Improve --no-empty-msgstr implementation (#321)

src/mdpo/md2po/__main__.py

@@ -33,6 +33,7 @@
 from mdpo.md2po import Md2Po
 from mdpo.md4c import DEFAULT_MD4C_GENERIC_PARSER_EXTENSIONS
 from mdpo.po import (
+    check_empty_msgstrs_in_filepaths,
     check_fuzzy_entries_in_filepaths,
     check_obsolete_entries_in_filepaths,
 )
@@ -274,17 +275,22 @@ def run(args=frozenset()):
                 exitcode = 4
 
         if opts.no_empty_msgstr:
-            for entry in pofile:
-                if not entry.msgstr:
-                    if not opts.quiet:
+            locations = list(check_empty_msgstrs_in_filepaths(
+                (opts.po_filepath,),
+            ))
+            if locations:
+                if len(locations) > 2:  # noqa PLR2004
+                    sys.stderr.write(
+                        f'Found {len(locations)} empty msgstrs:\n',
+                    )
+                    for location in locations:
+                        sys.stderr.write(f'{location}\n')
+                else:
+                    for location in locations:
                         sys.stderr.write(
-                            (
-                                f"Empty msgstr found at {opts.po_filepath}"
-                                " and passed '--no-empty-msgstr'\n"
-                            ),
+                            f'Found empty msgstr at {location}\n',
                         )
-                        exitcode = 5
-                    break
+                exitcode = 5
 
     return (pofile, exitcode)
 

src/mdpo/md2po2md/__init__.py

@@ -6,6 +6,7 @@
 from mdpo.md2po import Md2Po
 from mdpo.md4c import DEFAULT_MD4C_GENERIC_PARSER_EXTENSIONS
 from mdpo.po import (
+    check_empty_msgstrs_in_filepaths,
     check_fuzzy_entries_in_filepaths,
     check_obsolete_entries_in_filepaths,
 )
@@ -30,6 +31,7 @@ def markdown_to_pofile_to_markdown(
     _check_saved_files_changed=False,
     no_obsolete=False,
     no_fuzzy=False,
+    no_empty_msgstr=False,
 ):
     """Translate a set of Markdown files using PO files.
 
@@ -78,6 +80,7 @@ def markdown_to_pofile_to_markdown(
             ``pofile_to_markdown`` function.
         no_obsolete (bool): If ``True``, check for obsolete entries in PO files.
         no_fuzzy (bool): If ``True``, check for fuzzy entries in PO files.
+        no_empty_msgstr (bool): If ``True``, check for empty ``msgstr`` entries.
     """
     if '{lang}' not in output_paths_schema:
         raise ValueError(
@@ -107,7 +110,7 @@ def markdown_to_pofile_to_markdown(
     _saved_files_changed = None if not _check_saved_files_changed else False
     obsoletes = []
     fuzzies = []
-    empty = False
+    empties = []
 
     for filepath in input_paths_glob_:
         for lang in langs:
@@ -165,12 +168,6 @@ def markdown_to_pofile_to_markdown(
             if _check_saved_files_changed and _saved_files_changed is False:
                 _saved_files_changed = md2po._saved_files_changed
 
-            if not empty:
-                for entry in md2po.pofile:
-                    if not entry.msgstr:
-                        empty = True
-                        break
-
             # po2md
             po2md = Po2Md(
                 [po_filepath],
@@ -199,13 +196,9 @@ def markdown_to_pofile_to_markdown(
                     check_fuzzy_entries_in_filepaths([po_filepath]),
                 )
 
-            if not empty:
-                for pofile in po2md.pofiles:
-                    for entry in pofile:
-                        if not entry.msgstr:
-                            empty = True
-                            break
-                    if empty:
-                        break
+            if no_empty_msgstr:
+                empties.extend(
+                    check_empty_msgstrs_in_filepaths([po_filepath]),
+                )
 
-    return (_saved_files_changed, obsoletes, fuzzies, empty)
+    return (_saved_files_changed, obsoletes, fuzzies, empties)

src/mdpo/md2po2md/__main__.py

@@ -144,13 +144,14 @@ def run(args=frozenset()):
             '_check_saved_files_changed': opts.check_saved_files_changed,
             'no_obsolete': opts.no_obsolete,
             'no_fuzzy': opts.no_fuzzy,
+            'no_empty_msgstr': opts.no_empty_msgstr,
         }
 
         (
             _saved_files_changed,
             obsoletes,
             fuzzies,
-            empty,
+            empties,
         ) = markdown_to_pofile_to_markdown(
             opts.langs,
             opts.input_paths_glob,
@@ -189,14 +190,19 @@ def run(args=frozenset()):
 
             exitcode = 4
 
-        if opts.no_empty_msgstr and empty:
-            exitcode = 5
-
-            if not opts.quiet:
+        if opts.no_empty_msgstr:
+            if len(empties) > 2:  # noqa PLR2004
                 sys.stderr.write(
-                    "Empty msgstr found at PO files and"
-                    " passed '--no-empty-msgstr'\n",
+                    f'Found {len(empties)} empty msgstrs:\n',
                 )
+                for location in empties:
+                    sys.stderr.write(f'{location}\n')
+            else:
+                for location in empties:
+                    sys.stderr.write(
+                        f'Found empty msgstr at {location}\n',
+                    )
+            exitcode = 5
 
     return exitcode
 

src/mdpo/mdpo2html/__main__.py

@@ -23,6 +23,7 @@
 from mdpo.io import environ
 from mdpo.mdpo2html import MdPo2HTML
 from mdpo.po import (
+    check_empty_msgstrs_in_filepaths,
     check_fuzzy_entries_in_filepaths,
     check_obsolete_entries_in_filepaths,
     paths_or_globs_to_unique_pofiles,
@@ -168,16 +169,22 @@ def run(args=frozenset()):
                 exitcode = 4
 
         if opts.no_empty_msgstr:
-            for pofile in mdpo2html.pofiles:
-                for entry in pofile:
-                    if not entry.msgstr:
-                        if not opts.quiet:
-                            sys.stderr.write(
-                                f"Empty msgstr for msgid '{entry.msgid}'"
-                                " found at PO files and"
-                                " passed '--no-empty-msgstr'\n",
-                            )
-                        exitcode = 5
+            locations = list(check_empty_msgstrs_in_filepaths(
+                (opts.po_filepath,),
+            ))
+            if locations:
+                if len(locations) > 2:  # noqa PLR2004
+                    sys.stderr.write(
+                        f'Found {len(locations)} empty msgstrs:\n',
+                    )
+                    for location in locations:
+                        sys.stderr.write(f'{location}\n')
+                else:
+                    for location in locations:
+                        sys.stderr.write(
+                            f'Found empty msgstr at {location}\n',
+                        )
+                exitcode = 5
 
     return (output, exitcode)
 

src/mdpo/po.py

@@ -208,6 +208,25 @@ def check_fuzzy_entries_in_filepaths(filenames):
         )
 
 
+def check_empty_msgstrs_in_filepaths(filenames):
+    """Warns about all empty msgstr found in a set of PO files.
+
+    Args:
+        filenames (list): Set of file names to check.
+
+    Returns:
+        list(str): error messages produced.
+    """
+    for filename in filenames:
+        with open(filename, 'rb') as f:
+            content_lines = f.readlines()
+
+        yield from parse_empty_msgstr_from_content_lines(
+            content_lines,
+            location_prefix=f'{filename}:',
+        )
+
+
 def parse_obsoletes_from_content_lines(
     content_lines,
     location_prefix='line ',
@@ -247,3 +266,26 @@ def parse_fuzzy_from_content_lines(
     for i, line in enumerate(content_lines):
         if line.startswith(b'#,') and b'fuzzy' in line:
             yield f'{location_prefix}{i + 1}'
+
+
+def parse_empty_msgstr_from_content_lines(
+    content_lines,
+    location_prefix='line ',
+):
+    """Warns about all empty msgstr found in a set of PO files.
+
+    Args:
+        content_lines (list): Set of content lines to check.
+        location_prefix (str): Prefix to use in the location message.
+
+    Returns:
+        list(str): error locations found.
+    """
+    n_lines = len(content_lines)
+    for i, line in enumerate(content_lines):
+        if (
+            line.startswith(b'msgstr ""') or line.startswith(b'#~ msgstr ""')
+        ) and not content_lines[i - 1].startswith(b'msgid ""'):
+            next_i = i + 1
+            if next_i == n_lines or not content_lines[next_i].strip():
+                yield f'{location_prefix}{next_i}'

src/mdpo/po2md/__main__.py

@@ -26,6 +26,7 @@
 )
 from mdpo.io import environ
 from mdpo.po import (
+    check_empty_msgstrs_in_filepaths,
     check_fuzzy_entries_in_filepaths,
     check_obsolete_entries_in_filepaths,
     paths_or_globs_to_unique_pofiles,
@@ -192,17 +193,22 @@ def run(args=frozenset()):
                 exitcode = 4
 
         if opts.no_empty_msgstr:
-            for pofile in po2md.pofiles:
-                for entry in pofile:
-                    if not entry.msgstr:
-                        if not opts.quiet:
-                            sys.stderr.write(
-                                (
-                                    f"Empty msgstr found at {opts.po_filepath}"
-                                    " and passed '--no-empty-msgstr'\n"
-                                ),
-                            )
-                        exitcode = 5
+            locations = list(check_empty_msgstrs_in_filepaths(
+                (opts.po_filepath,),
+            ))
+            if locations:
+                if len(locations) > 2:  # noqa PLR2004
+                    sys.stderr.write(
+                        f'Found {len(locations)} empty msgstrs:\n',
+                    )
+                    for location in locations:
+                        sys.stderr.write(f'{location}\n')
+                else:
+                    for location in locations:
+                        sys.stderr.write(
+                            f'Found empty msgstr at {location}\n',
+                        )
+                exitcode = 5
 
     return (output, exitcode)
 

tests/test_unit/test_md2po/test_md2po_cli.py

@@ -893,14 +893,14 @@ def test_no_empty_mgstr(capsys, arg, tmp_file):
 
     with tmp_file(po_input, '.po') as filename:
         pofile, exitcode = run([arg, '-p', filename, '--no-location', 'Hello'])
-    stdout, stderr = capsys.readouterr()
+        stdout, stderr = capsys.readouterr()
 
-    assert exitcode == 5
-    assert f'{pofile}\n' == po_input
-    assert stdout == po_input
-    assert stderr == (
-        f"Empty msgstr found at {filename} and passed '--no-empty-msgstr'\n"
-    )
+        assert exitcode == 5
+        assert f'{pofile}\n' == po_input
+        assert stdout == po_input
+        assert stderr == (
+            f'Found empty msgstr at {filename}:6\n'
+        )
 
     po_input = '''#
 msgid ""