paperlint 14.3.0 → 14.4.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "paperlint",
-  "version": "14.3.0",
+  "version": "14.4.0",
   "description": "A linter for scientific papers written in LaTeX: catches mistakes before you submit to a conference or journal.",
   "type": "module",
   "license": "MIT",

package/skills/build-label-page/SKILL.md

@@ -4,7 +4,7 @@ description: Build a phone labelling page published as a claude.ai Artifact with
 allowed-tools: [Read, Write, Edit, Grep, Glob, Bash, Artifact, ArtifactData]
 ---
 
-<!-- vigiles:sha256:9d7a2c35df4e6a10 compiled from skills/build-label-page/SKILL.md.spec.ts -->
+<!-- vigiles:sha256:ed201d81c5934675 compiled from skills/build-label-page/SKILL.md.spec.ts -->
 
 # build-label-page — a page the person labels on a phone, and the labels back in the repo
 
@@ -57,7 +57,7 @@ building. In one line each:
    passing proves nothing about this.
 2. **Navigation never relies on in-page scrolling or toasts.** "Next unanswered" is a filter: it hides
    every finished item and group, and a note at the top lists what is left and what each still needs.
-3. **An item is done only when every question on it is answered.** The card names the missing
+3. **An item is done only when every question on it is answered.** The card asks the missing
    question; the progress counter counts complete items only.
 4. **Blind.** Neutral ids (`G001`…), order shuffled with a fixed seed, no model answers, verdicts or
    source labels on the page. The generator refuses to build when an item carries a field outside an

package/skills/build-label-page/SKILL.md.spec.ts

@@ -68,7 +68,7 @@ building. In one line each:
    passing proves nothing about this.
 2. **Navigation never relies on in-page scrolling or toasts.** "Next unanswered" is a filter: it hides
    every finished item and group, and a note at the top lists what is left and what each still needs.
-3. **An item is done only when every question on it is answered.** The card names the missing
+3. **An item is done only when every question on it is answered.** The card asks the missing
    question; the progress counter counts complete items only.
 4. **Blind.** Neutral ids (\`G001\`…), order shuffled with a fixed seed, no model answers, verdicts or
    source labels on the page. The generator refuses to build when an item carries a field outside an

package/skills/build-label-page/references/checklist.md

@@ -41,8 +41,8 @@ filter works in any frame, because the next thing to answer is simply the first
 
 ## 3. An item counts as done only when every question on it is answered
 
-**Do.** Done means all questions answered. The card shows which one is missing ("Still needs:
-boundary"). The progress counter counts complete items only, so "7 of 12" means 7 items with every
+**Do.** Done means all questions answered. The card asks the one still missing in its question
+box, and "Next unanswered" lists what each item still needs. The progress counter counts complete items only, so "7 of 12" means 7 items with every
 question answered. A group is finished when all its items are and its familiarity question (rule 5)
 is answered.
 
@@ -50,7 +50,7 @@ is answered.
 and surfaces weeks later as a gap in the data. A counter that counts partial items reports progress
 that the analysis cannot use.
 
-**Check.** Answer one question of a two-question item: the counter stays put and the card names the
+**Check.** Answer one question of a two-question item: the counter stays put and the card asks the
 other question.
 
 ## 4. Blind: the page tells the labeller nothing about the answer or the source

package/skills/build-label-page/references/closed-choice-generator.md

@@ -155,18 +155,22 @@ Any card may carry these blocks, on any `kind`, and each one is optional. The pa
 }
 ```
 
-- **Order on the card.** The card shows, in this order:
-  1. its place (`labels.place`);
-  2. the body, on a `labels.body` row when that label is set;
-  3. the repeat sentence;
-  4. the older fields (`code`, `unit`, `rules`, `readings`, `file`, `context`);
-  5. the situation's summary;
-  6. the excerpt;
-  7. **the question box**: the `quote` and its source, the question, and its answers with their `answer_notes`;
-  8. `references`, `more` and `table`, each in place under its heading. Nothing on a card is folded:
-     a labeller who has to open a block to answer reads half a card (#392).
-
-  The question therefore sits right under the code it names.
+- **Order on the card.** Each card is a bordered panel, with wide space before the next. It shows, in this order:
+  1. its place (`labels.place`), then its `title`;
+  2. the `quote` and its source;
+  3. the body, on a `labels.body` row when that label is set;
+  4. the repeat sentence;
+  5. the older fields (`code`, `unit`, `rules`, `readings`, `file`, `context`);
+  6. the situation's summary;
+  7. the excerpt;
+  8. **the question box**: the question, its answers with their `answer_notes`, and the wait bar of `min_seconds`;
+  9. the reason box of an answer with `ask_why`, then the status line. The status line never lists
+     what is missing (the question box asks it): it shows «Saved <time>», «Saving…», why a save did not happen, or nothing;
+  10. `references`, `more` and `table`, each in place under its heading. Nothing on a card is folded:
+      a labeller who has to open a block to answer reads half a card (#392).
+
+  A card is read once, top to bottom: the words asked about, the code they are about, the question
+  right under that code, its answers. A card without a `quote` has the same order without step 2.
 
 - **`situation`**: `{summary?, question}`.
   - `summary` is one sentence above the excerpt; optional.
@@ -202,7 +206,7 @@ Any card may carry these blocks, on any `kind`, and each one is optional. The pa
 }
 ```
 
-- **`quote`** shows the words being asked about at the top of the question box, `source` as a small line under them. A card needs `body` or `quote`.
+- **`quote`** shows the words being asked about at the top of the card, above its code, `source` as a small line under them. A card needs `body` or `quote`.
 - **`terms`**: each `term` must occur in the quote, its source or the annotation, inside one backtick span or outside all of them; the build refuses one it cannot find there, and a note entry with neither `quote` nor `code`. A term opens from its first occurrence in the quote, else the source, else the annotation. On the page it is a dotted button: a tap (or hover, where there is hover) opens one floating note anchored to it, at most 280 px wide (300 with a snippet), flipped above the term when there is no room below, kept inside the viewport. A tap outside closes it and does nothing else, so it never answers a card; so does Esc. Answer buttons carry no terms.
 - **`answer_notes`** puts a sub-line under an answer's label on this card: what the answer means here. A question or an answer the page does not have is refused.
 - Whether a note pushes the labeller toward one answer is not checked: the consumer lists each note for the study's author.

package/skills/build-label-page/scripts/label-page-blocks.test.ts

@@ -505,12 +505,12 @@ const unticked = (text: string): string => text.split("`").join("");
 const shownBy = (n: FakeNode): string => n.textContent.replace(/\n/g, "");
 /** The kinds of the card's parts, in the order they appear. */
 const KINDS = [
+  "quote",
+  "quote-source",
   "excerpt-path",
   "code-line",
   "gap",
   "annotation",
-  "quote",
-  "quote-source",
   "question",
   "answer",
   "answer-sub",
@@ -567,19 +567,18 @@ function fixtureFacts() {
       { question: V7_QUESTION.id, value: o.value },
       unticked(notes[o.value] ?? ""),
     ]),
-    status: [
-      [{ state: "missing", missing: [V7_QUESTION.id] }],
-      [`Still needs: ${V7_QUESTION.short}`],
-    ],
+    // The save button and the open question already say what is missing: the line keeps its values
+    // and shows nothing.
+    status: [[{ state: "missing", missing: [V7_QUESTION.id] }], [""]],
     order: [
+      "quote",
+      "quote-source",
       "excerpt-path",
       ...lines.flatMap((l, n) => [
         ...(l.n > (lines[n - 1]?.n ?? l.n) + 1 ? ["gap"] : []),
         "code-line",
         ...(l.mark ? ["annotation"] : []),
       ]),
-      "quote",
-      "quote-source",
       "question",
       ...V7_QUESTION.options.flatMap((o) =>
         o.value in notes ? ["answer", "answer-sub"] : ["answer"],
@@ -589,7 +588,7 @@ function fixtureFacts() {
   };
 }
 
-test("🔴 an approved card reads top to bottom: its path, the code with a gap row for what it leaves out, the marked line and the note under it, the quote, its source, the question, each answer with its sub-line, then what it still needs", async () => {
+test("🔴 an approved card reads top to bottom: the quote and its source, its path, the code with a gap row for what it leaves out, the marked line and the note under it, the question, each answer with its sub-line, then a status line that shows nothing while the card is unanswered", async () => {
   const { page: html, key } = build(setup("v7-face", v7()));
   const { app } = await openLive(html, { practice: V7_DONE });
   const id = Object.entries(key.ids).find(

package/skills/build-label-page/scripts/label-page-browser.test.ts

@@ -1357,3 +1357,64 @@ test(
   },
   BROWSER_TEST_MS,
 );
+
+/** Where a part of a card sits, by its `data-test`: the first one of that kind. */
+const boxOf = async (card: Locator, kind: string) =>
+  card.getByTestId(kind).first().boundingBox();
+const Panel = z.object({
+  border: z.number(),
+  radius: z.number(),
+  padding: z.number(),
+});
+test(
+  "🔴 a quote card reads once from top to bottom, quote, code, question, answers, wait bar, inside a panel of its own; its «still needs» line is not shown",
+  async () => {
+    await onV7Page(
+      "browser-v7-order",
+      async (page, cardOf) => {
+        const card = await cardOf("card-docstring");
+        const parts = ["quote", "excerpt-path", "question", "answer", "wait"];
+        const boxes = await Promise.all(parts.map((k) => boxOf(card, k)));
+        // Guards: each part starts below the one before it ends, so the eye never goes back up.
+        assert.deepEqual(
+          R.zip(boxes, boxes.slice(1)).map(
+            ([a, b]) => a !== null && b !== null && b.y >= a.y + a.height,
+          ),
+          [true, true, true, true],
+          JSON.stringify(boxes),
+        );
+        const status = card.getByTestId("status");
+        assert.deepEqual(
+          [await metaOf(status), await status.isVisible()],
+          [{ state: "missing", missing: ["forbid"] }, false],
+        );
+        // A panel: a border all round, rounded, padded, and wide space before the next card.
+        const panel = Panel.parse(
+          await card.evaluate((e) => {
+            const s = getComputedStyle(e);
+            return {
+              border: parseFloat(s.borderBottomWidth),
+              radius: parseFloat(s.borderTopLeftRadius),
+              padding: parseFloat(s.paddingLeft),
+            };
+          }),
+        );
+        assert.ok(
+          panel.border >= 1 && panel.radius >= 8 && panel.padding >= 12,
+          JSON.stringify(panel),
+        );
+        const cards = await page.getByTestId("item").all();
+        const [first, second] = await Promise.all(
+          cards.slice(0, 2).map(async (c) => c.boundingBox()),
+        );
+        assert.ok(first && second, "two cards drawn");
+        assert.ok(
+          second.y - (first.y + first.height) >= 32,
+          JSON.stringify([first, second]),
+        );
+      },
+      true,
+    );
+  },
+  BROWSER_TEST_MS,
+);

package/skills/build-label-page/scripts/label-page-html.test.ts

@@ -504,3 +504,32 @@ test("🔴 «All items are complete» waits for every group's familiarity too, a
   );
   assert.deepEqual([await said({}), await said(rated)], [false, true]);
 });
+
+test("🔴 the status line names nothing missing: empty while a card is unanswered, «Saved <time>» once part of it is saved, and its values keep what is missing", async () => {
+  const { page, key } = build(setup("status-missing", config()));
+  const [a, b] = Object.keys(key.ids);
+  assert.ok(a !== undefined && b !== undefined);
+  const savedAt = "2026-01-02T10:00:00Z";
+  const { app } = await openLive(page, {
+    answers: { [a]: { answers: { label: "yes" }, savedAt, source: "page" } },
+    groups: {},
+    practice: DONE_PRACTICE,
+  });
+  const statusOf = (id: string) => {
+    const card = app
+      .getByTestId("item")
+      .find((n) => z.object({ id: z.string() }).parse(n.meta()).id === id);
+    const line = card?.getByTestId("status")[0];
+    return [line?.meta(), line?.textContent];
+  };
+  assert.deepEqual(
+    [statusOf(a), statusOf(b)],
+    [
+      [
+        { state: "missing", missing: ["boundary"] },
+        `Saved ${new Date(savedAt).toLocaleTimeString()}`,
+      ],
+      [{ state: "missing", missing: ["label", "boundary"] }, ""],
+    ],
+  );
+});

package/skills/build-label-page/scripts/page-template.html

@@ -3,8 +3,8 @@
 <link rel="stylesheet" href="https://fonts.googleapis.com/css2?family=Roboto:wght@400;500;700&family=Roboto+Mono&display=swap">
 <style>
 /* One look, dark whatever the viewer's theme (Material Ocean, no purple): every color is a token below.
-   One column, sections apart by a hairline and wide space, items flat; the only raised block of an item is
-   the question being asked. Answer buttons are full-width rows with space between them, so a thumb cannot
+   One column, sections apart by a hairline and wide space; each item a bordered panel with wide space after
+   it, and the only raised block inside it the question being asked. Answer buttons are full-width rows with space between them, so a thumb cannot
    land between two of them, and the chosen one has a 2px accent border. */
 :root {
   color-scheme: dark;
@@ -28,8 +28,9 @@ p { margin: 0; max-width: var(--measure); }
 .muted { color: var(--muted); }
 .id { font-family: var(--mono); color: var(--muted); }
 section { display: flex; flex-direction: column; gap: var(--gap-item); }
-.item { display: flex; flex-direction: column; gap: 20px; padding-top: var(--gap-item);
-  border-top: 1px solid var(--line); }
+.item { display: flex; flex-direction: column; gap: 20px; padding: var(--gap-item);
+  border: 1px solid var(--line); border-radius: 14px; }
+.item + .item { margin-top: 24px; }
 /* The evidence of a card: one labelled row per field, the label a short word above its content. The two
    readings sit in the same place on every card: A on a solid accent rule, B on a dashed one. */
 .row { display: flex; flex-direction: column; gap: 4px; }
@@ -37,7 +38,6 @@ section { display: flex; flex-direction: column; gap: var(--gap-item); }
 .reading { border-left: 3px solid var(--accent); padding-left: 12px; }
 .reading.b { border-left: 3px dashed var(--text); }
 mark { background: transparent; color: var(--mark); font-weight: 500; }
-.item:first-of-type { border-top: 0; padding-top: 0; }
 .ask { background: var(--surface); border: 1px solid var(--line); border-radius: 12px; padding: 16px;
   display: flex; flex-direction: column; gap: 12px; }
 .q { color: var(--strong); font-weight: 500; }
@@ -569,11 +569,14 @@ button.more { width: auto; min-height: 44px; border: 0; color: var(--accent); pa
     } else if (state.inflight[key]) {
       as("saving");
       s.textContent = "Saving…";
+    } else if (missing.length && !savedAt) {
+      // The open question and its save button already say what is missing: nothing to show.
+      as("missing");
+      s.hidden = true;
     } else if (missing.length) {
       as("missing");
-      s.className = "status missing";
-      s.textContent = "Still needs: " + missing.map(function (q) { return q.short || q.text; }).join("; ") +
-        (savedAt ? ". Saved " + timeOf(savedAt) : "");
+      s.className = "status saved";
+      s.textContent = "Saved " + timeOf(savedAt);
     } else if (savedAt) {
       as("saved");
       s.className = "status saved";
@@ -591,7 +594,7 @@ button.more { width: auto; min-height: 44px; border: 0; color: var(--accent); pa
     evidence(card, item, ctx);
     var a = answersOf(item.id);
     var gate = locked(item.id);
-    var ask = askOne(card, a, state.editing[item.id], ctx, function (q, v) {
+    var ask = askOne(card, a, state.editing[item.id], function (q, v) {
       delete state.editing[item.id];
       answer(item.id, q.id, v);
     }, function (q) { state.editing[item.id] = q.id; render(); }, off || gate, false, item);
@@ -927,8 +930,10 @@ button.more { width: auto; min-height: 44px; border: 0; color: var(--accent); pa
     if (item.table) card.appendChild(table(item.table));
   }
   // What a card shows above its question, one labelled row per field (an item and a practice item alike).
+  // A card's quote comes first, under its title: the words asked about, then the code they are about.
   function evidence(card, item, ctx) {
     if (item.title) card.appendChild(rich("h3", "", item.title));
+    if (item.quote) card.appendChild(quoteBlock(item, ctx));
     if (item.body !== undefined) {
       var body = rich("p", "body", item.body);
       card.appendChild(DATA.labels.body ? row(DATA.labels.body, body) : body);
@@ -956,13 +961,10 @@ button.more { width: auto; min-height: 44px; border: 0; color: var(--accent); pa
   }
   // One question per screen: the answered ones stay as a line (with "Change"), and only one question
   // is asked: the one being changed, else the first unanswered.
-  // A card's quote sits at the top of its question box, or, with nothing left to ask, in the card itself.
-  function askOne(card, a, editing, ctx, onPick, onChange, off, withExamples, item) {
+  function askOne(card, a, editing, onPick, onChange, off, withExamples, item) {
     var open = DATA.questions.filter(function (q) { return !valid(DATA.allowed.questions[q.id], a[q.id]); });
     var asked = DATA.questions.filter(function (q) { return q.id === editing; })[0] || open[0];
     var situation = item.situation, notes = item.answer_notes || {};
-    var quote = item.quote ? quoteBlock(item, ctx) : null;
-    if (quote && !asked) card.appendChild(quote);
     DATA.questions.forEach(function (q) {
       if (q === asked || !valid(DATA.allowed.questions[q.id], a[q.id])) return;
       var line = el("p", "answered", (q.short || q.text) + ": " + labelOf(q, a[q.id]) + " ");
@@ -977,7 +979,6 @@ button.more { width: auto; min-height: 44px; border: 0; color: var(--accent); pa
     });
     if (!asked) return null;
     var ask = el("div", "ask");
-    if (quote) ask.appendChild(quote);
     // A card's situation asks the page's first question in its own words; the others keep theirs.
     ask.appendChild(mark(situation && asked === DATA.questions[0] ? situated("p", "q", situation.question, situation.line) : rich("p", "q", asked.text),
       "question", { question: asked.id }));
@@ -1013,7 +1014,7 @@ button.more { width: auto; min-height: 44px; border: 0; color: var(--accent); pa
     var ctx = termsOf(p);
     evidence(card, p, ctx);
     var a = practiceAnswers(p.id);
-    askOne(card, a, undefined, ctx, function (q, v) { answerPractice(p.id, q.id, v); }, null, off, true, p);
+    askOne(card, a, undefined, function (q, v) { answerPractice(p.id, q.id, v); }, null, off, true, p);
     if (practiceAnswered(p)) {
       var expected = DATA.questions.map(function (q) { return labelOf(q, p.expected[q.id]); }).join(" · ");
       card.appendChild(rich("p", practiceMatched(p) ? "status saved" : "quote",