[WIP] [ENH] implement bids-schema - part 1

+bids/+internal/append_to_structure.m

@@ -0,0 +1,62 @@
+function subject = append_to_structure(file, subject, modality)
+  % Copyright (C) 2021--, BIDS-MATLAB developers
+
+  p = bids.internal.parse_filename(file);
+  idx = find_suffix_group(modality, p.type);
+  if isempty(idx)
+    warning('append_to_structure:noMatchingSuffix', ...
+            'Skipping file with no valid suffix in schema: %s', file);
+    return
+  end
+
+  schema = bids.schema.load_schema();
+  entities = bids.schema.return_datatype_entities(schema.datatypes.(modality)(idx));
+  p = bids.internal.parse_filename(file, entities);
+
+  if ~isempty(subject.(modality))
+
+    missing_fields = setxor(fieldnames(subject.(modality)), fieldnames(p));
+
+    if ~isempty(missing_fields)
+      for iField = 1:numel(missing_fields)
+        p = add_missing_field(p, ...
+                              missing_fields{iField});
+        subject.(modality) = add_missing_field(subject.(modality), ...
+                                               missing_fields{iField});
+      end
+    end
+
+  end
+
+  subject.(modality) = [subject.(modality) p];
+
+end
+
+function structure = add_missing_field(structure, field)
+  if ~isfield(structure, field)
+    structure(1).(field) = '';
+  end
+end
+
+function idx = find_suffix_group(modality, suffix)
+
+  idx = [];
+
+  schema = bids.schema.load_schema();
+  suffix_groups = {schema.datatypes.(modality).suffixes}';
+
+  % the following loop could probably be improved with some cellfun magic
+  %   cellfun(@(x, y) any(strcmp(x,y)), {p.type}, suffix_groups)
+  for i = 1:numel(suffix_groups)
+    if any(strcmp(suffix, suffix_groups{i}))
+      idx = i;
+      break
+    end
+  end
+
+  if isempty(idx)
+    warning('findSuffix:noMatchingSuffix', ...
+            'No corresponding suffix in schema for %s for datatype %s', suffix, modality);
+  end
+
+end

+bids/+internal/parse_filename.m

@@ -4,20 +4,20 @@
   %
   % Example:
   %
-  % >> filename = '../sub-16/anat/sub-16_ses-mri_run-1_echo-2_FLASH.nii.gz';
+  % >> filename = '../sub-16/anat/sub-16_ses-mri_run-1_acq-hd_T1w.nii.gz';
   % >> bids.internal.parse_filename(filename)
   %
   % ans =
   %
   %   struct with fields:
   %
-  %     filename: 'sub-16_ses-mri_run-1_echo-2_FLASH.nii.gz'
-  %         type: 'FLASH'
+  %     filename: 'sub-16_ses-mri_run-1_acq-hd_T1w.nii.gz'
+  %         type: 'T1w'
   %          ext: '.nii.gz'
   %          sub: '16'
   %          ses: 'mri'
   %          run: '1'
-  %         echo: '2'
+  %         acq: 'hd'
   % __________________________________________________________________________
 
   % Copyright (C) 2016-2018, Guillaume Flandin, Wellcome Centre for Human Neuroimaging
@@ -50,7 +50,8 @@
     try
       p = orderfields(p, ['filename', 'ext', 'type', fields]);
     catch
-      warning('Ignoring file ''%s'' not matching template.', filename);
+      warning('bidsMatlab:noMatchingTemplate', ...
+              'Ignoring file %s not matching template.', filename);
       p = struct([]);
     end
   end

+bids/+internal/return_datatype_extensions.m

@@ -0,0 +1,14 @@
+function extensions = return_datatype_extensions(datatype)
+
+  extensions = '(';
+
+  for iExt = 1:numel(datatype.extensions)
+    if ~strcmp(datatype.extensions{iExt}, '.json')
+      extensions = [extensions,  datatype.extensions{iExt}, '|']; %#ok<AGROW>
+    end
+  end
+
+  % Replace final "|" by a "){1}"
+  extensions(end:end + 3) = '){1}';
+
+end

+bids/+internal/return_datatype_regular_expression.m

@@ -0,0 +1,8 @@
+function regular_expression = return_datatype_regular_expression(datatype)
+
+  suffixes = bids.internal.return_datatype_suffixes(datatype);
+  extensions = bids.internal.return_datatype_extensions(datatype);
+
+  regular_expression = ['^%s.*' suffixes extensions '$'];
+
+end

+bids/+internal/return_datatype_suffixes.m

@@ -0,0 +1,12 @@
+function suffixes = return_datatype_suffixes(datatype)
+
+  suffixes = '_(';
+
+  for iExt = 1:numel(datatype.suffixes)
+    suffixes = [suffixes,  datatype.suffixes{iExt}, '|']; %#ok<AGROW>
+  end
+
+  % Replace final "|" by a "){1}"
+  suffixes(end:end + 3) = '){1}';
+
+end

+bids/+schema/load_schema.m

@@ -0,0 +1,64 @@
+function schema = load_schema(SCHEMA_DIR)
+  % Loads a json schema by recursively looking through a folder structure.
+  %
+  % The nesting of the output structure reflects a combination of the folder structure and
+  % any eventual nesting within each json.
+  %
+  %
+  % Copyright (C) 2021--, BIDS-MATLAB developers
+
+  % TODO:
+  %  - folders that do not contain json files themselves but contain
+  %  subfolders that do, are not reflected in the output structure (they are
+  %  skipped). This can lead to "name conflicts". See "silenced" unit tests
+  %  for more info.
+
+  if nargin < 1
+    SCHEMA_DIR = fullfile(fileparts(mfilename('fullpath')), '..', '..', 'schema');
+  end
+
+  schema = struct();
+
+  [json_file_list, dirs] = bids.internal.file_utils('FPList', SCHEMA_DIR, '^*.json$');
+
+  schema = append_json_content_to_structure(schema, json_file_list);
+
+  schema = inspect_subdir(schema, dirs);
+
+end
+
+function structure = append_json_content_to_structure(structure, json_file_list)
+
+  for iFile = 1:size(json_file_list, 1)
+
+    file = deblank(json_file_list(iFile, :));
+
+    field_name = bids.internal.file_utils(file, 'basename');
+
+    structure.(field_name) = bids.util.jsondecode(file);
+  end
+
+end
+
+function structure = inspect_subdir(structure, subdir_list)
+  % recursively inspects subdirectory for json files and reflects folder
+  % hierarchy in the output structure.
+
+  for iDir = 1:size(subdir_list, 1)
+
+    directory = deblank(subdir_list(iDir, :));
+
+    [json_file_list, dirs] = bids.internal.file_utils('FPList', directory, '^*.json$');
+
+    if ~isempty(json_file_list)
+      field_name = bids.internal.file_utils(directory, 'basename');
+      structure.(field_name) = struct();
+      structure.(field_name) = append_json_content_to_structure(structure.(field_name), ...
+                                                                json_file_list);
+    end
+
+    structure = inspect_subdir(structure, dirs);
+
+  end
+
+end

+bids/+schema/return_datatype_entities.m

@@ -0,0 +1,11 @@
+function entities = return_datatype_entities(datatype)
+
+  schema = bids.schema.load_schema();
+
+  entity_names = fieldnames(datatype.entities);
+
+  for i = 1:size(entity_names, 1)
+    entities{1, i} = schema.entities.(entity_names{i}).entity; %#ok<*AGROW>
+  end
+
+end