yomitoku 0.9.2__py3-none-any.whl → 0.9.4__py3-none-any.whl

yomitoku/cli/main.py

@@ -49,10 +49,15 @@ def merge_all_pages(results):
             else:
                 out += "\n" + data
 
+        elif format == "pdf":
+            if out is None:
+                out = [data]
+            else:
+                out.append(data)
     return out
 
 
-def save_merged_file(out_path, args, out):
+def save_merged_file(out_path, args, out, imgs):
     if args.format == "json":
         save_json(out, out_path, args.encoding)
     elif args.format == "csv":
@@ -61,6 +66,13 @@ def save_merged_file(out_path, args, out):
         save_html(out, out_path, args.encoding)
     elif args.format == "md":
         save_markdown(out, out_path, args.encoding)
+    elif args.format == "pdf":
+        create_searchable_pdf(
+            imgs,
+            out,
+            output_path=out_path,
+            font_path=args.font_path,
+        )
 
 
 def validate_encoding(encoding):
@@ -82,12 +94,10 @@ def process_single_file(args, analyzer, path, format):
         imgs = load_image(path)
 
     format_results = []
-    results = []
     for page, img in enumerate(imgs):
         result, ocr, layout = analyzer(img)
         dirname = path.parent.name
         filename = path.stem
-        results.append(result)
 
         # cv2.imwrite(
         #    os.path.join(args.outdir, f"{dirname}_{filename}_p{page+1}.jpg"), img
@@ -148,6 +158,7 @@ def process_single_file(args, analyzer, path, format):
                     args.ignore_line_break,
                     img,
                     args.figure,
+                    args.figure_letter,
                     args.figure_dir,
                 )
             else:
@@ -157,6 +168,7 @@ def process_single_file(args, analyzer, path, format):
                     encoding=args.encoding,
                     img=img,
                     export_figure=args.figure,
+                    export_figure_letter=args.figure_letter,
                     figure_dir=args.figure_dir,
                 )
 
@@ -228,6 +240,21 @@ def process_single_file(args, analyzer, path, format):
                     "data": md,
                 }
             )
+        elif format == "pdf":
+            if not args.combine:
+                create_searchable_pdf(
+                    [img],
+                    [result],
+                    output_path=out_path,
+                    font_path=args.font_path,
+                )
+
+            format_results.append(
+                {
+                    "format": format,
+                    "data": result,
+                }
+            )
 
     out = merge_all_pages(format_results)
     if args.combine:
@@ -236,16 +263,8 @@ def process_single_file(args, analyzer, path, format):
             out_path,
             args,
             out,
-        )
-
-    if args.searchable_pdf:
-        pdf_path = os.path.join(args.outdir, f"{filename}.pdf")
-        create_searchable_pdf(
             imgs,
-            results,
-            output_path=pdf_path,
         )
-        logger.info(f"Output SearchablePDF: {pdf_path}")
 
 
 def main():
@@ -362,11 +381,11 @@ def main():
         choices=["auto", "left2right", "top2bottom", "right2left"],
     )
     parser.add_argument(
-        "--searchable_pdf",
-        action="store_true",
-        help="if set, create searchable PDF",
+        "--font_path",
+        default=None,
+        type=str,
+        help="Path to the font file(.ttf) for PDF output",
     )
-
     args = parser.parse_args()
 
     path = Path(args.arg1)
@@ -379,6 +398,13 @@ def main():
             f"Invalid output format: {args.format}. Supported formats are {SUPPORT_OUTPUT_FORMAT}"
         )
 
+    if (
+        args.font_path is not None
+        and not os.path.exists(args.font_path)
+        and format == "pdf"
+    ):
+        raise FileNotFoundError(f"Font file not found: {args.font_path}")
+
     validate_encoding(args.encoding)
 
     if format == "markdown":

yomitoku/constants.py

@@ -1,7 +1,7 @@
 import os
 
 ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
-SUPPORT_OUTPUT_FORMAT = ["json", "csv", "html", "markdown", "md"]
+SUPPORT_OUTPUT_FORMAT = ["json", "csv", "html", "markdown", "md", "pdf"]
 SUPPORT_INPUT_FORMAT = ["jpg", "jpeg", "png", "bmp", "tiff", "tif", "pdf"]
 MIN_IMAGE_SIZE = 32
 WARNING_IMAGE_SIZE = 720

yomitoku/export/export_csv.py

@@ -63,6 +63,7 @@ def convert_csv(
     ignore_line_break,
     img=None,
     export_figure: bool = True,
+    export_figure_letter: bool = False,
     figure_dir="figures",
 ):
     elements = []
@@ -89,6 +90,20 @@ def convert_csv(
             }
         )
 
+    if export_figure_letter:
+        for figure in inputs.figures:
+            paragraphs = sorted(figure.paragraphs, key=lambda x: x.order)
+            for paragraph in paragraphs:
+                contents = paragraph_to_csv(paragraph, ignore_line_break)
+                elements.append(
+                    {
+                        "type": "paragraph",
+                        "box": paragraph.box,
+                        "element": contents,
+                        "order": figure.order,
+                    }
+                )
+
     elements = sorted(elements, key=lambda x: x["order"])
 
     if export_figure:
@@ -109,6 +124,7 @@ def export_csv(
     encoding: str = "utf-8",
     img=None,
     export_figure: bool = True,
+    export_figure_letter: bool = False,
     figure_dir="figures",
 ):
     elements = convert_csv(
@@ -117,6 +133,7 @@ def export_csv(
         ignore_line_break,
         img,
         export_figure,
+        export_figure_letter,
         figure_dir,
     )
 

yomitoku/utils/searchable_pdf.py

@@ -14,7 +14,6 @@ import jaconv
 from ..constants import ROOT_DIR
 
 FONT_PATH = ROOT_DIR + "/resource/MPLUS1p-Medium.ttf"
-pdfmetrics.registerFont(TTFont("MPLUS1p-Medium", FONT_PATH))
 
 
 def _poly2rect(points):
@@ -62,14 +61,17 @@ def to_full_width(text):
     return jaconv_text
 
 
-def create_searchable_pdf(images, ocr_results, output_path):
+def create_searchable_pdf(images, ocr_results, output_path, font_path=None):
+    if font_path is None:
+        font_path = FONT_PATH
+
+    pdfmetrics.registerFont(TTFont("MPLUS1p-Medium", font_path))
+
     packet = BytesIO()
     c = canvas.Canvas(packet)
 
     for i, (image, ocr_result) in enumerate(zip(images, ocr_results)):
         image = Image.fromarray(image[:, :, ::-1])  # Convert BGR to RGB
-        pdfmetrics.registerFont(TTFont("MPLUS1p-Medium", FONT_PATH))
-
         image_path = f"tmp_{i}.png"
         image.save(image_path)
         w, h = image.size
@@ -97,7 +99,6 @@ def create_searchable_pdf(images, ocr_results, output_path):
 
             c.setFont("MPLUS1p-Medium", font_size)
             c.setFillColorRGB(1, 1, 1, alpha=0)  # 透明
-            # c.setFillColorRGB(0, 0, 0)
             if direction == "vertical":
                 base_y = h - y2 + (bbox_height - font_size)
                 for j, ch in enumerate(text):

{yomitoku-0.9.2.dist-info → yomitoku-0.9.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: yomitoku
-Version: 0.9.2
+Version: 0.9.4
 Summary: Yomitoku is an AI-powered document image analysis package designed specifically for the Japanese language.
 Author-email: Kotaro Kinoshita <kotaro.kinoshita@mlism.com>
 License: CC BY-NC-SA 4.0
@@ -87,7 +87,7 @@ yomitoku ${path_data} -f md -o results -v --figure --lite
 ```
 
 - `${path_data}` 解析対象の画像が含まれたディレクトリか画像ファイルのパスを直接して指定してください。ディレクトリを対象とした場合はディレクトリのサブディレクトリ内の画像も含めて処理を実行します。
-- `-f`, `--format` 出力形式のファイルフォーマットを指定します。(json, csv, html, md をサポート)
+- `-f`, `--format` 出力形式のファイルフォーマットを指定します。(json, csv, html, md, pdf(searchable-pdf) をサポート)
 - `-o`, `--outdir` 出力先のディレクトリ名を指定します。存在しない場合は新規で作成されます。
 - `-v`, `--vis` を指定すると解析結果を可視化した画像を出力します。
 - `-l`, `--lite` を指定すると軽量モデルで推論を実行します。通常より高速に推論できますが、若干、精度が低下する可能性があります。
@@ -98,7 +98,6 @@ yomitoku ${path_data} -f md -o results -v --figure --lite
 - `--encoding` エクスポートする出力ファイルの文字エンコーディングを指定します。サポートされていない文字コードが含まれる場合は、その文字を無視します。(utf-8, utf-8-sig, shift-jis, enc-jp, cp932)
 - `--combine` PDFを入力に与えたときに、複数ページが含まれる場合に、それらの予測結果を一つのファイルに統合してエクスポートします。
 - `--ignore_meta` 文章のheater, fotterなどの文字情報を出力ファイルに含めません。
-- `--searchable_pdf` 読み取った文字情報をPDFに埋め込み全文検索可能なPDFを出力します。
 
 その他のオプションに関しては、ヘルプを参照
 

{yomitoku-0.9.2.dist-info → yomitoku-0.9.4.dist-info}/RECORD

@@ -1,6 +1,6 @@
 yomitoku/__init__.py,sha256=kXOM8RbpwwLABG3p3vPT3dJWBk4JX2MFGrOeBEW0hKM,543
 yomitoku/base.py,sha256=9U3sfe69O6vuO430JzzKQQNkgPsLM9WdLfOUUhp3Ljs,3878
-yomitoku/constants.py,sha256=zlW5QRc_u_F3C2RAgBFWyHJZexBnJT5N15GC-9d3iLo,686
+yomitoku/constants.py,sha256=2jya14UflDkMdYWMKc-ZllkWbJW2qh59Cnt2brrgNb4,693
 yomitoku/document_analyzer.py,sha256=xliAelQdfsK64FtVuFvstDBr9uf2TwhqW31g2g91_CY,16888
 yomitoku/layout_analyzer.py,sha256=VhNf1ZQFoozj6WUGk5ll1p2p1jk5X3j-JPcDbTAoSl4,1856
 yomitoku/layout_parser.py,sha256=0MgbCsD90srQdsxkGEL0TgKm4rkmGzsQYx0sjKQ03yc,7718
@@ -10,7 +10,7 @@ yomitoku/table_structure_recognizer.py,sha256=tHjex6deT_FjRK5ePz9bUXA_QIhgv_vYtK
 yomitoku/text_detector.py,sha256=6IwEJJKp_F8YH0Oki0QV-Mqi--P2LGbNKo-_kxBB_eo,4383
 yomitoku/text_recognizer.py,sha256=eaxozNu-Ms6iv8efbKZzn8pJNW1Wo4f86bGhzSMtv3s,5992
 yomitoku/cli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-yomitoku/cli/main.py,sha256=7AaaFzMf33ER__XPDBNkrJkKwclne7QyVFWeBvpUYBY,12849
+yomitoku/cli/main.py,sha256=v1UYsnQdnylhLvDURuxLODU3IU-ssVGqOJT9r-TCVns,13623
 yomitoku/cli/mcp_server.py,sha256=WnWzxd13HaemC3b-5i9B9NVBGc3WGfum2nYhoBolEnk,5641
 yomitoku/configs/__init__.py,sha256=x5-ccjGiP6xxRtDPT7f1Enl7SsE0hSk0G8f7eF9V85I,886
 yomitoku/configs/cfg_layout_parser_rtdtrv2.py,sha256=8PRxB2Ar9UF7-DLtbgSokhrzdXb0veWI6Wc-X8qigRw,2329
@@ -25,7 +25,7 @@ yomitoku/data/__init__.py,sha256=KAofFc9rk9ZdTKBjemu9RM8Vj9XnKbWC2MPZ2RWtOdE,82
 yomitoku/data/dataset.py,sha256=lpBcpkMuQzRIyLJ4_mqtuhR9s2ZmzgBgc-XYuE_b2Sc,1326
 yomitoku/data/functions.py,sha256=RExCUxI3-gccIMw-H0ribX2jeGKkrJWhS4fNn_12c3Y,7878
 yomitoku/export/__init__.py,sha256=gmlikMHRXfzfJ_8q4fyDlnpGms-x1oggQOwJEWHMgBU,508
-yomitoku/export/export_csv.py,sha256=VY8mntUCPDbDco_dyvq5O0_Q4wga9_GTyjHCS-y4UiQ,3399
+yomitoku/export/export_csv.py,sha256=4U4KQ2RcBQmyUZ9O7a4uLoB6RUw80HPL1EEJUDwQlcI,4044
 yomitoku/export/export_html.py,sha256=LQDyZgbzmI0qJ0-FEK-54r9816H3L9hD10ChMcw0KyA,5620
 yomitoku/export/export_json.py,sha256=iNG37tdIuYG2x3NiiZemKaB6-X45WrhVPZhbX7RUzRI,2410
 yomitoku/export/export_markdown.py,sha256=KrdxDmKzVP_LbTKuDNGGsT31QOPKVsNNlb6wtLEW-1Q,4705
@@ -51,9 +51,9 @@ yomitoku/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 yomitoku/utils/graph.py,sha256=LKNB8ZhSQwOZMfeAimPMF5UCVVr2ZaUWoGDkz8z-uGU,456
 yomitoku/utils/logger.py,sha256=uOmtQDr0A0JD7wyFshedL08BiNrQorHnpktRXba8bjU,424
 yomitoku/utils/misc.py,sha256=r92x45kQR8lC5jO1MZaHBDtcCWBkQXg_WS9H4RXJzSY,4127
-yomitoku/utils/searchable_pdf.py,sha256=40JbcxWrHzYTtzvI9MPYHMrWqLWKiLWo4mWDNRFXwHY,3530
+yomitoku/utils/searchable_pdf.py,sha256=taZ-XtXN4RItePMDv4q0fRVlryusdkexA3TCXzwlXRo,3497
 yomitoku/utils/visualizer.py,sha256=DjDwHiAu1iFRKh96H3Egq4vuI2s_-9dLCDeykhKi8jo,5251
-yomitoku-0.9.2.dist-info/METADATA,sha256=vDEaaXAimCBfVwMeWmfyJBqzb7sXtZk4-ia3PXrtk7c,8966
-yomitoku-0.9.2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-yomitoku-0.9.2.dist-info/entry_points.txt,sha256=n3c8bQSj5Be5GHAOv_NZ8cldJFmWeigQxSmteFTmu_k,96
-yomitoku-0.9.2.dist-info/RECORD,,
+yomitoku-0.9.4.dist-info/METADATA,sha256=oDIp-lxMIQjIfVtrzQXBcY2PJFHlRwktVGFXndQRJZo,8872
+yomitoku-0.9.4.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+yomitoku-0.9.4.dist-info/entry_points.txt,sha256=n3c8bQSj5Be5GHAOv_NZ8cldJFmWeigQxSmteFTmu_k,96
+yomitoku-0.9.4.dist-info/RECORD,,