Extend paratext project settings class to support daughter translations

Author: Enkidu93

machine/corpora/paratext_project_settings.py

@@ -8,6 +8,7 @@
 
 @dataclass
 class ParatextProjectSettings:
+    guid: str
     name: str
     full_name: str
     encoding: str
@@ -20,6 +21,9 @@ class ParatextProjectSettings:
     biblical_terms_project_name: str
     biblical_terms_file_name: str
     language_code: Optional[str]
+    translation_type: str
+    parent_guid: Optional[str] = None
+    parent_name: Optional[str] = None
 
     def get_book_id(self, file_name: str) -> Optional[str]:
         """Returns None when the file name doesn't match the pattern of a book file name for the project."""
@@ -57,6 +61,18 @@ def get_all_scripture_book_ids(self) -> Iterable[str]:
         for book_id in get_scripture_books():
             yield book_id
 
+    def has_parent(self) -> bool:
+        return self.parent_guid is not None
+
+    def is_daughter_project_of(self, other_project: "ParatextProjectSettings") -> bool:
+        if not self.has_parent():
+            return False
+        return self.parent_guid == other_project.guid
+
+    def set_parent_project(self, other_project: "ParatextProjectSettings"):
+        # TODO anything else we should set? From what I can tell, stylesheets are not inherited
+        self.versification = other_project.versification
+
 
 def _get_book_file_name_digits(book_id: str) -> str:
     book_num = book_id_to_number(book_id)

machine/corpora/paratext_project_settings_parser_base.py

@@ -1,4 +1,5 @@
 from abc import ABC
+from typing import Optional
 from xml.etree import ElementTree
 
 from ..scripture.verse_ref import Versification
@@ -9,8 +10,13 @@
 
 
 class ParatextProjectSettingsParserBase(ABC):
-    def __init__(self, paratext_project_file_handler: ParatextProjectFileHandler):
+    def __init__(
+        self,
+        paratext_project_file_handler: ParatextProjectFileHandler,
+        parent_paratext_project_file_handler: Optional[ParatextProjectFileHandler] = None,
+    ):
         self._paratext_project_file_handler = paratext_project_file_handler
+        self._parent_paratext_project_file_handler = parent_paratext_project_file_handler
 
     def parse(self) -> ParatextProjectSettings:
         settings_file_name = "Settings.xml"
@@ -21,6 +27,7 @@ def parse(self) -> ParatextProjectSettings:
         with self._paratext_project_file_handler.open(settings_file_name) as stream:
             settings_tree = ElementTree.parse(stream)
 
+        guid = settings_tree.getroot().findtext("Guid", "")
         name = settings_tree.getroot().findtext("Name", "")
         full_name = settings_tree.getroot().findtext("FullName", "")
         encoding_str = settings_tree.getroot().findtext("Encoding", "65001")
@@ -36,7 +43,6 @@ def parse(self) -> ParatextProjectSettings:
         versification_type = int(settings_tree.getroot().findtext("Versification", "4"))
         versification = Versification.get_builtin(versification_type)
         if self._paratext_project_file_handler.exists("custom.vrs"):
-            guid = settings_tree.getroot().findtext("Guid", "")
             versification_name = f"{versification.name}-{guid}"
             versification = Versification.load(
                 self._paratext_project_file_handler.open("custom.vrs"),
@@ -77,12 +83,23 @@ def parse(self) -> ParatextProjectSettings:
             )
         language_code = None
         language_iso_code_setting = settings_tree.getroot().findtext("LanguageIsoCode", "")
-        if language_iso_code_setting:
-            language_iso_code_setting_parts = settings_tree.getroot().findtext("LanguageIsoCode", "").split(":")
+        if language_iso_code_setting is not None:
+            language_iso_code_setting_parts = language_iso_code_setting.split(":")
             if language_iso_code_setting_parts:
                 language_code = language_iso_code_setting_parts[0]
 
+        translation_info_setting = settings_tree.getroot().findtext("TranslationInfo")
+        translation_type = "Standard"
+        parent_name = None
+        parent_guid = None
+        if translation_info_setting is not None:
+            translation_info_setting_parts = translation_info_setting.split(":")
+            translation_type = translation_info_setting_parts[0]
+            parent_name = translation_info_setting_parts[1] if translation_info_setting_parts[1] != "" else None
+            parent_guid = translation_info_setting_parts[2] if translation_info_setting_parts[2] != "" else None
+
         return ParatextProjectSettings(
+            guid,
             name,
             full_name,
             encoding,
@@ -95,4 +112,7 @@ def parse(self) -> ParatextProjectSettings:
             parts[1],
             parts[2],
             language_code,
+            translation_type,
+            parent_guid,
+            parent_name,
         )

tests/corpora/test_file_paratext_project_settings_parser.py

@@ -9,3 +9,11 @@ def test_parse_custom_stylesheet() -> None:
     test_tag = settings.stylesheet.get_tag("test")
     assert test_tag.style_type is UsfmStyleType.CHARACTER
     assert test_tag.text_type is UsfmTextType.OTHER
+
+
+def test_is_daughter_project() -> None:
+    parser = FileParatextProjectSettingsParser(USFM_TEST_PROJECT_PATH)
+    settings = parser.parse()
+    assert settings.has_parent()
+    assert settings.is_daughter_project_of(settings)
+    assert settings.translation_type == "Standard"

tests/corpora/test_paratext_project_settings.py

@@ -115,6 +115,7 @@ def test_get_book_id_wrong_book_part_book_num_book_id() -> None:
 
 def _create_settings(file_name_form: str) -> ParatextProjectSettings:
     return ParatextProjectSettings(
+        "id",
         "Name",
         "Name",
         "utf-8",
@@ -127,4 +128,5 @@ def _create_settings(file_name_form: str) -> ParatextProjectSettings:
         "",
         "BiblicalTerms.xml",
         "en",
+        "Standard",
     )

tests/corpora/test_zip_paratext_project_settings_parser.py

@@ -18,6 +18,14 @@ def test_parse_custom_stylesheet() -> None:
         assert test_tag.text_type is UsfmTextType.OTHER
 
 
+def test_is_daughter_project() -> None:
+    with _TestEnvironment() as env:
+        settings = env.parser.parse()
+        assert settings.has_parent()
+        assert settings.is_daughter_project_of(settings)
+        assert settings.translation_type == "Standard"
+
+
 class _TestEnvironment(ContextManager["_TestEnvironment"]):
     def __init__(self) -> None:
         self._temp_dir = TemporaryDirectory()

tests/testutils/data/usfm/Tes/Settings.xml

@@ -24,7 +24,7 @@
   <AllowReadAccess>F</AllowReadAccess>
   <AllowSharingWithSLDR>F</AllowSharingWithSLDR>
   <Visibility>Public</Visibility>
-  <TranslationInfo>Standard::</TranslationInfo>
+  <TranslationInfo>Standard:Tes:a7e0b3ce0200736062f9f810a444dbfbe64aca35</TranslationInfo>
   <EncodingConverter />
   <UsfmVersion>3</UsfmVersion>
   <ParallelPassagesBooks>000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000</ParallelPassagesBooks>

tests/testutils/memory_paratext_project_file_handler.py

@@ -26,6 +26,7 @@ def create_stylesheet(self, file_name):
 class DefaultParatextProjectSettings(ParatextProjectSettings):
     def __init__(
         self,
+        guid: str = "id",
         name: str = "Test",
         full_name: str = "TestProject",
         encoding: Optional[str] = None,
@@ -38,9 +39,13 @@ def __init__(
         biblical_terms_project_name: str = "Test",
         biblical_terms_file_name: str = "ProjectBiblicalTerms.xml",
         language_code: str = "en",
+        translation_type: str = "Standard",
+        parent_guid: Optional[str] = None,
+        parent_name: Optional[str] = None,
     ):
 
         super().__init__(
+            guid,
             name,
             full_name,
             encoding if encoding is not None else "utf-8",
@@ -53,4 +58,7 @@ def __init__(
             biblical_terms_project_name,
             biblical_terms_file_name,
             language_code,
+            translation_type,
+            parent_guid,
+            parent_name,
         )