From a9e12a63da3e911ad1c4545b901ebbb661b0e5f6 Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Tue, 28 Apr 2026 12:18:29 +0200
Subject: [PATCH] wp

---
 pyproject.toml                     |  1 +
 src/eynollah/ocrd-tool.json        | 37 ++++++++++++
 src/eynollah/ocrd_cli_recognize.py | 91 ++++++++++++++++++++++++++++++
 3 files changed, 129 insertions(+)
 create mode 100644 src/eynollah/ocrd_cli_recognize.py

diff --git a/pyproject.toml b/pyproject.toml
index ca773a7..e15911f 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -42,6 +42,7 @@ classifiers = [
 eynollah = "eynollah.cli:main"
 eynollah-training = "eynollah.training.cli:main"
 ocrd-eynollah-segment = "eynollah.ocrd_cli_segment:main"
+ocrd-eynollah-recognize = "eynollah.ocrd_cli_recognize:main"
 ocrd-sbb-binarize = "eynollah.ocrd_cli_binarization:main"
 
 [project.urls]
diff --git a/src/eynollah/ocrd-tool.json b/src/eynollah/ocrd-tool.json
index fc61af7..c946541 100644
--- a/src/eynollah/ocrd-tool.json
+++ b/src/eynollah/ocrd-tool.json
@@ -163,5 +163,42 @@
         }
       ]
     }
+  },
+  "ocrd-eynollah-recognize": {
+    "executable": "ocrd-eynollah-recognize",
+    "categories": ["Text recognition and optimization"],
+    "steps": ["recognition/text-recognition"],
+    "input_file_grp_cardinality": 1,
+    "output_file_grp_cardinality": 1,
+    "parameters": {
+      "models": {
+        "type": "string",
+        "format": "uri",
+        "content-type": "text/directory",
+        "cacheable": true,
+        "description": "Directory containing the eynollah_models directory",
+        "required": true
+      },
+      "do_not_mask_with_textline_contour": {
+        "type": "boolean",
+        "description": "if this parameter set to true, cropped textline images will not be masked with textline contour.",
+        "default": false
+      },
+      "tr_ocr": {
+        "type": "boolean",
+        "description": "Whether to use (much more resource-intensive) transformer model",
+        "default": false
+      }
+    },
+    "resources": [
+      {
+        "url": "https://zenodo.org/records/17580627/files/models_ocr_v0_6_0.tar.gz?download=1",
+        "name": "models_ocr_v0_6_0",
+        "type": "archive",
+        "size": 6119874002,
+        "description": "Models for OCR",
+        "version_range": ">= v0.6.0"
+      }
+    ]
   }
 }
diff --git a/src/eynollah/ocrd_cli_recognize.py b/src/eynollah/ocrd_cli_recognize.py
new file mode 100644
index 0000000..63d9b8c
--- /dev/null
+++ b/src/eynollah/ocrd_cli_recognize.py
@@ -0,0 +1,91 @@
+from functools import cached_property
+from pathlib import Path
+from typing import Optional
+from ocrd.workspace import page_from_file
+from ocrd_models import OcrdFileType, OcrdPage
+
+from ocrd import Processor
+from ocrd_utils import (
+    make_file_id,
+)
+
+from eynollah.eynollah_ocr import Eynollah_ocr
+from eynollah.model_zoo.model_zoo import EynollahModelZoo
+from eynollah.utils.pil_cv2 import pil2cv
+from eynollah.utils.xml import etree_namespace_for_element_tag
+
+
+class EynollahRecognizeProcessor(Processor):
+
+    @cached_property
+    def executable(self):
+        return 'ocrd-eynollah-recognize'
+
+    def setup(self):
+        """
+        Load model, set predict function
+        """
+        assert self.parameter
+        model_zoo = EynollahModelZoo(basedir=self.parameter['models'])
+        assert self.parameter
+        self.eynollah_ocr = Eynollah_ocr(
+            model_zoo=model_zoo,
+            tr_ocr=self.parameter['tr_ocr'],
+            do_not_mask_with_textline_contour=self.parameter['do_not_mask_with_textline_contour'],
+            batch_size=self.parameter['batch_size'],
+            min_conf_value_of_textline_text=self.parameter['min_conf_value_of_textline_text'])
+
+    # FIXME: This is just a proof-of-concept, very inefficient and non-conformant
+    # TODO: OCR writing should use PAGE API once result dataclass mechanism is settled,
+    #       then simplify/port to proces_page_pcgts
+    def process_page_file(self, *input_files: Optional[OcrdFileType]) -> None:
+        assert self.workspace
+        page_file = input_files[0]
+        assert page_file
+        page = page_from_file(page_file)
+        assert page
+        page_image, page_coords, _ = self.workspace.image_from_page(
+            page, page_file.pageId,
+            feature_selector="")
+        page_ns = etree_namespace_for_element_tag(page.etree.getroot().tag)
+
+        img = pil2cv(page_image)
+        if self.eynollah_ocr.tr_ocr:
+            result = self.eynollah_ocr.run_trocr(
+                img=img,
+                page_tree=page.etree,
+                page_ns=page_ns,
+
+                tr_ocr_input_height_and_width = 384
+            )
+        else:
+            page_image_bin, _, _ = self.workspace.image_from_page(
+                page, page_file.pageId,
+                feature_selector="binarized")
+            result = self.eynollah_ocr.run_cnn( 
+                img=img,
+                page_tree=page.etree,
+                page_ns=page_ns,
+
+                img_bin=pil2cv(page_image_bin),
+                image_width=512,
+                image_height=32,
+            )
+        output_file_id = make_file_id(page_file, self.output_file_grp)
+        output_filename = Path(self.output_file_grp, output_file_id + '.xml')
+        output_filename.parent.mkdir()
+        self.eynollah_ocr.write_ocr(
+            result=result,
+            img=img,
+            page_tree=page.etree,
+            page_ns=page_ns,
+            out_file_ocr=str(output_filename),
+            out_image_with_text=None,
+        )
+        self.workspace.add_file(
+            file_id=output_file_id,
+            file_grp=self.output_file_grp,
+            page_id=page_file.pageId,
+            local_filename=output_filename,
+            mimetype=page_ns,
+        )