WIP: add PRONOM export

src/jsonid/export.py

@@ -6,20 +6,21 @@
 from datetime import timezone
 
 try:
+    import pronom
     import registry_data
     import version
 except ModuleNotFoundError:
     try:
-        from src.jsonid import registry_data, version
+        from src.jsonid import pronom, registry_data, version
     except ModuleNotFoundError:
-        from jsonid import registry_data, version
+        from jsonid import pronom, registry_data, version
 
 logger = logging.getLogger(__name__)
 
 
 def exportJSON() -> None:  # pylint: disable=C0103
     """Export to JSON."""
-    logger.debug("exporting registry ad JSON")
+    logger.debug("exporting registry as JSON")
     data = registry_data.registry()
     json_obj = []
     id_ = {
@@ -35,3 +36,62 @@ def exportJSON() -> None:  # pylint: disable=C0103
     for datum in data:
         json_obj.append(datum.json())
     print(json.dumps(json_obj, indent=2))
+
+
+def exportPRONOM() -> None:
+    """Export a PRONOM compatible set of signatures."""
+    logger.debug("exporting registry as PRONOM")
+    data = registry_data.registry()
+    all_sequences = []
+
+    formats = []
+
+    for datum in data:
+        id_ = datum.json()["identifier"]
+        name_ = datum.json()["name"][0]["@en"]
+        markers = datum.json()["markers"]
+
+        format_sequences = []
+
+        try:
+            sequences = pronom.process_markers(markers.copy())
+            all_sequences.append((id_, name_, sequences))
+            format_sequences.append(sequences)
+        except pronom.UnprocessableEntity as err:
+            logger.error(
+                "%s %s: cannot handle: %s",
+                id_,
+                name_,
+                err,
+            )
+            for marker in markers:
+                logger.debug("--- START ---")
+                logger.debug("marker: %s", marker)
+                logger.debug("---  END  ---")
+            continue
+
+        format = pronom.Format(
+            id=0,
+            name=name_,
+            version="",
+            puid=id_,
+            mime="TODO",
+            classification="structured text",  # TODO: magic
+            external_signatures=[
+                pronom.ExternalSignature(
+                    id=0,
+                    signature="JSON",
+                    type="TODO",
+                )
+            ],
+            internal_signatures=format_sequences[0],
+            priorities=[],
+        )
+
+        formats.append(format)
+
+    pronom.process_formats_and_save(formats, "abc.xml")
+
+
+def exportPRONOMXML() -> None:
+    """Export a PRONOM compatible set of signatures."""

src/jsonid/export_helpers.py

@@ -0,0 +1,30 @@
+"""Helpers for the export functions."""
+
+import datetime
+from datetime import timezone
+from typing import Final
+from xml.dom.minidom import parseString
+
+UTC_TIME_FORMAT: Final[str] = "%Y-%m-%dT%H:%M:%SZ"
+
+
+def get_utc_timestamp_now():
+    """Get a formatted UTC timestamp for 'now' that can be used when
+    a timestamp is needed.
+    """
+    return datetime.datetime.now(timezone.utc).strftime(UTC_TIME_FORMAT)
+
+
+def new_prettify(c):
+    """Remove excess newlines from DOM output.
+
+    via: https://stackoverflow.com/a/14493981
+    """
+    reparsed = parseString(c)
+    return "\n".join(
+        [
+            line
+            for line in reparsed.toprettyxml(indent=" " * 2).split("\n")
+            if line.strip()
+        ]
+    )

src/jsonid/helpers.py

@@ -2,7 +2,7 @@
 
 import logging
 import time
-from typing import Union
+from typing import Final, Union
 
 try:
     import htm_template
@@ -75,6 +75,15 @@ def format_marker(marker_text: str, marker: dict) -> str:
     return f"{marker_text}{marker_formatted}\n"
 
 
+TYPE_BOOL: Final[str] = "bool"
+TYPE_FLOAT: Final[str] = "float"
+TYPE_INTEGER: Final[str] = "integer"
+TYPE_LIST: Final[str] = "list"
+TYPE_NONE: Final[str] = "NoneType"
+TYPE_MAP: Final[str] = "map"
+TYPE_STRING: Final[str] = "string"
+
+
 def substitute_type_text(replace_me: Union[str, type]):
     """Output a text substitution for a type that will otherwise not
     pretty-print.
@@ -83,19 +92,19 @@ def substitute_type_text(replace_me: Union[str, type]):
     # pylint: disable=R0911
 
     if replace_me.__name__ == "dict":
-        return "map"
+        return TYPE_MAP
     if replace_me.__name__ == "int":
-        return "integer"
+        return TYPE_INTEGER
     if replace_me.__name__ == "list":
-        return "list"
+        return TYPE_LIST
     if replace_me.__name__ == "str":
-        return "string"
+        return TYPE_STRING
     if replace_me.__name__ == "float":
-        return "float"
+        return TYPE_NONE
     if replace_me.__name__ == "bool":
-        return "bool"
+        return TYPE_BOOL
     if replace_me.__name__ == "NoneType":
-        return "NoneType"
+        return TYPE_NONE
     if not isinstance(replace_me, type):
         pass
     return replace_me

src/jsonid/jsonid.py

@@ -160,6 +160,7 @@ def main() -> None:
         "--pronom",
         help="return a PRONOM-centric view of the results",
         required=False,
+        action="store_true",
     )
     parser.add_argument(
         "--export",
@@ -220,7 +221,8 @@ def main() -> None:
     if args.registry:
         raise NotImplementedError("custom registry is not yet available")
     if args.pronom:
-        raise NotImplementedError("pronom view is not yet implemented")
+        export.exportPRONOM()
+        sys.exit()
     if args.language:
         raise NotImplementedError("multiple languages are not yet implemented")
     if args.export: