dcicutils 8.14.0.1b15__py3-none-any.whl → 8.14.0.1b17__py3-none-any.whl

dcicutils/scripts/update_portal_object.py

@@ -17,10 +17,11 @@ import re
 import shutil
 import sys
 from typing import Callable, List, Optional, Tuple, Union
+from dcicutils.captured_output import captured_output
 from dcicutils.command_utils import yes_or_no
-from dcicutils.common import ORCHESTRATED_APPS, APP_SMAHT
+from dcicutils.common import ORCHESTRATED_APPS, APP_CGAP, APP_FOURFRONT, APP_SMAHT
 from dcicutils.ff_utils import delete_metadata, purge_metadata
-from dcicutils.misc_utils import get_error_message, ignored, PRINT, to_camel_case, to_snake_case
+from dcicutils.misc_utils import get_error_message, ignored, normalize_string, PRINT, to_camel_case, to_snake_case
 from dcicutils.portal_utils import Portal as PortalFromUtils
 from dcicutils.tmpfile_utils import temporary_directory
 
@@ -145,30 +146,16 @@ def main():
         parser.print_help()
         sys.exit(1)
 
-    if app := args.app:
-        if (app not in ORCHESTRATED_APPS) and ((app := app.lower()) not in ORCHESTRATED_APPS):
-            usage(f"ERROR: Unknown app name; must be one of: {' | '.join(ORCHESTRATED_APPS)}")
-    else:
-        app = APP_SMAHT
-
     if not (args.post or args.patch or args.upsert or args.delete or args.purge or args.load):
         usage()
 
+    if not (portal := _create_portal(env=args.env, ini=args.ini, app=args.app, load=args.load,
+                                     verbose=args.verbose, debug=args.debug, quiet=args.quiet)):
+        exit(1)
+
     if args.load:
-        if args.post or args.patch or args.upsert or args.delete or args.purge:
-            _print("Cannot use any other update option"
-                   "when using the --load option (to load data via snovault.loadxl).")
-            exit(1)
-        if args.env:
-            if args.ini:
-                _print("The --env is not used for the --load option (to load data via snovault.loadxl).")
-            args.ini = args.env
-        if not _load_data(load=args.load, ini_file=args.ini, explicit_schema_name=args.schema,
-                          verbose=args.verbose, debug=args.debug, noprogress=args.noprogress):
-            exit(1)
-        exit(0)
-
-    portal = _create_portal(env=args.env, app=app, verbose=args.verbose, debug=args.debug)
+        _load_data(portal=portal, load=args.load, ini_file=args.ini, explicit_schema_name=args.schema,
+                   verbose=args.verbose, debug=args.debug, noprogress=args.noprogress)
 
     if explicit_schema_name := args.schema:
         schema, explicit_schema_name = _get_schema(portal, explicit_schema_name)
@@ -402,17 +389,24 @@ def _upsert_data(portal: Portal, data: dict, schema_name: str,
         return
 
 
-def _load_data(load: str, ini_file: str, explicit_schema_name: Optional[str] = None,
+def _load_data(portal: Portal, load: str, ini_file: str, explicit_schema_name: Optional[str] = None,
                verbose: bool = False, debug: bool = False, noprogress: bool = False,
-               _portal: Optional[Portal] = None, _single_insert_file: Optional[str] = None) -> bool:
+               _single_insert_file: Optional[str] = None) -> bool:
 
+    import snovault.loadxl
     from snovault.loadxl import load_all_gen, LoadGenWrapper
-    from dcicutils.captured_output import captured_output
     from dcicutils.progress_bar import ProgressBar
 
+    loadxl_summary = {}
+    loadxl_unresolved = {}
+    loadxl_output = []
+    loadxl_total_item_count = 0
+    loadxl_total_error_count = 0
+
     def loadxl(portal: Portal, inserts_directory: str, schema_names_to_load: dict):
 
-        nonlocal LoadGenWrapper, load_all_gen, verbose, debug
+        nonlocal LoadGenWrapper, load_all_gen, loadxl_summary, verbose, debug
+        nonlocal loadxl_total_item_count, loadxl_total_error_count
         progress_total = sum(schema_names_to_load.values()) * 2  # loadxl does two passes
         progress_bar = ProgressBar(progress_total, interrupt_exit=True) if not noprogress else None
 
@@ -425,49 +419,75 @@ def _load_data(load: str, ini_file: str, explicit_schema_name: Optional[str] = N
                 return str_or_bytes.strip()
             return ""
 
+        def loadxl_print(arg):
+            if arg:
+                loadxl_output.append(normalize_string(str(arg)))
+
+        snovault.loadxl.print = loadxl_print
+
         LOADXL_RESPONSE_PATTERN = re.compile(r"^([A-Z]+):\s*([a-zA-Z\/\d_-]+)\s*(\S+)\s*(\S+)?\s*(.*)$")
         LOADXL_ACTION_NAME = {"POST": "Create", "PATCH": "Update", "SKIP": "Check",
                               "CHECK": "Validate", "ERROR": "Error"}
         current_item_type = None
         current_item_count = 0
         current_item_total = 0
-        total_item_count = 0
+
         for item in LoadGenWrapper(load_all_gen(testapp=portal.vapp, inserts=inserts_directory,
                                                 docsdir=None, overwrite=True, verbose=True)):
-            total_item_count += 1
+            loadxl_total_item_count += 1
             item = decode_bytes(item)
             match = LOADXL_RESPONSE_PATTERN.match(item)
             if not match or match.re.groups < 3:
                 continue
-            action = LOADXL_ACTION_NAME[match.group(1).upper()]
-            # response_value = match.group(0)
-            # identifying_value = match.group(2)
+            if (action := LOADXL_ACTION_NAME[match.group(1).upper()]) == "Error":
+                identifying_value = match.group(2)
+                # Example message for unresolved link:
+                # ERROR: /22813a02-906b-4b60-b2b2-4afaea24aa28 Bad response: 422 Unprocessable Entity
+                # (not 200 OK or 3xx redirect for http://localhost/file_set?skip_indexing=true)b\'{"@type":
+                # ["ValidationFailure", "Error"], "status": "error", "code": # 422, "title": "Unprocessable Entity",
+                # "description": "Failed validation", "errors": [{"location": "body", "name": # "Schema: ",
+                # "description": "Unable to resolve link: /Library/a4e8f79f-4d47-4e85-9707-c343c940a315"},
+                # {"location": "body", "name": "Schema: libraries.0",
+                # "description": "\\\'a4e8f79f-4d47-4e85-9707-c343c940a315\\\' not found"}]}\'
+                unresolved_link_error_message_prefix = "Unable to resolve link:"
+                if (i := item.find(unresolved_link_error_message_prefix)) > 0:
+                    unresolved_link = item[i + len(unresolved_link_error_message_prefix):].strip()
+                    if (i := unresolved_link.find("\"")) > 0:
+                        if (unresolved_link := unresolved_link[0:i]):
+                            if ((error_type := re.search(r"https?://.*/(.*)\?skip_indexing=.*", item)) and
+                                (len(error_type.groups()) == 1)):  # noqa
+                                error_type = to_camel_case(error_type.group(1))
+                                identifying_value = f"/{error_type}{identifying_value}"
+                                if not loadxl_unresolved.get(identifying_value):
+                                    loadxl_unresolved[identifying_value] = []
+                                loadxl_unresolved[identifying_value].append(unresolved_link)
+                loadxl_total_error_count += 1
+                continue
             item_type = match.group(3)
             if current_item_type != item_type:
                 if noprogress and debug and current_item_type is not None:
-                    print()
+                    _print()
                 current_item_type = item_type
                 current_item_count = 0
                 current_item_total = schema_names_to_load[item_type]
                 if progress_bar:
                     progress_bar.set_description(f"▶ {to_camel_case(current_item_type)}: {action}")
             current_item_count += 1
+            if loadxl_summary.get(current_item_type, None) is None:
+                loadxl_summary[current_item_type] = 0
+            loadxl_summary[current_item_type] += 1
             if progress_bar:
-                progress_bar.set_progress(total_item_count)
+                progress_bar.set_progress(loadxl_total_item_count)
             elif debug:
-                print(f"{current_item_type}: {current_item_count} or {current_item_total} ({action})")
+                _print(f"{current_item_type}: {current_item_count} or {current_item_total} ({action})")
         if progress_bar:
             progress_bar.set_description("▶ Load Complete")
-            print()
-
-    if not ini_file:
-        ini_file = _DEFAULT_INI_FILE_FOR_LOAD
-    if not os.path.isabs(ini_file := os.path.normpath(os.path.expanduser(ini_file))):
-        ini_file = os.path.normpath(os.path.join(os.getcwd(), ini_file))
-    if not os.path.exists(ini_file):
-        _print(f"The INI file required for --load is not found: {ini_file}")
-        exit(1)
+            if loadxl_total_item_count > loadxl_total_error_count:
+                _print()
 
+    if not portal.vapp:
+        _print("Must using INI based Portal object with --load (use --ini option to specify an INI file).")
+        return False
     if not os.path.isabs(load := os.path.normpath(os.path.expanduser(load))):
         load = os.path.normpath(os.path.join(os.getcwd(), load))
     if not os.path.exists(load):
@@ -480,10 +500,6 @@ def _load_data(load: str, ini_file: str, explicit_schema_name: Optional[str] = N
         inserts_directory = None
         inserts_file = load
 
-    if not (portal := _portal):
-        with captured_output(not debug):
-            portal = Portal(ini_file)
-
     if inserts_file:
         with io.open(inserts_file, "r") as f:
             try:
@@ -503,9 +519,9 @@ def _load_data(load: str, ini_file: str, explicit_schema_name: Optional[str] = N
                     file_name = os.path.join(tmpdir, f"{to_snake_case(schema_name)}.json")
                     with io.open(file_name, "w") as f:
                         json.dump(data, f)
-                    return _load_data(load=tmpdir, ini_file=ini_file, explicit_schema_name=schema_name,
+                    return _load_data(portal=portal, load=tmpdir, ini_file=ini_file, explicit_schema_name=schema_name,
                                       verbose=verbose, debug=debug, noprogress=noprogress,
-                                      _portal=portal, _single_insert_file=inserts_file)
+                                      _single_insert_file=inserts_file)
             elif isinstance(data, dict):
                 if schema_name := explicit_schema_name:
                     if _is_schema_name_list(portal, schema_names := list(data.keys())):
@@ -530,20 +546,20 @@ def _load_data(load: str, ini_file: str, explicit_schema_name: Optional[str] = N
                             json.dump(schema_data, f)
                         nfiles += 1
                     if nfiles > 0:
-                        return _load_data(load=tmpdir, ini_file=ini_file,
+                        return _load_data(portal=portal, load=tmpdir, ini_file=ini_file,
                                           verbose=verbose, debug=debug, noprogress=noprogress,
-                                          _portal=portal, _single_insert_file=inserts_file)
+                                          _single_insert_file=inserts_file)
                 return True
             else:
                 _print(f"Unrecognized JSON data in file: {inserts_file}")
                 return False
         return True
+
     if verbose:
         if _single_insert_file:
             _print(f"Loading data into Portal (via snovault.loadxl) from file: {_single_insert_file}")
         else:
             _print(f"Loading data into Portal (via snovault.loadxl) from directory: {inserts_directory}")
-        _print(f"Portal INI file for load is: {ini_file}")
 
     schema_names = list(_get_schemas(portal).keys())
     schema_snake_case_names = [to_snake_case(item) for item in schema_names]
@@ -585,11 +601,27 @@ def _load_data(load: str, ini_file: str, explicit_schema_name: Optional[str] = N
             loadxl(portal=portal, inserts_directory=tmpdir, schema_names_to_load=schema_names_to_load)
     else:
         loadxl(portal=portal, inserts_directory=inserts_directory, schema_names_to_load=schema_names_to_load)
+
     if verbose:
         if _single_insert_file:
             _print(f"Done loading data into Portal (via snovault.loadxl) from file: {_single_insert_file}")
         else:
             _print(f"Done loading data into Portal (via snovault.loadxl) from directory: {inserts_directory}")
+        _print(f"Total items loaded: {loadxl_total_item_count}"
+               f"{f' (errors: {loadxl_total_error_count})' if loadxl_total_error_count else ''}")
+        for item in sorted(loadxl_summary.keys()):
+            _print(f"▷ {to_camel_case(item)}: {loadxl_summary[item]}")
+    if loadxl_unresolved:
+        _print("✗ Unresolved references:")
+        for item in loadxl_unresolved:
+            _print(f"  ▶ {item}: {len(loadxl_unresolved[item])}")
+            for subitem in loadxl_unresolved[item]:
+                _print(f"    ▷ {subitem}")
+    if debug and loadxl_output:
+        _print("✗ Output from loadxl:")
+        for item in loadxl_output:
+            _print(f"  ▶ {item}")
+
     return True
 
 
@@ -611,25 +643,67 @@ def _prune_data_for_update(data: dict, noignore: bool = False, ignore: Optional[
     return {key: value for key, value in data.items() if key not in ignore_these_properties}
 
 
-def _create_portal(env: Optional[str] = None, app: Optional[str] = None,
-                   verbose: bool = False, debug: bool = False) -> Optional[Portal]:
+def _create_portal(env: Optional[str] = None, ini: Optional[str] = None, app: Optional[str] = None,
+                   load: Optional[str] = None, verbose: bool = False, debug: bool = False,
+                   quiet: bool = False) -> Optional[Portal]:
+
+    if app:
+        if (app not in ORCHESTRATED_APPS) and ((app := app.lower()) not in ORCHESTRATED_APPS):
+            _print(f"Unknown app name; must be one of: {' | '.join(ORCHESTRATED_APPS)}")
+            return None
+    elif APP_SMAHT in (env or os.environ.get(_SMAHT_ENV_ENVIRON_NAME) or ""):
+        app = APP_SMAHT
+    elif APP_CGAP in (env or ""):
+        app = APP_CGAP
+    elif APP_FOURFRONT in (env or ""):
+        app = APP_FOURFRONT
+
+    if ini:
+        if env:
+            if not quiet:
+                _print("Ignoring --env option when --ini option is given.")
+        elif (app == _SMAHT_ENV_ENVIRON_NAME) and (env := os.environ.get(_SMAHT_ENV_ENVIRON_NAME)):
+            if not quiet:
+                _print(f"Ignoring SMAHT_ENV environment variable ({env}) when --ini option is given.")
+        if not os.path.isabs(ini_file := os.path.normpath(os.path.expanduser(ini))):
+            ini_file = os.path.normpath(os.path.join(os.getcwd(), ini_file))
+        if not os.path.exists(ini_file):
+            _print(f"Specified Portal INI file not found: {ini_file}")
+            return None
+        with captured_output(not debug):
+            if not (portal := Portal(ini_file, app=app)):
+                _print(f"Cannot create INI based Portal object: {env} ({app})")
+                return None
+    else:
+        env_from_environ = False
+        if not env and app:
+            # If the --load option is specified, and no --ini option is specified, then do NOT default
+            # to using the SMAHT_ENV environment variable (if set) for an access-key based Portal
+            # object; rather default to the default INI file (i.e. development.ini).
+            if (not load) and (app == APP_SMAHT) and (env := os.environ.get(_SMAHT_ENV_ENVIRON_NAME)):
+                env_from_environ = True
+        if not env:
+            if os.path.exists(ini_file := os.path.normpath(os.path.join(os.getcwd(), _DEFAULT_INI_FILE_FOR_LOAD))):
+                return _create_portal(ini=ini_file, app=app, verbose=verbose, debug=debug)
+            return None
+        if not (portal := Portal(env, app=app) if env or app else None):
+            _print(f"Cannot create access-key based Portal object: {env}{f' ({app})' if app else ''}")
+            return None
+
+    if (ini_file := portal.ini_file):
+        if not quiet:
+            _print(f"Portal environment: {ini_file}")
+    elif (env := portal.env) or (env := os.environ.get(_SMAHT_ENV_ENVIRON_NAME)):
+        _print(f"Portal environment"
+               f"{f' (from {_SMAHT_ENV_ENVIRON_NAME})' if env_from_environ else ''}: {portal.env}")
+        if verbose:
+            if portal.keys_file:
+                _print(f"Portal keys file: {portal.keys_file}")
+            if portal.key_id:
+                _print(f"Portal key prefix: {portal.key_id[0:2]}******")
+            if portal.server:
+                _print(f"Portal server: {portal.server}")
 
-    env_from_environ = None
-    if not env and (app == APP_SMAHT):
-        if env := os.environ.get(_SMAHT_ENV_ENVIRON_NAME):
-            env_from_environ = True
-    if not (portal := Portal(env, app=app) if env or app else None):
-        return None
-    if verbose:
-        if (env := portal.env) or (env := os.environ(_SMAHT_ENV_ENVIRON_NAME)):
-            _print(f"Portal environment"
-                   f"{f' (from {_SMAHT_ENV_ENVIRON_NAME})' if env_from_environ else ''}: {portal.env}")
-        if portal.keys_file:
-            _print(f"Portal keys file: {portal.keys_file}")
-        if portal.key_id:
-            _print(f"Portal key prefix: {portal.key_id[0:2]}******")
-        if portal.server:
-            _print(f"Portal server: {portal.server}")
     return portal
 
 

{dcicutils-8.14.0.1b15.dist-info → dcicutils-8.14.0.1b17.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dcicutils
-Version: 8.14.0.1b15
+Version: 8.14.0.1b17
 Summary: Utility package for interacting with the 4DN Data Portal and other 4DN resources
 Home-page: https://github.com/4dn-dcic/utils
 License: MIT

{dcicutils-8.14.0.1b15.dist-info → dcicutils-8.14.0.1b17.dist-info}/RECORD

@@ -60,7 +60,7 @@ dcicutils/s3_utils.py,sha256=h2B9ftOo-kxqfiKth5ZDC_cAUFy1Pbu7BrVanFnE5Iw,28839
 dcicutils/schema_utils.py,sha256=GmRm-XqZKJ6qine16SQF1txcby9WougDav_sYmKNs9E,12400
 dcicutils/scripts/publish_to_pypi.py,sha256=sMd4WASQGlxlh7uLrt2eGkFRXYgONVmvIg8mClMS5RQ,13903
 dcicutils/scripts/run_license_checker.py,sha256=z2keYnRDZsHQbTeo1XORAXSXNJK5axVzL5LjiNqZ7jE,4184
-dcicutils/scripts/update_portal_object.py,sha256=LwqGSiR7Q0_8GP0ToNfaeVQ1hZtQ69EOLW21PyKBC7c,33950
+dcicutils/scripts/update_portal_object.py,sha256=MsNpF6eHGVxr7dUg8QrwZo9XfIjPjAavOWqyI_RrfpA,38397
 dcicutils/scripts/view_portal_object.py,sha256=lcgXWH9ooVf7tJDIRnoFGOgT0wYLGhiJlJW3a9w6A_c,36983
 dcicutils/secrets_utils.py,sha256=8dppXAsiHhJzI6NmOcvJV5ldvKkQZzh3Fl-cb8Wm7MI,19745
 dcicutils/sheet_utils.py,sha256=VlmzteONW5VF_Q4vo0yA5vesz1ViUah1MZ_yA1rwZ0M,33629
@@ -75,8 +75,8 @@ dcicutils/trace_utils.py,sha256=g8kwV4ebEy5kXW6oOrEAUsurBcCROvwtZqz9fczsGRE,1769
 dcicutils/validation_utils.py,sha256=cMZIU2cY98FYtzK52z5WUYck7urH6JcqOuz9jkXpqzg,14797
 dcicutils/variant_utils.py,sha256=2H9azNx3xAj-MySg-uZ2SFqbWs4kZvf61JnK6b-h4Qw,4343
 dcicutils/zip_utils.py,sha256=_Y9EmL3D2dUZhxucxHvrtmmlbZmK4FpSsHEb7rGSJLU,3265
-dcicutils-8.14.0.1b15.dist-info/LICENSE.txt,sha256=qnwSmfnEWMl5l78VPDEzAmEbLVrRqQvfUQiHT0ehrOo,1102
-dcicutils-8.14.0.1b15.dist-info/METADATA,sha256=h4m3Yg-a4hgUo52m_3UmmBaMXsbAYRP1A-HRrhuutec,3440
-dcicutils-8.14.0.1b15.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-dcicutils-8.14.0.1b15.dist-info/entry_points.txt,sha256=W6kEWdUJk9tQ4myAgpehPdebcwvCAZ7UgB-wyPgDUMg,335
-dcicutils-8.14.0.1b15.dist-info/RECORD,,
+dcicutils-8.14.0.1b17.dist-info/LICENSE.txt,sha256=qnwSmfnEWMl5l78VPDEzAmEbLVrRqQvfUQiHT0ehrOo,1102
+dcicutils-8.14.0.1b17.dist-info/METADATA,sha256=WxgzreGqLqRZT1QRIzME9quTpxihdyduDQKkxX9kdS0,3440
+dcicutils-8.14.0.1b17.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+dcicutils-8.14.0.1b17.dist-info/entry_points.txt,sha256=W6kEWdUJk9tQ4myAgpehPdebcwvCAZ7UgB-wyPgDUMg,335
+dcicutils-8.14.0.1b17.dist-info/RECORD,,