Parse HDF5 string padding

silx-kit · Jun 4, 2024 · 7786645 · 7786645
1 parent 6c1678e
commit 7786645
Show file tree

Hide file tree

Showing 15 changed files with 512 additions and 186 deletions.
diff --git a/package.json b/package.json
@@ -54,6 +54,9 @@
     "wait-on": "7.2.0"
   },
   "pnpm": {
+    "overrides": {
+      "eslint-config-galex>typescript": "$typescript"
+    },
     "peerDependencyRules": {
       "allowedVersions": {
         "@phenomnomnominal/tsquery>typescript": "5.x",

diff --git a/packages/app/src/providers/h5grove/models.ts b/packages/app/src/providers/h5grove/models.ts
@@ -88,6 +88,7 @@ export interface H5GroveTimeType extends H5GroveBaseType {
 export interface H5GroveStringType extends H5GroveBaseType {
   class: 3;
   cset: number;
+  strPad?: number; // optional for backwards compatibility with h5grove <= 2.1.0
   vlen: boolean;
 }
 

diff --git a/packages/app/src/providers/h5grove/utils.test.ts b/packages/app/src/providers/h5grove/utils.test.ts
@@ -1,4 +1,4 @@
-import { H5T_CSET, H5T_ORDER } from '@h5web/shared/h5t';
+import { H5T_CSET, H5T_ORDER, H5T_STR } from '@h5web/shared/h5t';
 import {
   arrayType,
   bitfieldType,
@@ -42,13 +42,13 @@ describe('parseDType', () => {
   it('should convert string types', () => {
     expect(
       parseDType({ class: 3, size: 6, cset: 0, vlen: false }),
-    ).toStrictEqual(strType(H5T_CSET.ASCII, 6));
+    ).toStrictEqual(strType(H5T_CSET.ASCII, undefined, 6));
     expect(
       parseDType({ class: 3, size: 6, cset: 0, vlen: true }),
     ).toStrictEqual(strType(H5T_CSET.ASCII));
     expect(
-      parseDType({ class: 3, size: 6, cset: 1, vlen: false }),
-    ).toStrictEqual(strType(H5T_CSET.UTF8, 6));
+      parseDType({ class: 3, size: 6, cset: 1, strPad: 1, vlen: false }),
+    ).toStrictEqual(strType(H5T_CSET.UTF8, H5T_STR.NULLPAD, 6));
     expect(
       parseDType({ class: 3, size: 6, cset: 1, vlen: true }),
     ).toStrictEqual(strType(H5T_CSET.UTF8));

diff --git a/packages/app/src/providers/h5grove/utils.ts b/packages/app/src/providers/h5grove/utils.ts
@@ -149,8 +149,8 @@ export function parseDType(type: H5GroveType): DType {
   }
 
   if (h5tClass === H5T_CLASS.STRING) {
-    const { cset, vlen } = type;
-    return strType(cset, vlen ? undefined : size);
+    const { cset, strPad, vlen } = type;
+    return strType(cset, strPad, vlen ? undefined : size);
   }
 
   if (h5tClass === H5T_CLASS.BITFIELD) {

diff --git a/packages/app/src/providers/hsds/models.ts b/packages/app/src/providers/hsds/models.ts
@@ -1,3 +1,4 @@
+import type { H5T_CSET, H5T_STR } from '@h5web/shared/h5t';
 import type { Entity } from '@h5web/shared/hdf5-models';
 
 /* --------------------- */
@@ -104,7 +105,8 @@ export interface HsdsNumericType {
 
 export interface HsdsStringType {
   class: 'H5T_STRING';
-  charSet: `H5T_CSET_${'ASCII' | 'UTF8'}`;
+  charSet: `H5T_CSET_${keyof typeof H5T_CSET}`;
+  strPad: `H5T_STR_${keyof typeof H5T_STR}`;
   length: number | 'H5T_VARIABLE';
 }
 

diff --git a/packages/app/src/providers/hsds/utils.test.ts b/packages/app/src/providers/hsds/utils.test.ts
@@ -1,4 +1,4 @@
-import { H5T_CSET, H5T_ORDER } from '@h5web/shared/h5t';
+import { H5T_CSET, H5T_ORDER, H5T_STR } from '@h5web/shared/h5t';
 import type { DType } from '@h5web/shared/hdf5-models';
 import {
   arrayType,
@@ -52,22 +52,26 @@ describe('convertHsdsType', () => {
     const asciiStr: HsdsStringType = {
       class: 'H5T_STRING',
       charSet: 'H5T_CSET_ASCII',
+      strPad: 'H5T_STR_NULLTERM',
       length: 25,
     };
 
     expect(convertHsdsType(asciiStr)).toStrictEqual(
-      strType(H5T_CSET.ASCII, 25),
+      strType(H5T_CSET.ASCII, H5T_STR.NULLTERM, 25),
     );
   });
 
   it('should convert variable-length UTF-8 string type', () => {
     const unicodeStr: HsdsStringType = {
       class: 'H5T_STRING',
       charSet: 'H5T_CSET_UTF8',
+      strPad: 'H5T_STR_NULLPAD',
       length: 'H5T_VARIABLE',
     };
 
-    expect(convertHsdsType(unicodeStr)).toStrictEqual(strType(H5T_CSET.UTF8));
+    expect(convertHsdsType(unicodeStr)).toStrictEqual(
+      strType(H5T_CSET.UTF8, H5T_STR.NULLPAD),
+    );
   });
 
   it('should convert integer types', () => {

diff --git a/packages/app/src/providers/hsds/utils.ts b/packages/app/src/providers/hsds/utils.ts
@@ -1,5 +1,5 @@
 import { assertArray, isGroup } from '@h5web/shared/guards';
-import { H5T_CSET, H5T_ORDER } from '@h5web/shared/h5t';
+import { H5T_CSET, H5T_ORDER, H5T_STR } from '@h5web/shared/h5t';
 import type {
   ArrayShape,
   Attribute,
@@ -138,11 +138,14 @@ export function convertHsdsType(hsdsType: HsdsType): DType {
       return convertHsdsCompoundType(hsdsType);
 
     case 'H5T_STRING': {
-      const { charSet, length } = hsdsType;
+      const { charSet, strPad, length } = hsdsType;
       return strType(
         H5T_CSET[
           charSet.slice(charSet.lastIndexOf('_') + 1) as keyof typeof H5T_CSET
         ],
+        H5T_STR[
+          strPad.slice(strPad.lastIndexOf('_') + 1) as keyof typeof H5T_STR
+        ],
         length === 'H5T_VARIABLE' ? undefined : length,
       );
     }

diff --git a/packages/app/src/providers/mock/mock-file.ts b/packages/app/src/providers/mock/mock-file.ts
@@ -1,4 +1,4 @@
-import { H5T_CSET } from '@h5web/shared/h5t';
+import { H5T_CSET, H5T_STR } from '@h5web/shared/h5t';
 import type { GroupWithChildren } from '@h5web/shared/hdf5-models';
 import {
   arrayType,
@@ -62,7 +62,7 @@ export function makeMockFile(): GroupWithChildren {
         }),
         scalar('scalar_compound', ['foo', 2], {
           type: compoundType({
-            str: strType(H5T_CSET.ASCII, 3),
+            str: strType(H5T_CSET.ASCII, H5T_STR.NULLPAD, 3),
             int: intType(8),
           }),
           attributes: [

diff --git a/packages/h5wasm/package.json b/packages/h5wasm/package.json
@@ -47,7 +47,7 @@
   },
   "dependencies": {
     "comlink": "4.4.1",
-    "h5wasm": "0.7.4",
+    "h5wasm": "0.7.5",
     "nanoid": "5.0.7"
   },
   "devDependencies": {