diff --git a/src/mol-io/writer/_spec/cif.spec.ts b/src/mol-io/writer/_spec/cif.spec.ts
new file mode 100644
index 0000000000000000000000000000000000000000..fba3f3c51ce81a7b14f07a9d420afe878fda4568
--- /dev/null
+++ b/src/mol-io/writer/_spec/cif.spec.ts
@@ -0,0 +1,183 @@
+import * as Data from '../../reader/cif/data-model'
+import { CifWriter } from '../cif';
+import decodeMsgPack from '../../common/msgpack/decode'
+import { EncodedFile, EncodedCategory } from '../../common/binary-cif';
+import Field from '../../reader/cif/binary/field';
+import * as C from '../cif/encoder';
+
+const cartn_x = Data.CifField.ofNumbers([1.001, 1.002, 1.003, 1.004, 1.005, 1.006, 1.007, 1.008, 1.009]);
+const cartn_y = Data.CifField.ofNumbers([-3.0, -2.666, -2.3333, -2.0, -1.666, -1.333, -1.0, -0.666, -0.333]);
+const cartn_z = Data.CifField.ofNumbers([1, 2, 3, 4, 5, 6, 7, 8, 9].map(i => Math.sqrt(i)));
+const label_seq_id = Data.CifField.ofNumbers([1, 2, 3, 6, 11, 23, 47, 106, 235]);
+const atom_site = Data.CifCategory.ofFields('atom_site', { 'Cartn_x': cartn_x, 'Cartn_y': cartn_y, 'Cartn_z': cartn_z, 'label_seq_id': label_seq_id });
+const field1 = Data.CifField.ofNumbers([1, 2, 3, 6, 11, 23, 47, 106, 235]);
+const field2 = Data.CifField.ofNumbers([-1, -2, -3, -6, -11, -23, -47, -106, -235]);
+const other_fields = Data.CifCategory.ofFields('other_fields', { 'field1': field1, 'field2': field2 });
+
+const encoding_aware_encoder = CifWriter.createEncoder({
+ binary: true,
+ binaryAutoClassifyEncoding: true,
+ binaryEncodingPovider: CifWriter.createEncodingProviderFromJsonConfig([
+ {
+ 'categoryName': 'atom_site',
+ 'columnName': 'Cartn_y',
+ 'encoding': 'rle',
+ 'precision': 0
+ },
+ {
+ 'categoryName': 'atom_site',
+ 'columnName': 'Cartn_z',
+ 'encoding': 'delta',
+ 'precision': 1
+ },
+ {
+ 'categoryName': 'atom_site',
+ 'columnName': 'label_seq_id',
+ 'encoding': 'delta-rle'
+ }
+ ])
+});
+
+describe('encoding-config', () => {
+ const decoded = process(encoding_aware_encoder);
+
+ const decoded_atom_site = decoded.blocks[0].categories['atom_site'];
+ const decoded_cartn_x = decoded_atom_site.getField('Cartn_x')!;
+ const decoded_cartn_y = decoded_atom_site.getField('Cartn_y')!;
+ const decoded_cartn_z = decoded_atom_site.getField('Cartn_z')!;
+ const decoded_label_seq_id = decoded_atom_site.getField('label_seq_id')!;
+
+ const delta = 0.001;
+ function assert(e: ArrayLike<number>, a: ArrayLike<number>) {
+ expect(e.length).toBe(a.length);
+ for (let i = 0; i < e.length; i++) {
+ expect(Math.abs(e[i] - a[i])).toBeLessThan(delta);
+ }
+ }
+
+ function join(field: Data.CifField) {
+ return field.binaryEncoding!.map(e => e.kind).join();
+ }
+
+ it('strategy', () => {
+ expect(join(decoded_cartn_x)).toBe('FixedPoint,Delta,IntegerPacking,ByteArray');
+ expect(join(decoded_cartn_y)).toBe('FixedPoint,RunLength,IntegerPacking,ByteArray');
+ expect(join(decoded_cartn_z)).toBe('FixedPoint,Delta,IntegerPacking,ByteArray');
+ expect(join(decoded_label_seq_id)).toBe('Delta,RunLength,IntegerPacking,ByteArray');
+ });
+
+ it('precision', () => {
+ assert(decoded_cartn_x.toFloatArray(), cartn_x.toFloatArray());
+ assert(decoded_cartn_y.toFloatArray(), cartn_y.toFloatArray().map(d => Math.round(d)));
+ assert(decoded_cartn_z.toFloatArray(), cartn_z.toFloatArray().map(d => Math.round(d * 10) / 10));
+ assert(decoded_label_seq_id.toIntArray(), label_seq_id.toIntArray());
+ });
+});
+
+const filter_aware_encoder1 = CifWriter.createEncoder({
+ binary: true,
+ binaryAutoClassifyEncoding: true
+});
+filter_aware_encoder1.setFilter(C.Category.filterOf('atom_site\n' +
+'\n' +
+'atom_site.Cartn_x\n' +
+'atom_site.Cartn_y\n'));
+
+const filter_aware_encoder2 = CifWriter.createEncoder({
+ binary: true
+});
+filter_aware_encoder2.setFilter(C.Category.filterOf('!atom_site\n' +
+'\n' +
+'!other_fields.field2\n'));
+
+describe('filtering-config', () => {
+ const decoded1 = process(filter_aware_encoder1);
+
+ const atom_site1 = decoded1.blocks[0].categories['atom_site'];
+ const cartn_x1 = atom_site1.getField('Cartn_x');
+ const cartn_y1 = atom_site1.getField('Cartn_y');
+ const cartn_z1 = atom_site1.getField('Cartn_z');
+ const label_seq_id1 = atom_site1.getField('label_seq_id');
+ const fields1 = decoded1.blocks[0].categories['other_fields'];
+
+ it('whitelist-filtering', () => {
+ expect(atom_site1).toBeDefined();
+ expect(cartn_x1).toBeDefined();
+ expect(cartn_y1).toBeDefined();
+ expect(cartn_z1).toBeUndefined();
+ expect(label_seq_id1).toBeUndefined();
+ expect(fields1).toBeUndefined();
+ });
+
+ const decoded2 = process(filter_aware_encoder2);
+
+ const atom_site2 = decoded2.blocks[0].categories['atom_site'];
+ const fields2 = decoded2.blocks[0].categories['other_fields'];
+ const field12 = fields2.getField('field1');
+ const field22 = fields2.getField('field2');
+
+ it('blacklist-filtering', () => {
+ expect(atom_site2).toBeUndefined();
+ expect(fields2).toBeDefined();
+ expect(field12).toBeDefined();
+ expect(field22).toBeUndefined();
+ });
+});
+
+function process(encoder: C.Encoder) {
+ encoder.startDataBlock('test');
+
+ for (const cat of [atom_site, other_fields]) {
+ const fields: CifWriter.Field[] = [];
+ for (const f of cat.fieldNames) {
+ fields.push(wrap(f, cat.getField(f)!))
+ }
+ encoder.writeCategory(getCategoryInstanceProvider(cat, fields));
+ }
+
+ const encoded = encoder.getData() as Uint8Array;
+
+ const unpacked = decodeMsgPack(encoded) as EncodedFile;
+ return Data.CifFile(unpacked.dataBlocks.map(block => {
+ const cats = Object.create(null);
+ for (const cat of block.categories) cats[cat.name.substr(1)] = Category(cat);
+ return Data.CifBlock(block.categories.map(c => c.name.substr(1)), cats, block.header);
+ }));
+}
+
+
+function getCategoryInstanceProvider(cat: Data.CifCategory, fields: CifWriter.Field[]): CifWriter.Category {
+ return {
+ name: cat.name,
+ instance: () => CifWriter.categoryInstance(fields, { data: cat, rowCount: cat.rowCount })
+ };
+}
+
+function wrap(name: string, field: Data.CifField): CifWriter.Field {
+ const type = Data.getCifFieldType(field);
+ if (type['@type'] === 'str') {
+ return { name, type: CifWriter.Field.Type.Str, value: field.str, valueKind: field.valueKind };
+ } else if (type['@type'] === 'float') {
+ return { name, type: CifWriter.Field.Type.Float, value: field.float, valueKind: field.valueKind };
+ } else {
+ return { name, type: CifWriter.Field.Type.Int, value: field.int, valueKind: field.valueKind };
+ }
+}
+
+function Category(data: EncodedCategory): Data.CifCategory {
+ const map = Object.create(null);
+ const cache = Object.create(null);
+ for (const col of data.columns) map[col.name] = col;
+ return {
+ rowCount: data.rowCount,
+ name: data.name.substr(1),
+ fieldNames: data.columns.map(c => c.name),
+ getField(name) {
+ const col = map[name];
+ if (!col) return void 0;
+ if (!!cache[name]) return cache[name];
+ cache[name] = Field(col);
+ return cache[name];
+ }
+ }
+}
\ No newline at end of file
diff --git a/src/mol-io/writer/cif.ts b/src/mol-io/writer/cif.ts
index c191d8a4bdeca25d9c377abe354980931bcad2f3..76298baf837774d20b1f727053960da10d2247fc 100644
--- a/src/mol-io/writer/cif.ts
+++ b/src/mol-io/writer/cif.ts
@@ -53,5 +53,64 @@ export namespace CifWriter {
return ff && ff.binaryEncoding ? ArrayEncoder.fromEncoding(ff.binaryEncoding) : void 0;
}
}
+ };
+
+ export function createEncodingProviderFromJsonConfig(hints: EncodingStrategyHint[]): EncodingProvider {
+ return {
+ get(c, f) {
+ for (let i = 0; i < hints.length; i++) {
+ const hint = hints[i];
+ if (hint.categoryName === c && hint.columnName === f) {
+ return resolveEncoding(hint);
+ }
+ }
+ }
+ }
+ }
+
+ function resolveEncoding(hint: EncodingStrategyHint): ArrayEncoder {
+ const precision: number | undefined = hint.precision;
+ if (precision !== void 0) {
+ const multiplier = Math.pow(10, precision);
+ const fixedPoint = E.by(E.fixedPoint(multiplier));
+ switch (hint.encoding) {
+ case 'pack':
+ return fixedPoint.and(E.integerPacking);
+ case 'rle':
+ return fixedPoint.and(E.runLength).and(E.integerPacking);
+ case 'delta':
+ return fixedPoint.and(E.delta).and(E.integerPacking);
+ case 'delta-rle':
+ return fixedPoint.and(E.delta).and(E.runLength).and(E.integerPacking);
+ };
+ } else {
+ switch (hint.encoding) {
+ case 'pack':
+ return E.by(E.integerPacking);
+ case 'rle':
+ return E.by(E.runLength).and(E.integerPacking);
+ case 'delta':
+ return E.by(E.delta).and(E.integerPacking);
+ case 'delta-rle':
+ return E.by(E.delta).and(E.runLength).and(E.integerPacking);
+ }
+ }
+ throw new Error('cannot be reached');
}
-}
\ No newline at end of file
+}
+
+/**
+ * Defines the information needed to encode certain fields: category and column name as well as encoding tag, precision is optional and identifies float columns.
+ */
+export interface EncodingStrategyHint {
+ categoryName: string,
+ columnName: string,
+ // TODO would be nice to infer strategy and precision if needed
+ encoding: EncodingType,
+ /**
+ * number of decimal places to keep - must be specified to float columns
+ */
+ precision?: number
+}
+
+type EncodingType = 'pack' | 'rle' | 'delta' | 'delta-rle'
\ No newline at end of file
diff --git a/src/mol-io/writer/cif/encoder.ts b/src/mol-io/writer/cif/encoder.ts
index f0d95d18b7543d92f48c50d7f5f7897ed371befa..9c0b8a743f3377edd90527b13977b95165fbfe03 100644
--- a/src/mol-io/writer/cif/encoder.ts
+++ b/src/mol-io/writer/cif/encoder.ts
@@ -132,6 +132,60 @@ export namespace Category {
includeField(categoryName: string, fieldName: string): boolean,
}
+ export function filterOf(directives: string): Filter {
+ const cat_whitelist: string[] = [];
+ const cat_blacklist: string[] = [];
+ const field_whitelist: string[] = [];
+ const field_blacklist: string[] = [];
+
+ for (let d of directives.split(/[\r\n]+/)) {
+ d = d.trim();
+ // allow for empty lines in config
+ if (d.length === 0) continue;
+ // let ! denote blacklisted entries
+ const blacklist = /^!/.test(d);
+ if (blacklist) d = d.substr(1);
+ const split = d.split(/\./);
+ const field = split[1];
+ const list = blacklist ? (field ? field_blacklist : cat_blacklist) : (field ? field_whitelist : cat_whitelist);
+
+ list[list.length] = d;
+
+ // ensure categories are aware about whitelisted columns
+ if (field && !cat_whitelist.includes(split[0])) {
+ cat_whitelist[cat_whitelist.length] = split[0];
+ }
+ }
+
+ const wlcatcol = field_whitelist.map(it => it.split('.')[0]);
+ // blacklist has higher priority
+ return {
+ includeCategory(cat) {
+ // block if category in black
+ if (cat_blacklist.includes(cat)) {
+ return false;
+ } else {
+ // if there is a whitelist, the category has to be explicitly allowed
+ return cat_whitelist.length <= 0 ||
+ // otherwise include if whitelist contains category
+ cat_whitelist.indexOf(cat) !== -1;
+ }
+ },
+ includeField(cat, field) {
+ // column names are assumed to follow the pattern 'category_name.column_name'
+ const full = cat + '.' + field;
+ if (field_blacklist.includes(full)) {
+ return false;
+ } else {
+ // if for this category no whitelist entries exist
+ return !wlcatcol.includes(cat) ||
+ // otherwise must be specifically allowed
+ field_whitelist.includes(full);
+ }
+ }
+ }
+ }
+
export const DefaultFilter: Filter = {
includeCategory(cat) { return true; },
includeField(cat, field) { return true; }