@instafy/cli 0.3.0 → 0.4.0

package/dist/bind/src/starters/ranked/app.js

@@ -4,7 +4,9 @@
  * data.json:
  *   title, lede     the page title and its one-line answer (the verdict, and the biggest unknown if there is one)
  *   basis           how it is ranked, one line: "Ranked by safety reviews, then price."
- *   filter (only for a constraint the person stated): {key: "price", max: 300, label: "Under $300"}
+ *   filter (only for a constraint the person stated): {key: "price", max: 300, label: "Under $300", unknown?:
+ *            "Price unknown"}. Only a price the page shows counts: a row without one is never listed as under $300;
+ *            it is listed apart, under "Price unknown".
  *   items: [{ref: "option:a", rank, name, why (one short line), badge?, specs: [fact] (at most 3), price?: fact,
  *            store?: {name, url, readable}, details?: [{label, fact}], sources?: [{name, url}]}]
  *   badge: only for a label the person must notice. "Our pick" only when it is your own pick and "why" says why;
@@ -12,8 +14,9 @@
  *          "BabyGearLab: best budget"), never a "Details" toggle. Nothing is added by itself.
  *   price: the price as a fact (not in specs), with "asof" when the source dates it; a live check replaces it. A
  *          review site's price is its list price: "list": true, so the row says "list price" and never reads as the
- *          store's price now. When every price comes from one source on one date, one line under the list names the
- *          source once ("List prices on BabyGearLab ↗ · seen 5 Aug"), and "Sources" does not name it again.
+ *          store's price now. When one page on one date gives every price, one line under the list names it once
+ *          ("List prices on BabyGearLab ↗ · seen 5 Aug"), and "Sources" does not name it again. A missing price reads
+ *          "Price unknown", then why ({"note": "the store blocks price checks"}).
  *   details: only facts the row does not already show.
  *   unchecked: ["what was not checked", …]
  *   fact: {"v": "$145", "src": "Example Store", "url": "https://…", "date": "2026-10-17", "est": false}
@@ -31,24 +34,36 @@
     render: render
   });
 
-  function priceOf(item) { return site.live(item.ref, "price") || item.price || null; }
+  /** The price a row shows: a live check, else the seed price (also when the check found the store blocked). */
+  function priceOf(item) {
+    var live = site.live(item.ref, "price");
+    if (live && site.blocked(item.ref)) return item.price || live;
+    return live || item.price || null;
+  }
   function valueFor(item, key) {
     if (key === "price") return priceOf(item);
     var d = (item.details || []).filter(function (x) { return x.key === key; })[0];
     return d ? d.fact : null;
   }
-  function fits(data, item) {
+  /** "Price unknown": the heading of the rows the filter cannot place. */
+  function unknownTitle(data) {
     var f = data.filter;
-    if (!f) return true;
-    var n = site.num(valueFor(item, f.key));
-    if (n == null) return true;
-    return (f.max == null || n <= f.max) && (f.min == null || n >= f.min);
+    if (f.unknown) return f.unknown;
+    if (f.key === "price") return "Price unknown";
+    var d = (data.items || []).map(function (i) { return (i.details || []).filter(function (x) { return x.key === f.key; })[0]; }).filter(Boolean)[0];
+    return ((d && d.label) || f.key) + " unknown";
   }
+  function lowerFirst(text) { text = String(text || ""); return text.charAt(0).toLowerCase() + text.slice(1); }
   function render(data, st) {
     site.head(data); // title, lede and the basis line ("Ranked by …") come first, before any button
     var items = (data.items || []).slice().sort(function (a, b) { return (a.rank || 0) - (b.rank || 0); });
-    var shown = st.show === "all" ? items : items.filter(function (i) { return fits(data, i); });
-    var hidden = items.length - shown.length;
+    // the stated limit shows the rows within it, then the rows nobody could place ("Price unknown", never counted as
+    // within it); "All" is the whole ranking
+    var parts = st.show === "all" || !data.filter ? { "in": items, unknown: [], out: [] }
+      : site.byFilter(items, data.filter, function (i) { return valueFor(i, data.filter.key); });
+    var shown = parts["in"];
+    var apart = parts.unknown;
+    var hidden = parts.out.length;
     var priced = items.some(function (i) { return i.price || i.store; });
     var shared = sharedPrice(items);
 
@@ -72,7 +87,14 @@
       // today's price is ("Open a store for today's price."), instead of a box about checks nobody was offered
       var inline = !!(shared && stores && site.storesBlocked(items));
       if (inline) stores = null;
-      wrap.appendChild(h("ol", { "class": "list ranked" }, shown.map(function (item) { return row(item, st, priced, shared); })));
+      if (shown.length) wrap.appendChild(h("ol", { "class": "list ranked" }, shown.map(function (item) { return row(item, st, priced, shared); })));
+      else wrap.appendChild(h("p", { "class": "cap", text: "None is known to be " + lowerFirst(data.filter.label || "within the limit") + "." }));
+      if (apart.length) {
+        wrap.appendChild(h("section", { "class": "group", attrs: { id: "group-unknown", "aria-label": unknownTitle(data) } },
+          h("div", { "class": "group-head" }, h("h2", { text: unknownTitle(data) }),
+            h("span", { "class": "cap", text: "Not counted as " + lowerFirst(data.filter.label || "within the limit") })),
+          h("ol", { "class": "list ranked" }, apart.map(function (item) { return row(item, st, priced, shared); }))));
+      }
       if (hidden > 0) {
         wrap.appendChild(h("p", { "class": "cap" }, hidden + " hidden · ",
           h("button", { "class": "link-btn", text: "Show all", attrs: { type: "button", "data-key": "show-all" }, on: { click: function () { site.setState({ show: "all" }); } } })));
@@ -93,11 +115,12 @@
     site.notes(items.length ? data.unchecked : [], items.length ? sources : []);
   }
 
-  /** The price fact every priced row shares (same source and date), else null. */
+  /** The price fact every priced row shares (one page on one date), else null. */
   function sharedPrice(items) {
     var facts = items.map(function (i) { return i.price; }).filter(Boolean);
     if (facts.length < 2) return null;
-    var key = function (f) { return f && f.src && !f.est && f.v != null ? [f.src, f.asof || f.date || "", f.list ? "list" : ""].join("|") : null; };
+    var page = function (f) { var u = site.safeUrl(f.url); return u ? u.split("#")[0] : "src:" + f.src; };
+    var key = function (f) { return f && f.src && !f.est && site.valueText(f) ? [page(f), f.asof || f.date || "", f.list ? "list" : ""].join("|") : null; };
     var k = key(facts[0]);
     return k && facts.every(function (f) { return key(f) === k; }) ? facts[0] : null;
   }
@@ -149,7 +172,7 @@
         item.why ? h("p", { "class": "why", text: item.why }) : null,
         specs.length ? h("p", { "class": "cap", text: specs.join(" · ") }) : null,
         // every priced row has the same price line, with or without a store
-        priced ? site.trust(live || item.price || null, { label: "Price", price: true, ref: item.ref, intent: "read_offer", store: item.store, seed: item.price || null, shared: !live && !!shared && !!item.price }) : null,
+        priced ? site.trust(live || item.price || null, { label: "Price", unknown: "Unknown", price: true, ref: item.ref, intent: "read_offer", store: item.store, seed: item.price || null, shared: !live && !!shared && !!item.price }) : null,
         open && hasMore(item) ? more(item) : null));
   }
 

package/dist/bind/src/starters/ranked/data.json

@@ -1,6 +1,6 @@
 {
   "title": "Top three options",
-  "lede": "Option A is the best all-round pick. Option C costs least, but its price was not checked.",
+  "lede": "Option A is the best all-round pick. Option C's price is unknown: its store blocks price checks.",
   "basis": "Ranked by review scores, then price.",
   "items": [
     {
@@ -106,7 +106,7 @@
       "ref": "option:option-c",
       "rank": 3,
       "name": "Option C",
-      "why": "The lowest price, fewer reviews",
+      "why": "Well reviewed, by fewer people",
       "specs": [
         {
           "v": "9.8 oz",

package/dist/bind/src/starters/site.js

@@ -9,12 +9,18 @@
  * - Rows: give each row id="row-<slug>" and data-ref="<kind>:<slug>" (site.rowAttrs(ref)). A "select" command from
  *   chat scrolls to the row and marks it; "highlight" fades the rows it names. "toast" and "reload" are the client's.
  * - Facts: one shape everywhere, {"v": "8.9 oz", "src": "Maker", "url": "https://…", "date": "2026-10-17", "est": false}.
- *   Optional "asof": the source's own date, shown in place of "date" ("seen 5 Aug"); a price with an "asof" over 30
- *   days old is muted and says "may have changed", like a live price past its time.
- *   site.trust(fact | record, opts) draws the one trust line: "Source ↗ · seen 17 Oct", "at Store ↗ · checked 4 min
- *   ago", "Estimate", "Price not checked" (nobody checked it), "Not listed · Source ↗" (a fact with a source and no
- *   value: the source was read and does not give it). A price fact with "list": true is that source's list price,
- *   never a store's price now: "$900 · list price on BabyGearLab ↗ · seen 5 Aug".
+ *   "url" is the page it was read on and "date" the day it was read; each value cites them: "checked 17 Oct on
+ *   maker.com ↗" (a price: "seen 17 Oct on shop.com ↗"). Optional "asof": the source's own date, shown in place of
+ *   "date" ("seen 5 Aug on …"); a price with an "asof" over 30 days old is muted and says "may have changed", like a
+ *   live price past its time. A value that only says nothing is known ("—", "n/a", "?", "unknown") is no value.
+ *   site.trust(fact | record, opts) draws the one trust line: "checked 17 Oct on maker.com ↗", "at Store ↗ · checked
+ *   4 min ago", "Estimate", "Price not checked" (nobody checked it), "Not listed · checked 16 Oct on maker.com ↗" (a
+ *   fact with a source and no value: the source was read and does not give it); in a table of options a value nobody
+ *   found reads opts.unknown, "Unknown · why" ("Price unknown"). A price fact with "list": true is that source's
+ *   list price, never a store's price now: "$900 · list price on BabyGearLab ↗ · seen 5 Aug".
+ * - Filters: site.within(value, filter) is "in", "out" or "unknown" for the page's one filter ({key, max, min}), and
+ *   site.byFilter(rows, filter, valueOf) splits rows into {in, unknown, out}. Only a value the page shows counts: a
+ *   row without one is never "under $140"; the starters list it apart ("Price unknown").
  * - Live checks: site.checkAll(rows) is the page's one primary button ("Check all prices", with two or more checkable
  *   rows). trust(value, {ref, intent, store, seed}) ends a row's caption in its one quiet action: "Check price" (only
  *   without "Check all prices"), "Try again" after an error, "Open store ↗" when the store can't be read. While a
@@ -147,24 +153,82 @@
     }
   }
   function isRecord(x) { return !!(x && typeof x === "object" && x.prov && x.type); }
-  /** The text of a fact or record value ("" when there is none). */
+  /** A value that only says nothing is known: blank, dashes, "?", "n/a", "unknown", "TBD". */
+  var NO_VALUE = /^(?:[\s\-\u2010-\u2015?.\u00b7\u2026/]*|n\s*\/\s*a|unknown|tb[ad])$/i;
+  /** The text of a fact or record value ("" when there is none, or it only says nothing is known: a dash in a table
+   * is never drawn as if it were a value). */
   function valueText(x) {
     if (!x) return "";
+    var text;
     if (isRecord(x)) {
       if (x.type === "offer") return money(x.price);
       if (x.value === true) return "Yes";
       if (x.value === false) return "No";
-      return x.value == null ? "" : String(x.value);
+      text = x.value == null ? "" : String(x.value);
+    } else {
+      text = x.v == null ? "" : String(x.v);
+    }
+    return NO_VALUE.test(text.trim()) ? "" : text;
+  }
+  /** The first number in a text, read the way shops write it: a ".", ",", space or "'" before exactly three digits
+   * groups thousands ("1.299", "1 299", "1'299", "1,299" are 1299), and the last "." or "," is the decimal mark
+   * otherwise ("1.299,00" and "1,299.00" are 1299, "549,99" is 549.99, "4.6" is 4.6, "0.125" is 0.125). */
+  function numberIn(text) {
+    var m = /(-?)(\d+)((?:[.,'’    ]\d+)*)/.exec(String(text == null ? "" : text));
+    if (!m) return null;
+    var head = m[2];
+    var parts = m[3] ? m[3].match(/[.,'’    ]\d+/g) : [];
+    var mark = function (p) { return p.charAt(0); };
+    var dec = -1;
+    parts.forEach(function (p, i) { if (mark(p) === "." || mark(p) === ",") dec = i; });
+    // the last "." or "," groups thousands when three digits follow it, one to three (not just 0) lead, and no other
+    // kind of decimal mark comes before it: "1.299", "1,299,000"
+    if (dec >= 0 && parts[dec].length === 4 && head.length <= 3 && head !== "0" &&
+      parts.slice(0, dec).every(function (p) { return (mark(p) !== "." && mark(p) !== ",") || mark(p) === mark(parts[dec]); })) dec = -1;
+    var whole = head;
+    var end = dec >= 0 ? dec : parts.length;
+    for (var i = 0; i < end; i++) {
+      // a mark that does not group thousands ends the number: "2 3-year plans" is 2
+      if (head.length > 3 || parts[i].length !== 4) return Number(m[1] + whole);
+      whole += parts[i].slice(1);
     }
-    return x.v == null ? "" : String(x.v);
+    return Number(m[1] + whole + (dec >= 0 ? "." + parts[dec].slice(1) : ""));
   }
-  /** A number to sort by: fact.n, an offer's amount, else the first number in the value text. */
+  /** A number to sort by: fact.n, an offer's amount, else the first number in the value text (numberIn). */
   function num(x) {
     if (!x) return null;
     if (isRecord(x) && x.type === "offer") return x.price && typeof x.price.amount === "number" ? x.price.amount : null;
     if (typeof x.n === "number") return x.n;
-    var m = /-?\d[\d,]*(?:\.\d+)?/.exec(valueText(x));
-    return m ? Number(m[0].replace(/,/g, "")) : null;
+    return numberIn(valueText(x));
+  }
+  /** The number of a value the page shows, else null. A value counts only when its trust line draws it: a live value,
+   * an estimate, or a fact with its source. An unsourced fact (drawn "Not checked"), a check that could not read it,
+   * and a dash are unknown, so no filter or sort counts them. */
+  function known(x) {
+    if (!x) return null;
+    if (isRecord(x)) {
+      if (x.type === "claim" && (x.status === "uncertain" || x.status === "not_found" || x.status === "error")) return null;
+      return valueText(x) ? num(x) : null;
+    }
+    if (!valueText(x) || !(x.src || x.est)) return null;
+    return num(x);
+  }
+  /** Where a value stands against the page's one filter ({key, max?, min?}): "in", "out", or "unknown" when the page
+   * has no number for it. Unknown is never "in": a row without a price is not "under $140". */
+  function within(x, filter) {
+    if (!filter) return "in";
+    var n = known(x);
+    if (n == null) return "unknown";
+    var max = filter.max == null || filter.max === "" ? null : Number(filter.max);
+    var min = filter.min == null || filter.min === "" ? null : Number(filter.min);
+    return (max == null || n <= max) && (min == null || n >= min) ? "in" : "out";
+  }
+  /** Rows split by the page's one filter, each part in the rows' own order: {in, unknown, out}. Without a filter every
+   * row is "in". valueOf(row) gives the row's value for the filter's key. */
+  function byFilter(rows, filter, valueOf) {
+    var parts = { "in": [], unknown: [], out: [] };
+    (rows || []).forEach(function (r) { parts[within(filter ? valueOf(r) : null, filter)].push(r); });
+    return parts;
   }
 
   // ---------- live values ----------
@@ -189,6 +253,30 @@
     var when = x && (x.asof || x.date);
     return when && date(when) ? "seen " + date(when) : "";
   }
+  /** Where and when a value was read, as one caption part: "checked 17 Oct on maker.com ↗" (the host links the page it
+   * was read on). A price, or a date the source gives ("asof"), is "seen": "seen 5 Aug on reviews.example ↗", never
+   * "checked", so it does not read as today's price. Without a page link it names the source: "Maker · checked 17
+   * Oct". o.price: the value is a price. */
+  function citeParts(x, o) {
+    o = o || {};
+    if (!x) return [];
+    var when = x.asof || x.date;
+    var day = when && date(when) ? ((o.price || x.asof ? "seen " : "checked ") + date(when)) : "";
+    var site = host(x.url);
+    if (site) return [h("span", { "class": "cite" }, day ? day + " on " : "", link(x.url, site))];
+    var out = [];
+    if (x.src) out.push(String(x.src));
+    if (day) out.push(h("span", { "class": "age", text: day }));
+    return out;
+  }
+  /** cite() as one inline piece, for a line under a list ("Weight: checked 17 Oct on maker.com ↗"). */
+  function cite(x, o) {
+    var parts = citeParts(x, o);
+    if (!parts.length) return null;
+    var el = h("span", null);
+    parts.forEach(function (p, i) { if (i) el.appendChild(document.createTextNode(" · ")); append(el, p); });
+    return el;
+  }
   function newest(list) {
     return list.slice().sort(function (a, b) { return (Date.parse(b.prov.observed_at) || 0) - (Date.parse(a.prov.observed_at) || 0); })[0] || null;
   }
@@ -233,12 +321,14 @@
 
   /**
    * The one trust line for a fact or a live record: the value, then a quiet caption.
-   *   value · source ↗ · seen 17 Oct          (a seed fact; the source's own date when it gives one)
+   *   value · checked 17 Oct on maker.com ↗    (a seed fact: the page it was read on and the day; "seen 5 Aug on …"
+   *                                            for a price or the source's own date)
    *   value at Store ↗ · checked 4 min ago     (a live check; " · may have changed" once it is old)
    *   value · Estimate · how ↗
    *   Price not checked · reason · where to check ↗
-   * opts: label ("Price": says what was not checked), price (a price: "at <store>"), shared (the source is said once
-   * for the whole column: value only), and for a row that can be checked live: ref, intent, store, params. Then the
+   * opts: label ("Price": says what was not checked), unknown (the word for a missing value: "Unknown" in a table of
+   * options, "To confirm" on a checklist), price (a price: "at <store>"), shared (the source is said once for the
+   * whole column: value only), and for a row that can be checked live: ref, intent, store, params. Then the
    * caption ends in the row's one action: "Check price" (only when there is no "Check all prices"), "Try again" after
    * an error, "Open store ↗" when the store can't be read. While a check runs the value stays and the caption says
    * "Checking…".
@@ -259,17 +349,21 @@
     var value = valueText(x);
     // "Not checked" has one meaning on every page: nobody checked it. A checklist says "To confirm" (opts.unknown),
     // because there "checked" is the tick.
+    // A table of options says "Unknown" ("Price unknown") for a value nobody found, then why; "Not listed" when the
+    // page that was read does not give it.
+    var word = opts.unknown || "Not checked";
     var notChecked = function (note, url) {
-      add((opts.label ? opts.label + " not checked" : (opts.unknown || "Not checked")) + (usefulNote(note) ? " · " + usefulNote(note) : ""));
+      add((opts.label ? opts.label + " " + word.toLowerCase() : word) + (usefulNote(note) ? " · " + usefulNote(note) : ""));
       if (safeUrl(url)) add(link(url, "where to check", "act"));
     };
-    // looked for and not there: the source was read and does not give it ("Not listed · Maker ↗ · seen 16 Oct"); the
-    // link to the page that was read stays even when the source is named once for the column (then without the date)
+    // looked for and not there: the source was read and does not give it ("Not listed · checked 16 Oct on maker.com
+    // ↗"); the link to the page that was read stays even when the source is named once for the column (then without
+    // the date)
     var notListed = function (note, src, url, fact) {
       add((opts.label ? opts.label + " not listed" : "Not listed") + (usefulNote(note) ? " · " + usefulNote(note) : ""));
-      if (src) add(safeUrl(url) ? link(url, src) : src);
-      var seen = !opts.shared && src ? seenText(fact) : "";
-      if (seen) add(h("span", { "class": "age", text: seen }));
+      if (fact && !opts.shared) citeParts(fact, { price: opts.price }).forEach(add);
+      else if (safeUrl(url)) add(link(url, host(url) || src));
+      else if (src) add(src);
     };
     var failed = live && !pending && checks.errors.has(opts.ref);
 
@@ -285,6 +379,9 @@
       } else if (x.type === "claim" && x.status === "error") {
         value = "";
         add(h("span", null, icon("error"), "Couldn't check" + (usefulNote(x.note) ? " · " + usefulNote(x.note) : "")));
+      } else if (!value) {
+        // a record whose value only says nothing is known ("—") is unknown, never a caption without a value
+        notChecked(x.note);
       } else if (ESTIMATE_BASIS.indexOf(x.basis) >= 0) {
         add("Estimate");
         if (safeUrl(url)) add(link(url, "how"));
@@ -293,7 +390,7 @@
         add(h("span", { "class": "age", text: "checked " + ago(x.prov.observed_at) }));
       }
       if (expired(x) && value) { el.classList.add("is-old"); add("may have changed"); }
-    } else if (x && x.est) {
+    } else if (x && x.est && value) {
       // an estimate names its basis: a source to check it against ("Estimate · Taiwan Railway ↗") or how ↗
       add("Estimate");
       if (x.src) add(link(x.url, x.src)); else if (safeUrl(x.url)) add(link(x.url, "how"));
@@ -305,9 +402,14 @@
       // under the list names the source: next to "Open Amazon ↗" a bare "$900" reads as Amazon's price today
       var list = opts.price && !atStore && x.list;
       if (!opts.shared) {
-        add(atStore ? h("span", null, "at ", link(x.url, x.src)) : list ? h("span", null, "list price on ", link(x.url, x.src)) : link(x.url, x.src));
-        var when = seenText(x);
-        if (when) add(h("span", { "class": "age", text: when }));
+        if (atStore || list) {
+          add(atStore ? h("span", null, "at ", link(x.url, x.src)) : h("span", null, "list price on ", link(x.url, x.src)));
+          var when = seenText(x);
+          if (when) add(h("span", { "class": "age", text: when }));
+        } else {
+          // every value cites the page it was read on and the day: "checked 17 Oct on maker.com ↗"
+          citeParts(x, { price: opts.price }).forEach(add);
+        }
         if (seedOld) add("may have changed");
       } else if (list) {
         add("list price");
@@ -872,10 +974,10 @@
   }
 
   window.site = {
-    version: "1.4.0",
+    version: "1.5.0",
     h: h, append: append, icon: icon, link: link, slug: slug, rowAttrs: rowAttrs, safeUrl: safeUrl,
     date: date, dayDate: dayDate, ago: ago, money: money, num: num, valueText: valueText, host: host, seen: seenText,
-    old: isOld, expired: expired,
+    cite: cite, known: known, within: within, byFilter: byFilter, old: isOld, expired: expired,
     live: live, trust: trust, rowAction: rowAction, checkButton: checkButton, checkAll: checkAll, canCheck: canCheck,
     anyPending: anyPending, priceNotice: priceNotice, storesBlocked: storesBlocked, blocked: blockedRecord, tidy: tidyCaptions, remember: remember, recall: recall,
     state: state, setState: setState, onState: onState, render: render, schedule: schedule,

package/dist/bind/src/web.mjs

@@ -5,7 +5,13 @@
 import { isUnavailable, pageFetch } from "./page-fetch.mjs";
 
 export const USER_AGENT = "Instafy-Agent/1.0 (+https://instafy.dev; reads one public page on a person's request)";
-export const AI_AGENTS = ["gptbot", "chatgpt-user", "oai-searchbot", "claudebot", "claude-user", "anthropic-ai", "perplexitybot", "ccbot", "google-extended"];
+// The AI crawlers and the user-initiated AI fetchers (the closest match to a read on a person's request). A site that
+// keeps any of them out of a path is not read there.
+export const AI_AGENTS = [
+  "gptbot", "chatgpt-user", "oai-searchbot", "claudebot", "claude-user", "claude-searchbot", "anthropic-ai", "perplexitybot",
+  "perplexity-user", "ccbot", "google-extended", "applebot-extended", "meta-externalagent", "meta-externalfetcher",
+  "bytespider", "amazonbot", "cohere-ai", "mistralai-user", "duckassistbot",
+];
 
 /** robots.txt → [{ agents: [lowercase], rules: [{ allow, path }] }] */
 export function parseRobots(txt) {