Separate cli and core logic

pyimorg/__init__.py

@@ -1,15 +1 @@
-from __future__ import annotations
-
-import click
-
-from .cli import diff, groupby
-
-__all__ = ['cli']
-
-@click.group()
-def cli():
-    """Command-line tool for organizing images."""
-    pass
-
-cli.command(diff)
-cli.command(groupby)
+from .pyimorg import *

pyimorg/__main__.py

@@ -1,3 +1,3 @@
-from . import cli
+from .cli import cli
 
 cli()

pyimorg/cli/__init__.py

@@ -1,2 +1,14 @@
-from .diff import *
-from .groupby import *
+import click
+
+from .diff import diff
+from .groupby import groupby
+
+__all__ = ['cli']
+
+@click.group()
+def cli():
+    """Command-line tool for organizing images."""
+    pass
+
+cli.command(diff)
+cli.command(groupby)

pyimorg/cli/diff.py

@@ -1,17 +1,12 @@
 from __future__ import annotations
 
-import logging
 from pathlib import Path
-from typing import Literal
 
 import click
 
-from ..filesystem import cksum_parallel, cp_p_parallel, is_image, mkdir_p_parallel
-from ..logger import set_logger_level
+from ..pyimorg.diff import Hasher, diff as _diff
 
-__all__ = ['Hasher', 'diff']
-
-Hasher = Literal['sha256', 'sha512']
+__all__ = ['diff']
 
 @click.argument('src1', type=click.Path(exists=True, file_okay=False, dir_okay=True))
 @click.argument('src2', type=click.Path(exists=True, file_okay=False, dir_okay=True))
@@ -27,8 +22,6 @@ def diff(
     *,
     hasher: Hasher,
     threads: int,
-    logger_level: int = logging.INFO,
-    disable_progbar: bool = False,
 ):
     """
     Given two sets of images contained in SRC1 and SRC2,
@@ -47,104 +40,4 @@ def diff(
     - `src2_only`: Contains the images that only exist in SRC2 but not in SRC1.
     The path of each output image is based on that in SRC2.
     """
-    if isinstance(src1, str):
-        src1 = Path(src1)
-    if isinstance(src2, str):
-        src2 = Path(src2)
-    if isinstance(dst, str):
-        dst = Path(dst)
-
-    if dst.exists():
-        msg = f'The destination directory ({dst}) already exists.'
-        raise ValueError(msg)
-
-    set_logger_level(logger_level)
-
-    src1_img_paths = [path for path in src1.rglob('*') if is_image(path)]
-    src1_img_hashes = cksum_parallel(
-        src1_img_paths,
-        digest=hasher,
-        n_jobs=threads,
-        desc='Hashing images from src1',
-        disable_progbar=disable_progbar,
-    )
-    src1_img_hash_to_path = {
-        img_hash: img_path
-        for img_path, img_hash in zip(src1_img_paths, src1_img_hashes)
-    }
-
-    src2_img_paths = [path for path in src2.rglob('*') if is_image(path)]
-    src2_img_hashes = cksum_parallel(
-        src2_img_paths,
-        digest=hasher,
-        n_jobs=threads,
-        desc='Hashing images from src2',
-        disable_progbar=disable_progbar,
-    )
-    src2_img_hash_to_path = {
-        img_hash: img_path
-        for img_path, img_hash in zip(src2_img_paths, src2_img_hashes)
-    }
-
-    hashes_in_src1 = set(src1_img_hashes)
-    hashes_in_src2 = set(src2_img_hashes)
-    hashes_in_both = hashes_in_src1 & hashes_in_src2
-    hashes_in_src1_only = hashes_in_src1 - hashes_in_src2
-    hashes_in_src2_only = hashes_in_src2 - hashes_in_src1
-
-    in_both_src_path_to_dst_path = {
-        src1_img_hash_to_path[img_hash]: dst / 'both' / src1_img_hash_to_path[img_hash].relative_to(src1)
-        for img_hash in hashes_in_both
-    }
-
-    mkdir_p_parallel(
-        {path.parent for path in in_both_src_path_to_dst_path.values()},
-        n_jobs=threads,
-        desc='Creating output directories for images that exist in both',
-        disable_progbar=disable_progbar,
-    )
-
-    cp_p_parallel(
-        in_both_src_path_to_dst_path.items(),
-        n_jobs=threads,
-        desc='Writing output images that exist in both',
-        disable_progbar=disable_progbar,
-    )
-
-    in_src1_only_src_path_to_dst_path = {
-        src1_img_hash_to_path[img_hash]: dst / 'src1_only' / src1_img_hash_to_path[img_hash].relative_to(src1)
-        for img_hash in hashes_in_src1_only
-    }
-
-    mkdir_p_parallel(
-        {path.parent for path in in_src1_only_src_path_to_dst_path.values()},
-        n_jobs=threads,
-        desc='Creating output directories for images that exist in src1 only',
-        disable_progbar=disable_progbar,
-    )
-
-    cp_p_parallel(
-        in_src1_only_src_path_to_dst_path.items(),
-        n_jobs=threads,
-        desc='Writing output images that exist in src1 only',
-        disable_progbar=disable_progbar,
-    )
-
-    in_src2_only_src_path_to_dst_path = {
-        src2_img_hash_to_path[img_hash]: dst / 'src2_only' / src2_img_hash_to_path[img_hash].relative_to(src2)
-        for img_hash in hashes_in_src2_only
-    }
-
-    mkdir_p_parallel(
-        {path.parent for path in in_src2_only_src_path_to_dst_path.values()},
-        n_jobs=threads,
-        desc='Creating output directories for images that exist in src2 only',
-        disable_progbar=disable_progbar,
-    )
-
-    cp_p_parallel(
-        in_src2_only_src_path_to_dst_path.items(),
-        n_jobs=threads,
-        desc='Writing output images that exist in src2 only',
-        disable_progbar=disable_progbar,
-    )
+    return _diff(src1, src2, dst, hasher=hasher, threads=threads)

pyimorg/cli/groupby.py

@@ -1,31 +1,14 @@
 from __future__ import annotations
 
-from collections import defaultdict
-from datetime import datetime
-import logging
 from pathlib import Path
 from typing import Literal
 
 import click
-from tqdm import tqdm
 
-from ..exif import read_captured_timestamp
-from ..filesystem import cp_p_parallel, is_image, mkdir_p_parallel
-from ..func import map_mt_with_tqdm
-from ..logger import set_logger_level
+from ..pyimorg import Group, groupby as _groupby
 
 __all__ = ['Group', 'groupby']
 
-Group = Literal['year', 'month', 'day']
-
-def _get_dst_dir_name(timestamp: datetime, *, groupby: Group) -> str:
-    if groupby == 'year':
-        return timestamp.strftime(r'%Y')
-    elif groupby == 'month':
-        return timestamp.strftime(r'%Y%m')
-    elif groupby == 'day':
-        return timestamp.strftime(r'%Y%m%d')
-
 @click.argument('src', type=click.Path(exists=True, file_okay=False, dir_okay=True))
 @click.argument('dst', type=click.Path(exists=False))
 @click.option('--group', '-g', type=click.Choice(['year', 'month', 'day']), default='month',
@@ -38,8 +21,6 @@ def groupby(
     *,
     group: Literal['year', 'month', 'day'],
     threads: int,
-    logger_level: int = logging.INFO,
-    disable_progbar: bool = False,
 ):
     """
     Given a set of images contained in SRC,
@@ -55,55 +36,4 @@ def groupby(
     - `day`: `YYYYMMDD` format, where `YYYY` is the year, `MM` is the month,
     and `DD` is the day of the month.
     """
-    if isinstance(src, str):
-        src = Path(src)
-    if isinstance(dst, str):
-        dst = Path(dst)
-
-    if dst.exists():
-        msg = f'The destination directory ({dst}) already exists.'
-        raise ValueError(msg)
-
-    set_logger_level(logger_level)
-
-    src_img_paths = [path for path in src.rglob('*') if is_image(path)]
-    src_img_captured_timestamps = map_mt_with_tqdm(
-        src_img_paths,
-        read_captured_timestamp,
-        n_jobs=threads,
-        desc=None if disable_progbar else 'Reading metadata of images',
-    )
-
-    dst_dir_name_to_src_img_paths: defaultdict[str, list[Path]] = defaultdict(list)
-    for src_img_path, captured_timestamp in tqdm(
-        zip(src_img_paths, src_img_captured_timestamps),
-        desc=f'Grouping images by {group}',
-        total=len(src_img_paths),
-        disable=disable_progbar,
-    ):
-        if captured_timestamp is None:
-            dst_dir_name = 'UNKNOWN'
-        else:
-            dst_dir_name = _get_dst_dir_name(captured_timestamp, groupby=group)
-
-        dst_dir_name_to_src_img_paths[dst_dir_name].append(src_img_path)
-
-    src_img_path_to_dst_img_path = {
-        src_img_path: dst / dst_dir_name / src_img_path.name
-        for dst_dir_name, src_img_paths in dst_dir_name_to_src_img_paths.items()
-        for src_img_path in src_img_paths
-    }
-
-    mkdir_p_parallel(
-        {path.parent for path in src_img_path_to_dst_img_path.values()},
-        n_jobs=threads,
-        desc='Creating output directories',
-        disable_progbar=disable_progbar,
-    )
-
-    cp_p_parallel(
-        src_img_path_to_dst_img_path.items(),
-        n_jobs=threads,
-        desc='Writing output images',
-        disable_progbar=disable_progbar,
-    )
+    return _groupby(src, dst, group=group, threads=threads)

pyimorg/pyimorg/__init__.py

@@ -0,0 +1,2 @@
+from .diff import *
+from .groupby import *