geometamaker 0.1.2__py3-none-any.whl → 0.2.1__py3-none-any.whl

geometamaker/geometamaker.py

@@ -2,6 +2,7 @@ import functools
 import hashlib
 import logging
 import os
+import re
 import requests
 from collections import defaultdict
 from datetime import datetime, timezone
@@ -13,13 +14,20 @@ import pygeoprocessing
 import yaml
 from osgeo import gdal
 from osgeo import osr
+from pathlib import Path
 from pydantic import ValidationError
+import tarfile
 
 from . import models
 from .config import Config
 
+logging.getLogger('chardet').setLevel(logging.INFO)  # DEBUG is just too noisy
 
-LOGGER = logging.getLogger(__name__)
+LOGGER = logging.getLogger('geometamaker')
+_NOT_FOR_CLI = 'not_for_cli'
+_LOG_EXTRA_NOT_FOR_CLI = {
+    _NOT_FOR_CLI: True
+}
 
 # URI schemes we support. A subset of fsspec.available_protocols()
 PROTOCOLS = [
@@ -31,6 +39,12 @@ PROTOCOLS = [
 DT_FMT = '%Y-%m-%d %H:%M:%S %Z'
 
 
+def _gdal_progress_callback(complete, message, data):
+    percentage = complete * 100
+    if (percentage > 0) & (percentage % 5 == 0):
+        LOGGER.info(f'{message} {percentage}%')
+
+
 # TODO: In the future we can remove these exception managers in favor of the
 # builtin gdal.ExceptionMgr. It was released in 3.7.0 and debugged in 3.9.1.
 # https://github.com/OSGeo/gdal/blob/v3.9.3/NEWS.md#gdalogr-391-release-notes
@@ -103,6 +117,80 @@ def _wkt_to_epsg_units_string(wkt_string):
     return crs_string, units_string
 
 
+def _list_files_with_depth(directory, depth, exclude_regex=None,
+                           exclude_hidden=True):
+    """List files in directory up to depth
+
+    Args:
+        directory (string): path to a directory
+        depth (int): maximum number of subdirectory levels to traverse when
+            walking through a directory. A value of 1 limits the walk to files
+            in the top-level ``directory`` only. A value of 2 allows
+            descending into immediate subdirectories, etc.
+        exclude_regex (str, optional): a regular expression to pattern-match
+            any files for which you do not want to create metadata.
+        exclude_hidden (bool, default True): whether to ignore hidden files
+
+    Returns:
+        list of relative filepaths in ``directory``
+
+    """
+    directory = Path(directory).resolve()
+    file_list = []
+
+    for path in directory.rglob("*"):
+        relative_path = path.relative_to(directory)
+        current_depth = len(relative_path.parts)
+        if current_depth > depth:
+            continue
+        if exclude_hidden and (
+                any(part.startswith('.') for part in relative_path.parts)):
+            continue
+        file_list.append(str(relative_path))
+
+    # remove excluded files based on regex
+    if exclude_regex is not None:
+        file_list = [f for f in file_list if not re.search(exclude_regex, f)]
+
+    return sorted(file_list)
+
+
+def _group_files_by_root(file_list):
+    """Get set of files (roots) and extensions by filename"""
+    root_set = set()
+    root_ext_map = defaultdict(set)
+    for filepath in file_list:
+        root, ext = os.path.splitext(filepath)
+        # tracking which files share a root name
+        # so we can check if these comprise a shapefile
+        root_ext_map[root].add(ext)
+        root_set.add(root)
+    return root_ext_map, sorted(list(root_set))
+
+
+def _get_collection_size_time_uid(directory):
+    """Get size of directory (in bytes), when it was last modified, and uid"""
+    total_bytes = 0
+    latest_mtime = 0
+
+    for root, _, files in os.walk(directory):
+        for file in files:
+            file_path = os.path.join(root, file)
+            stat = os.stat(file_path)
+            total_bytes += stat.st_size
+            latest_mtime = max(latest_mtime, stat.st_mtime)
+
+    last_modified = datetime.fromtimestamp(latest_mtime, tz=timezone.utc)
+    last_modified_str = last_modified.strftime('%Y-%m-%d %H:%M:%S %Z')
+
+    hash_func = hashlib.sha256()
+    hash_func.update(
+        f'{total_bytes}{last_modified_str}{directory}'.encode('utf-8'))
+    uid = f'sizetimestamp:{hash_func.hexdigest()}'
+
+    return total_bytes, last_modified_str, uid
+
+
 def detect_file_type(filepath, scheme):
     """Detect the type of resource contained in the file.
 
@@ -119,13 +207,16 @@ def detect_file_type(filepath, scheme):
     """
     # TODO: guard against classifying netCDF, HDF5, etc as GDAL rasters.
     # We'll likely want a different data model for multi-dimensional arrays.
-
     # Frictionless supports a wide range of formats. The quickest way to
     # determine if a file is recognized as a table or archive is to call list.
-    info = frictionless.list(filepath)[0]
+    try:
+        info = frictionless.list(filepath)[0]
+    except frictionless.FrictionlessException:
+        raise RuntimeError(f'Cannot detect file type of "{filepath}"')
     if info.type == 'table':
         return 'table'
-    if info.compression:
+    # Frictionless doesn't recognize .tgz compression (but does recognize .tar.gz)
+    if info.compression or info.format == "tgz":
         return 'archive'
     # GDAL considers CSV a vector, so check against frictionless first.
     try:
@@ -177,7 +268,7 @@ def describe_file(source_dataset_path, scheme):
     hash_func = hashlib.new('sha256')
     hash_func.update(
         f'{description["bytes"]}{description["last_modified"]}\
-        {description["path"]}'.encode('ascii'))
+        {description["path"]}'.encode('utf-8'))
     description['uid'] = f'sizetimestamp:{hash_func.hexdigest()}'
 
     # We don't have a use for including these attributes in our metadata:
@@ -186,37 +277,64 @@ def describe_file(source_dataset_path, scheme):
     return description
 
 
-def describe_archive(source_dataset_path, scheme):
+def describe_archive(source_dataset_path, scheme, **kwargs):
     """Describe file properties of a compressed file.
 
     Args:
         source_dataset_path (str): path to a file.
         scheme (str): the protocol prefix of the filepath
+        kwargs (dict): additional options when describing a dataset.
 
     Returns:
         dict
 
     """
+    def _list_tgz_contents(path):
+        """List contents of a .tar, .tgz, or .tar.gz archive."""
+        file_list = []
+        with fsspec.open(path, 'rb') as fobj:
+            with tarfile.open(fileobj=fobj, mode='r:*') as tar:
+                file_list = [member.name for member in tar.getmembers()
+                             if member.isfile()]
+        return file_list
+
+    def _list_zip_contents(path):
+        """List contents of a zip archive"""
+        file_list = []
+        ZFS = fsspec.get_filesystem_class('zip')
+        zfs = ZFS(path)
+        for dirpath, _, files in zfs.walk(zfs.root_marker):
+            for f in files:
+                file_list.append(os.path.join(dirpath, f))
+        return file_list
+
     description = describe_file(source_dataset_path, scheme)
     # innerpath is from frictionless and not useful because
     # it does not include all the files contained in the zip
     description.pop('innerpath', None)
 
-    ZFS = fsspec.get_filesystem_class('zip')
-    zfs = ZFS(source_dataset_path)
-    file_list = []
-    for dirpath, _, files in zfs.walk(zfs.root_marker):
-        for f in files:
-            file_list.append(os.path.join(dirpath, f))
+    if description.get("compression") == "zip":
+        file_list = _list_zip_contents(source_dataset_path)
+    elif description.get("format") in ["tgz", "tar"]:
+        file_list = _list_tgz_contents(source_dataset_path)
+        # 'compression' attr not auto-added by frictionless.describe for .tgz
+        # (but IS added for .tar.gz)
+        if source_dataset_path.endswith((".tgz")):
+            description["compression"] = "gz"
+    else:
+        raise ValueError(f"Unsupported archive format: {source_dataset_path}")
+
     description['sources'] = file_list
     return description
 
 
-def describe_vector(source_dataset_path, scheme):
+def describe_vector(source_dataset_path, scheme, **kwargs):
     """Describe properties of a GDAL vector file.
 
     Args:
         source_dataset_path (str): path to a GDAL vector.
+        scheme (str): the protocol prefix of the filepath
+        kwargs (dict): additional options when describing a dataset.
 
     Returns:
         dict
@@ -229,12 +347,18 @@ def describe_vector(source_dataset_path, scheme):
     vector = gdal.OpenEx(source_dataset_path, gdal.OF_VECTOR)
     layer = vector.GetLayer()
     fields = []
-    description['n_features'] = layer.GetFeatureCount()
     for fld in layer.schema:
         fields.append(
             models.FieldSchema(name=fld.name, type=fld.GetTypeName()))
+    layer_schema = models.LayerSchema(
+        name=layer.GetName(),
+        n_features=layer.GetFeatureCount(),
+        table=models.TableSchema(fields=fields),
+        gdal_metadata=layer.GetMetadata())
+    description['data_model'] = models.VectorSchema(
+        layers=[layer_schema],
+        gdal_metadata=vector.GetMetadata())
     vector = layer = None
-    description['data_model'] = models.TableSchema(fields=fields)
 
     info = pygeoprocessing.get_vector_info(source_dataset_path)
     bbox = models.BoundingBox(*info['bounding_box'])
@@ -248,33 +372,65 @@ def describe_vector(source_dataset_path, scheme):
     return description
 
 
-def describe_raster(source_dataset_path, scheme):
+def describe_raster(source_dataset_path, scheme, **kwargs):
     """Describe properties of a GDAL raster file.
 
     Args:
         source_dataset_path (str): path to a GDAL raster.
+        scheme (str): the protocol prefix of the filepath
+        kwargs (dict): additional options when describing a dataset:
+            * ``'compute_stats'`` (bool): whether to compute statistics
+              for each band in the raster. Default is False.
 
     Returns:
         dict
 
     """
+    compute_stats = kwargs.get('compute_stats', False)
     description = describe_file(source_dataset_path, scheme)
     if 'http' in scheme:
         source_dataset_path = f'/vsicurl/{source_dataset_path}'
     info = pygeoprocessing.get_raster_info(source_dataset_path)
+    raster = gdal.OpenEx(source_dataset_path)
+    raster_gdal_metadata = raster.GetMetadata()
     bands = []
     for i in range(info['n_bands']):
         b = i + 1
+        band = raster.GetRasterBand(b)
+        band_gdal_metadata = band.GetMetadata()
+        if compute_stats:
+            try:
+                if not 'STATISTICS_VALID_PERCENT' in band_gdal_metadata:
+                    # Sometimes some stats exist, but not all. If this one doesn't,
+                    # it's important enough that we want to force computation.
+                    _ = band.ComputeStatistics(0, callback=_gdal_progress_callback)
+                else:
+                    # 0=do not approximate stats, 1=calculate if they don't exist
+                    # If exact stats exist they will be retrieved without
+                    # computing them, otherwise, this forces computation.
+                    # https://github.com/OSGeo/gdal/blob/master/gcore/gdalrasterband.cpp
+                    _ = band.GetStatistics(0, 1)
+                band_gdal_metadata = band.GetMetadata()
+            except RuntimeError as e:
+                LOGGER.warning(
+                    f'Could not compute statistics for band {b} of '
+                    f'{source_dataset_path}: {e}')
+
         bands.append(models.BandSchema(
             index=b,
             gdal_type=gdal.GetDataTypeName(info['datatype']),
             numpy_type=numpy.dtype(info['numpy_type']).name,
-            nodata=info['nodata'][i]))
+            nodata=info['nodata'][i],
+            gdal_metadata=band_gdal_metadata))
+        band = None
+    raster = None
+
     description['data_model'] = models.RasterSchema(
         bands=bands,
         pixel_size=info['pixel_size'],
         raster_size={'width': info['raster_size'][0],
-                     'height': info['raster_size'][1]})
+                     'height': info['raster_size'][1]},
+        gdal_metadata=raster_gdal_metadata)
     # Some values of raster info are numpy types, which the
     # yaml dumper doesn't know how to represent.
     bbox = models.BoundingBox(*[float(x) for x in info['bounding_box']])
@@ -288,12 +444,13 @@ def describe_raster(source_dataset_path, scheme):
     return description
 
 
-def describe_table(source_dataset_path, scheme):
+def describe_table(source_dataset_path, scheme, **kwargs):
     """Describe properties of a tabular dataset.
 
     Args:
         source_dataset_path (str): path to a file representing a table.
         scheme (str): the protocol prefix of the filepath
+        kwargs (dict): additional options when describing a dataset.
 
     Returns:
         dict
@@ -305,7 +462,139 @@ def describe_table(source_dataset_path, scheme):
     return description
 
 
-DESRCIBE_FUNCS = {
+def describe_collection(directory, depth=numpy.iinfo(numpy.int16).max,
+                        exclude_regex=None, exclude_hidden=True,
+                        describe_files=False, backup=True, **kwargs):
+    """Create a single metadata document to describe a collection of files.
+
+    Describe all the files within a directory as members of a "collection".
+    The resulting metadata resource should include a list of all the files
+    included in the collection along with a description and metadata filepath
+    (or placeholder). Optionally create individual metadata files for each
+    supported file in a directory.
+
+    Args:
+        directory (str): path to collection
+        depth (int, optional): maximum number of subdirectory levels to
+            traverse when walking through ``directory`` to find files included
+            in the collection. A value of 1 limits the walk to files in the
+            top-level ``directory`` only. A value of 2 allows descending into
+            immediate subdirectories, etc. All files in all subdirectories in
+            the collection will be included by default.
+        exclude_regex (str, optional): a regular expression to pattern-match
+            any files you do not want included in the output metadata yml.
+        exclude_hidden (bool, default True): whether to exclude hidden files
+            (files that start with ".").
+        describe_files (bool, default False): whether to ``describe`` all
+            files, i.e., create individual metadata files for each supported
+            resource in the collection.
+        backup (bool): whether to write a backup of a pre-existing metadata
+            file before ovewriting it in cases where that file is not a valid
+            geometamaker document.
+        kwargs (dict): optional keyward arguments accepted by ``describe``.
+
+    Returns:
+        Collection metadata
+    """
+    directory = str(Path(directory).resolve())
+
+    file_list = _list_files_with_depth(directory, depth, exclude_regex,
+                                       exclude_hidden)
+
+    root_ext_map, root_list = _group_files_by_root(file_list)
+
+    items = []
+
+    for root in root_list:
+        extensions = root_ext_map[root]
+        if '.shp' in extensions:
+            # if we're dealing with a shapefile, we do not want to describe any
+            # of these other files with the same root name
+            extensions.difference_update(['.shx', '.sbn', '.sbx', '.prj', '.dbf', '.cpg'])
+        # Only drop .yml if its sidecar file, i.e. the corresponding data file
+        # (root) exists on disk
+        if '.yml' in extensions and os.path.exists(root):
+            extensions.discard('.yml')
+        for ext in extensions:
+            filepath = os.path.join(directory, f'{root}{ext}')
+            try:
+                this_desc = describe(filepath, **kwargs)
+            except ValueError:
+                # if file type isn't supported by geometamaker, e.g. pdf
+                # or if trying to describe a dir
+                this_desc = None
+
+            if describe_files and this_desc:
+                this_desc.write(backup=backup)
+
+            if ext and os.path.exists(filepath + '.yml'):
+                metadata_yml = f'{root}{ext}' + '.yml'
+            else:
+                metadata_yml = ''
+
+            this_resource = models.CollectionItemSchema(
+                path=f'{root}{ext}',
+                description=this_desc.description if this_desc else '',
+                metadata=metadata_yml
+            )
+            items.append(this_resource)
+
+    total_bytes, last_modified, uid = _get_collection_size_time_uid(directory)
+
+    resource = models.CollectionResource(
+        path=directory,
+        type='collection',
+        format='directory',
+        scheme=fsspec.utils.get_protocol(directory),
+        bytes=total_bytes,
+        last_modified=last_modified,
+        items=items,
+        uid=uid
+    )
+
+    # Check if there is existing metadata for the collection
+    try:
+        metadata_path = f'{directory}-metadata.yml'
+        existing_metadata = models.CollectionResource.load(metadata_path)
+
+        # Copy any existing item descriptions from existing yml to new metadata
+        # Note that descriptions in individual resources' ymls will take
+        # priority over item descriptions from preexisting collection metadata
+        for item in resource.items:
+            # Existing metadata's item desc will overwrite new metadata item
+            # desc if new item desc is ''
+            existing_item_desc = [
+                i.description for i in existing_metadata.items if (
+                    i.path == item.path)]
+            if item.description == '' and len(existing_item_desc) > 0:
+                item.description = existing_item_desc[0]
+
+        # Replace fields in existing yml if new metadata has existing value
+        resource = existing_metadata.replace(resource)
+
+    except (ValueError, ValidationError) as error:
+        LOGGER.warning(error)
+        LOGGER.warning(
+            f'Ignoring an existing YAML document: {metadata_path} because it'
+            f' is invalid or incompatible.')
+        LOGGER.warning(
+            'A subsequent call to `.write()` will replace this file, but it'
+            f' will be backed up to {metadata_path}.bak.\n'
+            f'Use `.write(backup=False)` to skip the backup.\n',
+            extra=_LOG_EXTRA_NOT_FOR_CLI)
+        resource._would_overwrite = True
+
+    except FileNotFoundError:
+        pass
+
+    # Add profile metadata
+    config = Config()
+    resource = resource.replace(config.profile)
+
+    return resource
+
+
+DESCRIBE_FUNCS = {
     'archive': describe_archive,
     'table': describe_table,
     'vector': describe_vector,
@@ -321,7 +610,7 @@ RESOURCE_MODELS = {
 
 
 @_osgeo_use_exceptions
-def describe(source_dataset_path, profile=None):
+def describe(source_dataset_path, compute_stats=False):
     """Create a metadata resource instance with properties of the dataset.
 
     Properties of the dataset are used to populate as many metadata
@@ -331,20 +620,23 @@ def describe(source_dataset_path, profile=None):
     Args:
         source_dataset_path (string): path or URL to dataset to which the
             metadata applies
-        profile (geometamaker.models.Profile): a profile object from
-            which to populate some metadata attributes
+        compute_stats (bool): whether to compute statistics
+            for each band in a raster.
 
     Returns:
         geometamaker.models.Resource: a metadata object
 
     """
-    config = Config()
-    user_profile = config.profile
-    if profile is not None:
-        user_profile = user_profile.replace(profile)
 
     metadata_path = f'{source_dataset_path}.yml'
 
+    if os.path.isdir(source_dataset_path):
+        raise ValueError(
+            f"Cannot `describe` {source_dataset_path} as it is a directory, "
+            "not a dataset. \nIf you are trying to create metadata for the "
+            "files within a directory and/or the directory itself, please use "
+            "`geometamaker.describe_collection` instead.")
+
     # Despite naming, this does not open a file that must be closed
     of = fsspec.open(source_dataset_path)
     if not of.fs.exists(source_dataset_path):
@@ -356,58 +648,63 @@ def describe(source_dataset_path, profile=None):
             f'Cannot describe {source_dataset_path}. {protocol} '
             f'is not one of the suppored file protocols: {PROTOCOLS}')
     resource_type = detect_file_type(source_dataset_path, protocol)
-    description = DESRCIBE_FUNCS[resource_type](
-        source_dataset_path, protocol)
+    description = DESCRIBE_FUNCS[resource_type](
+        source_dataset_path, protocol, compute_stats=compute_stats)
     description['type'] = resource_type
+    resource = RESOURCE_MODELS[resource_type](**description)
 
     # Load existing metadata file
     try:
+        # For the data model, use heuristic to decide if the new resource
+        # should inherit values from the existing resource.
+        # After that, take all non-empty values from the new resource
+        # and update the existing resource.
         existing_resource = RESOURCE_MODELS[resource_type].load(metadata_path)
-        if 'data_model' in description:
-            if isinstance(description['data_model'], models.RasterSchema):
-                # If existing band metadata still matches data_model of the file
-                # carry over existing metadata because it could include
-                # human-defined properties.
-                new_bands = []
-                for band in description['data_model'].bands:
-                    try:
-                        eband = existing_resource.get_band_description(band.index)
-                        # TODO: rewrite this as __eq__ of BandSchema?
-                        if (band.numpy_type, band.gdal_type, band.nodata) == (
-                                eband.numpy_type, eband.gdal_type, eband.nodata):
-                            updated_dict = band.model_dump() | eband.model_dump()
-                            band = models.BandSchema(**updated_dict)
-                    except IndexError:
-                        pass
-                    new_bands.append(band)
-                description['data_model'].bands = new_bands
-            if isinstance(description['data_model'], models.TableSchema):
-                # If existing field metadata still matches data_model of the file
-                # carry over existing metadata because it could include
-                # human-defined properties.
-                new_fields = []
-                for field in description['data_model'].fields:
-                    try:
-                        efield = existing_resource.get_field_description(
-                            field.name)
-                        # TODO: rewrite this as __eq__ of FieldSchema?
-                        if field.type == efield.type:
-                            updated_dict = field.model_dump() | efield.model_dump()
-                            field = models.FieldSchema(**updated_dict)
-                    except KeyError:
-                        pass
-                    new_fields.append(field)
-                description['data_model'].fields = new_fields
-        # overwrite properties that are intrinsic to the dataset
-        updated_dict = existing_resource.model_dump() | description
-        resource = RESOURCE_MODELS[resource_type](**updated_dict)
-
-    # Common path: metadata file does not already exist
-    # Or less common, ValueError if it exists but is incompatible
+        if resource_type == 'raster':
+            for band in resource.data_model.bands:
+                try:
+                    eband = existing_resource.get_band_description(band.index)
+                except IndexError:
+                    continue
+                if (band.numpy_type, band.gdal_type, band.nodata) == (
+                        eband.numpy_type, eband.gdal_type, eband.nodata):
+                    resource.set_band_description(
+                        band.index,
+                        title=eband.title,
+                        description=eband.description,
+                        units=eband.units)
+        if resource_type in ('vector', 'table'):
+            for field in resource._get_fields():
+                try:
+                    efield = existing_resource.get_field_description(field.name)
+                except KeyError:
+                    continue
+                if field.type == efield.type:
+                    resource.set_field_description(
+                        field.name,
+                        title=efield.title,
+                        description=efield.description,
+                        units=efield.units)
+        resource = existing_resource.replace(resource)
+
+    except (ValueError, ValidationError) as error:
+        LOGGER.warning(error)
+        LOGGER.warning(
+            f'Ignoring an existing YAML document: {metadata_path} because it'
+            f' is invalid or incompatible.')
+        LOGGER.warning(
+            'A subsequent call to `.write()` will replace this file, but it'
+            ' will be backed up to {metadata_path}.bak.\n'
+            f'Use `.write(backup=False)` to skip the backup.\n',
+            extra=_LOG_EXTRA_NOT_FOR_CLI)
+        resource._would_overwrite = True
+
     except FileNotFoundError:
-        resource = RESOURCE_MODELS[resource_type](**description)
+        # Common path: metadata file does not already exist
+        pass
 
-    resource = resource.replace(user_profile)
+    config = Config()
+    resource = resource.replace(config.profile)
     return resource
 
 
@@ -442,30 +739,20 @@ def validate(filepath):
         return error
 
 
-def validate_dir(directory, recursive=False):
+def validate_dir(directory, depth=numpy.iinfo(numpy.int16).max):
     """Validate all compatible yml documents in the directory.
 
     Args:
         directory (string): path to a directory
-        recursive (bool): whether or not to describe files
-            in all subdirectories
+        depth (int): maximum number of subdirectory levels to
+            traverse when walking through ``directory``.
 
     Returns:
         tuple (list, list): a list of the filepaths that were validated and
             an equal-length list of the validation messages.
 
     """
-    file_list = []
-    if recursive:
-        for path, dirs, files in os.walk(directory):
-            for file in files:
-                file_list.append(os.path.join(path, file))
-    else:
-        file_list.extend(
-            [os.path.join(directory, path)
-                for path in os.listdir(directory)
-                if os.path.isfile(os.path.join(directory, path))])
-
+    file_list = _list_files_with_depth(directory, depth)
     messages = []
     yaml_files = []
     for filepath in file_list:
@@ -473,7 +760,7 @@ def validate_dir(directory, recursive=False):
             yaml_files.append(filepath)
             msg = ''
             try:
-                error = validate(filepath)
+                error = validate(os.path.join(directory, filepath))
                 if error:
                     msg = error
             except ValueError:
@@ -484,48 +771,3 @@ def validate_dir(directory, recursive=False):
             messages.append(msg)
 
     return (yaml_files, messages)
-
-
-def describe_dir(directory, recursive=False):
-    """Describe all compatible datasets in the directory.
-
-    Take special care to only describe multifile datasets,
-    such as ESRI Shapefiles, one time.
-
-    Args:
-        directory (string): path to a directory
-        recursive (bool): whether or not to describe files
-            in all subdirectories
-
-    Returns:
-        None
-
-    """
-    root_set = set()
-    root_ext_map = defaultdict(set)
-    for path, dirs, files in os.walk(directory):
-        for file in files:
-            full_path = os.path.join(path, file)
-            root, ext = os.path.splitext(full_path)
-            # tracking which files share a root name
-            # so we can check if these comprise a shapefile
-            root_ext_map[root].add(ext)
-            root_set.add(root)
-        if not recursive:
-            break
-
-    for root in root_set:
-        extensions = root_ext_map[root]
-        if '.shp' in extensions:
-            # if we're dealing with a shapefile, we do not want to describe any
-            # of these other files with the same root name
-            extensions.difference_update(['.shx', '.sbn', '.sbx', '.prj', '.dbf'])
-        for ext in extensions:
-            filepath = f'{root}{ext}'
-            try:
-                resource = describe(filepath)
-            except ValueError as error:
-                LOGGER.debug(error)
-                continue
-            resource.write()
-            LOGGER.info(f'{filepath} described')