Support daughter Paratext projects

machine/corpora/file_paratext_project_settings_parser.py

@@ -1,8 +1,11 @@
+from typing import Optional
+
 from ..utils.typeshed import StrPath
 from .file_paratext_project_file_handler import FileParatextProjectFileHandler
+from .paratext_project_settings import ParatextProjectSettings
 from .paratext_project_settings_parser_base import ParatextProjectSettingsParserBase
 
 
 class FileParatextProjectSettingsParser(ParatextProjectSettingsParserBase):
-    def __init__(self, project_dir: StrPath) -> None:
-        super().__init__(FileParatextProjectFileHandler(project_dir))
+    def __init__(self, project_dir: StrPath, parent_settings: Optional[ParatextProjectSettings] = None) -> None:
+        super().__init__(FileParatextProjectFileHandler(project_dir), parent_settings)

machine/corpora/file_paratext_project_text_updater.py

@@ -1,16 +1,18 @@
 from pathlib import Path
-from typing import BinaryIO
+from typing import BinaryIO, Optional
 
 from ..utils.typeshed import StrPath
 from .file_paratext_project_file_handler import FileParatextProjectFileHandler
 from .file_paratext_project_settings_parser import FileParatextProjectSettingsParser
+from .paratext_project_settings import ParatextProjectSettings
 from .paratext_project_text_updater_base import ParatextProjectTextUpdaterBase
 
 
 class FileParatextProjectTextUpdater(ParatextProjectTextUpdaterBase):
-    def __init__(self, project_dir: StrPath) -> None:
+    def __init__(self, project_dir: StrPath, parent_settings: Optional[ParatextProjectSettings] = None) -> None:
         super().__init__(
-            FileParatextProjectFileHandler(project_dir), FileParatextProjectSettingsParser(project_dir).parse()
+            FileParatextProjectFileHandler(project_dir),
+            FileParatextProjectSettingsParser(project_dir, parent_settings).parse(),
         )
 
         self._project_dir = project_dir

machine/corpora/file_paratext_project_versification_error_detector.py

@@ -1,11 +1,15 @@
+from typing import Optional
+
 from ..utils.typeshed import StrPath
 from .file_paratext_project_file_handler import FileParatextProjectFileHandler
 from .file_paratext_project_settings_parser import FileParatextProjectSettingsParser
+from .paratext_project_settings import ParatextProjectSettings
 from .paratext_project_versification_error_detector_base import ParatextProjectVersificationErrorDetectorBase
 
 
 class FileParatextProjectVersificationErrorDetector(ParatextProjectVersificationErrorDetectorBase):
-    def __init__(self, project_dir: StrPath) -> None:
+    def __init__(self, project_dir: StrPath, parent_settings: Optional[ParatextProjectSettings] = None) -> None:
         super().__init__(
-            FileParatextProjectFileHandler(project_dir), FileParatextProjectSettingsParser(project_dir).parse()
+            FileParatextProjectFileHandler(project_dir),
+            FileParatextProjectSettingsParser(project_dir, parent_settings).parse(),
         )

machine/corpora/paratext_backup_text_corpus.py

@@ -1,4 +1,4 @@
-from typing import List
+from typing import List, Optional
 from zipfile import ZipFile
 
 from ..utils.typeshed import StrPath
@@ -8,9 +8,22 @@
 
 
 class ParatextBackupTextCorpus(ScriptureTextCorpus):
-    def __init__(self, filename: StrPath, include_markers: bool = False, include_all_text: bool = False) -> None:
+    def __init__(
+        self,
+        filename: StrPath,
+        include_markers: bool = False,
+        include_all_text: bool = False,
+        parent_filename: Optional[StrPath] = None,
+    ) -> None:
+
+        parent_settings = None
+        if parent_filename is not None:
+            with ZipFile(parent_filename, "r") as parent_archive:
+                parent_parser = ZipParatextProjectSettingsParser(parent_archive)
+                parent_settings = parent_parser.parse()
+
         with ZipFile(filename, "r") as archive:
-            parser = ZipParatextProjectSettingsParser(archive)
+            parser = ZipParatextProjectSettingsParser(archive, parent_settings)
             settings = parser.parse()
 
             versification = settings.versification

machine/corpora/paratext_project_settings.py

@@ -8,6 +8,7 @@
 
 @dataclass
 class ParatextProjectSettings:
+    guid: str
     name: str
     full_name: str
     encoding: str
@@ -20,6 +21,10 @@ 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
+    _parent: Optional["ParatextProjectSettings"] = 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 +62,26 @@ def get_all_scripture_book_ids(self) -> Iterable[str]:
         for book_id in get_scripture_books():
             yield book_id
 
+    @property
+    def has_parent(self) -> bool:
+        return self.parent_guid is not None
+
+    @property
+    def parent(self) -> Optional["ParatextProjectSettings"]:
+        return self._parent
+
+    @parent.setter
+    def parent(self, value: "ParatextProjectSettings"):
+        if not self.is_daughter_project_of(value):
+            raise ValueError(f"Project {value.name} is not the parent project of project {self.name}.")
+        self._parent = value
+        self.versification = value.versification
+
+    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 _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_settings: Optional[ParatextProjectSettings] = None,
+    ):
         self._paratext_project_file_handler = paratext_project_file_handler
+        self.parent_paratext_project_settings = parent_paratext_project_settings
 
     def parse(self) -> ParatextProjectSettings:
         settings_file_name = "Settings.xml"
@@ -21,9 +27,10 @@ def parse(self) -> ParatextProjectSettings:
         with self._paratext_project_file_handler.open(settings_file_name) as stream:
             settings_tree = ElementTree.parse(stream)
 
-        name = settings_tree.getroot().findtext("Name", "")
-        full_name = settings_tree.getroot().findtext("FullName", "")
-        encoding_str = settings_tree.getroot().findtext("Encoding", "65001")
+        guid: str = settings_tree.getroot().findtext("Guid", "")
+        name: str = settings_tree.getroot().findtext("Name", "")
+        full_name: str = settings_tree.getroot().findtext("FullName", "")
+        encoding_str: str = settings_tree.getroot().findtext("Encoding", "65001")
         code_page = parse_integer(encoding_str)
         if code_page is None:
             raise NotImplementedError(
@@ -36,14 +43,13 @@ 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"),
                 versification,
                 versification_name,
             )
-        stylesheet_file_name = settings_tree.getroot().findtext("StyleSheet", "usfm.sty")
+        stylesheet_file_name: str = settings_tree.getroot().findtext("StyleSheet", "usfm.sty")
         if (
             not self._paratext_project_file_handler.exists(stylesheet_file_name)
             and stylesheet_file_name != "usfm_sb.sty"
@@ -65,7 +71,7 @@ def parse(self) -> ParatextProjectSettings:
             post_part = naming_elem.get("PostPart")
             if post_part:
                 suffix = post_part
-        biblical_terms_list_setting = settings_tree.getroot().findtext("BiblicalTermsListSetting")
+        biblical_terms_list_setting: Optional[str] = settings_tree.getroot().findtext("BiblicalTermsListSetting")
         if biblical_terms_list_setting is None:
             # Default to Major::BiblicalTerms.xml to mirror Paratext behavior
             biblical_terms_list_setting = "Major::BiblicalTerms.xml"
@@ -76,13 +82,24 @@ def parse(self) -> ParatextProjectSettings:
                 f" is not in the expected format (e.g., Major::BiblicalTerms.xml) but is {biblical_terms_list_setting}."
             )
         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(":")
+        language_iso_code_setting: Optional[str] = settings_tree.getroot().findtext("LanguageIsoCode", "")
+        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]
 
-        return ParatextProjectSettings(
+        translation_info_setting: Optional[str] = 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
+
+        settings = ParatextProjectSettings(
+            guid,
             name,
             full_name,
             encoding,
@@ -95,4 +112,12 @@ def parse(self) -> ParatextProjectSettings:
             parts[1],
             parts[2],
             language_code,
+            translation_type,
+            parent_guid,
+            parent_name,
         )
+
+        if self.parent_paratext_project_settings is not None and settings.has_parent:
+            settings.parent = self.parent_paratext_project_settings
+
+        return settings

machine/corpora/paratext_text_corpus.py

@@ -1,5 +1,5 @@
 from pathlib import Path
-from typing import List
+from typing import List, Optional
 
 from ..utils.typeshed import StrPath
 from .file_paratext_project_settings_parser import FileParatextProjectSettingsParser
@@ -8,8 +8,20 @@
 
 
 class ParatextTextCorpus(ScriptureTextCorpus):
-    def __init__(self, project_dir: StrPath, include_markers: bool = False, include_all_text: bool = False) -> None:
-        parser = FileParatextProjectSettingsParser(project_dir)
+    def __init__(
+        self,
+        project_dir: StrPath,
+        include_markers: bool = False,
+        include_all_text: bool = False,
+        parent_project_dir: Optional[StrPath] = None,
+    ) -> None:
+
+        parent_settings = None
+        if parent_project_dir is not None:
+            parent_parser = FileParatextProjectSettingsParser(parent_project_dir)
+            parent_settings = parent_parser.parse()
+
+        parser = FileParatextProjectSettingsParser(project_dir, parent_settings)
         settings = parser.parse()
 
         versification = settings.versification