OCR-D · tdoan2010 · May 17, 2022 · May 17, 2022 · May 17, 2022 · May 17, 2022
diff --git a/.gitignore b/.gitignore
@@ -125,3 +125,5 @@ sanders*
 ws1
 *.doctree
 .vscode
+
+.idea/
diff --git a/ocrd/ocrd/cli/__init__.py b/ocrd/ocrd/cli/__init__.py
@@ -31,6 +31,7 @@ def get_help(self, ctx):
 from ocrd.decorators import ocrd_loglevel
 from .zip import zip_cli
 from .log import log_cli
+from .server import server_cli
 
 @click.group()
 @click.version_option()
@@ -48,3 +49,4 @@ def cli(**kwargs): # pylint: disable=unused-argument
 cli.add_command(validate_cli)
 cli.add_command(log_cli)
 cli.add_command(resmgr_cli)
+cli.add_command(server_cli)
diff --git a/ocrd/ocrd/cli/server.py b/ocrd/ocrd/cli/server.py
@@ -0,0 +1,48 @@
+"""
+OCR-D CLI: start the processing server
+
+.. click:: ocrd.cli.server:server_cli
+    :prog: ocrd server
+    :nested: full
+
+"""
+from subprocess import run, PIPE
+
+import click
+import uvicorn
+
+from ocrd.helpers import parse_server_input, parse_version_string
+from ocrd.server.main import ProcessorAPI
+from ocrd_utils import parse_json_string_with_comments, initLogging
+
+
+@click.command('processing-server')
+@click.argument('processor_name', required=True, type=click.STRING)
+@click.option('--address',
+              help='Host name/IP, port, and connection string to a Mongo DB in the format IP:PORT:MONGO_URL',
+              required=True,
+              type=click.STRING)
+def server_cli(processor_name, address):
+    try:
+        ip, port, mongo_url = parse_server_input(address)
+    except ValueError:
+        raise click.UsageError('The --server option must have the format IP:PORT:MONGO_URL')
+
+    ocrd_tool = parse_json_string_with_comments(
+        run([processor_name, '--dump-json'], stdout=PIPE, check=True, universal_newlines=True).stdout
+    )
+    version = parse_version_string(
+        run([processor_name, '--version'], stdout=PIPE, check=True, universal_newlines=True).stdout
+    )
+
+    initLogging()
+
+    # Start the server
+    app = ProcessorAPI(
+        title=ocrd_tool['executable'],
+        description=ocrd_tool['description'],
+        version=version,
+        ocrd_tool=ocrd_tool,
+        db_url=mongo_url
+    )
+    uvicorn.run(app, host=ip, port=port, access_log=False)
diff --git a/ocrd/ocrd/decorators/__init__.py b/ocrd/ocrd/decorators/__init__.py
@@ -1,40 +1,42 @@
-from os.path import isfile
 import sys
+from contextlib import redirect_stdout
+from io import StringIO
+from typing import Type
 
 import click
+import uvicorn
 
+from ocrd.server.main import ProcessorAPI
+from ocrd_utils import getLogger, initLogging
 from ocrd_utils import (
-    is_local_filename,
-    get_local_filename,
-    set_json_key_value_overrides,
+    set_json_key_value_overrides, parse_json_string_with_comments,
 )
-
-from ocrd_utils import getLogger, initLogging
 from ocrd_validators import WorkspaceValidator
+from ocrd.decorators.loglevel_option import ocrd_loglevel
+from ocrd.decorators.mets_find_options import mets_find_options
+from ocrd.decorators.ocrd_cli_options import ocrd_cli_options
+from ocrd.decorators.parameter_option import parameter_option, parameter_override_option
+from ocrd.helpers import parse_server_input, parse_version_string
+from ocrd.processor.base import run_processor, Processor
+from ocrd.resolver import Resolver
 
-from ..resolver import Resolver
-from ..processor.base import run_processor
-
-from .loglevel_option import ocrd_loglevel
-from .parameter_option import parameter_option, parameter_override_option
-from .ocrd_cli_options import ocrd_cli_options
-from .mets_find_options import mets_find_options
 
 def ocrd_cli_wrap_processor(
-    processorClass,
-    ocrd_tool=None,
-    mets=None,
-    working_dir=None,
-    dump_json=False,
-    dump_module_dir=False,
-    help=False, # pylint: disable=redefined-builtin
-    profile=False,
-    profile_file=None,
-    version=False,
-    overwrite=False,
-    show_resource=None,
-    list_resources=False,
-    **kwargs
+        processorClass: Type[Processor],
+        ocrd_tool=None,
+        mets=None,
+        working_dir=None,
+        address=None,
+        dump_json=False,
+        dump_module_dir=False,
+        help=False,  # pylint: disable=redefined-builtin
+        profile=False,
+        profile_file=None,
+        version=False,
+        overwrite=False,
+        show_resource=None,
+        list_resources=False,
+        **kwargs
 ):
     if not sys.argv[1:]:
         processorClass(workspace=None, show_help=True)
@@ -50,6 +52,37 @@ def ocrd_cli_wrap_processor(
             list_resources=list_resources
         )
         sys.exit()
+    if address:
+        try:
+            ip, port, mongo_url = parse_server_input(address)
+        except ValueError:
+            raise click.UsageError('The --server option must have the format IP:PORT:MONGO_URL')
+
+        initLogging()
+
+        # Read the ocrd_tool object
+        f1 = StringIO()
+        with redirect_stdout(f1):
+            processorClass(workspace=None, dump_json=True)
+        ocrd_tool = parse_json_string_with_comments(f1.getvalue())
+
+        # Read the version string
+        f2 = StringIO()
+        with redirect_stdout(f2):
+            processorClass(workspace=None, show_version=True)
+        version = parse_version_string(f2.getvalue())
+
+        # Start the server
+        app = ProcessorAPI(
+            title=ocrd_tool['executable'],
+            description=ocrd_tool['description'],
+            version=version,
+            ocrd_tool=ocrd_tool,
+            db_url=mongo_url,
+            processor_class=processorClass
+        )
+
+        uvicorn.run(app, host=ip, port=port, access_log=False)
     else:
         initLogging()
         LOG = getLogger('ocrd_cli_wrap_processor')
@@ -86,7 +119,8 @@ def ocrd_cli_wrap_processor(
         # XXX While https://github.com/OCR-D/core/issues/505 is open, set 'overwrite_mode' globally on the workspace
         if overwrite:
             workspace.overwrite_mode = True
-        report = WorkspaceValidator.check_file_grp(workspace, kwargs['input_file_grp'], '' if overwrite else kwargs['output_file_grp'], page_id)
+        report = WorkspaceValidator.check_file_grp(workspace, kwargs['input_file_grp'],
+                                                   '' if overwrite else kwargs['output_file_grp'], page_id)
         if not report.is_valid:
             raise Exception("Invalid input/output file grps:\n\t%s" % '\n\t'.join(report.errors))
         if profile or profile_file:
@@ -97,6 +131,7 @@ def ocrd_cli_wrap_processor(
             print("Profiling...")
             pr = cProfile.Profile()
             pr.enable()
+
             def exit():
                 pr.disable()
                 print("Profiling completed")
@@ -106,5 +141,6 @@ def exit():
                 s = io.StringIO()
                 pstats.Stats(pr, stream=s).sort_stats("cumulative").print_stats()
                 print(s.getvalue())
+
             atexit.register(exit)
         run_processor(processorClass, ocrd_tool, mets, workspace=workspace, **kwargs)
diff --git a/ocrd/ocrd/decorators/ocrd_cli_options.py b/ocrd/ocrd/decorators/ocrd_cli_options.py
@@ -1,3 +1,4 @@
+import click
 from click import option
 from .parameter_option import parameter_option, parameter_override_option
 from .loglevel_option import loglevel_option
@@ -26,6 +27,7 @@ def cli(mets_url):
         option('-O', '--output-file-grp', help='File group(s) used as output.', default='OUTPUT'),
         option('-g', '--page-id', help="ID(s) of the pages to process"),
         option('--overwrite', help="Overwrite the output file group or a page range (--page-id)", is_flag=True, default=False),
+        option('--address', help='Host name/IP, port, and connection string to a Mongo DB.', type=click.STRING),
         option('-C', '--show-resource', help='Dump the content of processor resource RESNAME', metavar='RESNAME'),
         option('-L', '--list-resources', is_flag=True, default=False, help='List names of processor resources'),
         parameter_option,

diff --git a/ocrd/ocrd/helpers.py b/ocrd/ocrd/helpers.py
@@ -0,0 +1,37 @@
+from typing import Tuple
+
+
+def parse_server_input(input_str: str) -> Tuple[str, int, str]:
+    """
+    Parse the string into 3 parts, IP address, port, and Mongo database connection string.
+
+    Args:
+        input_str (str): a string with the format ``ip:port:db``, where ``ip`` and ``port`` is where the server listens
+            on, and ``db`` is a connection string to a Mongo database.
+
+    Returns:
+        str, int, str: the IP, port, and Mongo DB connection string respectively.
+    """
+    elements = input_str.split(':', 2)
+    if len(elements) != 3:
+        raise ValueError
+    ip = elements[0]
+    port = int(elements[1])
+    mongo_url = elements[2]
+
+    return ip, port, mongo_url
+
+
+def parse_version_string(version_str: str) -> str:
+    """
+    Get the version number from the output of the :py:function:`~ocrd.processor.base.Processor.show_version`.
+
+    Args:
+        version_str (str): A string which looks like this ``Version %s, ocrd/core %s``
+
+    Returns:
+        str: the string between the word ``Version`` and the first comma
+    """
+    first_split = version_str.split(',')
+    second_split = first_split[0].split(' ')
+    return second_split[1]
diff --git a/ocrd/ocrd/processor/__init__.py b/ocrd/ocrd/processor/__init__.py
@@ -3,6 +3,7 @@
 )
 from .helpers import (
     run_cli,
+    run_cli_from_api,
     run_processor,
     generate_processor_help
 )
diff --git a/ocrd/ocrd/processor/builtin/ocrd-tool.json b/ocrd/ocrd/processor/builtin/ocrd-tool.json
@@ -3,6 +3,6 @@
   "description": "Bare-bones processor that copies file from input group to output group",
   "steps": ["preprocessing/optimization"],
   "categories": ["Image preprocessing"],
-  "input_file_grp": "DUMMY_INPUT",
-  "output_file_grp": "DUMMY_OUTPUT"
+  "input_file_grp": ["DUMMY_INPUT"],
+  "output_file_grp": ["DUMMY_OUTPUT"]
 }
-Original file line number
+Diff line change
@@ Expand Up / @@ -125,3 +125,5 @@ sanders* @@
     ws1
     *.doctree
     .vscode
+    .idea/