Merge pull request #49 from casework/change_local_uuid_setup_parameter

Author: kchason

case_utils/case_file/__init__.py

@@ -17,8 +17,10 @@
 
 __version__ = "0.3.0"
 
+import argparse
 import datetime
 import hashlib
+import logging
 import os
 import typing
 import warnings
@@ -207,10 +209,9 @@ def create_file_node(
 
 
 def main() -> None:
-    import argparse
-
     parser = argparse.ArgumentParser()
     parser.add_argument("--base-prefix", default=DEFAULT_PREFIX)
+    parser.add_argument("--debug", action="store_true")
     parser.add_argument("--disable-hashes", action="store_true")
     parser.add_argument("--disable-mtime", action="store_true")
     parser.add_argument(
@@ -220,6 +221,8 @@ def main() -> None:
     parser.add_argument("in_file")
     args = parser.parse_args()
 
+    logging.basicConfig(level=logging.DEBUG if args.debug else logging.INFO)
+
     case_utils.local_uuid.configure()
 
     NS_BASE = rdflib.Namespace(args.base_prefix)

case_utils/local_uuid.py

@@ -17,20 +17,91 @@
 
 __version__ = "0.2.0"
 
+import logging
 import os
+import pathlib
 import sys
+import typing
+import warnings
 import uuid
 
-USE_DEMO_UUID: bool = False
+DEMO_UUID_BASE: typing.Optional[str] = None
 
 DEMO_UUID_COUNTER: int = 0
 
+_logger = logging.getLogger(pathlib.Path(__file__).name)
+
 
 def configure() -> None:
-    global USE_DEMO_UUID
+    global DEMO_UUID_BASE
 
     if os.getenv("DEMO_UUID_REQUESTING_NONRANDOM") == "NONRANDOM_REQUESTED":
-        USE_DEMO_UUID = True
+        warnings.warn(
+            "Environment variable DEMO_UUID_REQUESTING_NONRANDOM is deprecated.  See case_utils.local_uuid.demo_uuid for usage notes on its replacement, CASE_DEMO_NONRANDOM_UUID_BASE.  Proceeding with random UUIDs.",
+            DeprecationWarning,
+        )
+        return
+
+    env_base_dir_name = os.getenv("CASE_DEMO_NONRANDOM_UUID_BASE")
+    if env_base_dir_name is None:
+        return
+
+    base_dir_original_path = pathlib.Path(env_base_dir_name)
+    if not base_dir_original_path.exists():
+        warnings.warn(
+            "Environment variable CASE_DEMO_NONRANDOM_UUID_BASE is expected to refer to an existing directory.  Proceeding with random UUIDs."
+        )
+        return
+    if not base_dir_original_path.is_dir():
+        warnings.warn(
+            "Environment variable CASE_DEMO_NONRANDOM_UUID_BASE is expected to refer to a directory.  Proceeding with random UUIDs."
+        )
+        return
+
+    # Component: An emphasis this is an example.
+    demo_uuid_base_parts = ["example.org"]
+
+    # Component: Present working directory, relative to CASE_DEMO_NONRANDOM_UUID_BASE if that environment variable is an ancestor of pwd.
+    base_dir_resolved_path = base_dir_original_path.resolve()
+    srcdir_original_path = pathlib.Path(os.getcwd())
+    srcdir_resolved_path = srcdir_original_path.resolve()
+    # _logger.debug("base_dir_resolved_path = %r.", base_dir_resolved_path)
+    # _logger.debug("srcdir_resolved_path = %r.", srcdir_resolved_path)
+    try:
+        srcdir_relative_path = srcdir_resolved_path.relative_to(base_dir_resolved_path)
+        # _logger.debug("srcdir_relative_path = %r.", srcdir_relative_path)
+        demo_uuid_base_parts.append(str(srcdir_relative_path))
+    except ValueError:
+        # If base_dir is not an ancestor directory of srcdir, default to srcdir.
+        # _logger.debug("PWD is not relative to base path.")
+        demo_uuid_base_parts.append(str(srcdir_resolved_path))
+
+    # Component: Command of argument vector.
+    env_venv_name = os.getenv("VIRTUAL_ENV")
+    if env_venv_name is None:
+        demo_uuid_base_parts.append(sys.argv[0])
+    else:
+        command_original_path = pathlib.Path(sys.argv[0])
+        command_resolved_path = command_original_path.resolve()
+        venv_original_path = pathlib.Path(env_venv_name)
+        venv_resolved_path = venv_original_path.resolve()
+        try:
+            command_relative_path = command_resolved_path.relative_to(
+                venv_resolved_path
+            )
+            # _logger.debug("command_relative_path = %r.", command_relative_path)
+            demo_uuid_base_parts.append(str(command_relative_path))
+        except ValueError:
+            # _logger.debug("Command path is not relative to virtual environment path.")
+            demo_uuid_base_parts.append(str(command_resolved_path))
+
+    if len(sys.argv) > 1:
+        # Component: Arguments of argument vector.
+        demo_uuid_base_parts.extend(sys.argv[1:])
+
+    # _logger.debug("demo_uuid_base_parts = %r.", demo_uuid_base_parts)
+
+    DEMO_UUID_BASE = "/".join(demo_uuid_base_parts)
 
 
 def demo_uuid() -> str:
@@ -39,38 +110,34 @@ def demo_uuid() -> str:
 
     WARNING: This function was developed for use ONLY for reducing (but not eliminating) version-control edits to identifiers in sample data.  It creates UUIDs that are decidedly NOT random, and should remain consistent on repeated calls to the importing script.
 
-    To prevent accidental non-random UUID usage, an environment variable must be set to an uncommon string, hard-coded in this function.
+    To prevent accidental non-random UUID usage, an environment variable must be set to a string provided by the caller.  The variable's required value is the path to some directory.  The variable's recommended value is the equivalent of the Make variable "top_srcdir" - that is, the root directory of the containing Git repository, some parent of the current process's current working directory.
     """
+    global DEMO_UUID_BASE
     global DEMO_UUID_COUNTER
 
-    if os.getenv("DEMO_UUID_REQUESTING_NONRANDOM") != "NONRANDOM_REQUESTED":
-        raise EnvironmentError(
-            "demo_uuid() called without DEMO_UUID_REQUESTING_NONRANDOM in environment."
+    if os.getenv("CASE_DEMO_NONRANDOM_UUID_BASE") is None:
+        raise ValueError(
+            "demo_uuid() called without CASE_DEMO_NONRANDOM_UUID_BASE in environment."
         )
 
-    # Component: An emphasis this is an example.
-    parts = ["example.org"]
+    if DEMO_UUID_BASE is None:
+        raise ValueError("demo_uuid() called with DEMO_UUID_BASE unset.")
+
+    parts = [DEMO_UUID_BASE]
 
     # Component: Incrementing counter.
     DEMO_UUID_COUNTER += 1
     parts.append(str(DEMO_UUID_COUNTER))
 
-    # Component: Present working directory, replacing $HOME with '~'.
-    env_HOME: str = os.getenv("HOME", "/nonexistent")
-    parts.append(os.getcwd().replace(env_HOME, "~"))
-
-    # Component: Argument vector.
-    parts.extend(sys.argv)
-
     return str(uuid.uuid5(uuid.NAMESPACE_URL, "/".join(parts)))
 
 
 def local_uuid() -> str:
     """
     Generate either a UUID4, or if requested via environment configuration, a non-random demo UUID.
     """
-    global USE_DEMO_UUID
-    if USE_DEMO_UUID:
-        return demo_uuid()
-    else:
+    global DEMO_UUID_BASE
+    if DEMO_UUID_BASE is None:
         return str(uuid.uuid4())
+    else:
+        return demo_uuid()

tests/case_utils/case_file/Makefile

@@ -114,9 +114,10 @@ sample.txt.json: \
   $(top_srcdir)/case_utils/namespace.py \
   sample.txt-nocompact.json
 	rm -f $@ _$@ __$@
-	export DEMO_UUID_REQUESTING_NONRANDOM=NONRANDOM_REQUESTED \
+	export CASE_DEMO_NONRANDOM_UUID_BASE="$(top_srcdir)" \
 	  && source $(tests_srcdir)/venv/bin/activate \
 	    && case_file \
+	      --debug \
 	      __$@ \
 	      sample.txt
 	source $(tests_srcdir)/venv/bin/activate \
@@ -146,9 +147,10 @@ sample.txt.ttl: \
   $(top_srcdir)/case_utils/namespace.py \
   sample.txt.done.log
 	rm -f _$@ __$@
-	export DEMO_UUID_REQUESTING_NONRANDOM=NONRANDOM_REQUESTED \
+	export CASE_DEMO_NONRANDOM_UUID_BASE="$(top_srcdir)" \
 	  && source $(tests_srcdir)/venv/bin/activate \
 	    && case_file \
+	      --debug \
 	      __$@ \
 	      sample.txt
 	java -jar $(RDF_TOOLKIT_JAR) \
@@ -169,9 +171,10 @@ sample.txt-disable_hashes.ttl: \
   $(top_srcdir)/case_utils/namespace.py \
   sample.txt.done.log
 	rm -f _$@ __$@
-	export DEMO_UUID_REQUESTING_NONRANDOM=NONRANDOM_REQUESTED \
+	export CASE_DEMO_NONRANDOM_UUID_BASE="$(top_srcdir)" \
 	  && source $(tests_srcdir)/venv/bin/activate \
 	    && case_file \
+	      --debug \
 	      --disable-hashes \
 	      __$@ \
 	      sample.txt
@@ -194,9 +197,10 @@ sample.txt-nocompact.json: \
   $(top_srcdir)/case_utils/namespace.py \
   sample.txt.done.log
 	rm -f _$@
-	export DEMO_UUID_REQUESTING_NONRANDOM=NONRANDOM_REQUESTED \
+	export CASE_DEMO_NONRANDOM_UUID_BASE="$(top_srcdir)" \
 	  && source $(tests_srcdir)/venv/bin/activate \
 	    && case_file \
+	      --debug \
 	      _$@ \
 	      sample.txt
 	# To avoid making noisy, uninformative updates from blank node identifiers, only move the new file into place if it is not isomorphic with the Git-tracked version of the target.

tests/case_utils/case_file/kb.json

@@ -9,7 +9,23 @@
     },
     "@graph": [
         {
-            "@id": "kb:file-1080c4cc-7886-5a52-bac1-f6a2b16c0ddb",
+            "@id": "kb:file-800784de-5c9e-5eb2-b843-0ac51a1bd4b9",
+            "@type": "uco-observable:File",
+            "uco-core:hasFacet": {
+                "@type": "uco-observable:FileFacet",
+                "uco-observable:fileName": "sample.txt",
+                "uco-observable:modifiedTime": {
+                    "@type": "xsd:dateTime",
+                    "@value": "2010-01-02T03:04:56+00:00"
+                },
+                "uco-observable:sizeInBytes": {
+                    "@type": "xsd:integer",
+                    "@value": "4"
+                }
+            }
+        },
+        {
+            "@id": "kb:file-ace6460a-92a9-58b9-83ea-a18ae87f6e04",
             "@type": "uco-observable:File",
             "uco-core:hasFacet": [
                 {
@@ -78,22 +94,6 @@
                     }
                 }
             ]
-        },
-        {
-            "@id": "kb:file-b5e8a943-c556-5964-a618-8f0d000822af",
-            "@type": "uco-observable:File",
-            "uco-core:hasFacet": {
-                "@type": "uco-observable:FileFacet",
-                "uco-observable:fileName": "sample.txt",
-                "uco-observable:modifiedTime": {
-                    "@type": "xsd:dateTime",
-                    "@value": "2010-01-02T03:04:56+00:00"
-                },
-                "uco-observable:sizeInBytes": {
-                    "@type": "xsd:integer",
-                    "@value": "4"
-                }
-            }
         }
     ]
 }

tests/case_utils/case_file/kb.ttl

@@ -8,7 +8,17 @@
 @prefix uco-vocabulary: <https://ontology.unifiedcyberontology.org/uco/vocabulary/> .
 @prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
 
-kb:file-1080c4cc-7886-5a52-bac1-f6a2b16c0ddb
+kb:file-800784de-5c9e-5eb2-b843-0ac51a1bd4b9
+	a uco-observable:File ;
+	uco-core:hasFacet [
+		a uco-observable:FileFacet ;
+		uco-observable:fileName "sample.txt" ;
+		uco-observable:modifiedTime "2010-01-02T03:04:56+00:00"^^xsd:dateTime ;
+		uco-observable:sizeInBytes "4"^^xsd:integer ;
+	] ;
+	.
+
+kb:file-ace6460a-92a9-58b9-83ea-a18ae87f6e04
 	a uco-observable:File ;
 	uco-core:hasFacet
 		[
@@ -46,13 +56,3 @@ kb:file-1080c4cc-7886-5a52-bac1-f6a2b16c0ddb
 		;
 	.
 
-kb:file-b5e8a943-c556-5964-a618-8f0d000822af
-	a uco-observable:File ;
-	uco-core:hasFacet [
-		a uco-observable:FileFacet ;
-		uco-observable:fileName "sample.txt" ;
-		uco-observable:modifiedTime "2010-01-02T03:04:56+00:00"^^xsd:dateTime ;
-		uco-observable:sizeInBytes "4"^^xsd:integer ;
-	] ;
-	.
-

tests/case_utils/case_file/sample.txt-disable_hashes.ttl

@@ -6,7 +6,7 @@
 @prefix uco-observable: <https://ontology.unifiedcyberontology.org/uco/observable/> .
 @prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
 
-kb:file-b5e8a943-c556-5964-a618-8f0d000822af
+kb:file-800784de-5c9e-5eb2-b843-0ac51a1bd4b9
 	a uco-observable:File ;
 	uco-core:hasFacet [
 		a uco-observable:FileFacet ;