Merge pull request #496 from JonStargaryen/master

Download CCD from Configurable URL

src/cli/chem-comp-dict/create-ions.ts

@@ -15,7 +15,7 @@ const writeFile = util.promisify(fs.writeFile);
 
 import { DatabaseCollection } from '../../mol-data/db';
 import { CCD_Schema } from '../../mol-io/reader/cif/schema/ccd';
-import { ensureDataAvailable, readCCD } from './util';
+import { DefaultDataOptions, ensureDataAvailable, readCCD } from './util';
 
 function extractIonNames(ccd: DatabaseCollection<CCD_Schema>) {
     const ionNames: string[] = [];
@@ -44,8 +44,8 @@ export const IonNames = new Set(${JSON.stringify(ionNames).replace(/"/g, "'").re
     writeFile(filePath, output);
 }
 
-async function run(out: string, forceDownload = false) {
-    await ensureDataAvailable(forceDownload);
+async function run(out: string, options = DefaultDataOptions) {
+    await ensureDataAvailable(options);
     const ccd = await readCCD();
     const ionNames = extractIonNames(ccd);
     if (!fs.existsSync(path.dirname(out))) {
@@ -65,10 +65,15 @@ parser.add_argument('--forceDownload', '-f', {
     action: 'store_true',
     help: 'Force download of CCD and PVCD.'
 });
+parser.add_argument('--ccdUrl', '-c', {
+    help: 'Fetch the CCD from a custom URL. This forces download of the CCD.',
+    required: false
+});
 interface Args {
     out: string,
     forceDownload?: boolean,
+    ccdUrl?: string
 }
 const args: Args = parser.parse_args();
 
-run(args.out, args.forceDownload);
+run(args.out, { forceDownload: args.forceDownload, ccdUrl: args.ccdUrl });

src/cli/chem-comp-dict/create-saccharides.ts

@@ -14,7 +14,7 @@ const writeFile = util.promisify(fs.writeFile);
 
 import { DatabaseCollection } from '../../mol-data/db';
 import { CCD_Schema } from '../../mol-io/reader/cif/schema/ccd';
-import { ensureDataAvailable, readCCD } from './util';
+import { DefaultDataOptions, ensureDataAvailable, readCCD } from './util';
 
 function extractSaccharideNames(ccd: DatabaseCollection<CCD_Schema>) {
     const saccharideNames: string[] = [];
@@ -47,8 +47,8 @@ export const SaccharideNames = new Set(${JSON.stringify(ionNames).replace(/"/g,
     writeFile(filePath, output);
 }
 
-async function run(out: string, forceDownload = false) {
-    await ensureDataAvailable(forceDownload);
+async function run(out: string, options = DefaultDataOptions) {
+    await ensureDataAvailable(options);
     const ccd = await readCCD();
     const saccharideNames = extractSaccharideNames(ccd);
     if (!fs.existsSync(path.dirname(out))) {
@@ -68,10 +68,15 @@ parser.add_argument('--forceDownload', '-f', {
     action: 'store_true',
     help: 'Force download of CCD and PVCD.'
 });
+parser.add_argument('--ccdUrl', '-c', {
+    help: 'Fetch the CCD from a custom URL. This forces download of the CCD.',
+    required: false
+});
 interface Args {
     out: string,
     forceDownload?: boolean,
+    ccdUrl?: string
 }
 const args: Args = parser.parse_args();
 
-run(args.out, args.forceDownload);
+run(args.out, { forceDownload: args.forceDownload, ccdUrl: args.ccdUrl });

src/cli/chem-comp-dict/create-table.ts

@@ -18,7 +18,7 @@ import { SetUtils } from '../../mol-util/set';
 import { DefaultMap } from '../../mol-util/map';
 import { mmCIF_chemCompBond_schema } from '../../mol-io/reader/cif/schema/mmcif-extras';
 import { ccd_chemCompAtom_schema } from '../../mol-io/reader/cif/schema/ccd-extras';
-import { ensureDataAvailable, getEncodedCif, readCCD, readPVCD } from './util';
+import { DefaultDataOptions, ensureDataAvailable, getEncodedCif, readCCD, readPVCD } from './util';
 
 type CCB = Table<CCD_Schema['chem_comp_bond']>
 type CCA = Table<CCD_Schema['chem_comp_atom']>
@@ -239,8 +239,8 @@ function createAtoms(ccd: DatabaseCollection<CCD_Schema>, pvcd: DatabaseCollecti
     );
 }
 
-async function run(out: string, binary = false, forceDownload = false, ccaOut?: string) {
-    await ensureDataAvailable(forceDownload);
+async function run(out: string, binary = false, options = DefaultDataOptions, ccaOut?: string) {
+    await ensureDataAvailable(options);
     const ccd = await readCCD();
     const pvcd = await readPVCD();
 
@@ -283,12 +283,22 @@ parser.add_argument('--ccaOut', '-a', {
     help: 'Optional generated file output path for chem_comp_atom data.',
     required: false
 });
+parser.add_argument('--ccdUrl', '-c', {
+    help: 'Fetch the CCD from a custom URL. This forces download of the CCD.',
+    required: false
+});
+parser.add_argument('--pvcdUrl', '-p', {
+    help: 'Fetch the PVCD from a custom URL. This forces download of the PVCD.',
+    required: false
+});
 interface Args {
     out: string,
     forceDownload?: boolean,
     binary?: boolean,
-    ccaOut?: string
+    ccaOut?: string,
+    ccdUrl?: string,
+    pvcdUrl?: string
 }
 const args: Args = parser.parse_args();
 
-run(args.out, args.binary, args.forceDownload, args.ccaOut);
+run(args.out, args.binary, { forceDownload: args.forceDownload, ccdUrl: args.ccdUrl, pvcdUrl: args.pvcdUrl }, args.ccaOut);

src/cli/chem-comp-dict/util.ts

@@ -35,9 +35,9 @@ export async function ensureAvailable(path: string, url: string, forceDownload =
     }
 }
 
-export async function ensureDataAvailable(forceDownload = false) {
-    await ensureAvailable(CCD_PATH, CCD_URL, forceDownload);
-    await ensureAvailable(PVCD_PATH, PVCD_URL, forceDownload);
+export async function ensureDataAvailable(options: DataOptions) {
+    await ensureAvailable(CCD_PATH, options.ccdUrl || CCD_URL, !!options.ccdUrl || options.forceDownload);
+    await ensureAvailable(PVCD_PATH, options.pvcdUrl || PVCD_URL, !!options.pvcdUrl || options.forceDownload);
 }
 
 export async function readFileAsCollection<S extends Database.Schema>(path: string, schema: S) {
@@ -68,6 +68,16 @@ export function getEncodedCif(name: string, database: Database<Database.Schema>,
     return encoder.getData();
 }
 
+export type DataOptions = {
+    ccdUrl?: string,
+    pvcdUrl?: string,
+    forceDownload?: boolean
+}
+
+export const DefaultDataOptions: DataOptions = {
+    forceDownload: false
+};
+
 const DATA_DIR = path.join(__dirname, '..', '..', '..', '..', 'build/data');
 const CCD_PATH = path.join(DATA_DIR, 'components.cif');
 const PVCD_PATH = path.join(DATA_DIR, 'aa-variants-v1.cif');