edfcore 0.4.511 → 0.4.513

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 = "0.4.511";
+export declare const VERSION = "0.4.513";
 //# 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 = '0.4.511';
+export const VERSION = '0.4.513';
 //# sourceMappingURL=constants.js.map

package/docs/CHANGELOG.md

@@ -6,6 +6,46 @@ 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.
 
+## 0.4.513
+
+- **Added** a bounds check on the evidence a diagnostic points at: the byte range is inside the
+  file, the signal index names a signal, and the record index is inside the records that were
+  read. 0.4.510 checks that `raw` matches the bytes at the offset, which settles the offset for
+  every diagnostic carrying text to compare. `PARTIAL_FINAL_RECORD` and `TRAILING_BYTES` carry
+  none — they point at samples, and quoting samples as text says nothing — so nothing reached
+  their offsets at all beyond three hand-picked sizes.
+- An offset past the end of the file is not a crash. `subarray` clamps, so an evidence block
+  renders short or empty and the reader concludes the bytes were empty. Both codes are now checked
+  across a range of truncations and appends, including one byte and one byte short of a whole
+  record, and both are asserted to end exactly at the end of the file — which is what "the rest of
+  the file after the last whole record" means.
+- The TAL side is read from record 2 rather than record 0. A TAL diagnostic reports
+  `records.start + position` while its buffer offset uses `position` alone, and on a read starting
+  at record 0 those are the same number — which is what almost every fixture does, so the one
+  arithmetic that can confuse a file position with a buffer position was exercised only where the
+  two agree.
+
+## 0.4.512
+
+- **Added** a read behind each of the four header-recovery diagnostics. When the header's numbers
+  disagree with the file, `parseHeader` recovers rather than refuses, and each recovery ends by
+  saying what a read will now do: which byte record 0 comes from, which records exist, which bytes
+  are never decoded. Every one of those is a claim about `decodeDigital` output, made in a module
+  that never calls it, and the existing tests stopped at the header fields the sentence mentions.
+- The gap is not hypothetical for `HEADER_SIZE_MISMATCH`. A three-signal file declaring a 512-byte
+  header has the same `recordCount` under either size, because the arithmetic that produced it used
+  whichever size the reader used; what separates the two readings is which 1024 bytes come back as
+  record 0. The declared offset lands in the middle of the per-signal block and decodes to numbers
+  rather than to an error, so the test asserts the ramp the writer wrote — and asserts alongside it
+  that the declared offset would have given something else, or the first assertion proves nothing.
+- `TRAILING_BYTES` and `PARTIAL_FINAL_RECORD` claim their bytes are never decoded and never padded
+  into existence. Both fixtures now write those bytes as `0x7F`, which decodes to 32639, a value
+  the sample ramp never produces — so "never decoded" is checked by looking for it in the output
+  rather than by trusting an offset. Reaching either by record number is refused, not returned.
+- `TRUNCATED_FILE` says the missing records are not readable and will not be fabricated. The
+  surviving records read exactly as they do in the intact file, and the pair the header claimed is
+  an `EdfRangeError` rather than zeros.
+
 ## 0.4.511
 
 - **Fixed** `physical-values.md` documenting the limitation 0.4.509 removed. Its note said a fifth

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.4.511",
+  "version": "0.4.513",
   "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 = '0.4.511';
+export const VERSION = '0.4.513';