Port https://github.com/sillsdev/machine/pull/353

machine/corpora/__init__.py

@@ -7,8 +7,10 @@
 from .dbl_bundle_text_corpus import DblBundleTextCorpus
 from .dictionary_alignment_corpus import DictionaryAlignmentCorpus
 from .dictionary_text_corpus import DictionaryTextCorpus
+from .file_paratext_project_file_handler import FileParatextProjectFileHandler
 from .file_paratext_project_settings_parser import FileParatextProjectSettingsParser
 from .file_paratext_project_text_updater import FileParatextProjectTextUpdater
+from .file_paratext_project_versification_error_detector import FileParatextProjectVersificationErrorDetector
 from .flatten import flatten
 from .memory_alignment_collection import MemoryAlignmentCollection
 from .memory_stream_container import MemoryStreamContainer
@@ -18,10 +20,12 @@
 from .parallel_text_row import ParallelTextRow
 from .paratext_backup_terms_corpus import ParatextBackupTermsCorpus
 from .paratext_backup_text_corpus import ParatextBackupTextCorpus
+from .paratext_project_file_handler import ParatextProjectFileHandler
 from .paratext_project_settings import ParatextProjectSettings
 from .paratext_project_settings_parser_base import ParatextProjectSettingsParserBase
 from .paratext_project_terms_parser_base import ParatextProjectTermsParserBase
 from .paratext_project_text_updater_base import ParatextProjectTextUpdaterBase
+from .paratext_project_versification_error_detector import ParatextProjectVersificationErrorDetector
 from .paratext_text_corpus import ParatextTextCorpus
 from .place_markers_usfm_update_block_handler import PlaceMarkersAlignmentInfo, PlaceMarkersUsfmUpdateBlockHandler
 from .scripture_element import ScriptureElement
@@ -70,16 +74,22 @@
 from .usfm_update_block import UsfmUpdateBlock
 from .usfm_update_block_element import UsfmUpdateBlockElement, UsfmUpdateBlockElementType
 from .usfm_update_block_handler import UsfmUpdateBlockHandler
+from .usfm_versification_error_detector import (
+    UsfmVersificationError,
+    UsfmVersificationErrorDetector,
+    UsfmVersificationErrorType,
+)
 from .usx_file_alignment_collection import UsxFileAlignmentCollection
 from .usx_file_alignment_corpus import UsxFileAlignmentCorpus
 from .usx_file_text import UsxFileText
 from .usx_file_text_corpus import UsxFileTextCorpus
 from .usx_memory_text import UsxMemoryText
 from .usx_zip_text import UsxZipText
+from .zip_paratext_project_file_handler import ZipParatextProjectFileHandler
 from .zip_paratext_project_settings_parser import ZipParatextProjectSettingsParser
-from .zip_paratext_project_settings_parser_base import ZipParatextProjectSettingsParserBase
 from .zip_paratext_project_terms_parser import ZipParatextProjectTermsParser
 from .zip_paratext_project_text_updater import ZipParatextProjectTextUpdater
+from .zip_paratext_project_versification_detector import ZipParatextProjectVersificationErrorDetector
 
 __all__ = [
     "AlignedWordPair",
@@ -95,8 +105,10 @@
     "EMPTY_SCRIPTURE_REF",
     "escape_spaces",
     "extract_scripture_corpus",
+    "FileParatextProjectFileHandler",
     "FileParatextProjectSettingsParser",
     "FileParatextProjectTextUpdater",
+    "FileParatextProjectVersificationErrorDetector",
     "flatten",
     "is_scripture",
     "lowercase",
@@ -113,14 +125,16 @@
     "ParallelTextRow",
     "ParatextBackupTermsCorpus",
     "ParatextBackupTextCorpus",
+    "ParatextProjectFileHandler",
     "ParatextProjectSettings",
     "ParatextProjectSettingsParserBase",
     "ParatextProjectTermsParserBase",
     "ParatextProjectTextUpdaterBase",
+    "ParatextProjectVersificationErrorDetector",
     "ParatextTextCorpus",
+    "parse_usfm",
     "PlaceMarkersAlignmentInfo",
     "PlaceMarkersUsfmUpdateBlockHandler",
-    "parse_usfm",
     "RtlReferenceOrder",
     "ScriptureElement",
     "ScriptureRef",
@@ -139,8 +153,8 @@
     "unescape_spaces",
     "UpdateUsfmMarkerBehavior",
     "UpdateUsfmParserHandler",
-    "UpdateUsfmTextBehavior",
     "UpdateUsfmRow",
+    "UpdateUsfmTextBehavior",
     "UsfmAttribute",
     "UsfmElementType",
     "UsfmFileText",
@@ -164,14 +178,18 @@
     "UsfmUpdateBlockElement",
     "UsfmUpdateBlockElementType",
     "UsfmUpdateBlockHandler",
+    "UsfmVersificationError",
+    "UsfmVersificationErrorDetector",
+    "UsfmVersificationErrorType",
     "UsxFileAlignmentCollection",
     "UsxFileAlignmentCorpus",
     "UsxFileText",
     "UsxFileTextCorpus",
     "UsxMemoryText",
     "UsxZipText",
+    "ZipParatextProjectFileHandler",
     "ZipParatextProjectSettingsParser",
-    "ZipParatextProjectSettingsParserBase",
     "ZipParatextProjectTermsParser",
     "ZipParatextProjectTextUpdater",
+    "ZipParatextProjectVersificationErrorDetector",
 ]

machine/corpora/file_paratext_project_file_handler.py

@@ -0,0 +1,27 @@
+from pathlib import Path
+from typing import BinaryIO, Optional
+
+from ..utils.typeshed import StrPath
+from .paratext_project_file_handler import ParatextProjectFileHandler
+from .usfm_stylesheet import UsfmStylesheet
+
+
+class FileParatextProjectFileHandler(ParatextProjectFileHandler):
+    def __init__(self, project_dir: StrPath) -> None:
+        self._project_dir = Path(project_dir)
+
+    def exists(self, file_name: str) -> bool:
+        return (self._project_dir / file_name).is_file()
+
+    def open(self, file_name: str) -> BinaryIO:
+        return open(self._project_dir / file_name, "rb")
+
+    def find(self, extension: str) -> Optional[Path]:
+        return next(self._project_dir.glob(f"*{extension}"), None)
+
+    def create_stylesheet(self, file_name: str) -> UsfmStylesheet:
+        custom_stylesheet_filename = self._project_dir / "custom.sty"
+        return UsfmStylesheet(
+            file_name,
+            custom_stylesheet_filename if custom_stylesheet_filename.is_file() else None,
+        )

machine/corpora/file_paratext_project_settings_parser.py

@@ -1,27 +1,8 @@
-from pathlib import Path
-from typing import BinaryIO, Optional
-
 from ..utils.typeshed import StrPath
+from .file_paratext_project_file_handler import FileParatextProjectFileHandler
 from .paratext_project_settings_parser_base import ParatextProjectSettingsParserBase
-from .usfm_stylesheet import UsfmStylesheet
 
 
 class FileParatextProjectSettingsParser(ParatextProjectSettingsParserBase):
     def __init__(self, project_dir: StrPath) -> None:
-        self._project_dir = Path(project_dir)
-
-    def _create_stylesheet(self, file_name: StrPath) -> UsfmStylesheet:
-        custom_stylesheet_filename = self._project_dir / "custom.sty"
-        return UsfmStylesheet(
-            file_name,
-            custom_stylesheet_filename if custom_stylesheet_filename.is_file() else None,
-        )
-
-    def _exists(self, file_name: StrPath) -> bool:
-        return (self._project_dir / file_name).is_file()
-
-    def _find(self, extension: str) -> Optional[Path]:
-        return next(self._project_dir.glob(f"*{extension}"), None)
-
-    def _open(self, file_name: StrPath) -> BinaryIO:
-        return open(self._project_dir / file_name, "rb")
+        super().__init__(FileParatextProjectFileHandler(project_dir))

machine/corpora/file_paratext_project_text_updater.py

@@ -2,13 +2,16 @@
 from typing import BinaryIO
 
 from ..utils.typeshed import StrPath
+from .file_paratext_project_file_handler import FileParatextProjectFileHandler
 from .file_paratext_project_settings_parser import FileParatextProjectSettingsParser
 from .paratext_project_text_updater_base import ParatextProjectTextUpdaterBase
 
 
 class FileParatextProjectTextUpdater(ParatextProjectTextUpdaterBase):
     def __init__(self, project_dir: StrPath) -> None:
-        super().__init__(FileParatextProjectSettingsParser(project_dir))
+        super().__init__(
+            FileParatextProjectFileHandler(project_dir), FileParatextProjectSettingsParser(project_dir).parse()
+        )
 
         self._project_dir = project_dir
 

machine/corpora/file_paratext_project_versification_error_detector.py

@@ -0,0 +1,11 @@
+from ..utils.typeshed import StrPath
+from .file_paratext_project_file_handler import FileParatextProjectFileHandler
+from .file_paratext_project_settings_parser import FileParatextProjectSettingsParser
+from .paratext_project_versification_error_detector import ParatextProjectVersificationErrorDetector
+
+
+class FileParatextProjectVersificationErrorDetector(ParatextProjectVersificationErrorDetector):
+    def __init__(self, project_dir: StrPath) -> None:
+        super().__init__(
+            FileParatextProjectFileHandler(project_dir), FileParatextProjectSettingsParser(project_dir).parse()
+        )

machine/corpora/paratext_project_file_handler.py

@@ -0,0 +1,18 @@
+from abc import ABC, abstractmethod
+from typing import BinaryIO
+
+from .usfm_stylesheet import UsfmStylesheet
+
+
+class ParatextProjectFileHandler(ABC):
+    @abstractmethod
+    def exists(self, file_name: str) -> bool: ...
+
+    @abstractmethod
+    def open(self, file_name: str) -> BinaryIO: ...
+
+    @abstractmethod
+    def find(self, extension: str) -> str: ...
+
+    @abstractmethod
+    def create_stylesheet(self, file_name: str) -> UsfmStylesheet: ...

machine/corpora/paratext_project_settings_parser_base.py

@@ -1,35 +1,24 @@
-from abc import ABC, abstractmethod
-from typing import BinaryIO
+from abc import ABC
 from xml.etree import ElementTree
 
 from ..scripture.verse_ref import Versification
 from ..utils.string_utils import parse_integer
 from .corpora_utils import get_encoding
+from .paratext_project_file_handler import ParatextProjectFileHandler
 from .paratext_project_settings import ParatextProjectSettings
-from .usfm_stylesheet import UsfmStylesheet
 
 
 class ParatextProjectSettingsParserBase(ABC):
-
-    @abstractmethod
-    def _exists(self, file_name: str) -> bool: ...
-
-    @abstractmethod
-    def _find(self, extension: str) -> str: ...
-
-    @abstractmethod
-    def _open(self, file_name: str) -> BinaryIO: ...
-
-    @abstractmethod
-    def _create_stylesheet(self, file_name: str) -> UsfmStylesheet: ...
+    def __init__(self, paratext_project_file_handler: ParatextProjectFileHandler):
+        self._paratext_project_file_handler = paratext_project_file_handler
 
     def parse(self) -> ParatextProjectSettings:
         settings_file_name = "Settings.xml"
-        if not self._exists(settings_file_name):
-            settings_file_name = self._find(".ssf")
+        if not self._paratext_project_file_handler.exists(settings_file_name):
+            settings_file_name = self._paratext_project_file_handler.find(".ssf")
         if not settings_file_name:
             raise ValueError("The project does not contain a settings file.")
-        with self._open(settings_file_name) as stream:
+        with self._paratext_project_file_handler.open(settings_file_name) as stream:
             settings_tree = ElementTree.parse(stream)
 
         name = settings_tree.getroot().findtext("Name", "")
@@ -46,18 +35,21 @@ def parse(self) -> ParatextProjectSettings:
 
         versification_type = int(settings_tree.getroot().findtext("Versification", "4"))
         versification = Versification.get_builtin(versification_type)
-        if self._exists("custom.vrs"):
+        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._open("custom.vrs"),
+                self._paratext_project_file_handler.open("custom.vrs"),
                 versification,
                 versification_name,
             )
         stylesheet_file_name = settings_tree.getroot().findtext("StyleSheet", "usfm.sty")
-        if not self._exists(stylesheet_file_name) and stylesheet_file_name != "usfm_sb.sty":
+        if (
+            not self._paratext_project_file_handler.exists(stylesheet_file_name)
+            and stylesheet_file_name != "usfm_sb.sty"
+        ):
             stylesheet_file_name = "usfm.sty"
-        stylesheet = self._create_stylesheet(stylesheet_file_name)
+        stylesheet = self._paratext_project_file_handler.create_stylesheet(stylesheet_file_name)
 
         prefix = ""
         form = "41MAT"

machine/corpora/paratext_project_terms_parser_base.py

@@ -1,12 +1,13 @@
 from __future__ import annotations
 
 import re
-from abc import ABC, abstractmethod
+from abc import ABC
 from collections import defaultdict
 from importlib.resources import open_binary
-from typing import BinaryIO, Dict, List, Optional, Sequence, Tuple, Union
+from typing import Dict, List, Optional, Sequence, Tuple, Union
 from xml.etree import ElementTree
 
+from .paratext_project_file_handler import ParatextProjectFileHandler
 from .paratext_project_settings import ParatextProjectSettings
 from .paratext_project_settings_parser_base import ParatextProjectSettingsParserBase
 
@@ -24,7 +25,12 @@
 
 
 class ParatextProjectTermsParserBase(ABC):
-    def __init__(self, settings: Union[ParatextProjectSettings, ParatextProjectSettingsParserBase]) -> None:
+    def __init__(
+        self,
+        paratext_project_file_handler: ParatextProjectFileHandler,
+        settings: Union[ParatextProjectSettings, ParatextProjectSettingsParserBase],
+    ) -> None:
+        self._paratext_project_file_handler = paratext_project_file_handler
         self._settings: ParatextProjectSettings
         if isinstance(settings, ParatextProjectSettingsParserBase):
             self._settings = settings.parse()
@@ -34,8 +40,8 @@ def __init__(self, settings: Union[ParatextProjectSettings, ParatextProjectSetti
     def parse(self, term_categories: Sequence[str], use_term_glosses: bool = True) -> List[Tuple[str, List[str]]]:
         biblical_terms_doc = None
         if self._settings.biblical_terms_list_type == "Project":
-            if self._exists(self._settings.biblical_terms_file_name):
-                with self._open(self._settings.biblical_terms_file_name) as stream:
+            if self._paratext_project_file_handler.exists(self._settings.biblical_terms_file_name):
+                with self._paratext_project_file_handler.open(self._settings.biblical_terms_file_name) as stream:
                     biblical_terms_doc = ElementTree.parse(stream)
                     term_id_to_category_dict = _get_category_per_id(biblical_terms_doc)
         elif self._settings.biblical_terms_list_type in _PREDEFINED_TERMS_LIST_TYPES:
@@ -60,8 +66,8 @@ def parse(self, term_categories: Sequence[str], use_term_glosses: bool = True) -
                 terms_glosses_doc = ElementTree.parse(stream)
 
         term_renderings_doc: Optional[ElementTree.ElementTree[ElementTree.Element]] = None
-        if self._exists("TermRenderings.xml"):
-            with self._open("TermRenderings.xml") as stream:
+        if self._paratext_project_file_handler.exists("TermRenderings.xml"):
+            with self._paratext_project_file_handler.open("TermRenderings.xml") as stream:
                 term_renderings_doc = ElementTree.parse(stream)
 
         terms_renderings: Dict[str, List[str]] = defaultdict(list)
@@ -94,12 +100,6 @@ def parse(self, term_categories: Sequence[str], use_term_glosses: bool = True) -
 
         return []
 
-    @abstractmethod
-    def _exists(self, file_name: str) -> bool: ...
-
-    @abstractmethod
-    def _open(self, file_name: str) -> BinaryIO: ...
-
 
 def _is_in_category(id: str, term_categories: Sequence[str], term_id_to_category_dict: Dict[str, str]) -> bool:
     category = term_id_to_category_dict.get(id)

machine/corpora/paratext_project_text_updater_base.py

@@ -1,7 +1,7 @@
-from abc import ABC, abstractmethod
-from typing import BinaryIO, Callable, Iterable, Optional, Sequence, Union
+from abc import ABC
+from typing import Callable, Iterable, Optional, Sequence, Union
 
-from ..utils.typeshed import StrPath
+from .paratext_project_file_handler import ParatextProjectFileHandler
 from .paratext_project_settings import ParatextProjectSettings
 from .paratext_project_settings_parser_base import ParatextProjectSettingsParserBase
 from .update_usfm_parser_handler import (
@@ -15,7 +15,12 @@
 
 
 class ParatextProjectTextUpdaterBase(ABC):
-    def __init__(self, settings: Union[ParatextProjectSettings, ParatextProjectSettingsParserBase]) -> None:
+    def __init__(
+        self,
+        paratext_project_file_handler: ParatextProjectFileHandler,
+        settings: Union[ParatextProjectSettings, ParatextProjectSettingsParserBase],
+    ) -> None:
+        self._paratext_project_file_handler = paratext_project_file_handler
         if isinstance(settings, ParatextProjectSettingsParserBase):
             self._settings = settings.parse()
         else:
@@ -37,9 +42,9 @@ def update_usfm(
         compare_segments: bool = False,
     ) -> Optional[str]:
         file_name: str = self._settings.get_book_file_name(book_id)
-        if not self._exists(file_name):
+        if not self._paratext_project_file_handler.exists(file_name):
             return None
-        with self._open(file_name) as sfm_file:
+        with self._paratext_project_file_handler.open(file_name) as sfm_file:
             usfm: str = sfm_file.read().decode(self._settings.encoding)
         handler = UpdateUsfmParserHandler(
             rows,
@@ -64,9 +69,3 @@ def update_usfm(
                 f". Error: '{e}'"
             )
             raise RuntimeError(error_message) from e
-
-    @abstractmethod
-    def _exists(self, file_name: StrPath) -> bool: ...
-
-    @abstractmethod
-    def _open(self, file_name: StrPath) -> BinaryIO: ...