streamlit-octostar-utils 0.6.5.dev2__tar.gz → 0.6.7__tar.gz

{streamlit_octostar_utils-0.6.5.dev2 → streamlit_octostar_utils-0.6.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: streamlit-octostar-utils
-Version: 0.6.5.dev2
+Version: 0.6.7
 Summary: 
 License: MIT
 License-File: LICENSE
@@ -24,6 +24,7 @@ Requires-Dist: iso639-lang (>=2.0.0,<3.0.0) ; extra == "nlp"
 Requires-Dist: nltk (>=3.8.0,<4.0.0) ; extra == "nlp"
 Requires-Dist: numpy (>=1.20.0)
 Requires-Dist: octostar-streamlit (>=0.1.25,<0.2.0)
+Requires-Dist: pillow (>=10.0.0)
 Requires-Dist: pottery (>=3.0.0,<4.0.0)
 Requires-Dist: presidio-analyzer (>=2.2.0,<3.0.0) ; extra == "nlp"
 Requires-Dist: psutil (>=5.9.0)

{streamlit_octostar_utils-0.6.5.dev2 → streamlit_octostar_utils-0.6.7}/pyproject.toml

@@ -5,7 +5,7 @@ include = '\.pyi?$'
 
 [tool.poetry]
 name = "streamlit-octostar-utils"
-version = "0.6.5-dev.2"
+version = "0.6.7"
 description = ""
 license = "MIT"
 authors = ["Octostar"]
@@ -43,6 +43,7 @@ pottery = "^3.0.0"
 slowapi = "^0.1.9"
 filetype = "^1.2.0"
 PyJWT = "^2.5.0"
+pillow = ">=10.0.0"
 #octostar-python-client = "^1.0.0"
 py3langid = { version = ">=0.2.0,<0.3.0", optional = true }
 spacy-download = { version = "==1.1.0", optional = true }

{streamlit_octostar_utils-0.6.5.dev2 → streamlit_octostar_utils-0.6.7}/streamlit_octostar_utils/core/opensearch_conversion.py

@@ -44,6 +44,7 @@ conversion_matrix = {
         False,
     ),
     ("str", "object"): lambda d: (json.loads(d), True),
+    ("str", "flat_object"): lambda d: (json.loads(d), True),
     ("str", "nested"): lambda d: (json.loads(d), True),
     ("bool", "text"): lambda d: (str(d), False),
     ("bool", "keyword"): lambda d: (str(d), False),
@@ -102,11 +103,13 @@ conversion_matrix = {
     ),
     ("bytes", "binary"): lambda d: (base64.b64encode(d).decode("utf-8"), False),
     ("bytes", "object"): lambda d: (json.loads(d.decode("utf-8")), True),
+    ("bytes", "flat_object"): lambda d: (json.loads(d.decode("utf-8")), True),
     ("bytes", "nested"): lambda d: (json.loads(d.decode("utf-8")), True),
     ("dict", "text"): lambda d: (json.dumps(d), False),
     ("dict", "keyword"): lambda d: (json.dumps(d), False),
     ("dict", "boolean"): lambda d: (bool(d), False),
     ("dict", "object"): lambda d: (d, True),
+    ("dict", "flat_object"): lambda d: (d, True),
     ("dict", "nested"): lambda d: (d, True),
 }
 
@@ -225,7 +228,7 @@ def handle_data_dict(data, curr_mapping, key):
     if (
         curr_mapping
         and has_opensearch_type(curr_mapping)
-        and curr_mapping.get("type") not in ["object", "nested"]
+        and curr_mapping.get("type") not in ["object", "nested", "flat_object"]
     ):
         convert_data_type(data, curr_mapping, key)
     else:

streamlit_octostar_utils-0.6.7/streamlit_octostar_utils/metadata/__init__.py

@@ -0,0 +1,13 @@
+from streamlit_octostar_utils.metadata.parent_meta import (
+    EXIF_PASSTHROUGH_FIELDS,
+    OS_PARENT_META_PREFIX,
+    resolve_with_aliases,
+    select_parent_meta_passthrough,
+)
+
+__all__ = [
+    "EXIF_PASSTHROUGH_FIELDS",
+    "OS_PARENT_META_PREFIX",
+    "resolve_with_aliases",
+    "select_parent_meta_passthrough",
+]

streamlit_octostar_utils-0.6.7/streamlit_octostar_utils/metadata/parent_meta.py

@@ -0,0 +1,121 @@
+"""Parent->child EXIF passthrough protocol (convention + selector + reader).
+
+Producer routes that create new derived files (face crops, video keyframes,
+extracted audio, PDF page splits, ...) carry the parent's EXIF date/GPS tags
+through to downstream extract-metadata by embedding
+``os_parent_meta_<canonical>`` keys into the child file's metadata container.
+
+This module owns the **convention**: which canonical EXIF fields propagate
+(:data:`EXIF_PASSTHROUGH_FIELDS`), what prefix wraps them
+(:data:`OS_PARENT_META_PREFIX`), how a parent's exiftool dict is reduced to a
+ready-to-embed payload (:func:`select_parent_meta_passthrough`), and how a
+reader recovers values regardless of which container surfaced them
+(:func:`resolve_with_aliases`).
+
+The actual **embedding** into PNG ``tEXt`` chunks, PDF Info dicts, or ffmpeg
+``-metadata`` flags lives in each producer app -- it requires container-
+specific libraries (PIL, PyMuPDF, ffmpeg) that don't belong in a shared
+utility lib. Each producer imports the convention/selector from here and
+implements its own stamper.
+
+Keep producer-side embedders and this reader on the same suffix protocol:
+:func:`resolve_with_aliases` is the only piece that knows what prefixes the
+various containers slap on tag names, and a new container format means
+changes here that both sides must see.
+"""
+
+from typing import Any, Dict
+
+
+OS_PARENT_META_PREFIX = "os_parent_meta_"
+
+_PDF_CUSTOM_PREFIX = "pdf:docinfo:custom:"
+
+EXIF_PASSTHROUGH_FIELDS = (
+    "EXIF:DateTimeOriginal",
+    "Composite:SubSecDateTimeOriginal",
+    "Composite:DateTimeOriginal",
+    "EXIF:CreateDate",
+    "Composite:SubSecCreateDate",
+    "QuickTime:CreateDate",
+    "QuickTime:MediaCreateDate",
+    "QuickTime:TrackCreateDate",
+    "PDF:CreateDate",
+    "XMP:CreateDate",
+    "XMP:DateCreated",
+    "IPTC:DateCreated",
+    "Composite:DateTimeCreated",
+    "EXIF:DateTime",
+    "EXIF:ModifyDate",
+    "Composite:SubSecModifyDate",
+    "PDF:ModDate",
+    "XMP:ModifyDate",
+    "Composite:GPSLatitude",
+    "EXIF:GPSLatitude",
+    "XMP:GPSLatitude",
+    "Composite:GPSLatitudeRef",
+    "EXIF:GPSLatitudeRef",
+    "XMP:GPSLatitudeRef",
+    "Composite:GPSLongitude",
+    "EXIF:GPSLongitude",
+    "XMP:GPSLongitude",
+    "Composite:GPSLongitudeRef",
+    "EXIF:GPSLongitudeRef",
+    "XMP:GPSLongitudeRef",
+    "Composite:GPSPosition",
+)
+
+
+def resolve_with_aliases(metadata: Dict[str, Any], key: str) -> Any:
+    """Return ``metadata[key]`` respecting ``os_parent_meta_<key>`` passthrough.
+
+    Lookup order:
+
+    1. ``os_parent_meta_<key>`` -- written directly by producer routes that
+       can place a key into the child container under an exact name.
+    2. ``pdf:docinfo:custom:os_parent_meta_<key>`` -- PDF custom property
+       container (mirrors the ``os_ContextBefore`` pattern used elsewhere).
+    3. Any key ending with ``:os_parent_meta_<key>`` -- catches the
+       format-specific group prefix that exiftool prepends when reading
+       container-native tags (``PNG:`` for PNG tEXt, ``ID3:`` for MP3 TXXX,
+       ``RIFF:`` for WAV INFO, etc.).
+    4. ``<key>`` -- canonical fallback (the value as it would appear if the
+       child carried real EXIF natively).
+    """
+    aliased = f"{OS_PARENT_META_PREFIX}{key}"
+    value = metadata.get(aliased)
+    if value is not None:
+        return value
+    value = metadata.get(f"{_PDF_CUSTOM_PREFIX}{aliased}")
+    if value is not None:
+        return value
+    suffix = f":{aliased}"
+    for k, v in metadata.items():
+        if v is not None and k.endswith(suffix):
+            return v
+    return metadata.get(key)
+
+
+def select_parent_meta_passthrough(metadata: Any) -> Dict[str, str]:
+    """Pick EXIF date/GPS tags worth embedding into derived children.
+
+    Input is a parent's raw exiftool metadata dict (as stored on the parent
+    entity's ``extract:metadata`` annotation). Output is keyed by
+    ``os_parent_meta_<canonical>`` with stringified values, ready to feed
+    into a container-specific embedder owned by the producer app.
+
+    Uses :func:`resolve_with_aliases` so a parent that itself inherited tags
+    from *its* parent (e.g. a keyframe of an inherited-meta video) still
+    propagates the original values forward.
+    """
+    if not isinstance(metadata, dict):
+        return {}
+    out: Dict[str, str] = {}
+    for key in EXIF_PASSTHROUGH_FIELDS:
+        value = resolve_with_aliases(metadata, key)
+        if value is None:
+            continue
+        if isinstance(value, (list, tuple)):
+            value = ",".join(str(v) for v in value)
+        out[f"{OS_PARENT_META_PREFIX}{key}"] = str(value)
+    return out