Rename file numbering

alto2txt2fixture/cli.py

@@ -15,7 +15,14 @@
 )
 from .settings import DATA_PROVIDER_INDEX, SETUP_TITLE, settings
 from .types import dotdict
-from .utils import check_newspaper_collection_configuration, console, gen_fixture_tables
+from .utils import (
+    FILE_NAME_0_PADDING_DEFAULT,
+    check_newspaper_collection_configuration,
+    console,
+    copy_dict_paths,
+    gen_fixture_tables,
+    glob_path_rename_by_0_padding,
+)
 
 cli = typer.Typer(pretty_exceptions_show_locals=False)
 
@@ -38,6 +45,9 @@ def plaintext(
     records_per_json: Annotated[
         int, typer.Option(help="Max records per json fixture")
     ] = DEFAULT_MAX_PLAINTEXT_PER_FIXTURE_FILE,
+    digit_padding: Annotated[
+        int, typer.Option(help="Padding '0's for indexing json fixture filenames")
+    ] = FILE_NAME_0_PADDING_DEFAULT,
 ) -> None:
     """Create a PlainTextFixture and save to `save_path`."""
     plaintext_fixture = PlainTextFixture(
@@ -47,15 +57,17 @@ def plaintext(
         export_directory=save_path,
         initial_pk=initial_pk,
         max_plaintext_per_fixture_file=records_per_json,
+        json_0_file_name_padding=digit_padding,
     )
     plaintext_fixture.info()
     while (
         not plaintext_fixture.compressed_files
         and not plaintext_fixture.plaintext_provided_uncompressed
     ):
         try_another_compressed_txt_source: bool = Confirm.ask(
-            f"No .txt files available from extract path: {plaintext_fixture.trunc_extract_path_str}\n"
-            "Would you like to extract fixtures from a different path?"
+            f"No .txt files available from extract path: "
+            f"{plaintext_fixture.trunc_extract_path_str}\n"
+            f"Would you like to extract fixtures from a different path?"
         )
         if try_another_compressed_txt_source:
             new_extract_path: str = Prompt.ask("Please enter a new extract path")
@@ -67,6 +79,49 @@ def plaintext(
     plaintext_fixture.export_to_json_fixtures()
 
 
+@cli.command()
+def reindex(
+    path: Annotated[Path, typer.Argument(help="Path to files to rename")],
+    folder: Annotated[Path, typer.Option(help="Path to save renamed files")] = Path(),
+    regex: Annotated[str, typer.Option(help="Regex to filter files to rename")] = "*",
+    padding: Annotated[
+        int, typer.Option(help="How many digits to pad by (guessed if blank)")
+    ] = FILE_NAME_0_PADDING_DEFAULT,
+    dry_run: Annotated[
+        bool, typer.Option(help="Show example paths without copy")
+    ] = True,
+) -> None:
+    """Rename files for ordering."""
+    rename_paths_dict: dict[os.PathLike, os.PathLike] = glob_path_rename_by_0_padding(
+        path=path,
+        output_path=folder,
+        glob_regex_str=regex,
+        padding=padding,
+    )
+
+    paths_table: Table = Table(title=f"Copy and rename folders")
+    paths_table.add_column("From Folder", justify="right", style="cyan")
+    paths_table.add_column("New Folder", style="magenta")
+    paths_table.add_row(str(path), str(folder))
+    console.print(paths_table)
+
+    file_names_table: Table = Table(title="Old and New File Names")
+    file_names_table.add_column("Current File Name", justify="right", style="cyan")
+    file_names_table.add_column("New File Name", style="magenta")
+    for old_path, new_path in rename_paths_dict.items():
+        file_names_table.add_row(Path(old_path).name, Path(new_path).name)
+    console.print(file_names_table)
+
+    make_copy: bool = False or not dry_run
+    if dry_run:
+        make_copy = Confirm.ask(
+            f"Would you like to copy these {len(rename_paths_dict)} "
+            f"files from Current:\n'{path}'\nto New:\n'{folder}\n'"
+        )
+    if make_copy:
+        copy_dict_paths(rename_paths_dict)
+
+
 def show_setup(clear: bool = True, title: str = SETUP_TITLE, **kwargs) -> None:
     """Generate a `rich.table.Table` for printing configuration to console."""
     if clear and os.name == "posix":

alto2txt2fixture/plaintext.py

@@ -17,6 +17,7 @@
     PlaintextFixtureFieldsDict,
 )
 from .utils import (
+    FILE_NAME_0_PADDING_DEFAULT,
     TRUNC_HEADS_PATH_DEFAULT,
     TRUNC_TAILS_PATH_DEFAULT,
     ZIP_FILE_EXTENSION,
@@ -128,6 +129,9 @@ class PlainTextFixture:
             Default begins at 1, can be set to another number if needed to
             add to add more to pre-existing set of records up to a given `pk`
 
+        json_0_file_name_padding:
+            Number of `0`s to prefix file name numbering.
+
         _disk_usage:
             Available harddrive space. Designed to help mitigate decompressing too
             many files for available disk space.
@@ -158,6 +162,7 @@ class PlainTextFixture:
         │ Data Provider       │ 'Living with Machines'         ...│
         │ Initial Primary Key │ 1                              ...│
         │ Max Rows Per JSON   │ 100                            ...│
+        │ JSON File Name 0s   │ 6                              ...│
         └─────────────────────┴────────────────────────────────...┘
         >>> plaintext_bl_lwm.free_hd_space_in_GB > 1
         True
@@ -190,6 +195,7 @@ class PlainTextFixture:
     saved_fixture_prefix: str = DEFAULT_PLAINTEXT_FILE_NAME_PREFIX
     export_directory: PathLike = DEFAULT_PLAINTEXT_FIXTURE_OUTPUT
     empty_info_default_str: str = "None"
+    json_0_file_name_padding: int = FILE_NAME_0_PADDING_DEFAULT
     _trunc_head_paths: int = TRUNC_HEADS_PATH_DEFAULT
     _trunc_tails_paths: int = TRUNC_TAILS_PATH_DEFAULT
     _trunc_tails_sub_paths: int = TRUNC_TAILS_SUBPATH_DEFAULT
@@ -256,7 +262,7 @@ def trunc_extract_path_str(self) -> str:
 
     @property
     def info_table(self) -> str:
-        """Generate a `rich.ltable.Table` of config information.
+        """Generate a `rich.table.Table` of config information.
 
         Example:
             ```pycon
@@ -278,6 +284,7 @@ def info_table(self) -> str:
         table.add_row("Data Provider", f"'{str(self.data_provider_name)}'")
         table.add_row("Initial Primary Key", str(self.initial_pk))
         table.add_row("Max Rows Per JSON", str(self.max_plaintext_per_fixture_file))
+        table.add_row("JSON File Name 0s", str(self.json_0_file_name_padding))
         return table
 
     def info(self) -> None:
@@ -401,7 +408,7 @@ def extract_path(self) -> Path:
 
     @property
     def compressed_files(self) -> tuple[PathLike, ...]:
-        """Return a tuple of all `self.files` with known archive filenames."""
+        """Return a tuple of all `self.files` with known archive file names."""
         return (
             tuple(sorted(valid_compression_files(files=self.files)))
             if self.files
@@ -593,7 +600,10 @@ def plaintext_paths_to_dicts(self) -> Generator[PlaintextFixtureDict, None, None
             )
 
     def export_to_json_fixtures(
-        self, output_path: PathLike | None = None, prefix: str | None = None
+        self,
+        output_path: PathLike | None = None,
+        prefix: str | None = None,
+        json_0_file_name_padding: int | None = None,
     ) -> None:
         """Iterate over `self.plaintext_paths` exporting to `json` `django` fixtures.
 
@@ -606,6 +616,8 @@ def export_to_json_fixtures(
                 Folder to save all `json` fixtures in.
             prefix:
                 Any `str` prefix for saved fixture files.
+            json_0_file_name_padding:
+                Number of `0`s to prefix file name numbering.
 
         Example:
             ```pycon
@@ -624,7 +636,7 @@ def export_to_json_fixtures(
             >>> len(plaintext_bl_lwm._exported_json_paths)
             1
             >>> plaintext_bl_lwm._exported_json_paths
-            (...Path(...plaintext_fixture-1.json...),)
+            (...Path(...plaintext_fixture-000001.json...),)
             >>> import json
             >>> exported_json = json.loads(
             ...     plaintext_bl_lwm._exported_json_paths[0].read_text()
@@ -652,12 +664,18 @@ def export_to_json_fixtures(
         """
         output_path = self.export_directory if not output_path else output_path
         prefix = self.saved_fixture_prefix if not prefix else prefix
+        json_0_file_name_padding = (
+            self.json_0_file_name_padding
+            if not json_0_file_name_padding
+            else json_0_file_name_padding
+        )
         save_fixture(
             self.plaintext_paths_to_dicts(),
             prefix=prefix,
             output_path=output_path,
             add_created=True,
             max_elements_per_file=self.max_plaintext_per_fixture_file,
+            file_name_0_padding=json_0_file_name_padding,
         )
         self._exported_json_paths = tuple(
             Path(path) for path in sorted(Path(output_path).glob(f"**/{prefix}*.json"))