megadetector 5.0.24__py3-none-any.whl → 5.0.25__py3-none-any.whl

megadetector/utils/wi_utils.py

@@ -27,8 +27,12 @@ from tqdm import tqdm
 
 from megadetector.utils.path_utils import insert_before_extension
 from megadetector.utils.ct_utils import split_list_into_n_chunks
+from megadetector.utils.ct_utils import round_floats_in_nested_dict
+from megadetector.utils.ct_utils import is_list_sorted
 from megadetector.utils.ct_utils import invert_dictionary
 from megadetector.utils.ct_utils import sort_list_of_dicts_by_key
+from megadetector.utils.ct_utils import sort_dictionary_by_value
+from megadetector.utils.ct_utils import sort_dictionary_by_key
 from megadetector.utils.path_utils import find_images
 from megadetector.postprocessing.validate_batch_results import \
     validate_batch_results, ValidateBatchResultsOptions
@@ -58,10 +62,28 @@ def is_valid_prediction_string(s):
     Returns:
         bool: True if this looks more or less like a WI prediction string
     """
-    
+
+    # Note to self... don't get tempted to remove spaces here; spaces are used
+    # to indicate subspecies.
     return isinstance(s,str) and (len(s.split(';')) == 7) and (s == s.lower())
 
 
+def is_valid_taxonomy_string(s):
+    """
+    Determine whether [s] is a valid 5-token WI taxonomy string.  Taxonmy strings look like:
+    
+    'mammalia;rodentia;;;;rodent'
+    'mammalia;chordata;canidae;canis;lupus dingo'
+    
+    Args:
+        s (str): the string to be tested for validity
+        
+    Returns:
+        bool: True if this looks more or less like a WI taxonomy string
+    """
+    return isinstance(s,str) and (len(s.split(';')) == 5) and (s == s.lower())
+
+    
 def wi_result_to_prediction_string(r):
     """
     Convert the dict [r] - typically loaded from a row in a downloaded .csv file - to
@@ -469,10 +491,14 @@ sample_update_payload = {
 
 blank_prediction_string = 'f1856211-cfb7-4a5b-9158-c0f72fd09ee6;;;;;;blank'
 no_cv_result_prediction_string = 'f2efdae9-efb8-48fb-8a91-eccf79ab4ffb;no cv result;no cv result;no cv result;no cv result;no cv result;no cv result'
-rodent_prediction_string = '90d950db-2106-4bd9-a4c1-777604c3eada;mammalia;rodentia;;;;rodent'
-mammal_prediction_string = 'f2d233e3-80e3-433d-9687-e29ecc7a467a;mammalia;;;;;mammal'
 animal_prediction_string = '1f689929-883d-4dae-958c-3d57ab5b6c16;;;;;;animal'
 human_prediction_string = '990ae9dd-7a59-4344-afcb-1b7b21368000;mammalia;primates;hominidae;homo;sapiens;human'
+vehicle_prediction_string = 'e2895ed5-780b-48f6-8a11-9e27cb594511;;;;;;vehicle'
+
+non_taxonomic_prediction_strings = [blank_prediction_string,
+                                    no_cv_result_prediction_string,
+                                    animal_prediction_string,
+                                    vehicle_prediction_string]
 
 process_cv_response_url = 'https://placeholder'
 
@@ -870,6 +896,7 @@ def get_kingdom(prediction_string):
         str: the kingdom field from the input string
     """
     tokens = prediction_string.split(';')
+    assert is_valid_prediction_string(prediction_string)
     return tokens[1]
 
 
@@ -912,7 +939,10 @@ def is_animal_classification(prediction_string):
     return True
 
 
-def generate_md_results_from_predictions_json(predictions_json_file,md_results_file,base_folder=None):
+def generate_md_results_from_predictions_json(predictions_json_file,
+                                              md_results_file,
+                                              base_folder=None,
+                                              max_decimals=5):
     """
     Generate an MD-formatted .json file from a predictions.json file.  Typically,
     MD results files use relative paths, and predictions.json files use absolute paths, so 
@@ -921,21 +951,38 @@ def generate_md_results_from_predictions_json(predictions_json_file,md_results_f
     Currently just applies the top classification category to every detection.  If the top classification 
     is "blank", writes an empty detection list.
     
-    wi_to_md.py is a command-line driver for this function.
+    speciesnet_to_md.py is a command-line driver for this function.
     
     Args:
-        predictions_json_file (str): path to a predictions.json file
+        predictions_json_file (str): path to a predictions.json file, or a dict
         md_results_file (str): path to which we should write an MD-formatted .json file
-        base_folder (str, optional): leading string to remove from each path in the predictions.json file
+        base_folder (str, optional): leading string to remove from each path in the 
+            predictions.json file
+        max_decimals (int, optional): number of decimal places to which we should round
+            all values
     """
         
     # Read predictions file    
-    with open(predictions_json_file,'r') as f:
-        predictions = json.load(f)
+    if isinstance(predictions_json_file,str):
+        with open(predictions_json_file,'r') as f:
+            predictions = json.load(f)
+    else:
+        assert isinstance(predictions_json_file,dict)
+        predictions = predictions_json_file
+    
+    # Round floating-point values (confidence scores, coordinates) to a
+    # reasonable number of decimal places
+    if max_decimals is not None and max_decimals > 0:
+        round_floats_in_nested_dict(predictions)
+        
     predictions = predictions['predictions']
     assert isinstance(predictions,list)
     
-    from megadetector.utils.ct_utils import is_list_sorted
+    # Convert backslashes to forward slashes in both filenames and the base folder string
+    for im in predictions:
+        im['filepath'] = im['filepath'].replace('\\','/')
+    if base_folder is not None:
+        base_folder = base_folder.replace('\\','/')
     
     detection_category_id_to_name = {}
     classification_category_name_to_id = {}
@@ -948,6 +995,8 @@ def generate_md_results_from_predictions_json(predictions_json_file,md_results_f
     # Create the output images list
     images_out = []
     
+    base_folder_replacements = 0
+    
     # im_in = predictions[0]
     for im_in in predictions:
         
@@ -957,6 +1006,7 @@ def generate_md_results_from_predictions_json(predictions_json_file,md_results_f
         fn = im_in['filepath']
         if base_folder is not None:
             if fn.startswith(base_folder):
+                base_folder_replacements += 1
                 fn = fn.replace(base_folder,'',1)
         
         im_out['file'] = fn
@@ -1056,6 +1106,11 @@ def generate_md_results_from_predictions_json(predictions_json_file,md_results_f
         
     # ...for each image
     
+    if base_folder is not None:
+        if base_folder_replacements == 0:
+            print('Warning: you supplied {} as the base folder, but I made zero replacements'.format(
+                base_folder))
+            
     # Fix the 'unknown' category
     
     if len(all_unknown_detections) > 0:
@@ -1075,7 +1130,8 @@ def generate_md_results_from_predictions_json(predictions_json_file,md_results_f
     
     # Prepare friendly classification names
     
-    classification_category_descriptions = invert_dictionary(classification_category_name_to_id)
+    classification_category_descriptions = \
+        invert_dictionary(classification_category_name_to_id)
     classification_categories_out = {}
     for category_id in classification_category_descriptions.keys():
         category_name = classification_category_descriptions[category_id].split(';')[-1]
@@ -1105,7 +1161,9 @@ def generate_md_results_from_predictions_json(predictions_json_file,md_results_f
 # ...def generate_md_results_from_predictions_json(...)
         
 
-def generate_predictions_json_from_md_results(md_results_file,predictions_json_file,base_folder=None):
+def generate_predictions_json_from_md_results(md_results_file,
+                                              predictions_json_file,
+                                              base_folder=None):
     """
     Generate a predictions.json file from the MD-formatted .json file [md_results_file].  Typically,
     MD results files use relative paths, and predictions.json files use absolute paths, so 
@@ -1165,13 +1223,16 @@ def generate_predictions_json_from_md_results(md_results_file,predictions_json_f
         
 # ...def generate_predictions_json_from_md_results(...)
 
+default_tokens_to_ignore = ['$RECYCLE.BIN']
 
 def generate_instances_json_from_folder(folder,
                                         country=None,
+                                        admin1_region=None,
                                         lat=None,
                                         lon=None,
                                         output_file=None,
-                                        filename_replacements=None):
+                                        filename_replacements=None,
+                                        tokens_to_ignore=default_tokens_to_ignore):
     """
     Generate an instances.json record that contains all images in [folder], optionally
     including location information, in a format suitable for run_model.py.  Optionally writes
@@ -1186,6 +1247,8 @@ def generate_instances_json_from_folder(folder,
         filename_replacements (dict, optional): str --> str dict indicating filename substrings
             that should be replaced with other strings.  Replacement occurs *after* converting
             backslashes to forward slashes.
+        tokens_to_ignore (list, optional): ignore any images with these tokens in their
+            names, typically used to avoid $RECYCLE.BIN.  Can be None.
         
     Returns:
         dict: dict with at least the field "instances"
@@ -1195,6 +1258,13 @@ def generate_instances_json_from_folder(folder,
         
     image_files_abs = find_images(folder,recursive=True,return_relative_paths=False)
     
+    if tokens_to_ignore is not None:
+        n_images_before_ignore_tokens = len(image_files_abs)
+        for token in tokens_to_ignore:
+            image_files_abs = [fn for fn in image_files_abs if token not in fn]
+        print('After ignoring {} tokens, kept {} of {} images'.format(
+            len(tokens_to_ignore),len(image_files_abs),n_images_before_ignore_tokens))
+    
     instances = []
     
     # image_fn_abs = image_files_abs[0]
@@ -1206,6 +1276,8 @@ def generate_instances_json_from_folder(folder,
                 instance['filepath'] = instance['filepath'].replace(s,filename_replacements[s])
         if country is not None:
             instance['country'] = country
+        if admin1_region is not None:
+            instance['admin1_region'] = admin1_region
         if lat is not None:
             assert lon is not None, 'Latitude provided without longitude'
             instance['latitude'] = lat
@@ -1226,14 +1298,243 @@ def generate_instances_json_from_folder(folder,
 # ...def generate_instances_json_from_folder(...)
 
 
-#%% Functions related to geofencing and taxonomy mapping
+def split_instances_into_n_batches(instances_json,n_batches,output_files=None):
+    """
+    Given an instances.json file, split it into batches of equal size.
+    
+    Args:
+        instances_json (str): input .json file in
+        n_batches (int): number of new files to generate
+        output_files (list, optional): output .json files for each
+            batch.  If supplied, should have length [n_batches].  If not 
+            supplied, filenames will be generated based on [instances_json].
+            
+    Returns:
+        list: list of output files that were written; identical to [output_files]
+        if it was supplied as input.
+    """
+    
+    with open(instances_json,'r') as f:
+        instances = json.load(f)
+    assert isinstance(instances,dict) and 'instances' in instances
+    instances = instances['instances']
+    
+    if output_files is not None:
+        assert len(output_files) == n_batches, \
+            'Expected {} output files, received {}'.format(
+                n_batches,len(output_files))
+    else:
+        output_files = []
+        for i_batch in range(0,n_batches):
+            batch_string = 'batch_{}'.format(str(i_batch).zfill(3))
+            output_files.append(insert_before_extension(instances_json,batch_string))
+                    
+    batches = split_list_into_n_chunks(instances, n_batches)
+    
+    for i_batch,batch in enumerate(batches):
+        batch_dict = {'instances':batch}
+        with open(output_files[i_batch],'w') as f:
+            json.dump(batch_dict,f,indent=1)
+    
+    print('Wrote {} batches to file'.format(n_batches))
+    
+    return output_files
+    
+    
+def merge_prediction_json_files(input_prediction_files,output_prediction_file):
+    """
+    Merge all predictions.json files in [files] into a single .json file.
+    
+    Args:
+        files (list): list of predictions.json files to merge
+        output_file (str): output .json file        
+    """
+    
+    predictions = []
+    image_filenames_processed = set()
+    
+    # input_json_fn = input_prediction_files[0]
+    for input_json_fn in tqdm(input_prediction_files):
+        
+        assert os.path.isfile(input_json_fn), \
+            'Could not find prediction file {}'.format(input_json_fn)            
+        with open(input_json_fn,'r') as f:
+            results_this_file = json.load(f)
+        assert isinstance(results_this_file,dict)
+        predictions_this_file = results_this_file['predictions']
+        for prediction in predictions_this_file:
+            image_fn = prediction['filepath']
+            assert image_fn not in image_filenames_processed
+        predictions.extend(predictions_this_file)
+        
+    output_dict = {'predictions':predictions}
+    
+    os.makedirs(os.path.dirname(output_prediction_file),exist_ok=True)
+    with open(output_prediction_file,'w') as f:
+        json.dump(output_dict,f,indent=1)
+
+# ...def merge_prediction_json_files(...)    
+
+
+def validate_predictions_file(fn,instances=None,verbose=True):
+    """
+    Validate the predictions.json file [fn].
+    
+    Args:
+        fn (str): a .json file in predictions.json (SpeciesNet) format
+        instances (str or list, optional): a folder, instances.json file,
+            or dict loaded from an instances.json file.  If supplied, this
+            function will verify that [fn] contains the same number of 
+            images as [instances].
+        verbose (bool, optional): enable additional debug output
+        
+    Returns:
+        dict: the contents of [fn]        
+    """
+    
+    with open(fn,'r') as f:
+        d = json.load(f)
+    predictions = d['predictions']
+        
+    failures = []
+
+    for im in predictions:
+        if 'failures' in im:
+            failures.append(im)
+
+    if verbose:
+        print('Read detector results for {} images, with {} failure(s)'.format(
+            len(d['predictions']),len(failures)))
+        
+    if instances is not None:
+        
+        if isinstance(instances,str):
+            if os.path.isdir(instances):
+                instances = generate_instances_json_from_folder(folder=instances)
+            elif os.path.isfile(instances):
+                with open(instances,'r') as f:
+                    instances = json.load(f)
+            else:
+                raise ValueError('Could not find instances file/folder {}'.format(
+                    instances))
+        assert isinstance(instances,dict)
+        assert 'instances' in instances
+        instances = instances['instances']
+        if verbose:
+            print('Expected results for {} files'.format(len(instances)))
+        assert len(instances) == len(predictions), \
+            '{} instances expected, {} found'.format(
+                len(instances),len(predictions))
+        
+        expected_files = set([instance['filepath'] for instance in instances])
+        found_files = set([prediction['filepath'] for prediction in predictions])
+        assert expected_files == found_files
+        
+    # ...if a list of instances was supplied
+
+    return d
+
+# ...def validate_predictions_file(...)        
+
+
+def find_geofence_adjustments(ensemble_json_file,use_latin_names=False):
+    """
+    Count the number of instances of each unique change made by the geofence.
+    
+    Args:
+        ensemble_json_file (str): SpeciesNet-formatted .json file produced
+            by the full ensemble.
+        use_latin_names (bool, optional): return a mapping using binomial names
+            rather than common names.
+    
+    Returns:
+        dict: maps strings that look like "puma,felidae family" to integers, 
+            where that entry would indicate the number of times that "puma" was 
+            predicted, but mapped to family level by the geofence.  Sorted in 
+            descending order by count.
+    """
+    
+    ensemble_results = validate_predictions_file(ensemble_json_file)
+    
+    assert isinstance(ensemble_results,dict)
+    predictions = ensemble_results['predictions']
+    
+    # Maps comma-separated pairs of common names (or binomial names) to
+    # the number of times that transition (first --> second) happened
+    rollup_pair_to_count = defaultdict(int)
+    
+    # prediction = predictions[0]
+    for prediction in tqdm(predictions):
+        
+        if 'failures' in prediction and \
+            prediction['failures'] is not None and \
+            len(prediction['failures']) > 0:
+                continue
+            
+        assert 'prediction_source' in prediction, \
+            'Prediction present without [prediction_source] field, are you sure this ' + \
+            'is an ensemble output file?'
+        
+        if 'geofence' in prediction['prediction_source']:
+            
+            classification_taxonomy_string = \
+                prediction['classifications']['classes'][0]
+            prediction_taxonomy_string = prediction['prediction']
+            assert is_valid_prediction_string(classification_taxonomy_string)
+            assert is_valid_prediction_string(prediction_taxonomy_string)
+
+            # Typical examples:
+            # '86f5b978-4f30-40cc-bd08-be9e3fba27a0;mammalia;rodentia;sciuridae;sciurus;carolinensis;eastern gray squirrel'
+            # 'e4d1e892-0e4b-475a-a8ac-b5c3502e0d55;mammalia;rodentia;sciuridae;;;sciuridae family'
+            classification_common_name = classification_taxonomy_string.split(';')[-1]
+            prediction_common_name = prediction_taxonomy_string.split(';')[-1]
+            classification_binomial_name = classification_taxonomy_string.split(';')[-2]
+            prediction_binomial_name = prediction_taxonomy_string.split(';')[-2]
+            
+            input_name = classification_binomial_name if use_latin_names else \
+                classification_common_name
+            output_name = prediction_binomial_name if use_latin_names else \
+                prediction_common_name
+                
+            rollup_pair = input_name.strip() + ',' + output_name.strip()
+            rollup_pair_to_count[rollup_pair] += 1
+            
+        # ...if we made a geofencing change
+        
+    # ...for each prediction
+           
+    rollup_pair_to_count = sort_dictionary_by_value(rollup_pair_to_count,reverse=True)
+    
+    return rollup_pair_to_count
+
+# ...def find_geofence_adjustments(...)    
+
+
+#%% Module-level globals related to taxonomy mapping and geofencing    
 
 # This maps a taxonomy string (e.g. mammalia;cetartiodactyla;cervidae;odocoileus;virginianus) to 
 # a dict with keys taxon_id, common_name, kingdom, phylum, class, order, family, genus, species
 taxonomy_string_to_taxonomy_info = None
+
+# Maps a binomial name (possibly three tokens, if it's a subspecies) to the same dict
+# described above.
 binomial_name_to_taxonomy_info = None
+
+# Maps a common name to the same dict described above
 common_name_to_taxonomy_info = None
 
+# Dict mapping 5-token semicolon-delimited taxonomy strings to geofencing rules
+taxonomy_string_to_geofencing_rules = None
+
+# Maps lower-case country names to upper-case country codes
+country_to_country_code = None
+
+# Maps upper-case country codes to lower-case country names
+country_code_to_country = None
+
+
+#%% Functions related to geofencing and taxonomy mapping
+
 def taxonomy_info_to_taxonomy_string(taxonomy_info):
     """
     Convert a taxonomy record in dict format to a semicolon-delimited string
@@ -1258,12 +1559,16 @@ def initialize_taxonomy_info(taxonomy_file,force_init=False,encoding='cp1252'):
     [common_name_to_taxonomy_info].
     
     Args:
-        taxonomy_file (str): .json file containing WI taxonomy information
+        taxonomy_file (str): .json file containing mappings from the short taxonomy strings 
+            to the longer strings with GUID and common name, see example below.            
         force_init (bool, optional): if the output dicts already exist, should we
             re-initialize anyway?
         encoding (str, optional): character encoding to use when opening the .json file
     """
     
+    if encoding is None:
+        encoding = 'cp1252'
+        
     global taxonomy_string_to_taxonomy_info
     global binomial_name_to_taxonomy_info
     global common_name_to_taxonomy_info
@@ -1326,22 +1631,159 @@ def initialize_taxonomy_info(taxonomy_file,force_init=False,encoding='cp1252'):
             # print('Warning: no binomial name for {}'.format(taxonomy_string))
             pass
         else:
+            # strip(), but don't remove spaces from the species name; 
+            # subspecies are separated with a space, e.g. canis;lupus dingo
             binomial_name = tokens[4].strip() + ' ' + tokens[5].strip()
             binomial_name_to_taxonomy_info[binomial_name] = taxon_info
+    
+    print('Created {} records in taxonomy_string_to_taxonomy_info'.format(
+        len(taxonomy_string_to_taxonomy_info)))
         
 # ...def initialize_taxonomy_info(...)
 
 
-#%% Geofencing functions
-
-# Dict mapping semicolon-delimited taxonomy strings to geofencing rules
-taxonomy_string_to_geofencing_rules = None
+def _parse_code_list(codes):
+    """
+    Turn a list of country or state codes in string, delimited string, or list format
+    into a list.  Also does basic validity checking.
+    """
+    
+    if not isinstance(codes,list):
+        
+        assert isinstance(codes,str)
+        
+        codes = codes.strip()
+        
+        # This is just a single codes
+        if ',' not in codes:
+            codes = [codes]
+        else:
+            codes = codes.split(',')
+        codes = [c.strip() for c in codes]
+        
+    assert isinstance(codes,list)
+    
+    codes = [c.upper().strip() for c in codes]
+    
+    for c in codes:
+        assert len(c) in (2,3)
+        
+    return codes
+        
+        
+def _generate_csv_rows_to_block_all_countries_except(
+        species_string,
+        block_except_list):
+    """
+    Generate rows in the format expected by geofence_fixes.csv, representing a list of
+    allow and block rules to block all countries currently allowed for this species
+    except [allow_countries], and add allow rules these countries.
+    """
+    
+    assert is_valid_taxonomy_string(species_string), \
+        '{} is not a valid taxonomy string'.format(species_string)
+        
+    global taxonomy_string_to_taxonomy_info
+    global binomial_name_to_taxonomy_info
+    global common_name_to_taxonomy_info
+    
+    assert taxonomy_string_to_geofencing_rules is not None, \
+        'Initialize geofencing prior to species lookup'
+    assert taxonomy_string_to_taxonomy_info is not None, \
+        'Initialize taxonomy lookup prior to species lookup'
+        
+    geofencing_rules_this_species = \
+        taxonomy_string_to_geofencing_rules[species_string]
+        
+    allowed_countries = []
+    if 'allow' in geofencing_rules_this_species:
+        allowed_countries.extend(geofencing_rules_this_species['allow'])
+        
+    blocked_countries = []
+    if 'block' in geofencing_rules_this_species:
+        blocked_countries.extend(geofencing_rules_this_species['block'])
+        
+    block_except_list = _parse_code_list(block_except_list)
+    
+    countries_to_block = []
+    countries_to_allow = []
+    
+    # country = allowed_countries[0]
+    for country in allowed_countries:
+        if country not in block_except_list and country not in blocked_countries:
+            countries_to_block.append(country)
+    
+    for country in block_except_list:
+        if country in blocked_countries:
+            raise ValueError("I can't allow a country that has already been blocked")
+        if country not in allowed_countries:
+            countries_to_allow.append(country)
+            
+    rows = generate_csv_rows_for_species(species_string,
+                                         allow_countries=countries_to_allow,
+                                         block_countries=countries_to_block)
+    
+    return rows
+   
+# ...def _generate_csv_rows_to_block_all_countries_except(...)
+    
+    
+def generate_csv_rows_for_species(species_string,
+                                  allow_countries=None,
+                                  block_countries=None,
+                                  allow_states=None,
+                                  block_states=None,
+                                  blockexcept_countries=None):
+    """
+    Generate rows in the format expected by geofence_fixes.csv, representing a list of
+    allow and/or block rules for the specified species and countries/states.  Does not check 
+    that the rules make sense; e.g. nothing will stop you in this function from both allowing
+    and blocking a country.
+    
+    Args:
+        species_string (str): string in semicolon-delimited WI taxonomy format
+        allow_countries (optional, list or str): three-letter country codes, list of
+            country codes, or comma-separated list of country codes to allow
+        block_countries (optional, list or str): three-letter country codes, list of
+            country codes, or comma-separated list of country codes to block
+        allow_states (optional, list or str): two-letter state codes, list of
+            state codes, or comma-separated list of state codes to allow
+        block_states (optional, list or str): two-letter state code, list of
+            state codes, or comma-separated list of state codes to block        
+            
+    Returns:
+        list of str: lines ready to be pasted into geofence_fixes.csv
+    """
+    
+    assert is_valid_taxonomy_string(species_string), \
+        '{} is not a valid taxonomy string'.format(species_string)
+    
+    lines = []
+    
+    if allow_countries is not None:
+        allow_countries = _parse_code_list(allow_countries)
+        for country in allow_countries:
+            lines.append(species_string + ',allow,' + country + ',')
+    
+    if block_countries is not None:
+        block_countries = _parse_code_list(block_countries)
+        for country in block_countries:
+            lines.append(species_string + ',block,' + country + ',')
+            
+    if allow_states is not None:
+        allow_states = _parse_code_list(allow_states)
+        for state in allow_states:
+            lines.append(species_string + ',allow,USA,' + state)
+    
+    if block_states is not None:
+        block_states = _parse_code_list(block_states)
+        for state in block_states:
+            lines.append(species_string + ',block,USA,' + state)
+         
+    return lines
 
-# Maps lower-case country names to upper-case country codes
-country_to_country_code = None
+# ...def generate_csv_rows_for_species(...)
 
-# Maps upper-case country codes to lower-case country names
-country_code_to_country = None
 
 def initialize_geofencing(geofencing_file,country_code_file,force_init=False):
     """
@@ -1351,10 +1793,13 @@ def initialize_geofencing(geofencing_file,country_code_file,force_init=False):
     
     Args:
         geofencing_file (str): .json file with geofencing rules
-        country_code_file (str): .csv file with country code mappings
+        country_code_file (str): .csv file with country code mappings, in columns 
+            called "name" and "alpha-3", e.g. from
+            https://github.com/lukes/ISO-3166-Countries-with-Regional-Codes/blob/master/all/all.csv
         force_init (bool, optional): if the output dicts already exist, should we
             re-initialize anyway?
     """
+    
     global taxonomy_string_to_geofencing_rules
     global country_to_country_code
     global country_code_to_country
@@ -1427,35 +1872,21 @@ def initialize_geofencing(geofencing_file,country_code_file,force_init=False):
 # ...def initialize_geofencing(...)
     
 
-def species_allowed_in_country(species,country,state=None,return_status=False):
+def _species_string_to_canonical_species_string(species):
+    """
+    Convert a string that may be a 5-token species string, a binomial name,
+    or a common name into a 5-token species string.
     """
-    Determines whether [species] is allowed in [country], according to 
-    already-initialized geofencing rules.
     
-    Args:
-        species (str): can be a common name, a binomial name, or a species string
-        country (str): country name or three-letter code
-        state (str, optional): two-letter US state code
-        return_status (bool, optional): by default, this function returns a bool;
-            if you want to know *why* [species] is allowed/not allowed, settings
-            return_status to True will return additional information.        
+    global taxonomy_string_to_taxonomy_info
+    global binomial_name_to_taxonomy_info
+    global common_name_to_taxonomy_info
     
-    Returns:
-        bool or str: typically returns True if [species] is allowed in [country], else
-        False.  Returns a more detailed string if return_status is set.
-    """
-
     assert taxonomy_string_to_geofencing_rules is not None, \
         'Initialize geofencing prior to species lookup'
     assert taxonomy_string_to_taxonomy_info is not None, \
         'Initialize taxonomy lookup prior to species lookup'
-        
-    # species = 'mammalia;cetartiodactyla;cervidae;odocoileus;virginianus'
-    # species = 'didelphis marsupialis'
-    # country = 'Guatemala'
-    
-    # species = 'common opossum'
-    
+            
     species = species.lower()
     
     # Turn "species" into a taxonomy string
@@ -1463,8 +1894,8 @@ def species_allowed_in_country(species,country,state=None,return_status=False):
     # If this is already a taxonomy string...    
     if len(species.split(';')) == 5:
         pass
-    # If this is a binomial name...
-    elif len(species.split(' ')) == 2 and (species in binomial_name_to_taxonomy_info):
+    # If this is a binomial name (which may include a subspecies)...
+    elif (len(species.split(' ')) in (2,3)) and (species in binomial_name_to_taxonomy_info):
         taxonomy_info = binomial_name_to_taxonomy_info[species]
         taxonomy_string = taxonomy_info_to_taxonomy_string(taxonomy_info)        
     # If this is a common name...
@@ -1474,6 +1905,37 @@ def species_allowed_in_country(species,country,state=None,return_status=False):
     else:
         raise ValueError('Could not find taxonomic information for {}'.format(species))
         
+    return taxonomy_string
+            
+
+def species_allowed_in_country(species,country,state=None,return_status=False):
+    """
+    Determines whether [species] is allowed in [country], according to 
+    already-initialized geofencing rules.
+    
+    Args:
+        species (str): can be a common name, a binomial name, or a species string
+        country (str): country name or three-letter code
+        state (str, optional): two-letter US state code
+        return_status (bool, optional): by default, this function returns a bool;
+            if you want to know *why* [species] is allowed/not allowed, settings
+            return_status to True will return additional information.        
+    
+    Returns:
+        bool or str: typically returns True if [species] is allowed in [country], else
+        False.  Returns a more detailed string if return_status is set.
+    """
+
+    global taxonomy_string_to_taxonomy_info
+    global binomial_name_to_taxonomy_info
+    global common_name_to_taxonomy_info
+    
+    assert taxonomy_string_to_geofencing_rules is not None, \
+        'Initialize geofencing prior to species lookup'
+    assert taxonomy_string_to_taxonomy_info is not None, \
+        'Initialize taxonomy lookup prior to species lookup'
+        
+    taxonomy_string = _species_string_to_canonical_species_string(species)
     
     # Normalize [state]
     
@@ -1515,6 +1977,8 @@ def species_allowed_in_country(species,country,state=None,return_status=False):
         blocked_countries = list(geofencing_rules_this_species['block'])
     
     status = None
+    
+    # The convention is that block rules win over allow rules
     if country_code in blocked_countries:
         status = 'blocked'
     elif country_code in allowed_countries:
@@ -1565,12 +2029,459 @@ def species_allowed_in_country(species,country,state=None,return_status=False):
 # ...def species_allowed_in_country(...)
 
 
+def restrict_to_taxa_list(taxa_list,
+                          speciesnet_taxonomy_file,
+                          input_file,
+                          output_file,
+                          allow_walk_down=False):
+    """
+    Given a prediction file in MD .json format, likely without having had
+    a geofence applied, apply a custom taxa list.
+    
+    Args:
+        taxa_list (str or list): list of latin names, or a text file containing
+            a list of latin names.  Optionally may contain a second (comma-delimited)
+            column containing common names, used only for debugging.  Latin names
+            must exist in the SpeciesNet taxonomy.
+        taxonomy_file (str): taxonomy filename, in the same format used for model
+            release (with 7-token taxonomy entries)            
+        output_file (str): .json file to write, in MD format      
+        allow_walk_down (bool, optional): should we walk down the taxonomy tree
+            when making mappings if a parent has only a single allowable child?
+            For example, if only a single felid species is allowed, should other
+            felid predictions be mapped to that species, as opposed to being mapped
+            to the family?
+    """
+
+    ##%% Read target taxa list
+    
+    if isinstance(taxa_list,str):
+        assert os.path.isfile(taxa_list), \
+            'Could not find taxa list file {}'.format(taxa_list)
+        with open(taxa_list,'r') as f:
+            taxa_list = f.readlines()
+    
+    taxa_list = [s.strip().lower() for s in taxa_list]
+    taxa_list = [s for s in taxa_list if len(s) > 0]
+    
+    target_latin_to_common = {}
+    for s in taxa_list:
+        if s.strip().startswith('#'):
+            continue
+        tokens = s.split(',')
+        assert len(tokens) <= 2
+        binomial_name = tokens[0]
+        assert len(binomial_name.split(' ')) in (1,2,3), \
+            'Illegal binomial name in species list: {}'.format(binomial_name)
+        if len(tokens) > 0:
+            common_name = tokens[1].strip().lower()
+        else:
+            common_name = None
+        assert binomial_name not in target_latin_to_common
+        target_latin_to_common[binomial_name] = common_name
+    
+
+    ##%% Read taxonomy file
+    
+    with open(speciesnet_taxonomy_file,'r') as f:
+        speciesnet_taxonomy_list = f.readlines()
+    speciesnet_taxonomy_list = [s.strip() for s in \
+                                speciesnet_taxonomy_list if len(s.strip()) > 0]
+    
+    # Maps the latin name of every taxon to the corresponding full taxon string
+    #
+    # For species, the key is a binomial name
+    speciesnet_latin_name_to_taxon_string = {}
+    speciesnet_common_name_to_taxon_string = {}
+    
+    def _insert_taxonomy_string(s):
+        
+        tokens = s.split(';')
+        assert len(tokens) == 7
+        
+        guid = tokens[0] # noqa
+        class_name = tokens[1]
+        order = tokens[2]
+        family = tokens[3]
+        genus = tokens[4]
+        species = tokens[5]        
+        common_name = tokens[6]
+        
+        if len(class_name) == 0:
+            assert common_name in ('animal','vehicle','blank')
+            return
+        
+        if len(species) > 0:
+            assert all([len(s) > 0 for s in [genus,family,order]])
+            binomial_name = genus + ' ' + species
+            if binomial_name not in speciesnet_latin_name_to_taxon_string:
+                speciesnet_latin_name_to_taxon_string[binomial_name] = s
+        elif len(genus) > 0:
+            assert all([len(s) > 0 for s in [family,order]])
+            if genus not in speciesnet_latin_name_to_taxon_string:
+                speciesnet_latin_name_to_taxon_string[genus] = s
+        elif len(family) > 0:
+            assert len(order) > 0
+            if family not in speciesnet_latin_name_to_taxon_string:
+                speciesnet_latin_name_to_taxon_string[family] = s
+        elif len(order) > 0:
+            if order not in speciesnet_latin_name_to_taxon_string:
+                speciesnet_latin_name_to_taxon_string[order] = s
+        else:
+            if class_name not in speciesnet_latin_name_to_taxon_string:
+                speciesnet_latin_name_to_taxon_string[class_name] = s
+        
+        if len(common_name) > 0:
+            if common_name not in speciesnet_common_name_to_taxon_string:
+                speciesnet_common_name_to_taxon_string[common_name] = s
+    
+    for s in speciesnet_taxonomy_list:
+        
+        _insert_taxonomy_string(s)
+    
+    
+    ##%% Make sure all parent taxa are represented in the taxonomy
+    
+    # In theory any taxon that appears as the parent of another taxon should
+    # also be in the taxonomy, but this isn't always true, so we fix it here.
+    
+    new_taxon_string_to_missing_tokens = defaultdict(list)
+    
+    # latin_name = next(iter(speciesnet_latin_name_to_taxon_string.keys()))
+    for latin_name in speciesnet_latin_name_to_taxon_string.keys():
+        
+        if 'no cv result' in latin_name:
+            continue
+        
+        taxon_string = speciesnet_latin_name_to_taxon_string[latin_name]
+        tokens = taxon_string.split(';')
+        
+        # Don't process GUID, species, or common name
+        # i_token = 6
+        for i_token in range(1,len(tokens)-2):
+            
+            test_token = tokens[i_token]            
+            if len(test_token) == 0:
+                continue
+            
+            # Do we need to make up a taxon for this token?
+            if test_token not in speciesnet_latin_name_to_taxon_string:
+                
+                new_tokens = [''] * 7
+                new_tokens[0] = 'fake_guid'
+                for i_copy_token in range(1,i_token+1):
+                    new_tokens[i_copy_token] = tokens[i_copy_token]
+                new_tokens[-1] = test_token + ' species'
+                assert new_tokens[-2] == ''
+                new_taxon_string = ';'.join(new_tokens)
+                # assert new_taxon_string not in new_taxon_strings
+                new_taxon_string_to_missing_tokens[new_taxon_string].append(test_token)
+                
+        # ...for each token
+        
+    # ...for each taxon
+        
+    print('Found {} taxa that need to be inserted to make the taxonomy valid:\n'.format(
+        len(new_taxon_string_to_missing_tokens)))
+    
+    new_taxon_string_to_missing_tokens = \
+        sort_dictionary_by_key(new_taxon_string_to_missing_tokens)
+    for taxon_string in new_taxon_string_to_missing_tokens:
+        missing_taxa = ','.join(new_taxon_string_to_missing_tokens[taxon_string])
+        print('{} ({})'.format(taxon_string,missing_taxa))
+        
+    for new_taxon_string in new_taxon_string_to_missing_tokens:
+        _insert_taxonomy_string(new_taxon_string)
+        
+        
+    ##%% Make sure all species on the allow-list are in the taxonomy
+    
+    n_failed_mappings = 0
+    
+    for target_taxon_latin_name in target_latin_to_common.keys():
+        if target_taxon_latin_name not in speciesnet_latin_name_to_taxon_string:
+            common_name = target_latin_to_common[target_taxon_latin_name]
+            s = '{} ({}) not in speciesnet taxonomy'.format(
+                target_taxon_latin_name,common_name)
+            if common_name in speciesnet_common_name_to_taxon_string:
+                s += ' (common name maps to {})'.format(
+                    speciesnet_common_name_to_taxon_string[common_name])
+            print(s)
+            n_failed_mappings += 1
+    
+    if n_failed_mappings > 0:
+        raise ValueError('Cannot continue with geofence generation')
+    
+    
+    ##%% For the allow-list, map each parent taxon to a set of allowable child taxa
+    
+    # Maps parent names to all allowed child names, or None if this is the 
+    # lowest-level allowable taxon on this path
+    allowed_parent_taxon_to_child_taxa = defaultdict(set)
+    
+    # latin_name = next(iter(target_latin_to_common.keys()))
+    for latin_name in target_latin_to_common:
+        
+        taxon_string = speciesnet_latin_name_to_taxon_string[latin_name]
+        tokens = taxon_string.split(';')
+        assert len(tokens) == 7
+        
+        # Remove GUID and common mame
+        #
+        # This is now always class/order/family/genus/species
+        tokens = tokens[1:-1]
+        
+        child_taxon = None
+        
+        # If this is a species
+        if len(tokens[-1]) > 0:
+            binomial_name = tokens[-2] + ' ' + tokens[-1]
+            assert binomial_name == latin_name
+            allowed_parent_taxon_to_child_taxa[binomial_name].add(None)
+            child_taxon = binomial_name
+            
+        # The first candidate parent is the genus        
+        parent_token_index = len(tokens) - 2
+
+        while(parent_token_index >= 0):
+            
+            parent_taxon = tokens[parent_token_index]
+            allowed_parent_taxon_to_child_taxa[parent_taxon].add(child_taxon)
+            child_taxon = parent_taxon
+            parent_token_index -= 1                
+    
+    # ...for each allowed latin name
+    
+    allowed_parent_taxon_to_child_taxa = \
+        sort_dictionary_by_key(allowed_parent_taxon_to_child_taxa)
+    
+    
+    ##%% Map all predictions that exist in this dataset...
+    
+    # ...to the prediction we should generate.
+    
+    with open(input_file,'r') as f:
+        input_data = json.load(f)
+    
+    input_category_id_to_common_name = input_data['classification_categories'] #noqa
+    input_category_id_to_taxonomy_string = \
+        input_data['classification_category_descriptions']
+    
+    input_category_id_to_output_taxon_string = {}
+    
+    # input_category_id = next(iter(input_category_id_to_taxonomy_string.keys()))
+    for input_category_id in input_category_id_to_taxonomy_string.keys():
+        
+        input_taxon_string = input_category_id_to_taxonomy_string[input_category_id]
+        input_taxon_tokens = input_taxon_string.split(';')
+        assert len(input_taxon_tokens) == 7
+    
+        # Don't mess with blank/no-cv-result/animal/human
+        if (input_taxon_string in non_taxonomic_prediction_strings) or \
+           (input_taxon_string == human_prediction_string):
+            input_category_id_to_output_taxon_string[input_category_id] = \
+                input_taxon_string
+            continue
+    
+        # Remove GUID and common mame
+        #
+        # This is now always class/order/family/genus/species
+        input_taxon_tokens = input_taxon_tokens[1:-1]
+        
+        test_index = len(input_taxon_tokens) - 1
+        target_taxon = None
+        
+        # Start at the species level, and see whether each taxon is allowed
+        while((test_index >= 0) and (target_taxon is None)):
+            
+            # Species are represented as binomial names
+            if (test_index == (len(input_taxon_tokens) - 1)) and \
+                (len(input_taxon_tokens[-1]) > 0):
+                test_taxon_name = \
+                    input_taxon_tokens[-2] + ' ' + input_taxon_tokens[-1]
+            else:
+                test_taxon_name = input_taxon_tokens[test_index]
+            
+            # If we haven't yet found the level at which this taxon is non-empty,
+            # keep going up
+            if len(test_taxon_name) == 0:            
+                test_index -= 1
+                continue
+            
+            assert test_taxon_name in speciesnet_latin_name_to_taxon_string
+            
+            # Is this taxon allowed according to the custom species list?
+            if test_taxon_name in allowed_parent_taxon_to_child_taxa:
+                
+                allowed_child_taxa = allowed_parent_taxon_to_child_taxa[test_taxon_name]
+                assert allowed_child_taxa is not None
+                
+                # If this is the lowest-level allowable token or there is not a 
+                # unique child, don't walk any further, even if walking down
+                # is enabled.
+                if (None in allowed_child_taxa):
+                    assert len(allowed_child_taxa) == 1
+                
+                if (None in allowed_child_taxa) or (len(allowed_child_taxa) > 1):
+                    target_taxon = test_taxon_name
+                elif not allow_walk_down:
+                    target_taxon = test_taxon_name
+                else:
+                    # If there's a unique child, walk back *down* the allowable
+                    # taxa until we run out of unique children
+                    while ((next(iter(allowed_child_taxa)) is not None) and \
+                          (len(allowed_child_taxa) == 1)):
+                        candidate_taxon = next(iter(allowed_child_taxa))
+                        assert candidate_taxon in allowed_parent_taxon_to_child_taxa
+                        assert candidate_taxon in speciesnet_latin_name_to_taxon_string
+                        allowed_child_taxa = \
+                            allowed_parent_taxon_to_child_taxa[candidate_taxon]
+                    target_taxon = candidate_taxon
+                    
+            # ...if this is an allowed taxon
+                    
+            test_index -= 1
+    
+        # ...for each token
+        
+        if target_taxon is None:
+            output_taxon_string = animal_prediction_string                        
+        else:
+            output_taxon_string = speciesnet_latin_name_to_taxon_string[target_taxon]
+        input_category_id_to_output_taxon_string[input_category_id] = output_taxon_string        
+    
+    # ...for each category
+        
+    
+    ##%% Build the new tables
+    
+    input_category_id_to_output_category_id = {}
+    output_taxon_string_to_category_id = {}
+    output_category_id_to_common_name = {}
+    
+    for input_category_id in input_category_id_to_output_taxon_string:
+        
+        original_common_name = \
+            input_category_id_to_common_name[input_category_id]
+        original_taxon_string = \
+            input_category_id_to_taxonomy_string[input_category_id]
+        output_taxon_string = \
+            input_category_id_to_output_taxon_string[input_category_id]
+    
+        output_common_name = output_taxon_string.split(';')[-1]
+        
+        # Do we need to create a new output category?
+        if output_taxon_string not in output_taxon_string_to_category_id:
+            output_category_id = str(len(output_taxon_string_to_category_id))
+            output_taxon_string_to_category_id[output_taxon_string] = \
+                output_category_id
+            output_category_id_to_common_name[output_category_id] = \
+                output_common_name            
+        else:
+            output_category_id = \
+                output_taxon_string_to_category_id[output_taxon_string]
+        
+        input_category_id_to_output_category_id[input_category_id] = \
+            output_category_id
+            
+        if False:
+            print('Mapping {} ({}) to:\n{} ({})\n'.format(
+                original_common_name,original_taxon_string,
+                output_common_name,output_taxon_string))
+        if False:    
+            print('Mapping {} to {}'.format(
+                original_common_name,output_common_name,))
+    
+    # ...for each category
+    
+    
+    ##%% Remap all category labels
+    
+    assert len(set(output_taxon_string_to_category_id.keys())) == \
+           len(set(output_taxon_string_to_category_id.values()))
+           
+    output_category_id_to_taxon_string = \
+        invert_dictionary(output_taxon_string_to_category_id)
+        
+    with open(input_file,'r') as f:
+        output_data = json.load(f)
+        
+    for im in tqdm(output_data['images']):
+        if 'detections' in im and im['detections'] is not None:
+            for det in im['detections']:
+                if 'classifications' in det:
+                    for classification in det['classifications']:
+                        classification[0] = \
+                            input_category_id_to_output_category_id[classification[0]]
+    
+    output_data['classification_categories'] = output_category_id_to_common_name
+    output_data['classification_category_descriptions'] = \
+        output_category_id_to_taxon_string
+        
+        
+    ##%% Write output
+    
+    with open(output_file,'w') as f:
+        json.dump(output_data,f,indent=1)
+    
+    
 #%% Interactive driver(s)
 
 if False:
     
     pass
 
+    #%% Shared cell to initialize geofencing and taxonomy information
+    
+    from megadetector.utils.wi_utils import species_allowed_in_country # noqa
+    from megadetector.utils.wi_utils import initialize_geofencing, initialize_taxonomy_info # noqa
+    from megadetector.utils.wi_utils import _species_string_to_canonical_species_string # noqa
+    from megadetector.utils.wi_utils import generate_csv_rows_for_species # noqa
+    from megadetector.utils.wi_utils import _generate_csv_rows_to_block_all_countries_except # noqa
+
+    from megadetector.utils.wi_utils import taxonomy_string_to_geofencing_rules # noqa
+    from megadetector.utils.wi_utils import taxonomy_string_to_taxonomy_info # noqa
+    
+    geofencing_file = r'c:\git\cameratrapai\data\geofence_base.json'
+    
+    country_code_file = r'g:\temp\country-codes.csv'
+    # encoding = 'cp1252'; taxonomy_file = r'g:\temp\taxonomy_mapping-' + encoding + '.json'    
+    encoding = None; taxonomy_file = r'g:\temp\taxonomy_mapping.json'
+    
+    initialize_geofencing(geofencing_file, country_code_file, force_init=True)
+    initialize_taxonomy_info(taxonomy_file, force_init=True, encoding=encoding)
+    
+    
+    #%% Test driver for geofence_fixes.csv function
+    
+    block_except_list = 'AUS, PNG, THA, IDN, MYS'
+    species = 'dingo'
+    species_string = _species_string_to_canonical_species_string(species)
+    rows = _generate_csv_rows_to_block_all_countries_except(species_string,block_except_list)
+        
+    import clipboard; clipboard.copy('\n'.join(rows))
+    
+    
+    #%%
+    
+    generate_csv_rows_for_species(species_string=species_string,
+                                  allow_countries=None,
+                                  block_countries=None,
+                                  allow_states=None,
+                                  block_states=None,
+                                  blockexcept_countries=None)
+    
+    
+    _generate_csv_rows_to_block_all_countries_except(species_string,'AUS')
+    
+    
+    #%% Test the effects of geofence changes
+
+    species = 'canis lupus dingo'
+    country = 'guatemala'
+    species_allowed_in_country(species,country,state=None,return_status=False)
+        
+    
     #%% instances.json generation test
     
     from megadetector.utils.wi_utils import generate_instances_json_from_folder # noqa
@@ -1604,13 +2515,6 @@ if False:
     
     #%% Geofencing tests
     
-    geofencing_file = r'g:\temp\geofence_mapping.json'
-    country_code_file = r'G:/temp/country-codes.csv'    
-    encoding = 'cp1252'; taxonomy_file = r'g:\temp\taxonomy_mapping-' + encoding + '.json'
-    
-    initialize_taxonomy_info(taxonomy_file, force_init=True, encoding=encoding)    
-    initialize_geofencing(geofencing_file, country_code_file, force_init=True)
-    
     species = 'didelphis marsupialis'
     print(binomial_name_to_taxonomy_info[species])
     country = 'Guatemala'
@@ -1624,13 +2528,6 @@ if False:
     
     #%% Test several species
     
-    geofencing_file = r'g:\temp\geofence_mapping.json'
-    country_code_file = r'G:/temp/country-codes.csv'    
-    encoding = 'cp1252'; taxonomy_file = r'g:\temp\taxonomy_mapping-' + encoding + '.json'
-    
-    initialize_taxonomy_info(taxonomy_file, force_init=True, encoding=encoding)    
-    initialize_geofencing(geofencing_file, country_code_file, force_init=True)
-    
     if True:
         
         # Make sure some Guatemalan species are allowed in Guatemala