dcicutils 8.14.0.1b9__py3-none-any.whl → 8.14.0.1b11__py3-none-any.whl

dcicutils/scripts/update_portal_object.py

@@ -14,13 +14,15 @@ import io
 import json
 import os
 import re
+import shutil
 import sys
 from typing import Callable, List, Optional, Tuple, Union
 from dcicutils.command_utils import yes_or_no
 from dcicutils.common import ORCHESTRATED_APPS, APP_SMAHT
 from dcicutils.ff_utils import delete_metadata, purge_metadata
-from dcicutils.misc_utils import get_error_message, ignored, PRINT
+from dcicutils.misc_utils import get_error_message, ignored, PRINT, to_camel_case, to_snake_case
 from dcicutils.portal_utils import Portal as PortalFromUtils
+from dcicutils.tmpfile_utils import temporary_directory
 
 
 class Portal(PortalFromUtils):
@@ -131,6 +133,8 @@ def main():
     parser.add_argument("--confirm", action="store_true", required=False, default=False, help="Confirm before action.")
     parser.add_argument("--verbose", action="store_true", required=False, default=False, help="Verbose output.")
     parser.add_argument("--quiet", action="store_true", required=False, default=False, help="Quiet output.")
+    parser.add_argument("--noprogress", action="store_true", required=False, default=False,
+                        help="No progress bar output for --load.")
     parser.add_argument("--debug", action="store_true", required=False, default=False, help="Debugging output.")
     args = parser.parse_args()
 
@@ -158,27 +162,8 @@ def main():
             _print("The --env is not used for the --load option (to load data via snovault.loadxl).")
         if args.schema:
             _print("The --schema is not used for the --load option (to load data via snovault.loadxl).")
-        from snovault.loadxl import load_data
-        from dcicutils.captured_output import captured_output
-        if args.ini:
-            ini_file = args.ini
-        else:
-            ini_file = _DEFAULT_INI_FILE_FOR_LOAD
-        if not os.path.exists(ini_file):
-            _print(f"The INI file required for --load is not found: {ini_file}")
-            exit(1)
-        if not os.path.isdir(args.load):
-            _print(f"Load directory does not exist: {args.load}")
-            exit(1)
-        portal = None
-        with captured_output(not args.debug):
-            portal = Portal(ini_file)
-        if args.verbose:
-            _print(f"Loading data files into Portal (via snovault.loadxl) from: {args.load}")
-            _print(f"Portal INI file for load is: {ini_file}")
-        load_data(portal.vapp, indir=args.load, overwrite=True, use_master_inserts=False)
-        if args.verbose:
-            _print(f"Done loading data into Portal (via snovault.loadxl) files from: {args.load}")
+        _load_data(load=args.load, ini_file=args.ini,
+                   verbose=args.verbose, debug=args.debug, noprogress=args.noprogress)
         exit(0)
 
     portal = _create_portal(env=args.env, app=app, verbose=args.verbose, debug=args.debug)
@@ -192,7 +177,7 @@ def main():
         _post_or_patch_or_upsert(portal=portal,
                                  file_or_directory=args.post,
                                  explicit_schema_name=explicit_schema_name,
-                                 update_function=post_data,
+                                 update_function=_post_data,
                                  update_action_name="POST",
                                  noignore=args.noignore, ignore=args.ignore,
                                  confirm=args.confirm, verbose=args.verbose, quiet=args.quiet, debug=args.debug)
@@ -200,7 +185,7 @@ def main():
         _post_or_patch_or_upsert(portal=portal,
                                  file_or_directory=args.patch,
                                  explicit_schema_name=explicit_schema_name,
-                                 update_function=patch_data,
+                                 update_function=_patch_data,
                                  update_action_name="PATCH",
                                  patch_delete_fields=args.delete,
                                  noignore=args.noignore, ignore=args.ignore,
@@ -210,7 +195,7 @@ def main():
         _post_or_patch_or_upsert(portal=portal,
                                  file_or_directory=args.upsert,
                                  explicit_schema_name=explicit_schema_name,
-                                 update_function=upsert_data,
+                                 update_function=_upsert_data,
                                  update_action_name="UPSERT",
                                  patch_delete_fields=args.delete,
                                  noignore=args.noignore, ignore=args.ignore,
@@ -241,14 +226,6 @@ def _post_or_patch_or_upsert(portal: Portal, file_or_directory: str,
                              confirm: bool = False, verbose: bool = False,
                              quiet: bool = False, debug: bool = False) -> None:
 
-    def is_schema_name_list(portal: Portal, keys: list) -> bool:
-        if isinstance(keys, list):
-            for key in keys:
-                if portal.get_schema(key) is None:
-                    return False
-                return True
-        return False
-
     def post_or_patch_or_upsert(portal: Portal, file: str, schema_name: Optional[str],
                                 patch_delete_fields: Optional[str] = None,
                                 confirm: bool = False, verbose: bool = False,
@@ -266,7 +243,7 @@ def _post_or_patch_or_upsert(portal: Portal, file_or_directory: str,
                                     patch_delete_fields=patch_delete_fields,
                                     noignore=noignore, ignore=ignore,
                                     confirm=confirm, verbose=verbose, debug=debug)
-                elif is_schema_name_list(portal, list(data.keys())):
+                elif _is_schema_name_list(portal, list(data.keys())):
                     if debug:
                         _print(f"DEBUG: File ({file}) contains a dictionary of schema names.")
                     for schema_name in data:
@@ -329,11 +306,11 @@ def _impose_special_ordering(data: List[dict], schema_name: str) -> List[dict]:
     return data
 
 
-def post_data(portal: Portal, data: dict, schema_name: str,
-              file: Optional[str] = None, index: int = 0,
-              patch_delete_fields: Optional[str] = None,
-              noignore: bool = False, ignore: Optional[List[str]] = None,
-              confirm: bool = False, verbose: bool = False, debug: bool = False) -> None:
+def _post_data(portal: Portal, data: dict, schema_name: str,
+               file: Optional[str] = None, index: int = 0,
+               patch_delete_fields: Optional[str] = None,
+               noignore: bool = False, ignore: Optional[List[str]] = None,
+               confirm: bool = False, verbose: bool = False, debug: bool = False) -> None:
     ignored(patch_delete_fields)
     if not (identifying_path := portal.get_identifying_path(data, portal_type=schema_name)):
         if isinstance(file, str) and isinstance(index, int):
@@ -359,11 +336,11 @@ def post_data(portal: Portal, data: dict, schema_name: str,
         return
 
 
-def patch_data(portal: Portal, data: dict, schema_name: str,
-               file: Optional[str] = None, index: int = 0,
-               patch_delete_fields: Optional[str] = None,
-               noignore: bool = False, ignore: Optional[List[str]] = None,
-               confirm: bool = False, verbose: bool = False, debug: bool = False) -> None:
+def _patch_data(portal: Portal, data: dict, schema_name: str,
+                file: Optional[str] = None, index: int = 0,
+                patch_delete_fields: Optional[str] = None,
+                noignore: bool = False, ignore: Optional[List[str]] = None,
+                confirm: bool = False, verbose: bool = False, debug: bool = False) -> None:
     if not (identifying_path := portal.get_identifying_path(data, portal_type=schema_name)):
         if isinstance(file, str) and isinstance(index, int):
             _print(f"ERROR: Item for PATCH has no identifying property: {file} (#{index + 1})")
@@ -390,11 +367,11 @@ def patch_data(portal: Portal, data: dict, schema_name: str,
         return
 
 
-def upsert_data(portal: Portal, data: dict, schema_name: str,
-                file: Optional[str] = None, index: int = 0,
-                patch_delete_fields: Optional[str] = None,
-                noignore: bool = False, ignore: Optional[List[str]] = None,
-                confirm: bool = False, verbose: bool = False, debug: bool = False) -> None:
+def _upsert_data(portal: Portal, data: dict, schema_name: str,
+                 file: Optional[str] = None, index: int = 0,
+                 patch_delete_fields: Optional[str] = None,
+                 noignore: bool = False, ignore: Optional[List[str]] = None,
+                 confirm: bool = False, verbose: bool = False, debug: bool = False) -> None:
     if not (identifying_path := portal.get_identifying_path(data, portal_type=schema_name)):
         if isinstance(file, str) and isinstance(index, int):
             _print(f"ERROR: Item for UPSERT has no identifying property: {file} (#{index + 1})")
@@ -423,6 +400,188 @@ 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,
+               verbose: bool = False, debug: bool = False, noprogress: bool = False) -> bool:
+
+    from snovault.loadxl import load_all_gen, LoadGenWrapper
+    from dcicutils.captured_output import captured_output
+    from dcicutils.progress_bar import ProgressBar
+
+    def loadxl(portal: Portal, inserts_directory: str, schema_names_to_load: dict):
+
+        nonlocal LoadGenWrapper, load_all_gen, verbose, debug
+        progress_total = sum(schema_names_to_load.values()) * 2  # loadxl does two passes
+        progress_bar = ProgressBar(progress_total) if not noprogress else None
+
+        def decode_bytes(str_or_bytes: Union[str, bytes], *, encoding: str = "utf-8") -> str:
+            if not isinstance(encoding, str):
+                encoding = "utf-8"
+            if isinstance(str_or_bytes, bytes):
+                return str_or_bytes.decode(encoding).strip()
+            elif isinstance(str_or_bytes, str):
+                return str_or_bytes.strip()
+            return ""
+
+        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
+            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)
+            item_type = match.group(3)
+            if current_item_type != item_type:
+                if noprogress and debug and current_item_type is not None:
+                    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 progress_bar:
+                progress_bar.set_progress(total_item_count)
+            elif debug:
+                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.expanduser(ini_file)):
+        ini_file = 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 not os.path.isabs(load := os.path.expanduser(load)):
+        load = os.path.join(os.getcwd(), load)
+    if not os.path.exists(load):
+        return False
+
+    if os.path.isdir(load):
+        inserts_directory = load
+        inserts_file = None
+    else:
+        inserts_directory = None
+        inserts_file = load
+
+    portal = None
+    with captured_output(not debug):
+        portal = Portal(ini_file)
+
+    if inserts_file:
+        with io.open(inserts_file, "r") as f:
+            try:
+                data = json.load(f)
+            except Exception:
+                _print(f"Cannot load JSON data from file: {inserts_file}")
+                return False
+            if isinstance(data, list):
+                if not (schema_name := explicit_schema_name):
+                    if not (schema_name := _get_schema_name_from_schema_named_json_file_name(portal, inserts_file)):
+                        _print("Unable to determine schema name for JSON data file: {inserts_file}")
+                        return False
+                with temporary_directory() as tmpdir:
+                    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=explicit_schema_name,
+                                      verbose=verbose, debug=debug, noprogress=noprogress)
+            elif isinstance(data, dict):
+                _print("DICT IN FILE FOR LOAD NOT YET SUPPPORTED")
+                if not _is_schema_name_list(portal, schema_names := list(data.keys())):
+                    _print(f"Unrecognized types in JSON data file: {inserts_file}")
+                    return False
+                with temporary_directory() as tmpdir:
+                    nfiles = 0
+                    for schema_name in schema_names:
+                        if not isinstance(schema_data := data[schema_name], list):
+                            _print(f"Unexpected value for data type ({schema_name})"
+                                   f" in JSON data file: {inserts_file} ▶ ignoring")
+                            continue
+                        file_name = os.path.join(tmpdir, f"{to_snake_case(schema_name)}.json")
+                        with io.open(file_name, "w") as f:
+                            json.dump(schema_data, f)
+                        nfiles += 1
+                    if nfiles > 0:
+                        return _load_data(load=tmpdir, ini_file=ini_file,
+                                          verbose=verbose, debug=debug, noprogress=noprogress)
+                # TODO
+                return True
+            else:
+                _print(f"Unrecognized JSON data in file: {inserts_file}")
+                return False
+        return True
+    if verbose:
+        _print(f"Loading data files into Portal (via snovault.loadxl) from: {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]
+    schema_names_to_load = {}
+
+    copy_to_temporary_directory = False
+    for json_file_path in glob.glob(os.path.join(inserts_directory, "*.json")):
+        json_file_name = os.path.basename(json_file_path)
+        schema_name = os.path.basename(json_file_name)[:-len(".json")]
+        if (schema_name not in schema_snake_case_names) and (schema_name not in schema_names):
+            _print(f"File is not named for a known schema: {json_file_name} ▶ ignoring")
+            copy_to_temporary_directory = True
+        else:
+            try:
+                with io.open(json_file_path, "r") as f:
+                    if not isinstance(data := json.load(f), list):
+                        _print("Data JSON file does not contain an array: {json_file_path} ▶ ignoring")
+                        copy_to_temporary_directory = True
+                    elif (nobjects := len(data)) < 1:
+                        _print("Data JSON file contains no items: {json_file_path} ▶ ignoring")
+                        copy_to_temporary_directory = True
+                    else:
+                        schema_names_to_load[schema_name] = nobjects
+            except Exception:
+                _print("Cannot load JSON data from file: {json_file_path} ▶ ignoring")
+                copy_to_temporary_directory = True
+    if not schema_names_to_load:
+        _print("Directory contains no valid data: {inserts_directory}")
+        return False
+    if copy_to_temporary_directory:
+        with temporary_directory() as tmpdir:
+            if debug:
+                _print(f"Using temporary directory: {tmpdir}")
+            for json_file_path in glob.glob(os.path.join(inserts_directory, "*.json")):
+                json_file_name = os.path.basename(json_file_path)
+                schema_name = os.path.basename(json_file_name)[:-len(".json")]
+                if (schema_name in schema_snake_case_names) or (schema_name in schema_names):
+                    shutil.copy(json_file_path, tmpdir)
+            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:
+        _print(f"Done loading data into Portal (via snovault.loadxl) files from: {inserts_directory}")
+    return True
+
+
+def _is_schema_name_list(portal: Portal, keys: list) -> bool:
+    if isinstance(keys, list):
+        for key in keys:
+            if portal.get_schema(key) is None:
+                return False
+            return True
+    return False
+
+
 def _prune_data_for_update(data: dict, noignore: bool = False, ignore: Optional[List[str]] = None) -> dict:
     ignore_these_properties = [] if noignore is True else _IGNORE_PROPERTIES_ON_UPDATE
     if isinstance(ignore, list):
@@ -498,17 +657,21 @@ def _parse_delete_fields(value: str) -> str:
 
 
 def _get_schema_name_from_schema_named_json_file_name(portal: Portal, value: str) -> Optional[str]:
-    try:
-        if not value.endswith(".json"):
-            return None
-        _, schema_name = _get_schema(portal, os.path.basename(value[:-5]))
-        return schema_name
-    except Exception:
-        return False
+    if isinstance(value, str) and value:
+        try:
+            if value.endswith(".json"):
+                value = value[:-5]
+            _, schema_name = _get_schema(portal, os.path.basename(value))
+            return schema_name
+        except Exception:
+            pass
+    return False
 
 
 @lru_cache(maxsize=1)
 def _get_schemas(portal: Portal) -> Optional[dict]:
+    if portal.vapp:
+        return portal.vapp.get("/profiles/?frame=raw").json
     return portal.get_schemas()
 
 

{dcicutils-8.14.0.1b9.dist-info → dcicutils-8.14.0.1b11.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dcicutils
-Version: 8.14.0.1b9
+Version: 8.14.0.1b11
 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-8.14.0.1b9.dist-info → dcicutils-8.14.0.1b11.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=9_ZpfKwIJUDbyEI0Xqu_9keMxTIVZ_CyxX8WeGrFI14,24376
+dcicutils/scripts/update_portal_object.py,sha256=fLgsPgnugKLS18A8JwP8O_UdmeFqNIjUIhrEFlIhfgE,32589
 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-8.14.0.1b9.dist-info/LICENSE.txt,sha256=qnwSmfnEWMl5l78VPDEzAmEbLVrRqQvfUQiHT0ehrOo,1102
-dcicutils-8.14.0.1b9.dist-info/METADATA,sha256=6dzBhVwY1DNx4-lCS_3FSaztYt-JC8bNFU9o8VoyycA,3439
-dcicutils-8.14.0.1b9.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-dcicutils-8.14.0.1b9.dist-info/entry_points.txt,sha256=W6kEWdUJk9tQ4myAgpehPdebcwvCAZ7UgB-wyPgDUMg,335
-dcicutils-8.14.0.1b9.dist-info/RECORD,,
+dcicutils-8.14.0.1b11.dist-info/LICENSE.txt,sha256=qnwSmfnEWMl5l78VPDEzAmEbLVrRqQvfUQiHT0ehrOo,1102
+dcicutils-8.14.0.1b11.dist-info/METADATA,sha256=HFNj87yKbAwTA98cVDJm_0r9aVJssg7zo5dIsqngVrg,3440
+dcicutils-8.14.0.1b11.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+dcicutils-8.14.0.1b11.dist-info/entry_points.txt,sha256=W6kEWdUJk9tQ4myAgpehPdebcwvCAZ7UgB-wyPgDUMg,335
+dcicutils-8.14.0.1b11.dist-info/RECORD,,