From e9c6d07386f154b5a2761c09c78d281bfb18f8b7 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <unixprog@gmail.com>
Date: Mon, 14 Feb 2022 10:14:22 +0100
Subject: [PATCH] Processor.resolve_resource: support on-demand download of URL
 values

---
 ocrd/ocrd/processor/base.py   | 24 ++++++++++++++++++++++--
 ocrd/ocrd/resource_manager.py |  5 ++---
 2 files changed, 24 insertions(+), 5 deletions(-)

diff --git a/ocrd/ocrd/processor/base.py b/ocrd/ocrd/processor/base.py
index 5e7ab6e9bb..06f228b5db 100644
--- a/ocrd/ocrd/processor/base.py
+++ b/ocrd/ocrd/processor/base.py
@@ -25,12 +25,14 @@
     getLogger,
     initLogging,
     list_resource_candidates,
+    nth_url_segment,
     pushd_popd,
     list_all_resources,
     get_processor_resource_types
 )
 from ocrd_validators import ParameterValidator
 from ocrd_models.ocrd_page import MetadataItemType, LabelType, LabelsType
+from ocrd.resource_manager import OcrdResourceManager
 
 # XXX imports must remain for backwards-compatibilty
 from .helpers import run_cli, run_processor, generate_processor_help # pylint: disable=unused-import
@@ -217,9 +219,27 @@ def resolve_resource(self, val):
         if ret:
             log.debug("Resolved %s to absolute path %s" % (val, ret[0]))
             return ret[0]
-        log.error("Could not find resource '%s' for executable '%s'. Try 'ocrd resmgr download %s %s' to download this resource.",
+        elif (val.startswith('http://') or val.startswith('https://')):
+            resmgr = OcrdResourceManager()
+            reslist = resmgr.find_resources(executable, url=val)
+            if reslist:
+                _, resdict = reslist[0]
+                log.info("Found registered resource for %s: '%s' (%s)." % (executable, val, resdict))
+            else:
+                resdict = {}
+                log.info("Not a registered resource for %s: '%s'." % (executable, val))
+            return str(resmgr.download(
+                executable,
+                val,
+                basedir = resmgr.location_to_resource_dir('data'),
+                name=resdict.get('name', nth_url_segment(val)),
+                path_in_archive=resdict.get('path_in_archive', '.'),
+                resource_type=resdict.get('type', 'file')
+                ))
+        else:
+            log.error("Could not find resource '%s' for executable '%s'. Try 'ocrd resmgr download %s %s' to download this resource or use a URL for the parameter value.",
                 val, executable, executable, val)
-        sys.exit(1)
+            sys.exit(1)
 
     def list_all_resources(self):
         """
diff --git a/ocrd/ocrd/resource_manager.py b/ocrd/ocrd/resource_manager.py
index 06832e0fd2..6d87ac5202 100644
--- a/ocrd/ocrd/resource_manager.py
+++ b/ocrd/ocrd/resource_manager.py
@@ -10,7 +10,7 @@
 from yaml import safe_load, safe_dump
 
 from ocrd_validators import OcrdResourceListValidator
-from ocrd_utils import getLogger
+from ocrd_utils import getLogger, nth_url_segment
 from ocrd_utils.os import get_processor_resource_types, list_all_resources, pushd_popd
 from .constants import RESOURCE_LIST_FILENAME, RESOURCE_USER_LIST_COMMENT
 
@@ -235,8 +235,7 @@ def download(
         log = getLogger('ocrd.resource_manager.download')
         destdir = Path(basedir) if no_subdir else Path(basedir, executable)
         if not name:
-            url_parsed = urlparse(url)
-            name = Path(unquote(url_parsed.path)).name
+            name = nth_url_segment(url)
         fpath = Path(destdir, name)
         is_url = url.startswith('https://') or url.startswith('http://')
         if fpath.exists() and not overwrite: