Fix: 1257 - Load ocrd tool json locally

requirements.txt

@@ -1,6 +1,7 @@
 atomicwrites >= 1.3.0
 beanie~=1.7
 click >=7
+cryptography < 43.0.0
 Deprecated == 1.2.0
 docker
 fastapi>=0.78.0
@@ -32,4 +33,3 @@ requests_unixsocket2
 shapely
 uvicorn
 uvicorn>=0.17.6
-

requirements_test.txt

@@ -1,4 +1,5 @@
 autopep8
+cryptography < 43.0.0
 pytest >= 4.0.0
 generateDS == 2.35.20
 pytest-benchmark >= 3.2.3

src/ocrd/ocrd-all-tool.json

@@ -0,0 +1,21 @@
+{
+  "ocrd-dummy": {
+    "executable": "ocrd-dummy",
+    "description": "Bare-bones processor creates PAGE-XML and optionally copies file from input group to output group",
+    "steps": [
+      "preprocessing/optimization"
+    ],
+    "categories": [
+      "Image preprocessing"
+    ],
+    "input_file_grp": "DUMMY_INPUT",
+    "output_file_grp": "DUMMY_OUTPUT",
+    "parameters": {
+      "copy_files": {
+        "type": "boolean",
+        "default": false,
+        "description": "Whether to actually copy files (true) or just create PAGE-XML as a side effect (false)"
+      }
+    }
+  }
+}

src/ocrd_network/constants.py

@@ -6,7 +6,8 @@
 DOCKER_RABBIT_MQ_FEATURES = "quorum_queue,implicit_default_bindings,classic_mirrored_queue_version"
 
 NETWORK_PROTOCOLS = ["http://", "https://"]
-OCRD_ALL_JSON_TOOLS_URL = "https://ocr-d.de/js/ocrd-all-tool.json"
+OCRD_ALL_TOOL_JSON = "ocrd-all-tool.json"
+OCRD_ALL_TOOL_JSON_URL = "https://ocr-d.de/js/ocrd-all-tool.json"
 # Used as a placeholder to lock all pages when no page_id is specified
 SERVER_ALL_PAGES_PLACEHOLDER = "all_pages"
 

src/ocrd_network/processing_server.py

@@ -10,7 +10,7 @@
 
 from ocrd.task_sequence import ProcessorTask
 from ocrd_utils import initLogging, getLogger
-from .constants import AgentType, JobState, OCRD_ALL_JSON_TOOLS_URL, ServerApiTags
+from .constants import AgentType, JobState, ServerApiTags
 from .database import (
     initiate_database,
     db_get_processing_job,
@@ -58,7 +58,7 @@
 )
 from .tcp_to_uds_mets_proxy import MetsServerProxy
 from .utils import (
-    download_ocrd_all_tool_json,
+    load_ocrd_all_tool_json,
     expand_page_ids,
     generate_id,
     generate_workflow_content,
@@ -90,8 +90,8 @@ def __init__(self, config_path: str, host: str, port: int) -> None:
         log_file = get_processing_server_logging_file_path(pid=getpid())
         configure_file_handler_with_formatter(self.log, log_file=log_file, mode="a")
 
-        self.log.info(f"Downloading ocrd all tool json")
-        self.ocrd_all_tool_json = download_ocrd_all_tool_json(ocrd_all_url=OCRD_ALL_JSON_TOOLS_URL)
+        self.log.info(f"Loading ocrd all tool json")
+        self.ocrd_all_tool_json = load_ocrd_all_tool_json()
         self.hostname = host
         self.port = port
 

src/ocrd_network/utils.py

@@ -3,6 +3,7 @@
 from fastapi import UploadFile
 from functools import wraps
 from hashlib import md5
+from json import loads
 from pathlib import Path
 from re import compile as re_compile, split as re_split
 from requests import get as requests_get, Session as Session_TCP
@@ -14,7 +15,8 @@
 from ocrd.resolver import Resolver
 from ocrd.workspace import Workspace
 from ocrd.mets_server import MpxReq
-from ocrd_utils import config, generate_range, REGEX_PREFIX, safe_filename, getLogger
+from ocrd_utils import config, generate_range, REGEX_PREFIX, safe_filename, getLogger, resource_string
+from .constants import OCRD_ALL_TOOL_JSON, OCRD_ALL_TOOL_JSON_URL
 from .rabbitmq_utils import OcrdResultMessage
 
 
@@ -92,14 +94,12 @@ def is_url_responsive(url: str, tries: int = 1, wait_time: int = 3) -> bool:
     return False
 
 
-def download_ocrd_all_tool_json(ocrd_all_url: str):
-    if not ocrd_all_url:
-        raise ValueError(f"The URL of ocrd all tool json is empty")
-    headers = {"Accept": "application/json"}
-    response = Session_TCP().get(ocrd_all_url, headers=headers)
-    if not response.status_code == 200:
-        raise ValueError(f"Failed to download ocrd all tool json from: '{ocrd_all_url}'")
-    return response.json()
+def load_ocrd_all_tool_json():
+    try:
+        ocrd_all_tool_json = loads(resource_string('ocrd', OCRD_ALL_TOOL_JSON))
+    except Exception as error:
+        raise ValueError(f"Failed to load ocrd all tool json from: '{OCRD_ALL_TOOL_JSON}', {error}")
+    return ocrd_all_tool_json
 
 
 def post_to_callback_url(logger, callback_url: str, result_message: OcrdResultMessage):

src/ocrd_utils/ocrd_logging.conf

@@ -11,7 +11,7 @@
 # each logger requires a corresponding configuration section below
 #
 [loggers]
-keys=root,ocrd,ocrd_network,ocrd_tensorflow,ocrd_shapely_geos,ocrd_PIL,uvicorn,uvicorn_access,uvicorn_error,multipart
+keys=root,ocrd,ocrd_network,ocrd_tensorflow,ocrd_shapely_geos,ocrd_PIL,uvicorn,uvicorn_access,uvicorn_error,multipart,paramiko,paramiko_transport
 
 #
 # mandatory handlers section
@@ -91,6 +91,21 @@ level=INFO
 handlers=consoleHandler
 qualname=PIL
 
+#
+# paramiko loggers
+#
+[logger_paramiko]
+level=INFO
+handlers=consoleHandler
+qualname=paramiko
+propagate=0
+
+[logger_paramiko_transport]
+level=INFO
+handlers=consoleHandler
+qualname=paramiko.transport
+propagate=0
+
 #
 # uvicorn loggers
 #