edfcore 0.5.14 → 0.5.16

package/dist/constants.d.ts

@@ -109,5 +109,5 @@ export declare const SIGNAL_FIELD_BLOCK_OFFSETS: {
     readonly reserved: 224;
 };
 /** Published package version. Kept in sync with package.json by a test. */
-export declare const VERSION = "0.5.14";
+export declare const VERSION = "0.5.16";
 //# sourceMappingURL=constants.d.ts.map

package/dist/constants.js

@@ -79,5 +79,5 @@ export const SIGNAL_FIELD_BLOCK_OFFSETS = {
     reserved: 224,
 };
 /** Published package version. Kept in sync with package.json by a test. */
-export const VERSION = '0.5.14';
+export const VERSION = '0.5.16';
 //# sourceMappingURL=constants.js.map

package/docs/CHANGELOG.md

@@ -6,6 +6,57 @@ alone does not tell you whether you were affected.
 edfcore is pre-1.0. Patch releases have carried behaviour changes where the old behaviour was a
 defect; those are called out below.
 
+## 0.5.16
+
+- **Added** the last program on `annotations.md`, run. That page ends with the conversion it exists
+  to teach — an event onset to a sample index, in integers, because `Math.round(onset * sampleRateHz)`
+  drifts and the rate is `undefined` outright when the record duration is zero. It is thirty lines,
+  it prints two lines of output, and it is the thing a reader copies. `annotations-page.test.ts`
+  runs the top of the page and stops before it.
+- The fixture is the page's own file, reconstructed from numbers the page already prints elsewhere
+  on it: `region.recordByteOffset` is 768, so the data signals hold 384 two-byte samples a record,
+  and the worked example's `256n` at one second fixes `EEG Fpz-Cz` at 256 of them, leaving 128 for
+  the second channel. One fixture now produces the four-event transcript at the top and the two
+  output lines at the bottom, both compared against text read out of the page.
+- The printed float turns out to be load-bearing. `0.007629510948348211` is `bitValue * offset` on a
+  ±500 µV signal against the standard 16-bit digital range — the physical value of digital zero,
+  which is the first sample of every record — and it is printed twice because both events land on a
+  record boundary. It is asserted as that expression rather than as sixteen copied digits.
+- Four claims in the prose under the program are checked with it. The `floorDiv` the page tells you
+  to write is compared character for character against the body of `floorDiv` in `src/tal/ticks.ts`.
+  Step 4's own recommendation — that `firstSampleIndex` comes back as `recordIndex x samplesPerRecord`
+  — is asserted rather than assumed. The floor "matters for negative onsets" is checked off the
+  sample grid, where the sentence has content: the page's own `-0.75` is an exact multiple at 256
+  samples a second, so both roundings agree on it, and a `-0.7501` onset is where truncation lands
+  one sample later, toward the file start, exactly as the page says.
+- The Warning at the end is checked too, and it is the sharpest of the four. On an EDF+D file with a
+  five-second hole the formula answers for an instant inside it with `640n` — an ordinary index,
+  well inside the samples the file holds — while `index.locate(2.5)` correctly returns `undefined`.
+  For an instant past the gap that IS in the file, the formula names record 7 in a four-record file
+  and the scanned index names record 2.
+
+## 0.5.15
+
+- **Added** direct tests for `floorDiv` and `ceilDiv`, which had none. `tal/ticks.ts` owns them, and
+  `ticks.test.ts` — the file named for that module — covers the onset parsers and the tick
+  conversions and neither of these. Five modules import them: `envelope.ts`, `biosemi.ts`,
+  `format-annotations.ts`, `sample-locate.ts` and `time/window.ts`. Every assertion they had was
+  incidental, made through a caller that happens to divide.
+- The whole content of both functions is the negative side. Bigint `/` truncates toward zero, so on
+  a negative numerator it is neither floor nor ceiling. Both are now checked against `Math.floor`
+  and `Math.ceil` over a grid crossing zero — 85 cells, small enough that float division is an exact
+  oracle — with `ceilDiv(a, b) === -floorDiv(-a, b)` asserted across it so the pair cannot drift
+  apart. The operator is then shown to disagree on every negative non-multiple **and only there**,
+  which is what stops the grid from proving nothing.
+- The docblock's reason for their living in `ticks.ts` — "three modules had grown their own copies
+  of the same four lines" — is enumerated out of `src/`: exactly one module declares them, five
+  call them, and each of the five imports them rather than redeclaring.
+- The last block pins the operand that makes this ordinary rather than an edge case. `trimToWindow`
+  selects sample `j` through `floorDiv((R - 1) * S, D) + 1`, and a window beginning at its own
+  chunk's start makes `R` zero — so the numerator is `-S`, negative for every file, with no unusual
+  geometry and nothing pre-stimulus about it. Truncation there returns `1` where the floor returns
+  `0`, and the window loses the first sample the caller aligned it to.
+
 ## 0.5.14
 
 - **Added** the CLI's default print cap driven at all four places it is applied.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.5.14",
+  "version": "0.5.16",
   "description": "Modern, typed, zero-dependency reader for EDF, EDF+, BDF and BDF+ biosignal files. Works in browsers and Node with true random access.",
   "keywords": [
     "edf",

package/src/constants.ts

@@ -93,4 +93,4 @@ export const SIGNAL_FIELD_BLOCK_OFFSETS = {
 } as const;
 
 /** Published package version. Kept in sync with package.json by a test. */
-export const VERSION = '0.5.14';
+export const VERSION = '0.5.16';