edf2csv 0.5.88 → 0.5.90

package/CHANGELOG.md

@@ -3,6 +3,67 @@
 Notable changes to edf2csv. Versions follow [semantic versioning](https://semver.org); while the
 major version is 0, a minor bump may contain breaking changes.
 
+## 0.5.90
+
+### Fixed: the duration warnings described rows a window had excluded
+
+```
+$ edf2csv rec.edf --start 2 --end 3 --out out --strict
+warning: 1 annotation states a duration below zero ... The value is written to annotations.csv as the file gave it ... check these rows before using the durations.
+warning: 1 annotation states a duration that is not a number, so its duration_s cell is empty.
+--strict: 2 warnings raised, so this run is reported as a failure.
+
+$ cat out/annotations.csv
+onset_s,duration_s,description,record_index
+2.5,0.25,clean event,2
+```
+
+One row, with a populated, positive duration. There is no such value, no such cell and no such
+rows — and `--strict` failed the run over two events it never wrote.
+
+Both warnings are mine, from 0.5.55 and 0.5.58, and both were raised from the counts the
+decoder accumulates over the whole file while `annotations.csv` is filtered to the requested
+window. `--info` had the other half of it: on an EDF+D recording it printed "The value is
+written to annotations.csv as the file gave it" while writing no files at all.
+
+Counted from the events themselves now, after the same filter the writer applies, so the count
+and the sentence describe the same rows. An unreadable duration is carried on the event as
+`durationUnreadable`, because `duration: null` cannot say whether the file gave one — which is
+the ambiguity those warnings exist to flag, and a library caller now has it too. A negative
+duration needs no flag; the value is right there.
+
+`--info` raises them against the window it was given, so it predicts what the conversion will
+say rather than a different set. On a continuous EDF+ it still says nothing, because it does not
+read the whole annotation channel — that is the documented limit of a header read, not a
+regression.
+
+## 0.5.89
+
+### Fixed: the getting-started example asked for a channel the recording does not have
+
+```bash
+edf2csv sleep-study.edf --start 1h --duration 5m \
+  --channels "EEG Fpz-Cz,ECG" --out ./epoch-42
+```
+
+exits 2 with `No channel named "ECG"`. The `--info` table forty lines up the same page lists
+that recording's five channels — EEG Fpz-Cz, EEG Pz-Oz, EOG horizontal, Resp oro-nasal, Temp
+rectal — and none of them is ECG. It is the page's one example of combining a window, a filter
+and a destination, and it is the third command a new reader runs.
+
+The same pair appears on four pages: getting-started, recipes, and faq twice. The fifth is in
+warnings-and-errors, demonstrating the mixed-rate warning, where the quoted rates were wrong for
+this recording too — "2 different sampling rates (256 Hz, 128 Hz)" against a file whose channels
+run at 100, 10 and 1 Hz. That one now pairs `EEG Fpz-Cz` with `Temp rectal` and quotes 100 Hz
+and 1 Hz, which is what the file gives.
+
+api.md's JavaScript examples have been executed against a fixture for a long time. The shell
+examples were checked by nobody, and this is the part of them a test can check without a shell:
+a channel named for this recording either exists in it or does not. It reads every page and the
+README, and it took two attempts — the first missed getting-started's command entirely, because
+that one wraps with a trailing backslash and its `--channels` is on the second line. It joins
+continuations first now.
+
 ## 0.5.88
 
 ### Fixed: a worker killed from outside made the batch exit 2, "the command line is the problem"

package/dist/cli.js

@@ -18,7 +18,7 @@ import process from 'node:process';
 import { EdfError } from './edf/errors.js';
 import { EdfFile } from './edf/reader.js';
 import { buildPlan, withoutFileRateWarning } from './convert/plan.js';
-import { ConversionError, USAGE_ERROR_CODES, convert, defaultOutputDir, stdoutRefusal, } from './convert/run.js';
+import { ConversionError, USAGE_ERROR_CODES, convert, defaultOutputDir, durationDiagnostics, stdoutRefusal, } from './convert/run.js';
 import { ChannelSelectionError } from './convert/channels.js';
 // Shared with the library so a bad option is the same error whichever way it arrived.
 import { OptionError } from './convert/options.js';
@@ -687,6 +687,17 @@ async function showInfo(input, shared, asJson, jsonIndent, batch = false, toStdo
           from describing the recording — and being told the command will not work is exactly
           what was asked. The conversion's own guard supplies the words, so there is one wording.
         */
+        /*
+          What the conversion would say about the durations it would write.
+    
+          Same window the conversion applies, so --info predicts the warnings rather than a
+          different set. Raised here rather than in the parser because the counts have to be of
+          the rows that reach annotations.csv — see durationDiagnostics.
+        */
+        plan.diagnostics.push(...durationDiagnostics(annotationData.annotations, {
+            from: plan.range.startSeconds,
+            to: plan.range.endSeconds,
+        }));
         if (toStdout) {
             const refusal = stdoutRefusal(file, plan);
             if (refusal) {