(catalogs, message)
@@ -295,7 +293,6 @@ class Builder(object):
             [path.join(self.srcdir, x) for x in self.config.locale_dirs],
             self.config.language,
             charset=self.config.source_encoding,
-            gettext_compact=self.config.gettext_compact,
             excluded=Matcher(['**/.?**']))
         message = __('targets for %d po files that are out of date') % len(catalogs)
         self.compile_catalogs(catalogs, message)

sphinx/util/i18n.py

@@ -12,6 +12,7 @@ import gettext
 import io
 import os
 import re
+import warnings
 from collections import namedtuple
 from datetime import datetime
 from os import path
@@ -20,6 +21,7 @@ import babel.dates
 from babel.messages.mofile import write_mo
 from babel.messages.pofile import read_po
 
+from sphinx.deprecation import RemovedInSphinx30Warning
 from sphinx.errors import SphinxError
 from sphinx.locale import __
 from sphinx.util import logging
@@ -103,7 +105,7 @@ def find_catalog_files(docname, srcdir, locale_dirs, lang, compaction):
     return files  # type: ignore
 
 
-def find_catalog_source_files(locale_dirs, locale, domains=None, gettext_compact=False,
+def find_catalog_source_files(locale_dirs, locale, domains=None, gettext_compact=None,
                               charset='utf-8', force_all=False,
                               excluded=Matcher([])):
     # type: (List[unicode], unicode, List[unicode], bool, unicode, bool, Matcher) -> Set[CatalogInfo]  # NOQA
@@ -115,14 +117,15 @@ def find_catalog_source_files(locale_dirs, locale, domains=None, gettext_compact
     :param str locale: a language as `'en'`
     :param list domains: list of domain names to get. If empty list or None
        is specified, get all domain names. default is None.
-    :param boolean gettext_compact:
-       * False: keep domains directory structure (default).
-       * True: domains in the sub directory will be merged into 1 file.
     :param boolean force_all:
        Set True if you want to get all catalogs rather than updated catalogs.
        default is False.
     :return: [CatalogInfo(), ...]
     """
+    if gettext_compact is not None:
+        warnings.warn('gettext_compact argument for find_catalog_source_files() '
+                      'is deprecated.', RemovedInSphinx30Warning)
+
     catalogs = set()  # type: Set[CatalogInfo]
 
     if not locale:
@@ -143,10 +146,7 @@ def find_catalog_source_files(locale_dirs, locale, domains=None, gettext_compact
                 if excluded(path.join(relpath(dirpath, base_dir), filename)):
                     continue
                 base = path.splitext(filename)[0]
-                domain = relpath(path.join(dirpath, base), base_dir)
+                domain = relpath(path.join(dirpath, base), base_dir).replace(path.sep, SEP)
-                if gettext_compact and path.sep in domain:
-                    domain = path.split(domain)[0]
-                domain = domain.replace(path.sep, SEP)
                 if domains and domain not in domains:
                     continue
                 cat = CatalogInfo(base_dir, domain, charset)

tests/test_util_i18n.py

@@ -154,7 +154,7 @@ def test_get_catalogs_with_compact(tempdir):
 
     catalogs = i18n.find_catalog_source_files([tempdir / 'loc1'], 'xx', gettext_compact=True)
     domains = set(c.domain for c in catalogs)
-    assert domains == set(['test1', 'test2', 'sub'])
+    assert domains == set(['test1', 'test2', 'sub/test3', 'sub/test4'])
 
 
 def test_get_catalogs_excluded(tempdir):