sebastien/Copies
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

améliorations diverses

Browse Source
This commit is contained in:
sebastien
2026-09-17 22:19:05 +02:00
parent 5b8215e7f5
commit 0a453403cf
11 changed files with 1072 additions and 81 deletions
Download Patch File Download Diff File Expand all files Collapse all files
copienator/commands/plotting.py
+112 -21
View File
@@ -1,7 +1,9 @@
from __future__ import annotations
import argparse
import copy
import queue
import re
import threading
import tkinter as tk
from collections.abc import Sequence
@@ -24,6 +26,9 @@ from copienator.utils import natural_key, read_all_labels
# --- Configuration & Globals ---
padding = 60
MISSING_LABEL_COLOR = "orange"
COMMON_MISSING_LABEL_COLOR = "#403a00"
COMMON_MISSING_THRESHOLD = 0.66
try:
font = ImageFont.truetype("DejaVuSans.ttf", size=30)
@@ -81,13 +86,81 @@ def normalized_labels(entries):
if str(value["label"]) != "_"
]
def prepare_image(image_path: str, bounding_boxes, all_labels, nb_pages, last_label_index):
def frequently_missing_labels(
copies_dir: Path,
all_labels: list[str],
threshold: float = COMMON_MISSING_THRESHOLD,
) -> set[str]:
"""Return labels absent from at least ``threshold`` of detected copies."""
labels_by_copy: dict[str, set[str]] = {}
for json_path in copies_dir.glob("*.json"):
match = re.fullmatch(r"(.+)_\d+", json_path.stem)
if match is None:
continue
try:
data = read_json(json_path)
entries = data["list"]
if not isinstance(entries, list):
continue
present = set(normalized_labels(entries))
except (OSError, KeyError, TypeError, ValueError):
continue
labels_by_copy.setdefault(match.group(1), set()).update(present)
copy_count = len(labels_by_copy)
if copy_count == 0:
return set()
return {
label
for label in all_labels
if sum(label not in present for present in labels_by_copy.values())
>= threshold * copy_count
}
def label_color(
label: str | None,
all_labels: list[str],
last_label_index: int,
common_missing: set[str],
) -> tuple[str, int]:
"""Choose a label color and return the updated chronological index."""
color = "black"
if not label or label not in all_labels:
return color, last_label_index
current_index = all_labels.index(label)
if current_index < last_label_index or (
last_label_index == -1 and current_index != 0
):
color = "red"
elif current_index > last_label_index + 1:
only_previous_is_missing = current_index == last_label_index + 2
previous_label = all_labels[current_index - 1]
color = (
COMMON_MISSING_LABEL_COLOR
if only_previous_is_missing and previous_label in common_missing
else MISSING_LABEL_COLOR
)
return color, current_index
def prepare_image(
image_path: str,
bounding_boxes,
all_labels,
nb_pages,
last_label_index,
common_missing: set[str] | None = None,
):
im = Image.open(image_path)
im.load()
width, height = im.size
new_im = Image.new(im.mode, (width + padding, height), "white")
new_im.paste(im, (0, 0))
draw = ImageDraw.Draw(new_im)
common_missing = common_missing or set()
for bbox in sort_bounding_boxes(bounding_boxes, nb_pages):
raw_y_min = int(bbox["box_2d"][0] * height / 1000)
@@ -99,15 +172,13 @@ def prepare_image(image_path: str, bounding_boxes, all_labels, nb_pages, last_la
abs_y_max = min(height, raw_y_max + 10)
abs_x_max = min(width, raw_x_max + 10)
color = "black"
label = bbox.get("label")
if label and label in all_labels:
current_index = all_labels.index(label)
if current_index < last_label_index or (last_label_index == -1 and current_index != 0):
color = "red"
elif current_index > last_label_index + 1:
color = "orange"
last_label_index = current_index
color, last_label_index = label_color(
label,
all_labels,
last_label_index,
common_missing,
)
draw.rectangle(((abs_x_min, abs_y_min), (abs_x_max, abs_y_max)), outline=color, width=4)
if label:
@@ -126,6 +197,7 @@ def _worker_items(base_dir, files_to_process, all_labels, output_queue):
"""
previous_copie = None
last_label_index = None
common_missing = frequently_missing_labels(base_dir / "Copies", all_labels)
for img_path in files_to_process:
json_path = base_dir / "Copies" / f"{img_path.stem}.json"
copie_part = int(img_path.stem[-2:])
@@ -158,7 +230,14 @@ def _worker_items(base_dir, files_to_process, all_labels, output_queue):
try:
print(f"Buffering {img_path.name}...")
(pil_image, last_label_index) = \
prepare_image(str(img_path), bb_list, all_labels, nb_pages, last_label_index)
prepare_image(
str(img_path),
bb_list,
all_labels,
nb_pages,
last_label_index,
common_missing,
)
error_msg = None
except Exception as e: # noqa: BLE001 - keep the item editable in the GUI
@@ -271,7 +350,6 @@ class ImageViewer:
# Start new batch
self.active_copie_name = metadata["copie"]
self.accumulated_results = {"name": metadata["name"], "list": []}
self.history.clear()
self.display_image(pil_image, json_path, metadata)
except queue.Empty:
@@ -293,16 +371,23 @@ class ImageViewer:
def on_previous(self, event):
if self.is_viewing and self.history:
print("Going back to previous image...")
prev_pil, prev_json, prev_meta, num_added = self.history.pop()
# Undo the accumulation to prevent duplicates when we hit Enter again
if self.accumulated_results and num_added > 0:
self.accumulated_results["list"] = self.accumulated_results["list"][:-num_added]
(
prev_pil,
prev_json,
prev_meta,
previous_copie_name,
previous_results,
) = self.history.pop()
# Push current image to the forward stack so we don't lose it
self.forward_stack.append((self.current_pil_image,
self.current_json_path, self.current_meta))
# Restore the aggregation exactly as it was before validating the
# previous image. This also makes navigation across copies safe.
self.active_copie_name = previous_copie_name
self.accumulated_results = previous_results
# Display the previous image immediately
self.display_image(prev_pil, prev_json, prev_meta)
def display_image(self, pil_image, json_path, metadata):
@@ -329,7 +414,8 @@ class ImageViewer:
def on_enter(self, event):
if self.is_viewing:
print(f"Committing data for {self.current_json_path.name}...")
num_added = 0 # ADD THIS LINE
previous_copie_name = self.active_copie_name
previous_results = copy.deepcopy(self.accumulated_results)
try:
current_data = read_json(self.current_json_path)
@@ -352,8 +438,6 @@ class ImageViewer:
print(msg)
messagebox.showerror("Label Error", msg)
return
num_added = len(converted_items)
# Add to accumulator
if self.accumulated_results:
self.accumulated_results["list"].extend(converted_items)
@@ -368,8 +452,15 @@ class ImageViewer:
messagebox.showerror("JSON Error", msg)
return # Abort advancement
self.history.append((self.current_pil_image, self.current_json_path,
self.current_meta, num_added))
self.history.append(
(
self.current_pil_image,
self.current_json_path,
self.current_meta,
previous_copie_name,
previous_results,
)
)
# Advance UI
self.is_viewing = False