Upload 85 files
Browse files- MM_tools/__pycache__/config.cpython-314.pyc +0 -0
- MM_tools/app/blueprints/ledger/__pycache__/routes.cpython-314.pyc +0 -0
- MM_tools/app/blueprints/ledger/routes.py +57 -51
- MM_tools/app/blueprints/ledger/templates/ledger/index.html +37 -12
- MM_tools/app/blueprints/ocr/__pycache__/routes.cpython-314.pyc +0 -0
- MM_tools/app/blueprints/ocr/routes.py +7 -6
- MM_tools/app/blueprints/ocr/templates/ocr/index.html +33 -9
- MM_tools/app/blueprints/vlm/__pycache__/routes.cpython-314.pyc +0 -0
- MM_tools/app/blueprints/vlm/routes.py +6 -4
- MM_tools/app/blueprints/vlm/templates/vlm/index.html +35 -10
- MM_tools/app/shared/__pycache__/utils.cpython-314.pyc +0 -0
- MM_tools/app/shared/utils.py +60 -0
- MM_tools/app/static/exports/whisper_text_time.txt +1 -1
MM_tools/__pycache__/config.cpython-314.pyc
ADDED
|
Binary file (1.26 kB). View file
|
|
|
MM_tools/app/blueprints/ledger/__pycache__/routes.cpython-314.pyc
CHANGED
|
Binary files a/MM_tools/app/blueprints/ledger/__pycache__/routes.cpython-314.pyc and b/MM_tools/app/blueprints/ledger/__pycache__/routes.cpython-314.pyc differ
|
|
|
MM_tools/app/blueprints/ledger/routes.py
CHANGED
|
@@ -18,7 +18,7 @@ from PIL import Image
|
|
| 18 |
|
| 19 |
from . import ledger_bp
|
| 20 |
from app.shared.models import get_qwen_model
|
| 21 |
-
from app.shared.utils import sse_event
|
| 22 |
|
| 23 |
# 出力先ディレクトリ(プロジェクトルートの output/)
|
| 24 |
OUTPUT_DIR = Path(__file__).parent.parent.parent.parent / "output"
|
|
@@ -28,29 +28,19 @@ OUTPUT_DIR.mkdir(exist_ok=True)
|
|
| 28 |
_model_lock = threading.Lock()
|
| 29 |
|
| 30 |
# 表データ抽出プロンプト
|
| 31 |
-
OCR_PROMPT = """
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
]
|
| 45 |
-
}
|
| 46 |
-
```
|
| 47 |
-
|
| 48 |
-
注意事項:
|
| 49 |
-
- 数字は半角で出力してください
|
| 50 |
-
- 日付は記載通りに出力してください
|
| 51 |
-
- 判読できない箇所は「???」と記載してください
|
| 52 |
-
- 表が複数ある場合は tables 配列に複数のオブジェクトを含めてください
|
| 53 |
-
- JSONのみを出力してください(前後の説明文は不要です)"""
|
| 54 |
|
| 55 |
|
| 56 |
@ledger_bp.route("/")
|
|
@@ -80,25 +70,27 @@ def process():
|
|
| 80 |
# SSEジェネレータの遅延実行時にリクエストストリームが閉じられる問題を防ぐため、
|
| 81 |
# ここでバイト列として先読みしておく
|
| 82 |
image_bytes: bytes | None = None
|
|
|
|
| 83 |
if image_file and image_file.filename:
|
| 84 |
image_bytes = image_file.read()
|
|
|
|
| 85 |
|
| 86 |
def generate():
|
| 87 |
with _model_lock:
|
| 88 |
try:
|
| 89 |
-
# ── 画像読み込み ──
|
| 90 |
-
yield sse_event({"progress": "
|
| 91 |
|
| 92 |
if image_bytes:
|
| 93 |
-
image =
|
| 94 |
elif image_path_str:
|
| 95 |
p = Path(image_path_str)
|
| 96 |
if not p.exists():
|
| 97 |
yield sse_event({"error": f"ファイルが見つかりません: {image_path_str}"})
|
| 98 |
return
|
| 99 |
-
image =
|
| 100 |
else:
|
| 101 |
-
yield sse_event({"error": "画像をアップロードするか、パスを入力してください。"})
|
| 102 |
return
|
| 103 |
|
| 104 |
# ── モデルロード ──
|
|
@@ -257,6 +249,7 @@ def _try_numeric(value: str):
|
|
| 257 |
def _save_to_excel(data: dict, output_path: Path) -> None:
|
| 258 |
"""
|
| 259 |
JSON データを整形した Excel ファイルに書き出す。
|
|
|
|
| 260 |
|
| 261 |
Args:
|
| 262 |
data: 抽出済み JSON 辞書 (title, tables キーを持つ)
|
|
@@ -269,25 +262,35 @@ def _save_to_excel(data: dict, output_path: Path) -> None:
|
|
| 269 |
ROW_FILL_ODD = PatternFill(start_color="DCE6F1", end_color="DCE6F1", fill_type="solid")
|
| 270 |
|
| 271 |
wb = openpyxl.Workbook()
|
|
|
|
|
|
|
|
|
|
| 272 |
book_title = data.get("title")
|
| 273 |
tables = data.get("tables", [])
|
| 274 |
|
| 275 |
if not tables:
|
| 276 |
raise ValueError("tables キーが空です。")
|
| 277 |
|
| 278 |
-
|
| 279 |
-
|
| 280 |
-
ws.title = f"帳簿{idx + 1}"
|
| 281 |
-
row_ptr = 1
|
| 282 |
|
| 283 |
-
|
| 284 |
-
|
| 285 |
-
|
| 286 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 287 |
|
| 288 |
headers = table.get("headers", [])
|
| 289 |
rows = table.get("rows", [])
|
| 290 |
|
|
|
|
|
|
|
|
|
|
| 291 |
for col_idx, header in enumerate(headers, start=1):
|
| 292 |
cell = ws.cell(row=row_ptr, column=col_idx, value=header)
|
| 293 |
cell.font = Font(bold=True, color="FFFFFF")
|
|
@@ -295,6 +298,7 @@ def _save_to_excel(data: dict, output_path: Path) -> None:
|
|
| 295 |
cell.alignment = Alignment(horizontal="center", vertical="center", wrap_text=True)
|
| 296 |
row_ptr += 1
|
| 297 |
|
|
|
|
| 298 |
for row_num, row_data in enumerate(rows):
|
| 299 |
fill = ROW_FILL_ODD if row_num % 2 == 0 else None
|
| 300 |
for col_idx, raw_value in enumerate(row_data, start=1):
|
|
@@ -307,20 +311,22 @@ def _save_to_excel(data: dict, output_path: Path) -> None:
|
|
| 307 |
cell.alignment = Alignment(horizontal="right", vertical="center")
|
| 308 |
row_ptr += 1
|
| 309 |
|
| 310 |
-
|
| 311 |
-
|
| 312 |
-
|
| 313 |
-
|
| 314 |
-
|
| 315 |
-
|
| 316 |
-
|
| 317 |
-
|
| 318 |
-
|
| 319 |
-
|
| 320 |
-
|
| 321 |
-
|
| 322 |
-
|
| 323 |
-
|
|
|
|
|
|
|
| 324 |
|
| 325 |
wb.save(str(output_path))
|
| 326 |
|
|
|
|
| 18 |
|
| 19 |
from . import ledger_bp
|
| 20 |
from app.shared.models import get_qwen_model
|
| 21 |
+
from app.shared.utils import sse_event, open_image_or_pdf
|
| 22 |
|
| 23 |
# 出力先ディレクトリ(プロジェクトルートの output/)
|
| 24 |
OUTPUT_DIR = Path(__file__).parent.parent.parent.parent / "output"
|
|
|
|
| 28 |
_model_lock = threading.Lock()
|
| 29 |
|
| 30 |
# 表データ抽出プロンプト
|
| 31 |
+
OCR_PROMPT = """You are a table OCR assistant. Extract all tabular data from this image and output ONLY valid JSON. No explanation, no markdown, no code fences — just raw JSON starting with { and ending with }.
|
| 32 |
+
|
| 33 |
+
Output format:
|
| 34 |
+
{"title": "table title or null", "tables": [{"headers": ["col1", "col2"], "rows": [["val1", "val2"]]}]}
|
| 35 |
+
|
| 36 |
+
Rules:
|
| 37 |
+
- Output ONLY the JSON object. Do not write anything before or after it.
|
| 38 |
+
- Use half-width digits for numbers.
|
| 39 |
+
- Keep dates exactly as written.
|
| 40 |
+
- Use "???" for unreadable characters.
|
| 41 |
+
- If multiple tables exist, include all in the "tables" array.
|
| 42 |
+
- If no title is visible, set "title" to null.
|
| 43 |
+
- Every row must have the same number of cells as headers."""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
|
| 45 |
|
| 46 |
@ledger_bp.route("/")
|
|
|
|
| 70 |
# SSEジェネレータの遅延実行時にリクエストストリームが閉じられる問題を防ぐため、
|
| 71 |
# ここでバイト列として先読みしておく
|
| 72 |
image_bytes: bytes | None = None
|
| 73 |
+
image_filename: str = ""
|
| 74 |
if image_file and image_file.filename:
|
| 75 |
image_bytes = image_file.read()
|
| 76 |
+
image_filename = image_file.filename # PDF 判定用
|
| 77 |
|
| 78 |
def generate():
|
| 79 |
with _model_lock:
|
| 80 |
try:
|
| 81 |
+
# ── 画像読み込み(PDF は1ページ目を変換)──
|
| 82 |
+
yield sse_event({"progress": "ファイルを読み込み中...", "step": 1})
|
| 83 |
|
| 84 |
if image_bytes:
|
| 85 |
+
image = open_image_or_pdf(image_bytes, image_filename)
|
| 86 |
elif image_path_str:
|
| 87 |
p = Path(image_path_str)
|
| 88 |
if not p.exists():
|
| 89 |
yield sse_event({"error": f"ファイルが見つかりません: {image_path_str}"})
|
| 90 |
return
|
| 91 |
+
image = open_image_or_pdf(p.read_bytes(), p.name)
|
| 92 |
else:
|
| 93 |
+
yield sse_event({"error": "画像または PDF をアップロードするか、パスを入力してください。"})
|
| 94 |
return
|
| 95 |
|
| 96 |
# ── モデルロード ──
|
|
|
|
| 249 |
def _save_to_excel(data: dict, output_path: Path) -> None:
|
| 250 |
"""
|
| 251 |
JSON データを整形した Excel ファイルに書き出す。
|
| 252 |
+
複数テーブルがある場合は 1 シートに縦に並べ、テーブル間に 1 行空白を入れる。
|
| 253 |
|
| 254 |
Args:
|
| 255 |
data: 抽出済み JSON 辞書 (title, tables キーを持つ)
|
|
|
|
| 262 |
ROW_FILL_ODD = PatternFill(start_color="DCE6F1", end_color="DCE6F1", fill_type="solid")
|
| 263 |
|
| 264 |
wb = openpyxl.Workbook()
|
| 265 |
+
ws = wb.active
|
| 266 |
+
ws.title = "帳簿"
|
| 267 |
+
|
| 268 |
book_title = data.get("title")
|
| 269 |
tables = data.get("tables", [])
|
| 270 |
|
| 271 |
if not tables:
|
| 272 |
raise ValueError("tables キーが空です。")
|
| 273 |
|
| 274 |
+
row_ptr = 1
|
| 275 |
+
first_header_row = None # freeze_panes 用に最初のヘッダ行を記録
|
|
|
|
|
|
|
| 276 |
|
| 277 |
+
# タイトル行(あれば)
|
| 278 |
+
if book_title:
|
| 279 |
+
title_cell = ws.cell(row=row_ptr, column=1, value=book_title)
|
| 280 |
+
title_cell.font = Font(bold=True, size=14)
|
| 281 |
+
row_ptr += 2
|
| 282 |
+
|
| 283 |
+
for idx, table in enumerate(tables):
|
| 284 |
+
# テーブル間の空白行(2つ目以降)
|
| 285 |
+
if idx > 0:
|
| 286 |
+
row_ptr += 1 # 1行空ける
|
| 287 |
|
| 288 |
headers = table.get("headers", [])
|
| 289 |
rows = table.get("rows", [])
|
| 290 |
|
| 291 |
+
# ヘッダ行
|
| 292 |
+
if first_header_row is None:
|
| 293 |
+
first_header_row = row_ptr
|
| 294 |
for col_idx, header in enumerate(headers, start=1):
|
| 295 |
cell = ws.cell(row=row_ptr, column=col_idx, value=header)
|
| 296 |
cell.font = Font(bold=True, color="FFFFFF")
|
|
|
|
| 298 |
cell.alignment = Alignment(horizontal="center", vertical="center", wrap_text=True)
|
| 299 |
row_ptr += 1
|
| 300 |
|
| 301 |
+
# データ行
|
| 302 |
for row_num, row_data in enumerate(rows):
|
| 303 |
fill = ROW_FILL_ODD if row_num % 2 == 0 else None
|
| 304 |
for col_idx, raw_value in enumerate(row_data, start=1):
|
|
|
|
| 311 |
cell.alignment = Alignment(horizontal="right", vertical="center")
|
| 312 |
row_ptr += 1
|
| 313 |
|
| 314 |
+
# 行高・列幅の自動調整
|
| 315 |
+
for row in ws.iter_rows():
|
| 316 |
+
ws.row_dimensions[row[0].row].height = 18
|
| 317 |
+
|
| 318 |
+
for col in ws.columns:
|
| 319 |
+
max_len = 0
|
| 320 |
+
col_letter = col[0].column_letter
|
| 321 |
+
for cell in col:
|
| 322 |
+
if cell.value:
|
| 323 |
+
w = sum(2 if ord(c) > 127 else 1 for c in str(cell.value))
|
| 324 |
+
max_len = max(max_len, w)
|
| 325 |
+
ws.column_dimensions[col_letter].width = min(max_len + 4, 50)
|
| 326 |
+
|
| 327 |
+
# 最初のヘッダ行の次行でフリーズ
|
| 328 |
+
if first_header_row is not None:
|
| 329 |
+
ws.freeze_panes = ws.cell(row=first_header_row + 1, column=1)
|
| 330 |
|
| 331 |
wb.save(str(output_path))
|
| 332 |
|
MM_tools/app/blueprints/ledger/templates/ledger/index.html
CHANGED
|
@@ -46,7 +46,7 @@
|
|
| 46 |
<div>
|
| 47 |
<p class="text-sm font-semibold text-emerald-800 mb-1">使い方</p>
|
| 48 |
<ol class="text-xs text-emerald-700 space-y-1 list-decimal list-inside">
|
| 49 |
-
<li>帳簿・表が写った画像をアップロード(またはパスを入力)</li>
|
| 50 |
<li>「Excel を生成」ボタンをクリック</li>
|
| 51 |
<li>解析完了後、Excel ファイルをダウンロード</li>
|
| 52 |
</ol>
|
|
@@ -57,14 +57,14 @@
|
|
| 57 |
<!-- 画像アップロードエリア -->
|
| 58 |
<div>
|
| 59 |
<label class="block text-sm font-semibold text-gray-600 mb-2">
|
| 60 |
-
画像のアップロード
|
| 61 |
</label>
|
| 62 |
<div id="drop-zone"
|
| 63 |
class="relative border-2 border-dashed border-emerald-300 rounded-xl bg-emerald-50
|
| 64 |
flex flex-col items-center justify-center min-h-[260px] cursor-pointer
|
| 65 |
transition-colors duration-200 hover:border-emerald-500 hover:bg-emerald-100">
|
| 66 |
|
| 67 |
-
<input type="file" id="image-input" accept="image/*" class="hidden" />
|
| 68 |
<img id="preview" class="hidden max-w-full max-h-52 object-contain rounded-lg shadow-sm" />
|
| 69 |
|
| 70 |
<div id="upload-placeholder" class="text-center px-6 py-8 select-none">
|
|
@@ -75,7 +75,7 @@
|
|
| 75 |
</svg>
|
| 76 |
</div>
|
| 77 |
<p class="text-emerald-800 font-medium text-sm">クリックまたはドラッグ&ドロップ</p>
|
| 78 |
-
<p class="text-gray-400 text-xs mt-1">PNG, JPG, WEBP など</p>
|
| 79 |
</div>
|
| 80 |
|
| 81 |
<button id="change-btn"
|
|
@@ -203,14 +203,16 @@
|
|
| 203 |
|
| 204 |
<!-- 対応形式 -->
|
| 205 |
<div class="bg-white border border-gray-100 rounded-xl p-4 shadow-sm">
|
| 206 |
-
<h3 class="text-xs font-semibold text-gray-500 uppercase tracking-wide mb-3">対応
|
| 207 |
<div class="flex flex-wrap gap-2">
|
| 208 |
<span class="text-xs bg-gray-100 text-gray-600 px-2 py-1 rounded-md font-mono">JPG</span>
|
| 209 |
<span class="text-xs bg-gray-100 text-gray-600 px-2 py-1 rounded-md font-mono">PNG</span>
|
| 210 |
<span class="text-xs bg-gray-100 text-gray-600 px-2 py-1 rounded-md font-mono">WEBP</span>
|
| 211 |
<span class="text-xs bg-gray-100 text-gray-600 px-2 py-1 rounded-md font-mono">BMP</span>
|
| 212 |
<span class="text-xs bg-gray-100 text-gray-600 px-2 py-1 rounded-md font-mono">TIFF</span>
|
|
|
|
| 213 |
</div>
|
|
|
|
| 214 |
<h3 class="text-xs font-semibold text-gray-500 uppercase tracking-wide mt-3 mb-2">精度を上げるコツ</h3>
|
| 215 |
<ul class="text-xs text-gray-500 space-y-1">
|
| 216 |
<li>• 表全体が収まるように撮影</li>
|
|
@@ -266,7 +268,7 @@
|
|
| 266 |
e.preventDefault();
|
| 267 |
dropZone.classList.remove('drag-over');
|
| 268 |
const file = e.dataTransfer.files[0];
|
| 269 |
-
if (file && file.type.startsWith('image/')) {
|
| 270 |
showPreview(file);
|
| 271 |
const dt = new DataTransfer();
|
| 272 |
dt.items.add(file);
|
|
@@ -275,14 +277,37 @@
|
|
| 275 |
});
|
| 276 |
|
| 277 |
function showPreview(file) {
|
| 278 |
-
const
|
| 279 |
-
|
| 280 |
-
preview.
|
| 281 |
-
preview.classList.remove('hidden');
|
| 282 |
uploadPh.classList.add('hidden');
|
| 283 |
changeBtn.classList.remove('hidden');
|
| 284 |
-
|
| 285 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 286 |
}
|
| 287 |
|
| 288 |
// ── ステップ表示 ────────────────────────────────────────────────
|
|
|
|
| 46 |
<div>
|
| 47 |
<p class="text-sm font-semibold text-emerald-800 mb-1">使い方</p>
|
| 48 |
<ol class="text-xs text-emerald-700 space-y-1 list-decimal list-inside">
|
| 49 |
+
<li>帳簿・表が写った画像または PDF をアップロード(またはパスを入力)</li>
|
| 50 |
<li>「Excel を生成」ボタンをクリック</li>
|
| 51 |
<li>解析完了後、Excel ファイルをダウンロード</li>
|
| 52 |
</ol>
|
|
|
|
| 57 |
<!-- 画像アップロードエリア -->
|
| 58 |
<div>
|
| 59 |
<label class="block text-sm font-semibold text-gray-600 mb-2">
|
| 60 |
+
画像 / PDF のアップロード
|
| 61 |
</label>
|
| 62 |
<div id="drop-zone"
|
| 63 |
class="relative border-2 border-dashed border-emerald-300 rounded-xl bg-emerald-50
|
| 64 |
flex flex-col items-center justify-center min-h-[260px] cursor-pointer
|
| 65 |
transition-colors duration-200 hover:border-emerald-500 hover:bg-emerald-100">
|
| 66 |
|
| 67 |
+
<input type="file" id="image-input" accept="image/*,.pdf" class="hidden" />
|
| 68 |
<img id="preview" class="hidden max-w-full max-h-52 object-contain rounded-lg shadow-sm" />
|
| 69 |
|
| 70 |
<div id="upload-placeholder" class="text-center px-6 py-8 select-none">
|
|
|
|
| 75 |
</svg>
|
| 76 |
</div>
|
| 77 |
<p class="text-emerald-800 font-medium text-sm">クリックまたはドラッグ&ドロップ</p>
|
| 78 |
+
<p class="text-gray-400 text-xs mt-1">PNG, JPG, WEBP, PDF など</p>
|
| 79 |
</div>
|
| 80 |
|
| 81 |
<button id="change-btn"
|
|
|
|
| 203 |
|
| 204 |
<!-- 対応形式 -->
|
| 205 |
<div class="bg-white border border-gray-100 rounded-xl p-4 shadow-sm">
|
| 206 |
+
<h3 class="text-xs font-semibold text-gray-500 uppercase tracking-wide mb-3">対応ファイル形式</h3>
|
| 207 |
<div class="flex flex-wrap gap-2">
|
| 208 |
<span class="text-xs bg-gray-100 text-gray-600 px-2 py-1 rounded-md font-mono">JPG</span>
|
| 209 |
<span class="text-xs bg-gray-100 text-gray-600 px-2 py-1 rounded-md font-mono">PNG</span>
|
| 210 |
<span class="text-xs bg-gray-100 text-gray-600 px-2 py-1 rounded-md font-mono">WEBP</span>
|
| 211 |
<span class="text-xs bg-gray-100 text-gray-600 px-2 py-1 rounded-md font-mono">BMP</span>
|
| 212 |
<span class="text-xs bg-gray-100 text-gray-600 px-2 py-1 rounded-md font-mono">TIFF</span>
|
| 213 |
+
<span class="text-xs bg-emerald-100 text-emerald-700 px-2 py-1 rounded-md font-mono">PDF</span>
|
| 214 |
</div>
|
| 215 |
+
<p class="text-xs text-gray-400 mt-2">※ PDF は1ページ目のみ処理します</p>
|
| 216 |
<h3 class="text-xs font-semibold text-gray-500 uppercase tracking-wide mt-3 mb-2">精度を上げるコツ</h3>
|
| 217 |
<ul class="text-xs text-gray-500 space-y-1">
|
| 218 |
<li>• 表全体が収まるように撮影</li>
|
|
|
|
| 268 |
e.preventDefault();
|
| 269 |
dropZone.classList.remove('drag-over');
|
| 270 |
const file = e.dataTransfer.files[0];
|
| 271 |
+
if (file && (file.type.startsWith('image/') || file.type === 'application/pdf' || file.name.toLowerCase().endsWith('.pdf'))) {
|
| 272 |
showPreview(file);
|
| 273 |
const dt = new DataTransfer();
|
| 274 |
dt.items.add(file);
|
|
|
|
| 277 |
});
|
| 278 |
|
| 279 |
function showPreview(file) {
|
| 280 |
+
const isPdf = file.type === 'application/pdf' || file.name.toLowerCase().endsWith('.pdf');
|
| 281 |
+
if (isPdf) {
|
| 282 |
+
preview.classList.add('hidden');
|
|
|
|
| 283 |
uploadPh.classList.add('hidden');
|
| 284 |
changeBtn.classList.remove('hidden');
|
| 285 |
+
let pdfLabel = dropZone.querySelector('#pdf-label');
|
| 286 |
+
if (!pdfLabel) {
|
| 287 |
+
pdfLabel = document.createElement('div');
|
| 288 |
+
pdfLabel.id = 'pdf-label';
|
| 289 |
+
pdfLabel.className = 'flex flex-col items-center gap-2 text-emerald-700 py-4';
|
| 290 |
+
dropZone.appendChild(pdfLabel);
|
| 291 |
+
}
|
| 292 |
+
pdfLabel.innerHTML = `
|
| 293 |
+
<svg class="w-12 h-12 text-emerald-400" fill="none" viewBox="0 0 24 24" stroke="currentColor" stroke-width="1.5">
|
| 294 |
+
<path stroke-linecap="round" stroke-linejoin="round"
|
| 295 |
+
d="M19.5 14.25v-2.625a3.375 3.375 0 00-3.375-3.375h-1.5A1.125 1.125 0 0113.5 7.125v-1.5a3.375 3.375 0 00-3.375-3.375H8.25m0 12.75h7.5m-7.5 3H12M10.5 2.25H5.625c-.621 0-1.125.504-1.125 1.125v17.25c0 .621.504 1.125 1.125 1.125h12.75c.621 0 1.125-.504 1.125-1.125V11.25a9 9 0 00-9-9z" />
|
| 296 |
+
</svg>
|
| 297 |
+
<p class="text-sm font-semibold">${file.name}</p>
|
| 298 |
+
<p class="text-xs text-gray-400">${(file.size / 1024).toFixed(1)} KB(1ページ目を処理)</p>`;
|
| 299 |
+
} else {
|
| 300 |
+
const existingLabel = dropZone.querySelector('#pdf-label');
|
| 301 |
+
if (existingLabel) existingLabel.remove();
|
| 302 |
+
const reader = new FileReader();
|
| 303 |
+
reader.onload = (e) => {
|
| 304 |
+
preview.src = e.target.result;
|
| 305 |
+
preview.classList.remove('hidden');
|
| 306 |
+
uploadPh.classList.add('hidden');
|
| 307 |
+
changeBtn.classList.remove('hidden');
|
| 308 |
+
};
|
| 309 |
+
reader.readAsDataURL(file);
|
| 310 |
+
}
|
| 311 |
}
|
| 312 |
|
| 313 |
// ── ステップ表示 ────────────────────────────────────────────────
|
MM_tools/app/blueprints/ocr/__pycache__/routes.cpython-314.pyc
CHANGED
|
Binary files a/MM_tools/app/blueprints/ocr/__pycache__/routes.cpython-314.pyc and b/MM_tools/app/blueprints/ocr/__pycache__/routes.cpython-314.pyc differ
|
|
|
MM_tools/app/blueprints/ocr/routes.py
CHANGED
|
@@ -15,7 +15,7 @@ from PIL import Image
|
|
| 15 |
|
| 16 |
from . import ocr_bp
|
| 17 |
from app.shared.models import get_glm_model, get_lfm, is_glm_loaded, is_lfm_loaded, get_lfm_lock
|
| 18 |
-
from app.shared.utils import sse_event, resize_image_to_base64
|
| 19 |
|
| 20 |
|
| 21 |
@ocr_bp.route("/")
|
|
@@ -46,10 +46,11 @@ def run_ocr():
|
|
| 46 |
|
| 47 |
if "image" not in request.files or request.files["image"].filename == "":
|
| 48 |
def _err():
|
| 49 |
-
yield sse_event({"error": "画像をアップロードしてください。"})
|
| 50 |
return Response(stream_with_context(_err()), mimetype="text/event-stream")
|
| 51 |
|
| 52 |
image_file = request.files["image"]
|
|
|
|
| 53 |
prompt = request.form.get("prompt", "").strip()
|
| 54 |
if not prompt:
|
| 55 |
prompt = "画像内のテキストをすべて読み取ってHTMLとして出力してください。"
|
|
@@ -59,17 +60,17 @@ def run_ocr():
|
|
| 59 |
image_bytes = image_file.read()
|
| 60 |
except Exception as exc:
|
| 61 |
def _err():
|
| 62 |
-
yield sse_event({"error": f"
|
| 63 |
return Response(stream_with_context(_err()), mimetype="text/event-stream")
|
| 64 |
|
| 65 |
def generate():
|
| 66 |
"""モデルロード → OCR → SSE ストリーミングを行うジェネレータ。"""
|
| 67 |
-
# ── 画像を開く ──────────────────────
|
| 68 |
try:
|
| 69 |
-
image =
|
| 70 |
image_base64 = resize_image_to_base64(image)
|
| 71 |
except Exception as exc:
|
| 72 |
-
yield sse_event({"error": f"
|
| 73 |
return
|
| 74 |
|
| 75 |
# ── モデルロード状態を通知 ──────────────────────────────────
|
|
|
|
| 15 |
|
| 16 |
from . import ocr_bp
|
| 17 |
from app.shared.models import get_glm_model, get_lfm, is_glm_loaded, is_lfm_loaded, get_lfm_lock
|
| 18 |
+
from app.shared.utils import sse_event, resize_image_to_base64, open_image_or_pdf
|
| 19 |
|
| 20 |
|
| 21 |
@ocr_bp.route("/")
|
|
|
|
| 46 |
|
| 47 |
if "image" not in request.files or request.files["image"].filename == "":
|
| 48 |
def _err():
|
| 49 |
+
yield sse_event({"error": "画像または PDF をアップロードしてください。"})
|
| 50 |
return Response(stream_with_context(_err()), mimetype="text/event-stream")
|
| 51 |
|
| 52 |
image_file = request.files["image"]
|
| 53 |
+
image_filename = image_file.filename # PDF 判定用(ジェネレータ内で使用)
|
| 54 |
prompt = request.form.get("prompt", "").strip()
|
| 55 |
if not prompt:
|
| 56 |
prompt = "画像内のテキストをすべて読み取ってHTMLとして出力してください。"
|
|
|
|
| 60 |
image_bytes = image_file.read()
|
| 61 |
except Exception as exc:
|
| 62 |
def _err():
|
| 63 |
+
yield sse_event({"error": f"ファイルの読み込みに失敗しました: {exc}"})
|
| 64 |
return Response(stream_with_context(_err()), mimetype="text/event-stream")
|
| 65 |
|
| 66 |
def generate():
|
| 67 |
"""モデルロード → OCR → SSE ストリーミングを行うジェネレータ。"""
|
| 68 |
+
# ── 画像を開く(PDF は1ページ目を変換)──────────────────────
|
| 69 |
try:
|
| 70 |
+
image = open_image_or_pdf(image_bytes, image_filename)
|
| 71 |
image_base64 = resize_image_to_base64(image)
|
| 72 |
except Exception as exc:
|
| 73 |
+
yield sse_event({"error": f"ファイルの読み込みに失敗しました: {exc}"})
|
| 74 |
return
|
| 75 |
|
| 76 |
# ── モデルロード状態を通知 ──────────────────────────────────
|
MM_tools/app/blueprints/ocr/templates/ocr/index.html
CHANGED
|
@@ -47,7 +47,7 @@
|
|
| 47 |
flex flex-col items-center justify-center min-h-[360px] cursor-pointer
|
| 48 |
transition-colors duration-200 hover:border-indigo-500 hover:bg-indigo-100">
|
| 49 |
|
| 50 |
-
<input type="file" id="image-input" accept="image/*" class="hidden" />
|
| 51 |
<img id="preview" class="hidden max-w-full max-h-80 object-contain rounded-lg shadow-sm" />
|
| 52 |
|
| 53 |
<div id="upload-placeholder" class="text-center px-6 py-8 select-none">
|
|
@@ -58,7 +58,7 @@
|
|
| 58 |
</svg>
|
| 59 |
</div>
|
| 60 |
<p class="text-indigo-800 font-medium text-sm">クリックまたはドラッグ&ドロップ</p>
|
| 61 |
-
<p class="text-gray-400 text-xs mt-1">PNG, JPG, WEBP など</p>
|
| 62 |
</div>
|
| 63 |
|
| 64 |
<button id="change-btn"
|
|
@@ -365,7 +365,7 @@
|
|
| 365 |
e.preventDefault();
|
| 366 |
dropZone.classList.remove('drag-over');
|
| 367 |
const file = e.dataTransfer.files[0];
|
| 368 |
-
if (file && file.type.startsWith('image/')) {
|
| 369 |
showPreview(file);
|
| 370 |
const dt = new DataTransfer();
|
| 371 |
dt.items.add(file);
|
|
@@ -374,14 +374,38 @@
|
|
| 374 |
});
|
| 375 |
|
| 376 |
function showPreview(file) {
|
| 377 |
-
const
|
| 378 |
-
|
| 379 |
-
preview.
|
| 380 |
-
preview.classList.remove('hidden');
|
| 381 |
placeholder.classList.add('hidden');
|
| 382 |
changeBtn.classList.remove('hidden');
|
| 383 |
-
|
| 384 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 385 |
}
|
| 386 |
|
| 387 |
// ── SSE 共通パーサー ────────────────────────────────────────────
|
|
|
|
| 47 |
flex flex-col items-center justify-center min-h-[360px] cursor-pointer
|
| 48 |
transition-colors duration-200 hover:border-indigo-500 hover:bg-indigo-100">
|
| 49 |
|
| 50 |
+
<input type="file" id="image-input" accept="image/*,.pdf" class="hidden" />
|
| 51 |
<img id="preview" class="hidden max-w-full max-h-80 object-contain rounded-lg shadow-sm" />
|
| 52 |
|
| 53 |
<div id="upload-placeholder" class="text-center px-6 py-8 select-none">
|
|
|
|
| 58 |
</svg>
|
| 59 |
</div>
|
| 60 |
<p class="text-indigo-800 font-medium text-sm">クリックまたはドラッグ&ドロップ</p>
|
| 61 |
+
<p class="text-gray-400 text-xs mt-1">PNG, JPG, WEBP, PDF など</p>
|
| 62 |
</div>
|
| 63 |
|
| 64 |
<button id="change-btn"
|
|
|
|
| 365 |
e.preventDefault();
|
| 366 |
dropZone.classList.remove('drag-over');
|
| 367 |
const file = e.dataTransfer.files[0];
|
| 368 |
+
if (file && (file.type.startsWith('image/') || file.type === 'application/pdf' || file.name.toLowerCase().endsWith('.pdf'))) {
|
| 369 |
showPreview(file);
|
| 370 |
const dt = new DataTransfer();
|
| 371 |
dt.items.add(file);
|
|
|
|
| 374 |
});
|
| 375 |
|
| 376 |
function showPreview(file) {
|
| 377 |
+
const isPdf = file.type === 'application/pdf' || file.name.toLowerCase().endsWith('.pdf');
|
| 378 |
+
if (isPdf) {
|
| 379 |
+
preview.classList.add('hidden');
|
|
|
|
| 380 |
placeholder.classList.add('hidden');
|
| 381 |
changeBtn.classList.remove('hidden');
|
| 382 |
+
// PDF の場合はファイル名を表示
|
| 383 |
+
let pdfLabel = dropZone.querySelector('#pdf-label');
|
| 384 |
+
if (!pdfLabel) {
|
| 385 |
+
pdfLabel = document.createElement('div');
|
| 386 |
+
pdfLabel.id = 'pdf-label';
|
| 387 |
+
pdfLabel.className = 'flex flex-col items-center gap-2 text-indigo-700 py-4';
|
| 388 |
+
dropZone.appendChild(pdfLabel);
|
| 389 |
+
}
|
| 390 |
+
pdfLabel.innerHTML = `
|
| 391 |
+
<svg class="w-12 h-12 text-indigo-400" fill="none" viewBox="0 0 24 24" stroke="currentColor" stroke-width="1.5">
|
| 392 |
+
<path stroke-linecap="round" stroke-linejoin="round"
|
| 393 |
+
d="M19.5 14.25v-2.625a3.375 3.375 0 00-3.375-3.375h-1.5A1.125 1.125 0 0113.5 7.125v-1.5a3.375 3.375 0 00-3.375-3.375H8.25m0 12.75h7.5m-7.5 3H12M10.5 2.25H5.625c-.621 0-1.125.504-1.125 1.125v17.25c0 .621.504 1.125 1.125 1.125h12.75c.621 0 1.125-.504 1.125-1.125V11.25a9 9 0 00-9-9z" />
|
| 394 |
+
</svg>
|
| 395 |
+
<p class="text-sm font-semibold">${file.name}</p>
|
| 396 |
+
<p class="text-xs text-gray-400">${(file.size / 1024).toFixed(1)} KB(1ページ目を処理)</p>`;
|
| 397 |
+
} else {
|
| 398 |
+
const existingLabel = dropZone.querySelector('#pdf-label');
|
| 399 |
+
if (existingLabel) existingLabel.remove();
|
| 400 |
+
const reader = new FileReader();
|
| 401 |
+
reader.onload = (e) => {
|
| 402 |
+
preview.src = e.target.result;
|
| 403 |
+
preview.classList.remove('hidden');
|
| 404 |
+
placeholder.classList.add('hidden');
|
| 405 |
+
changeBtn.classList.remove('hidden');
|
| 406 |
+
};
|
| 407 |
+
reader.readAsDataURL(file);
|
| 408 |
+
}
|
| 409 |
}
|
| 410 |
|
| 411 |
// ── SSE 共通パーサー ────────────────────────────────────────────
|
MM_tools/app/blueprints/vlm/__pycache__/routes.cpython-314.pyc
CHANGED
|
Binary files a/MM_tools/app/blueprints/vlm/__pycache__/routes.cpython-314.pyc and b/MM_tools/app/blueprints/vlm/__pycache__/routes.cpython-314.pyc differ
|
|
|
MM_tools/app/blueprints/vlm/routes.py
CHANGED
|
@@ -20,6 +20,7 @@ from PIL import Image
|
|
| 20 |
|
| 21 |
from . import vlm_bp
|
| 22 |
from app.shared.models import get_qwen_model
|
|
|
|
| 23 |
|
| 24 |
# Qwen モデルは ledger と共用するため排他ロック
|
| 25 |
_model_lock = threading.Lock()
|
|
@@ -54,7 +55,7 @@ def process():
|
|
| 54 |
|
| 55 |
if not image_file or not image_file.filename:
|
| 56 |
def _err():
|
| 57 |
-
yield _sse({"error": "画像をアップロードしてください。"})
|
| 58 |
return Response(stream_with_context(_err()), mimetype="text/event-stream")
|
| 59 |
|
| 60 |
if not prompt:
|
|
@@ -64,18 +65,19 @@ def process():
|
|
| 64 |
|
| 65 |
# SSE ジェネレータの遅延実行でリクエストストリームが閉じられる問題を防ぐため先読み
|
| 66 |
image_bytes = image_file.read()
|
|
|
|
| 67 |
|
| 68 |
def generate():
|
| 69 |
from transformers import TextIteratorStreamer
|
| 70 |
with _model_lock:
|
| 71 |
try:
|
| 72 |
-
# ── 画像読み込み ──
|
| 73 |
yield _sse({
|
| 74 |
"status": "processing",
|
| 75 |
-
"message": "
|
| 76 |
"detail": "",
|
| 77 |
})
|
| 78 |
-
image =
|
| 79 |
|
| 80 |
# ── モデルロード ──
|
| 81 |
yield _sse({
|
|
|
|
| 20 |
|
| 21 |
from . import vlm_bp
|
| 22 |
from app.shared.models import get_qwen_model
|
| 23 |
+
from app.shared.utils import open_image_or_pdf
|
| 24 |
|
| 25 |
# Qwen モデルは ledger と共用するため排他ロック
|
| 26 |
_model_lock = threading.Lock()
|
|
|
|
| 55 |
|
| 56 |
if not image_file or not image_file.filename:
|
| 57 |
def _err():
|
| 58 |
+
yield _sse({"error": "画像または PDF をアップロードしてください。"})
|
| 59 |
return Response(stream_with_context(_err()), mimetype="text/event-stream")
|
| 60 |
|
| 61 |
if not prompt:
|
|
|
|
| 65 |
|
| 66 |
# SSE ジェネレータの遅延実行でリクエストストリームが閉じられる問題を防ぐため先読み
|
| 67 |
image_bytes = image_file.read()
|
| 68 |
+
image_filename = image_file.filename # PDF 判定用
|
| 69 |
|
| 70 |
def generate():
|
| 71 |
from transformers import TextIteratorStreamer
|
| 72 |
with _model_lock:
|
| 73 |
try:
|
| 74 |
+
# ── 画像読み込み(PDF は1ページ目を変換)──
|
| 75 |
yield _sse({
|
| 76 |
"status": "processing",
|
| 77 |
+
"message": "ファイルを読み込み中...",
|
| 78 |
"detail": "",
|
| 79 |
})
|
| 80 |
+
image = open_image_or_pdf(image_bytes, image_filename)
|
| 81 |
|
| 82 |
# ── モデルロード ──
|
| 83 |
yield _sse({
|
MM_tools/app/blueprints/vlm/templates/vlm/index.html
CHANGED
|
@@ -67,7 +67,7 @@
|
|
| 67 |
<div id="drop-zone"
|
| 68 |
class="relative rounded-xl bg-amber-50/50 cursor-pointer flex flex-col items-center justify-center gap-3 p-6 min-h-[200px]"
|
| 69 |
onclick="document.getElementById('file-input').click()">
|
| 70 |
-
<input id="file-input" type="file" accept="image/*" class="hidden" />
|
| 71 |
<!-- プレビュー(非表示時) -->
|
| 72 |
<div id="upload-placeholder" class="flex flex-col items-center gap-2 text-amber-600 pointer-events-none">
|
| 73 |
<svg class="w-12 h-12 opacity-60" fill="none" viewBox="0 0 24 24" stroke="currentColor" stroke-width="1.5">
|
|
@@ -75,7 +75,7 @@
|
|
| 75 |
d="M2.25 15.75l5.159-5.159a2.25 2.25 0 013.182 0l5.159 5.159m-1.5-1.5l1.409-1.409a2.25 2.25 0 013.182 0l2.909 2.909M3 16.5V18a2.25 2.25 0 002.25 2.25h13.5A2.25 2.25 0 0021 18v-1.5M16.5 6.75a2.25 2.25 0 11-4.5 0 2.25 2.25 0 014.5 0z" />
|
| 76 |
</svg>
|
| 77 |
<p class="text-sm font-medium">クリックまたはドラッグ&ドロップ</p>
|
| 78 |
-
<p class="text-xs text-gray-400">PNG / JPG / WEBP など</p>
|
| 79 |
</div>
|
| 80 |
<!-- 画像プレビュー -->
|
| 81 |
<img id="preview-img" src="" alt="プレビュー"
|
|
@@ -161,15 +161,40 @@
|
|
| 161 |
const fileInfo = document.getElementById('file-info');
|
| 162 |
|
| 163 |
function showPreview(file) {
|
| 164 |
-
const
|
| 165 |
-
|
| 166 |
-
previewImg.
|
| 167 |
-
previewImg.classList.remove('hidden');
|
| 168 |
uploadPlaceholder.classList.add('hidden');
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 169 |
fileInfo.textContent = `${file.name} (${(file.size / 1024).toFixed(1)} KB)`;
|
| 170 |
fileInfo.classList.remove('hidden');
|
| 171 |
-
}
|
| 172 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 173 |
}
|
| 174 |
|
| 175 |
fileInput.addEventListener('change', () => {
|
|
@@ -185,7 +210,7 @@
|
|
| 185 |
e.preventDefault();
|
| 186 |
dropZone.classList.remove('drag-over');
|
| 187 |
const file = e.dataTransfer.files[0];
|
| 188 |
-
if (file && file.type.startsWith('image/')) {
|
| 189 |
const dt = new DataTransfer();
|
| 190 |
dt.items.add(file);
|
| 191 |
fileInput.files = dt.files;
|
|
@@ -230,7 +255,7 @@
|
|
| 230 |
|
| 231 |
runBtn.addEventListener('click', async () => {
|
| 232 |
if (!fileInput.files[0]) {
|
| 233 |
-
alert('画像をアップロードしてください。');
|
| 234 |
return;
|
| 235 |
}
|
| 236 |
const prompt = document.getElementById('prompt-input').value.trim();
|
|
|
|
| 67 |
<div id="drop-zone"
|
| 68 |
class="relative rounded-xl bg-amber-50/50 cursor-pointer flex flex-col items-center justify-center gap-3 p-6 min-h-[200px]"
|
| 69 |
onclick="document.getElementById('file-input').click()">
|
| 70 |
+
<input id="file-input" type="file" accept="image/*,.pdf" class="hidden" />
|
| 71 |
<!-- プレビュー(非表示時) -->
|
| 72 |
<div id="upload-placeholder" class="flex flex-col items-center gap-2 text-amber-600 pointer-events-none">
|
| 73 |
<svg class="w-12 h-12 opacity-60" fill="none" viewBox="0 0 24 24" stroke="currentColor" stroke-width="1.5">
|
|
|
|
| 75 |
d="M2.25 15.75l5.159-5.159a2.25 2.25 0 013.182 0l5.159 5.159m-1.5-1.5l1.409-1.409a2.25 2.25 0 013.182 0l2.909 2.909M3 16.5V18a2.25 2.25 0 002.25 2.25h13.5A2.25 2.25 0 0021 18v-1.5M16.5 6.75a2.25 2.25 0 11-4.5 0 2.25 2.25 0 014.5 0z" />
|
| 76 |
</svg>
|
| 77 |
<p class="text-sm font-medium">クリックまたはドラッグ&ドロップ</p>
|
| 78 |
+
<p class="text-xs text-gray-400">PNG / JPG / WEBP / PDF など</p>
|
| 79 |
</div>
|
| 80 |
<!-- 画像プレビュー -->
|
| 81 |
<img id="preview-img" src="" alt="プレビュー"
|
|
|
|
| 161 |
const fileInfo = document.getElementById('file-info');
|
| 162 |
|
| 163 |
function showPreview(file) {
|
| 164 |
+
const isPdf = file.type === 'application/pdf' || file.name.toLowerCase().endsWith('.pdf');
|
| 165 |
+
if (isPdf) {
|
| 166 |
+
previewImg.classList.add('hidden');
|
|
|
|
| 167 |
uploadPlaceholder.classList.add('hidden');
|
| 168 |
+
// PDF アイコン表示
|
| 169 |
+
let pdfLabel = dropZone.querySelector('#pdf-label');
|
| 170 |
+
if (!pdfLabel) {
|
| 171 |
+
pdfLabel = document.createElement('div');
|
| 172 |
+
pdfLabel.id = 'pdf-label';
|
| 173 |
+
pdfLabel.className = 'flex flex-col items-center gap-2 text-amber-600 pointer-events-none';
|
| 174 |
+
dropZone.appendChild(pdfLabel);
|
| 175 |
+
}
|
| 176 |
+
pdfLabel.innerHTML = `
|
| 177 |
+
<svg class="w-12 h-12 opacity-80" fill="none" viewBox="0 0 24 24" stroke="currentColor" stroke-width="1.5">
|
| 178 |
+
<path stroke-linecap="round" stroke-linejoin="round"
|
| 179 |
+
d="M19.5 14.25v-2.625a3.375 3.375 0 00-3.375-3.375h-1.5A1.125 1.125 0 0113.5 7.125v-1.5a3.375 3.375 0 00-3.375-3.375H8.25m0 12.75h7.5m-7.5 3H12M10.5 2.25H5.625c-.621 0-1.125.504-1.125 1.125v17.25c0 .621.504 1.125 1.125 1.125h12.75c.621 0 1.125-.504 1.125-1.125V11.25a9 9 0 00-9-9z" />
|
| 180 |
+
</svg>
|
| 181 |
+
<p class="text-sm font-semibold">${file.name}</p>
|
| 182 |
+
<p class="text-xs text-gray-400">1ページ目を処理します</p>`;
|
| 183 |
fileInfo.textContent = `${file.name} (${(file.size / 1024).toFixed(1)} KB)`;
|
| 184 |
fileInfo.classList.remove('hidden');
|
| 185 |
+
} else {
|
| 186 |
+
const existingLabel = dropZone.querySelector('#pdf-label');
|
| 187 |
+
if (existingLabel) existingLabel.remove();
|
| 188 |
+
const reader = new FileReader();
|
| 189 |
+
reader.onload = e => {
|
| 190 |
+
previewImg.src = e.target.result;
|
| 191 |
+
previewImg.classList.remove('hidden');
|
| 192 |
+
uploadPlaceholder.classList.add('hidden');
|
| 193 |
+
fileInfo.textContent = `${file.name} (${(file.size / 1024).toFixed(1)} KB)`;
|
| 194 |
+
fileInfo.classList.remove('hidden');
|
| 195 |
+
};
|
| 196 |
+
reader.readAsDataURL(file);
|
| 197 |
+
}
|
| 198 |
}
|
| 199 |
|
| 200 |
fileInput.addEventListener('change', () => {
|
|
|
|
| 210 |
e.preventDefault();
|
| 211 |
dropZone.classList.remove('drag-over');
|
| 212 |
const file = e.dataTransfer.files[0];
|
| 213 |
+
if (file && (file.type.startsWith('image/') || file.type === 'application/pdf' || file.name.toLowerCase().endsWith('.pdf'))) {
|
| 214 |
const dt = new DataTransfer();
|
| 215 |
dt.items.add(file);
|
| 216 |
fileInput.files = dt.files;
|
|
|
|
| 255 |
|
| 256 |
runBtn.addEventListener('click', async () => {
|
| 257 |
if (!fileInput.files[0]) {
|
| 258 |
+
alert('画像または PDF をアップロードしてください。');
|
| 259 |
return;
|
| 260 |
}
|
| 261 |
const prompt = document.getElementById('prompt-input').value.trim();
|
MM_tools/app/shared/__pycache__/utils.cpython-314.pyc
CHANGED
|
Binary files a/MM_tools/app/shared/__pycache__/utils.cpython-314.pyc and b/MM_tools/app/shared/__pycache__/utils.cpython-314.pyc differ
|
|
|
MM_tools/app/shared/utils.py
CHANGED
|
@@ -20,6 +20,66 @@ def sse_event(data: dict) -> str:
|
|
| 20 |
return f"data: {json.dumps(data, ensure_ascii=False)}\n\n"
|
| 21 |
|
| 22 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
def resize_image_to_base64(image: Image.Image, max_size: int = 512) -> str:
|
| 24 |
"""
|
| 25 |
PIL 画像をリサイズして Base64 エンコード済み JPEG 文字列を返す。
|
|
|
|
| 20 |
return f"data: {json.dumps(data, ensure_ascii=False)}\n\n"
|
| 21 |
|
| 22 |
|
| 23 |
+
def open_image_or_pdf(file_bytes: bytes, filename: str = "") -> Image.Image:
|
| 24 |
+
"""
|
| 25 |
+
画像バイト列または PDF バイト列から PIL Image (RGB) を返す。
|
| 26 |
+
PDF の場合は1ページ目を 2x 解像度でレンダリングして返す。
|
| 27 |
+
|
| 28 |
+
Args:
|
| 29 |
+
file_bytes: 画像または PDF のバイト列
|
| 30 |
+
filename: 元ファイル名(拡張子で PDF を判別する際に使用)
|
| 31 |
+
|
| 32 |
+
Returns:
|
| 33 |
+
PIL.Image: RGB 形式の画像
|
| 34 |
+
|
| 35 |
+
Raises:
|
| 36 |
+
ImportError: PDF 処理に pymupdf が必要なのにインストールされていない場合
|
| 37 |
+
ValueError: PDF にページが存在しない場合
|
| 38 |
+
"""
|
| 39 |
+
is_pdf = (
|
| 40 |
+
filename.lower().endswith(".pdf")
|
| 41 |
+
or file_bytes[:4] == b"%PDF"
|
| 42 |
+
)
|
| 43 |
+
if is_pdf:
|
| 44 |
+
return _pdf_first_page_to_image(file_bytes)
|
| 45 |
+
return Image.open(io.BytesIO(file_bytes)).convert("RGB")
|
| 46 |
+
|
| 47 |
+
|
| 48 |
+
def _pdf_first_page_to_image(pdf_bytes: bytes) -> Image.Image:
|
| 49 |
+
"""
|
| 50 |
+
PDF バイト列の1ページ目を PIL Image に変換する(内部用)。
|
| 51 |
+
|
| 52 |
+
Args:
|
| 53 |
+
pdf_bytes: PDF ファイルのバイト列
|
| 54 |
+
|
| 55 |
+
Returns:
|
| 56 |
+
PIL.Image: 1ページ目を 2x 解像度でレンダリングした RGB 画像
|
| 57 |
+
|
| 58 |
+
Raises:
|
| 59 |
+
ImportError: pymupdf がインストールされていない場合
|
| 60 |
+
ValueError: PDF にページが存在しない場合
|
| 61 |
+
"""
|
| 62 |
+
try:
|
| 63 |
+
import fitz # pymupdf
|
| 64 |
+
except ImportError as exc:
|
| 65 |
+
raise ImportError(
|
| 66 |
+
"PDF の処理には pymupdf が必要です: pip install pymupdf"
|
| 67 |
+
) from exc
|
| 68 |
+
|
| 69 |
+
doc = fitz.open(stream=pdf_bytes, filetype="pdf")
|
| 70 |
+
if doc.page_count == 0:
|
| 71 |
+
doc.close()
|
| 72 |
+
raise ValueError("PDF にページが存在しません。")
|
| 73 |
+
|
| 74 |
+
page = doc[0]
|
| 75 |
+
mat = fitz.Matrix(2.0, 2.0) # 2x 解像度でレンダリング
|
| 76 |
+
pix = page.get_pixmap(matrix=mat)
|
| 77 |
+
img_bytes = pix.tobytes("png")
|
| 78 |
+
doc.close()
|
| 79 |
+
|
| 80 |
+
return Image.open(io.BytesIO(img_bytes)).convert("RGB")
|
| 81 |
+
|
| 82 |
+
|
| 83 |
def resize_image_to_base64(image: Image.Image, max_size: int = 512) -> str:
|
| 84 |
"""
|
| 85 |
PIL 画像をリサイズして Base64 エンコード済み JPEG 文字列を返す。
|
MM_tools/app/static/exports/whisper_text_time.txt
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
[1s → 26s] サーチングカット切削油は、非作成の悪いステンレス、シタン、コバール、インコネル、銅といった難作材加工油としてユーザー様にご指示をいただき、精度を重視する自動車部品、医療機器部品、衛生関連部品、航空機部品、弱電部品などの金属加工油として実績がございます。
|
|
|
|
| 1 |
+
[1s → 26s] 搜索切削油是不制造好的不良钢材、锯齿、铸铁、铬、铜等难加工材料的润滑油,作为精密加工的汽车零部件、医疗设备、卫生相关零部件、航空机零部件及弱电部件的金属加工油,拥有丰富的经验。 (原文: サーチングカット切削油は、非作成の悪いステンレス、シタン、コバール、インコネル、銅といった難作材加工油としてユーザー様にご指示をいただき、精度を重視する自動車部品、医療機器部品、衛生関連部品、航空機部品、弱電部品などの金属加工油として実績がございます。)
|