strom-research 1.17.0 → 1.18.0-beta.1

package/CHANGELOG.cs.md

@@ -5,6 +5,24 @@
 Změny každé verze ve zkratce. Úplné poznámky a instalační řádky jsou na
 [GitHub Releases](https://github.com/ACiDekCZ/strom-research/releases) (anglicky).
 
+## 1.18.0
+
+*Připravuje se: bety do 1.18.0-beta.1. Oddíl bude úplný s vydáním.*
+
+- **Fotky z telefonu nastojato.** Fotka otočená podle údaje EXIF se čte tak, jak se má dívat, nastojato; místa na ní
+  vyznačená (výřezy, poloviny dvoustrany, uložená část) jsou místa obrázku, jak se ukazuje. Fotka sama zůstává, jak
+  byla.
+- **Fotky HEIC a dokumenty PDF jdou přečíst.** Fotky HEIC z telefonu a stránky PDF se pro čtení převádějí na JPEG —
+  na macOS, Windows i Linuxu — a originál zůstává, jak přišel. Fotka HEIC jde do aplikace Strom jako výřez JPEG.
+- **Řečeno, když obrázek převést nejde.** Kde fotku nebo stránku PDF na počítači nic nepřevede, řekne to jedna věta
+  i s tím, co by pomohlo, v jazyce výzkumu.
+- **Převedené obrázky v pořádku.** `strom tidy` převedené obrázky po 30 dnech nebo nad 500 MB uklidí; když jsou
+  potřeba, vzniknou znovu.
+- **Souhlas v jednom jazyce.** Souhlas, který dává jen člověk (rozbalení výzkumu, odebrání stromu, odmítnuté okno),
+  zazní celý v jazyce výzkumu, nic z něj nezůstane anglicky.
+- **Odmítnutá aktualizace.** Když se `strom update` odmítne, řekne, že se nic nenainstalovalo a která verze zůstává,
+  a skončí chybovým kódem; nepovedená instalace se řekne také v jazyce výzkumu.
+
 ## 1.17.0
 
 *19. 10. 2026 · patří k aplikaci Strom 3.13.1.*

package/CHANGELOG.de.md

@@ -5,6 +5,27 @@
 Die Änderungen jeder Version in Kürze. Die vollständigen Hinweise und die Installationszeilen stehen bei
 [GitHub Releases](https://github.com/ACiDekCZ/strom-research/releases) (auf Englisch).
 
+## 1.18.0
+
+*In Vorbereitung: die Betas bis 1.18.0-beta.1. Der Abschnitt ist mit dem Release vollständig.*
+
+- **Handyfotos aufrecht.** Ein Foto, das nach seiner EXIF-Ausrichtung gedreht ist, wird gelesen, wie es gesehen
+  werden soll, aufrecht; die darauf markierten Stellen (Ausschnitte, die Hälften einer Doppelseite, ein gespeicherter
+  Teil) sind die des Bildes, wie es gezeigt wird. Das Foto selbst bleibt, wie es war.
+- **HEIC-Fotos und PDF-Dokumente lesbar.** HEIC-Fotos vom Handy und die Seiten einer PDF werden zum Lesen in JPEG
+  umgewandelt — unter macOS, Windows und Linux —, das Original bleibt, wie es kam. Ein HEIC-Foto geht als
+  JPEG-Ausschnitt in die Strom-App.
+- **Gesagt, wenn ein Bild nicht umgewandelt werden kann.** Wo nichts auf dem Computer ein Foto oder eine PDF-Seite
+  umwandelt, sagt es ein Satz, auch was helfen würde, in der Sprache der Forschung.
+- **Umgewandelte Bilder in Ordnung gehalten.** `strom tidy` räumt umgewandelte Bilder nach 30 Tagen oder über 500 MB
+  weg; bei Bedarf entstehen sie neu.
+- **Eine Zustimmung in einer Sprache.** Eine Zustimmung, die nur ein Mensch gibt (eine Forschung entpacken, einen
+  Stammbaum entfernen, ein abgelehntes Fenster), steht ganz in der Sprache der Forschung, nichts davon bleibt
+  Englisch.
+- **Ein abgelehntes Update.** Wird `strom update` abgelehnt, sagt es, dass nichts installiert wurde und welche
+  Version bleibt, und endet mit einem Fehlercode; eine misslungene Installation wird auch in der Sprache der
+  Forschung gesagt.
+
 ## 1.17.0
 
 *19.10.2026 · gehört zur Strom-App 3.13.1.*

package/CHANGELOG.md

@@ -5,6 +5,25 @@ English · [Čeština](CHANGELOG.cs.md) · [Deutsch](CHANGELOG.de.md)
 The changes of each version in short. The full notes and the install lines are on
 [GitHub Releases](https://github.com/ACiDekCZ/strom-research/releases).
 
+## 1.18.0
+
+*In preparation: the betas up to 1.18.0-beta.1. The section is complete with the release.*
+
+- **Phone photos upright.** A photo turned by its EXIF orientation is read as it is meant to be seen, upright; the
+  places marked on it (excerpts, the halves of a spread, a cut-out part) are of the picture as it is shown. The
+  photo itself stays as it was.
+- **HEIC photos and PDF documents can be read.** HEIC photos from a phone and the pages of a PDF are converted to
+  JPEG for reading — on macOS, Windows and Linux — and the original is kept as it came. A HEIC photo goes to the
+  Strom app as a JPEG excerpt.
+- **Said when a picture cannot be converted.** Where nothing on the computer converts a photo or a PDF page, one
+  sentence says so and what would help, in the research's language.
+- **Converted pictures kept in order.** `strom tidy` clears the converted pictures away after 30 days or above
+  500 MB; they are made again when needed.
+- **A consent in one language.** A consent only a person gives (unpacking a research, removing a tree, a window said
+  no to) is said whole in the research's language, nothing of it left in English.
+- **An update said no to.** When `strom update` is declined, it says nothing was installed and which version stays,
+  and ends with an error code; a failed install is said in the research's language too.
+
 ## 1.17.0
 
 *2026-10-19 · goes with the Strom app 3.13.1.*

package/assets/lang/cs.json

@@ -59,7 +59,6 @@
   "ui.dialog.title": "Strom výzkum",
   "ui.dialog.yes": "Povolit",
   "ui.dialog.no": "Nepovolit",
-  "ui.dialog.refused": "Nepovoleno – nic se nezměnilo.",
   "ui.link.isolated": "Tohle je izolovaná instalace (STROM_ISOLATED): odkazy strom-research:// nezapisuje – zůstávají běžné instalaci stromu.",
   "ui.doc.isolated": "izolovaná instalace v {folder}",
   "ui.doc.second": "druhá instalace: příkaz {command}, ve složce {folder}, její odkazy {scheme}://",
@@ -760,9 +759,10 @@
   "ui.update.available": "Vyšla nová verze: {version} (tato je {current}) – nainstaluje ji strom update.",
   "ui.update.npm": "Vyšla nová verze: {version}. strom je tu nainstalovaný přes npm: npm install -g strom-research@{tag}",
   "ui.update.sure": "Aktualizovat strom z {current} na {version}? Výzkumy zůstanou, jak jsou.",
-  "ui.update.later": "Teď ne – kdykoli strom update.",
+  "ui.update.later": "Neaktualizováno – nic se nenainstalovalo, strom zůstává ve verzi {current}. Kdykoli znovu: {command}",
   "ui.update.downloading": "Stahuji strom {version}…",
-  "ui.update.failed": "Aktualizace se nepovedla: {detail}. strom zůstává, jak byl.",
+  "ui.error.update.failed": "aktualizace se nepovedla: {detail}. strom zůstává ve verzi {current}",
+  "ui.error.update.failed.hint": "znovu {command} – nebo spustit instalátor znovu",
   "ui.update.done": "✓ strom je aktualizovaný: {previous} → {version}.",
   "ui.update.whatsnew": "Co je nového: {url}",
   "ui.update.restart": "strom je aktualizovaný — tahle nabídka ale ještě běží ve staré verzi. Po novém spuštění stromu (nebo zástupce „{shortcut}“) poběží nová verze.",
@@ -1022,6 +1022,7 @@
   "ui.tidy.kind.views": "Náhledy skenů (udělají se znovu, až budou potřeba)",
   "ui.tidy.kind.excerpts": "Výřezy ze skenů (udělají se znovu, až budou potřeba)",
   "ui.tidy.kind.fetch": "Pracovní složky přerušených stahování",
+  "ui.tidy.kind.converted": "Fotky a strany PDF převedené pro čtení (vytvoří se znovu, až budou potřeba)",
   "ui.tidy.kind.kept": "Starší poslání z aplikace Strom, už zapsaná (změny jsou v historii)",
   "ui.tidy.kind.inbox": "Dávná poslání z aplikace Strom, která ve schránce nechal starší strom",
   "ui.tidy.kind.pointers": "Zbytky skončených pracovních sezení",
@@ -1040,6 +1041,8 @@
   "ui.tidy.hint": "Vedle výzkumu si strom drží {strom}, z toho jde uvolnit {size}: strom tidy (v menu Nastavení → Údržba → Místo na disku)",
   "ui.error.consent": "tady je potřeba souhlas – {question}",
   "ui.error.consent.hint": "spustit ve vlastním terminálu: {set}",
+  "ui.error.consent.refused": "nepovoleno – nic se nezměnilo",
+  "ui.error.consent.refused.hint": "kdykoli znovu: {set}",
   "ui.status.disk": "Na disku: výzkum {research} · vedle něj si strom drží {strom}",
   "ui.status.disk.frees": " (strom tidy uvolní {size})",
   "ui.status.refused": "{host} odmítl automatický přístup stromu — snímky ručně; zrušit: strom allow host {host} --unblock",
@@ -1215,6 +1218,19 @@
   "ui.error.tree.empty": "soubor je prázdný, nejsou v něm žádné osoby",
   "ui.error.tree.empty.hint": "soubor vyexportovaný z aplikace Strom nebo z jiného programu pro rodokmeny",
   "ui.error.tree.unreadable": "tohle není soubor GEDCOM ani rodokmen z aplikace Strom",
+  "ui.error.read.heic": "{file}: fotku HEIC nejde na tomto počítači převést pro čtení – originál zůstává beze změny",
+  "ui.error.read.heic.windows.hint": "rozšíření HEIF a HEVC z Microsoft Storu, nebo fotku uložit jako JPEG",
+  "ui.error.read.heic.mac.hint": "fotku uložit jako JPEG (Náhled: Soubor → Exportovat)",
+  "ui.error.read.heic.linux.hint": "nainstalovat heif-convert (libheif) nebo ImageMagick, nebo fotku uložit jako JPEG",
+  "ui.error.read.pdf-render": "{file}: toto PDF nejde na tomto počítači převést na obrázky pro čtení – originál zůstává beze změny",
+  "ui.error.read.pdf-render.hint": "strany uložit jako JPEG",
+  "ui.error.read.pdf-render.linux.hint": "nainstalovat pdftoppm (poppler-utils), nebo strany uložit jako JPEG",
+  "ui.error.read.convert": "{file} se nepodařilo převést pro čtení – originál zůstává beze změny",
+  "ui.error.read.convert.hint": "soubor uložit jako JPEG (umí to každý prohlížeč obrázků)",
+  "ui.error.read.type": "{file} není obrázek, který jde přečíst (JPEG, PNG, HEIC nebo PDF) – originál zůstává beze změny",
+  "ui.error.read.type.hint": "soubor uložit jako JPEG (umí to každý prohlížeč obrázků)",
+  "ui.error.read.page": "{file} má {pages, plural, one {jednu stranu} few {# strany} other {# stran}}, strana {page} v něm není",
+  "ui.error.read.page.hint": "strana od 1 do {pages}",
   "ui.error.adopt.none": "tento výzkum na žádný strom nečeká",
   "ui.error.tree.unknown": "žádný rodokmen „{name}“ tu není",
   "ui.error.tree.unknown.hint": "rodokmeny, které tu jsou: strom trees",

package/assets/lang/de.json

@@ -59,7 +59,6 @@
   "ui.dialog.title": "Strom Ahnenforschung",
   "ui.dialog.yes": "Erlauben",
   "ui.dialog.no": "Nicht erlauben",
-  "ui.dialog.refused": "Nicht erlaubt – nichts wurde geändert.",
   "ui.link.isolated": "Dies ist eine isolierte Installation (STROM_ISOLATED): sie registriert keine strom-research://-Links – diese bleiben beim eigenen strom.",
   "ui.doc.isolated": "isolierte Installation in {folder}",
   "ui.doc.second": "zweite Installation: der Befehl {command}, in {folder}, ihre Links {scheme}://",
@@ -760,9 +759,10 @@
   "ui.update.available": "Eine neue Version ist da: {version} (diese ist {current}) – strom update installiert sie.",
   "ui.update.npm": "Eine neue Version ist da: {version}. strom ist hier mit npm installiert: npm install -g strom-research@{tag}",
   "ui.update.sure": "strom von {current} auf {version} aktualisieren? Die Forschungen bleiben, wie sie sind.",
-  "ui.update.later": "Nicht jetzt – jederzeit strom update.",
+  "ui.update.later": "Nicht aktualisiert – nichts wurde installiert, strom bleibt bei {current}. Jederzeit erneut: {command}",
   "ui.update.downloading": "strom {version} wird heruntergeladen…",
-  "ui.update.failed": "Die Aktualisierung hat nicht geklappt: {detail}. strom bleibt, wie er war.",
+  "ui.error.update.failed": "die Aktualisierung hat nicht geklappt: {detail}. strom bleibt bei {current}",
+  "ui.error.update.failed.hint": "erneut {command} – oder das Installationsprogramm erneut ausführen",
   "ui.update.done": "✓ strom ist aktualisiert: {previous} → {version}.",
   "ui.update.whatsnew": "Was ist neu: {url}",
   "ui.update.restart": "strom ist aktualisiert — dieses Menü läuft aber noch in der alten Version. Nach einem Neustart von strom (oder der Verknüpfung „{shortcut}“) läuft die neue Version.",
@@ -1022,6 +1022,7 @@
   "ui.tidy.kind.views": "Ansichten von Scans (werden bei Bedarf neu erstellt)",
   "ui.tidy.kind.excerpts": "Ausschnitte aus Scans (werden bei Bedarf neu erstellt)",
   "ui.tidy.kind.fetch": "Arbeitsordner abgebrochener Downloads",
+  "ui.tidy.kind.converted": "Fotos und PDF-Seiten, zum Lesen umgewandelt (bei Bedarf neu erstellt)",
   "ui.tidy.kind.kept": "Ältere Sendungen der Strom-App, schon geschrieben (die Änderungen sind in der Geschichte)",
   "ui.tidy.kind.inbox": "Längst vergangene Sendungen der Strom-App, die ein älterer strom im Eingang ließ",
   "ui.tidy.kind.pointers": "Reste beendeter Arbeitssitzungen",
@@ -1040,6 +1041,8 @@
   "ui.tidy.hint": "Neben der Forschung hält strom {strom}, davon lassen sich {size} freigeben: strom tidy (im Menü Einstellungen → Wartung → Speicherplatz)",
   "ui.error.consent": "dafür braucht es eine Zustimmung – {question}",
   "ui.error.consent.hint": "im eigenen Terminal ausführen: {set}",
+  "ui.error.consent.refused": "nicht erlaubt – nichts wurde geändert",
+  "ui.error.consent.refused.hint": "jederzeit erneut: {set}",
   "ui.status.disk": "Auf der Festplatte: die Forschung {research} · daneben hält strom {strom}",
   "ui.status.disk.frees": " (strom tidy gibt {size} frei)",
   "ui.status.refused": "{host} hat den automatischen Zugriff von strom abgelehnt — Bilder von Hand; aufheben: strom allow host {host} --unblock",
@@ -1215,6 +1218,19 @@
   "ui.error.tree.empty": "die Datei ist leer, es sind keine Personen darin",
   "ui.error.tree.empty.hint": "eine Datei, aus der Strom-App oder einem anderen Stammbaumprogramm exportiert",
   "ui.error.tree.unreadable": "das ist weder eine GEDCOM-Datei noch ein Stammbaum der Strom-App",
+  "ui.error.read.heic": "{file}: ein HEIC-Foto lässt sich auf diesem Computer nicht zum Lesen umwandeln – das Original bleibt unverändert",
+  "ui.error.read.heic.windows.hint": "die Erweiterungen HEIF und HEVC aus dem Microsoft Store, oder das Foto als JPEG speichern",
+  "ui.error.read.heic.mac.hint": "das Foto als JPEG speichern (Vorschau: Ablage → Exportieren)",
+  "ui.error.read.heic.linux.hint": "heif-convert (libheif) oder ImageMagick installieren, oder das Foto als JPEG speichern",
+  "ui.error.read.pdf-render": "{file}: dieses PDF lässt sich auf diesem Computer nicht in Bilder zum Lesen umwandeln – das Original bleibt unverändert",
+  "ui.error.read.pdf-render.hint": "die Seiten als JPEG speichern",
+  "ui.error.read.pdf-render.linux.hint": "pdftoppm (poppler-utils) installieren, oder die Seiten als JPEG speichern",
+  "ui.error.read.convert": "{file} ließ sich nicht zum Lesen umwandeln – das Original bleibt unverändert",
+  "ui.error.read.convert.hint": "die Datei als JPEG speichern (jeder Bildbetrachter kann das)",
+  "ui.error.read.type": "{file} ist kein Bild, das sich lesen lässt (JPEG, PNG, HEIC oder PDF) – das Original bleibt unverändert",
+  "ui.error.read.type.hint": "die Datei als JPEG speichern (jeder Bildbetrachter kann das)",
+  "ui.error.read.page": "{file} hat {pages, plural, one {eine Seite} other {# Seiten}}, keine Seite {page}",
+  "ui.error.read.page.hint": "eine Seite von 1 bis {pages}",
   "ui.error.adopt.none": "diese Forschung wartet auf keinen Stammbaum",
   "ui.error.tree.unknown": "keinen Stammbaum „{name}“ gibt es hier",
   "ui.error.tree.unknown.hint": "die Stammbäume hier: strom trees",

package/dist/cli/context.js

@@ -209,8 +209,10 @@ export class Context {
             const answer = this.io.dialog ? this.io.dialog(text) : systemDialog(text, this.env);
             if (answer === true)
                 return "window";
+            // said no: an agent reads the English and what not to do; a person (the app, a script) their language — nothing
+            // changed and how to ask again (printError: ui.error.consent.refused), never the agent's English under it
             if (answer === false)
-                throw new StromError(ui(lang, "ui.dialog.refused"), { hint: `the user said no to: ${question} — do not ask again unless they want to` });
+                throw new StromError(ui("en", "ui.error.consent.refused"), { hint: `the user said no to: ${question} — do not ask again unless they want to`, code: "consent.refused", params: { set } });
         }
         // a person (a script, no terminal) reads it in their language; an agent the English it goes by
         // a run nobody watches: nobody to say yes before it ends — a task for it, for the user to see, and on another way

package/dist/cli/ui.js

@@ -15,7 +15,6 @@ export const UI = {
     "ui.dialog.title": "Strom research",
     "ui.dialog.yes": "Allow",
     "ui.dialog.no": "Don't allow",
-    "ui.dialog.refused": "Not allowed — nothing was changed.",
     "ui.link.isolated": "This is an isolated installation (STROM_ISOLATED): it registers no strom-research:// links — they stay the regular strom's.",
     "ui.doc.isolated": "isolated installation in {folder}",
     "ui.doc.second": "a second installation: the command {command}, in {folder}, its links {scheme}://",
@@ -687,9 +686,10 @@ export const UI = {
     "ui.update.available": "A new version is out: {version} (this is {current}) — strom update installs it.",
     "ui.update.npm": "A new version is out: {version}. strom is installed with npm here: npm install -g strom-research@{tag}",
     "ui.update.sure": "Update strom from {current} to {version}? The research stays as it is.",
-    "ui.update.later": "Not now — strom update any time.",
+    "ui.update.later": "Not updated — nothing was installed, strom stays at {current}. Any time again: {command}",
     "ui.update.downloading": "Downloading strom {version}…",
-    "ui.update.failed": "The update did not work: {detail}. strom stays as it was.",
+    "ui.error.update.failed": "the update did not work: {detail}. strom stays at {current}",
+    "ui.error.update.failed.hint": "{command} again — or run the installer again",
     "ui.update.done": "✓ strom is updated: {previous} → {version}.",
     "ui.update.whatsnew": "What's new: {url}",
     "ui.update.restart": "strom is updated — this menu still runs the old version. Start strom again (or the shortcut \"{shortcut}\"), and the new version is at work.",
@@ -926,6 +926,7 @@ export const UI = {
     "ui.tidy.kind.views": "Views of scans (made again when needed)",
     "ui.tidy.kind.excerpts": "Excerpts of scans (made again when needed)",
     "ui.tidy.kind.fetch": "Work folders of downloads cut short",
+    "ui.tidy.kind.converted": "Photos and PDF pages converted for reading (made again when needed)",
     "ui.tidy.kind.kept": "Sends of the Strom app written long ago (their changes are in the history)",
     "ui.tidy.kind.inbox": "Sends of the Strom app long ago, left in the inbox by an older strom",
     "ui.tidy.kind.pointers": "Leftovers of work sessions that ended",
@@ -944,6 +945,8 @@ export const UI = {
     "ui.tidy.hint": "Beside the research strom keeps {strom}, {size} of it can go: strom tidy (the menu: Settings → Maintenance → Disk space)",
     "ui.error.consent": "this needs a yes — {question}",
     "ui.error.consent.hint": "run in a terminal of its own: {set}",
+    "ui.error.consent.refused": "not allowed — nothing was changed",
+    "ui.error.consent.refused.hint": "any time again: {set}",
     "ui.status.disk": "On disk: the research {research} · strom's own beside it {strom}",
     "ui.status.disk.frees": " (strom tidy frees {size})",
     "ui.status.refused": "{host} refused strom's automated access — images by hand; to lift it: strom allow host {host} --unblock",
@@ -1128,6 +1131,19 @@ export const UI = {
     "ui.error.tree.empty": "the file is empty: no people in it",
     "ui.error.tree.empty.hint": "a file exported from the Strom app or another family-tree program",
     "ui.error.tree.unreadable": "this is neither a GEDCOM file nor a family tree of the Strom app",
+    "ui.error.read.heic": "{file}: a HEIC photo cannot be converted for reading on this computer — the original stays as it is",
+    "ui.error.read.heic.windows.hint": "the HEIF and HEVC extensions from the Microsoft Store, or the photo saved as JPEG",
+    "ui.error.read.heic.mac.hint": "the photo saved as JPEG (Preview: File → Export)",
+    "ui.error.read.heic.linux.hint": "heif-convert (libheif) or ImageMagick installed, or the photo saved as JPEG",
+    "ui.error.read.pdf-render": "{file}: this PDF cannot be turned into pictures for reading on this computer — the original stays as it is",
+    "ui.error.read.pdf-render.hint": "the pages saved as JPEG",
+    "ui.error.read.pdf-render.linux.hint": "pdftoppm (poppler-utils) installed, or the pages saved as JPEG",
+    "ui.error.read.convert": "{file} could not be converted for reading — the original stays as it is",
+    "ui.error.read.convert.hint": "the file saved as JPEG (any image viewer can)",
+    "ui.error.read.type": "{file} is no picture that can be read (JPEG, PNG, HEIC or PDF) — the original stays as it is",
+    "ui.error.read.type.hint": "the file saved as JPEG (any image viewer can)",
+    "ui.error.read.page": "{file} has {pages, plural, one {one page} other {# pages}}, no page {page}",
+    "ui.error.read.page.hint": "a page from 1 to {pages}",
     "ui.error.adopt.none": "this research waits for no tree",
     "ui.error.tree.unknown": "no tree \"{name}\"",
     "ui.error.tree.unknown.hint": "strom trees",

package/dist/commands/clips.js

@@ -10,6 +10,8 @@ import { counted, lines, truncate } from "../cli/format.js";
 import { UsageError } from "../core/errors.js";
 import { normId, requireRecord, update } from "../core/records.js";
 import { makeView } from "../core/views.js";
+import { shownToFile } from "../image/image.js";
+import { exifOrientation } from "../image/index.js";
 import { clipText, MAX_CLIPS, regionText } from "../core/media.js";
 import { withMargin } from "../core/excerpt.js";
 import { batches, READER_MINUTES_HELP } from "../core/reader.js";
@@ -161,10 +163,12 @@ register({
         // what the readers cost goes to the session that started them (apart from the agent's own)
         addReaders(tree, ctx.env, ctx.refs, { usd: r0.cost(), runs: r0.runs(), partial: r0.partial() });
         // 3. What passed becomes the source's clips.
+        // the readers saw each image upright; a clip is kept as the file lies (the app turns it as it turns the image)
         const clipsOf = new Map();
+        const asFile = (l) => shownToFile(l.region, exifOrientation(path.join(shared, all.get(l.media).file)));
         for (const l of hits)
             if (passed.has(`${l.source} ${l.media}`))
-                clipsOf.set(l.source, [...(clipsOf.get(l.source) ?? []), { media: l.media, region: l.region }].slice(0, MAX_CLIPS));
+                clipsOf.set(l.source, [...(clipsOf.get(l.source) ?? []), { media: l.media, region: asFile(l) }].slice(0, MAX_CLIPS));
         for (const [id, clips] of clipsOf)
             update(tree, id, "source", (s) => ({ ...s, clips }), { op: "source.edit", summary: `${id} clips ${clips.map(clipText).join(" ")} (found by a reader, checked)` });
         // The reports are working notes of the research: committed with it.

package/dist/commands/media.js

@@ -11,9 +11,10 @@ import { notFound, UsageError } from "../core/errors.js";
 import { makeNote } from "../core/actions.js";
 import { clipText, collectFiles, coveringPart, fileSha256, findImage, imageNumbers, imageOfRef, imageRange, IMAGE_FILE_EXT, inboxFolders, inputPath, isWhole, mimeOf, onlyParts, otherCopies, parseImageList, regionText, sharperPart, storeShared, wholeSizeOf } from "../core/media.js";
 import { create, normId, requireRecord, update } from "../core/records.js";
-import { imageSizeOfFile } from "../image/index.js";
+import { exifOrientation, imageSizeOfFile } from "../image/index.js";
+import { fileToShown, shownToFile, turnsAcross } from "../image/image.js";
 import { imageOf, pageOf } from "../core/calibration.js";
-import { cropOf, describeView, ENLARGED_HINT, halves, makeView, OVERVIEW_HINT, parseCrop, partRegion, readInHalves, REDUCED_HINT, SHARPER_SCAN, viewLine, viewRegion, viewSize } from "../core/views.js";
+import { cropOf, describeView, ENLARGED_HINT, halves, makeView, mediaOrient, OVERVIEW_HINT, parseCrop, partRegion, readInHalves, REDUCED_HINT, SHARPER_SCAN, viewLine, viewRegion, viewSize } from "../core/views.js";
 import { IMAGE_MAX, IMAGE_MAX_LARGE, OVERVIEW_MAX } from "../agents/images.js";
 import { viewSizesFor } from "../core/viewsizes.js";
 import { PROFILES } from "../agents/profiles.js";
@@ -141,13 +142,18 @@ function resolveTarget(tree, shared, ref, opts) {
             throw new UsageError(`image ${image} of ${b.id} is not registered`, { hint: `strom media list --recordset ${b.id} · strom media add <files> --recordset ${b.id}` });
         return { key: m.id, file: path.join(shared, m.file), media: m };
     }
-    const id = normId(ref);
+    // a page of a document the user gave: I0002:3
+    const paged = /^([Ii]\d+):(\d+)$/.exec(ref.trim());
+    const id = normId(paged ? paged[1] : ref);
     if (id.startsWith("I")) {
         const i = requireRecord(tree, id, "input");
         const file = inputPath(tree, i);
         if (!file)
             throw new UsageError(`${i.id} has no file`);
-        return { key: i.id, file };
+        const page = paged ? Number(paged[2]) : undefined;
+        if (page !== undefined && page < 1)
+            throw new UsageError(`no page ${page}`, { hint: `${i.id}:1 is its first page` });
+        return { key: i.id, file, ...(page !== undefined ? { page } : {}) };
     }
     const m = requireRecord(tree, id, "media");
     return { key: m.id, file: path.join(shared, m.file), media: m };
@@ -243,8 +249,10 @@ function partSpecs(t, p, max, inHalves = false) {
     const size = ofParts ?? (t.media?.width && t.media.height ? { width: t.media.width, height: t.media.height } : imageSizeOfFile(t.file));
     if (!size)
         return [{ half: p.half, crop: p.crop, label: p.label }]; // no image there: the view says why
-    const W = size.width;
-    const H = size.height;
+    // the boxes are of the picture as the view shows it: a photo turned a quarter by its EXIF has its sides swapped
+    const across = !ofParts && turnsAcross(exifOrientation(t.file));
+    const W = across ? size.height : size.width;
+    const H = across ? size.width : size.height;
     let boxes;
     if (p.both) {
         // the halves rule: the two pages of a spread only where they are sharper than one view of it — a book read worse
@@ -489,11 +497,11 @@ function resolveTargets(tree, shared, all, args, opts, parts, maxViews = VIEW_MA
         }
         else {
             const r = resolveTarget(tree, shared, w.ref, {});
-            t = { ...r, ref: r.media && !r.media.part && r.media.recordset && r.media.image !== undefined ? `${r.media.recordset}:${r.media.image}` : r.key };
+            t = { ...r, ref: r.media && !r.media.part && r.media.recordset && r.media.image !== undefined ? `${r.media.recordset}:${r.media.image}` : r.page ? `${r.key}:${r.page}` : r.key };
         }
-        if (seen.has(t.key))
+        if (seen.has(`${t.key}:${t.page ?? 1}`))
             continue;
-        seen.add(t.key);
+        seen.add(`${t.key}:${t.page ?? 1}`);
         targets.push(t);
     }
     if (!targets.length && !several) {
@@ -513,24 +521,29 @@ function oneView(ctx, tree, shared, all, target, asked, meta = {}, noSharper = f
     let v;
     let sharper;
     if (t.media?.width && t.media.height && !t.media.part) {
-        const px = viewRegion(spec, t.media.width, t.media.height);
-        const s = sharperPart(all, t.media, { x: px.x / t.media.width, y: px.y / t.media.height, w: px.w / t.media.width, h: px.h / t.media.height });
+        // the place asked is of the picture shown (upright by its EXIF); parts are kept as the file lies
+        const o = mediaOrient(shared, t.media);
+        const [W, H] = turnsAcross(o) ? [t.media.height, t.media.width] : [t.media.width, t.media.height];
+        const px = viewRegion(spec, W, H);
+        const s = sharperPart(all, t.media, shownToFile({ x: px.x / W, y: px.y / H, w: px.w / W, h: px.h / H }, o));
         if (s) {
             sharper = s;
             shown = s.part;
             const whole = !s.part.part && !spec.crop && !spec.half;
-            v = makeView(tree, path.join(shared, s.part.file), s.part.id, { ...spec, half: undefined, crop: whole ? undefined : regionText(s.crop) }, meta);
+            v = makeView(tree, path.join(shared, s.part.file), s.part.id, { ...spec, half: undefined, crop: whole ? undefined : regionText(fileToShown(s.crop, mediaOrient(shared, s.part))) }, meta);
         }
     }
-    v ??= makeView(tree, t.file, t.key, spec, meta);
+    v ??= makeView(tree, t.file, t.key, t.page ? { ...spec, page: t.page } : spec, meta);
     const page = t.media ? pageLabel(tree, t.media) : "";
     // still too little detail: a connector that fetched the image may fetch this part of it sharper
     const base = shown?.part ?? { x: 0, y: 0, w: 1, h: 1 };
+    // what the view shows, as the picture lies in its file (a photo shown turned by its EXIF): where clips are kept
+    const inFile = shownToFile({ x: v.region.x / v.original.width, y: v.region.y / v.original.height, w: v.region.w / v.original.width, h: v.region.h / v.original.height }, v.orient);
     const inWhole = {
-        x: base.x + (v.region.x / v.original.width) * base.w,
-        y: base.y + (v.region.y / v.original.height) * base.h,
-        w: (v.region.w / v.original.width) * base.w,
-        h: (v.region.h / v.original.height) * base.h,
+        x: base.x + inFile.x * base.w,
+        y: base.y + inFile.y * base.h,
+        w: inFile.w * base.w,
+        h: inFile.h * base.h,
     };
     const whole = t.media?.recordset !== undefined && t.media.image !== undefined ? findImage(all, t.media.recordset, t.media.image) : undefined;
     const from = whole?.fetched ? listConnectors(shared).find((c) => c.name === whole.fetched.connector) : undefined;
@@ -538,11 +551,12 @@ function oneView(ctx, tree, shared, all, target, asked, meta = {}, noSharper = f
     const fetchPart = canPart && !noSharper && v.scale > 1.25 && inWhole.w * inWhole.h < 0.9
         ? `strom fetch ${from.name} --recordset ${whole.recordset} --images ${whole.image} --crop ${regionText(inWhole)}`
         : undefined;
-    // The same part as a clip of the source read in it: of the whole image where it is registered.
+    // The same part as a clip of the source read in it: of the whole image where it is registered — as --clip takes it,
+    // of the picture as its view shows it (source add keeps it as the file lies)
     const clip = !t.media || (!spec.crop && !spec.half)
         ? undefined
         : whole && !whole.part
-            ? clipText({ media: whole.id, region: inWhole })
+            ? clipText({ media: whole.id, region: fileToShown(inWhole, mediaOrient(shared, whole)) })
             : clipText({ media: t.key, region: { x: v.region.x / v.original.width, y: v.region.y / v.original.height, w: v.region.w / v.original.width, h: v.region.h / v.original.height } });
     return { v, page, ...(t !== target ? { part: t.media } : {}), ...(sharper ? { sharper } : {}), ...(fetchPart ? { fetchPart } : {}), ...(clip ? { clip } : {}) };
 }
@@ -697,7 +711,8 @@ register({
             if (it.image === undefined)
                 throw new UsageError("which image is it a part of? --image <n>", { hint: "strom media add <file> --recordset B… --image <n> --half left" });
             const whole = findImage(tree.list("media").filter((m) => !m.part), recordset, it.image);
-            it.part = partRegion(opts, whole);
+            // of the whole image as its view shows it (upright by its EXIF)
+            it.part = partRegion(opts, whole, mediaOrient(shared, whole));
         }
         if (recordset) {
             // an image of a book without its number cannot be cited — nor found again
@@ -845,7 +860,7 @@ register({
         "--half right; B0001 --image 2 --crop … B0001 --image 7 --crop …; B0001 2 --crop … 7 --crop …); parts after the\n" +
         "last image are of every image. At most\n" +
         `${VIEW_MAX_IMAGES} images and ${VIEW_MAX_VIEWS} views in one call: every view stays in the context that opens it.`,
-    args: [{ name: "image", description: "M0012, B0001:57 (record set:image), a range B0001:57-60, B0001 with --page, or an input I0002; several = several images", required: true, variadic: true }],
+    args: [{ name: "image", description: "M0012, B0001:57 (record set:image), a range B0001:57-60, B0001 with --page, an input I0002, or a page of an input document I0002:3 (a PDF; HEIC and PDF are converted for reading, the original kept); several = several images", required: true, variadic: true }],
     options: [
         { name: "crop", type: "string", multiple: true, value: "<x,y,w,h>", description: "part of the image: fractions (0.1,0.35,0.4,0.2) or pixels; repeated = several parts; after one image of several: of that image alone (--half, --split, --grid too)" },
         { name: "half", type: "string", multiple: true, value: "<side>", description: "left or right page of a double page (top, bottom); both = the two pages, overlapping at the gutter" },
@@ -870,6 +885,7 @@ register({
         "strom media view B0001:2 --crop 0.05,0.10,0.45,0.30 --crop 0.05,0.40,0.45,0.30",
         "strom media view B0001:2 --crop 0.05,0.10,0.45,0.30 B0001:3 --crop 0.50,0.60,0.45,0.20",
         "strom media view B0001 --page 112 113 --split 2x2",
+        "strom media view I0005:2 --half top",
     ],
     run(ctx, { args, opts, seq }) {
         // the whole image is what a view is without a part of it (strom read --whole: each image in one view)

package/dist/commands/pack.js

@@ -81,8 +81,9 @@ register({
     async run(ctx, { args }) {
         const lang = ctx.uiLang();
         const src = ctx.resolvePath(args[0]);
+        // the English an agent (and a program) goes by; a person (a script, a launcher) reads it in their language
         if (!ctx.interactive)
-            throw new NeedsConsentError([{ key: "unpack", kind: "consent", question: ui(lang, "ui.unpack.person"), set: `strom unpack "${src}"` }]);
+            throw new NeedsConsentError([{ key: "unpack", kind: "consent", question: ui("en", "ui.unpack.person"), set: `strom unpack "${src}"` }], ui(lang, "ui.unpack.person"));
         // someone new to strom (a launcher of the package installed it just now): the setup wizard first
         if (!ctx.settings.home() && !isAgent(ctx.env)) {
             await setupWizard(ctx);

package/dist/commands/read.js

@@ -17,7 +17,8 @@ import { IMAGE_MAX, IMAGE_MAX_LARGE } from "../agents/images.js";
 import { viewSizesFor } from "../core/viewsizes.js";
 import { currentSession } from "../core/session.js";
 import { BATCH, BATCH_MAX, batches, NEXT_BLOCK, parseReport, READER_MINUTES_HELP, readerPrompt, readerSettings, viewCount, withoutNextBlock } from "../core/reader.js";
-import { imageSize } from "../image/index.js";
+import { exifOrientation, shownSize } from "../image/index.js";
+import { turnsAcross } from "../image/image.js";
 import { RUNNERS } from "../runners/index.js";
 import { which, withoutAgentMarks } from "../core/which.js";
 import { permissionPath } from "../agents/files.js";
@@ -306,13 +307,17 @@ register({
             const file = path.join(ctx.settings.shared().value, m.file);
             const page = m.page ?? (b && m.image !== undefined ? pageOf(b, m.image) : undefined);
             const contrast = Boolean(opts.contrast);
-            if (typeof opts.crop === "string" && m.width && m.height) {
+            // the image as a reader sees it: upright (a photo turned by its EXIF has its sides swapped)
+            const across = fs.existsSync(file) && turnsAcross(exifOrientation(file));
+            const W = across ? m.height : m.width;
+            const H = across ? m.width : m.height;
+            if (typeof opts.crop === "string" && W && H) {
                 // a crop is read at full resolution — in overlapping parts when it is bigger than one look
-                const all = tiles(parseCrop(partCrops.get(m.id) ?? opts.crop, m.width, m.height), max);
+                const all = tiles(parseCrop(partCrops.get(m.id) ?? opts.crop, W, H), max);
                 const parts = all.map((t) => ({ label: t.label, view: view(file, m, { crop: t.crop, max, contrast }, all.length > 1 ? "split" : "crop").file }));
                 return { id: m.id, image: m.image, page, view: parts[0].view, ...(parts.length > 1 ? { parts } : {}) };
             }
-            const size = !opts.whole && fs.existsSync(file) ? (m.width && m.height ? { width: m.width, height: m.height } : imageSize(new Uint8Array(fs.readFileSync(file)))) : undefined;
+            const size = !opts.whole && fs.existsSync(file) ? (W && H ? { width: W, height: H } : shownSize(new Uint8Array(fs.readFileSync(file)))) : undefined;
             if (size && half) {
                 // one page of a spread, with the strip past the gutter (a scan is seldom centred on it)
                 const v = view(file, m, { crop: cropOf(halves(size.width, size.height)[half]), max, contrast }, "half");

package/dist/commands/setup.js

@@ -45,7 +45,7 @@ import { agentSettingsLine, chooseEffort } from "../cli/model-choice.js";
 import { calibrationKey, calibrationLabel, calibrationOffer, forgetCalibration, researchKey, sizesText, viewModel, viewSizesFor } from "../core/viewsizes.js";
 import { bookTuned, bookViewSizes, forgetTuning, keyTuningOf, readingOf, selfTune, treeTuning } from "../core/tune.js";
 import { loadAliases } from "../core/modelkey.js";
-import { NeedsInputError, StromError, UsageError } from "../core/errors.js";
+import { EXIT, NeedsInputError, StromError, UsageError } from "../core/errors.js";
 import { worsenings } from "../core/tunereset.js";
 import { resetReading } from "../cli/tunereset.js";
 import { loadRollup, manyWeb } from "../core/readstats.js";
@@ -646,13 +646,16 @@ register({
             : newer
                 ? t("ui.update.sure", { version: newest, current: VERSION })
                 : t("ui.update.node.sure", { from: inst.node ?? "?", to: node.version });
+        const again = to ? `strom update --channel ${to}` : "strom update";
         if (isAgent(ctx.env) || !ctx.yes) {
+            // said no: said so, nothing installed and the version it stays at, and not ok (exit 1) — never a quiet 0 that
+            // reads as done (found: strom update --channel beta ended 0 at the old version, nobody knew why)
             if (!isAgent(ctx.env) && ctx.interactive) {
                 if (!(await ctx.confirm(question, true)))
-                    return { text: t("ui.update.later"), data: { current: VERSION, latest: newest, updated: false, ...extra } };
+                    return { text: t("ui.update.later", { current: VERSION, command: again }), data: { current: VERSION, latest: newest, updated: false, refused: true, ...extra }, exitCode: EXIT.error };
             }
             else
-                ctx.requireHuman(`update strom to ${newest}${to ? ` (--channel ${to})` : ""}`, to ? `strom update --channel ${to}` : "strom update", "update", question);
+                ctx.requireHuman(`update strom to ${newest}${to ? ` (--channel ${to})` : ""}`, again, "update", question);
         }
         // Another channel, or an older version: every research and the settings backed up before anything is installed —
         // none made, nothing installed (Milan, 2026-10-07: "přechod beta ↔ produkce nikdy nepřijde o data")
@@ -666,7 +669,9 @@ register({
             done = await installUpdate(ctx.env, inst.root, process.platform, { release: latest, keep: ctx.researchFolders(), ...(switching ? { channel: to } : {}) });
         }
         catch (e) {
-            throw new StromError(t("ui.update.failed", { detail: e.message }), { hint: "strom update — or run the installer again" });
+            // the person's language for a person (ui.error.update.failed and its hint), the English for an agent and a program
+            const params = { detail: e.message, current: VERSION, command: again };
+            throw new StromError(ui("en", "ui.error.update.failed", params), { hint: ui("en", "ui.error.update.failed.hint", params), code: "update.failed", params });
         }
         // The agents learn what the new version tells them.
         const { command, args } = stromLauncher();

package/dist/commands/sources.js

@@ -11,7 +11,8 @@ import { makeNote, makeNotes, parseDate } from "../core/actions.js";
 import { calibrationLine, fitCalibration } from "../core/calibration.js";
 import { foldText } from "../core/text.js";
 import { clipNote, clipText, imageOfRef, MAX_CLIPS, transcriptNote } from "../core/media.js";
-import { partRegion } from "../core/views.js";
+import { mediaOrient, partRegion } from "../core/views.js";
+import { fileToShown } from "../image/image.js";
 import { formatName } from "../core/people.js";
 import { yearsOption, yearsOverlap } from "../core/years.js";
 import { levelWords } from "../core/archiveasks.js";
@@ -49,8 +50,9 @@ function mediaRefs(tree, refs) {
 /**
  * Where the entry is on an image, as its view was cut: "M0012@0.05,0.4,0.45,0.18", "B0001:57@left:0.1,0.4,0.8,0.1"
  * (a half, then a crop within it), pixels of the registered image, or "M0012" alone (the entry is the whole image).
+ * Of the picture as the view showed it (a photo upright, turned by its EXIF); kept as the picture lies in its file.
  */
-function clipRefs(tree, specs) {
+function clipRefs(tree, specs, shared) {
     return specs.map((spec) => {
         const m = /^([^@]+?)(?:@(.+))?$/.exec(spec.trim()) ?? [];
         let [ref, where] = [m[1] ?? "", m[2]];
@@ -62,7 +64,8 @@ function clipRefs(tree, specs) {
         if (!where)
             return { media, region: { x: 0, y: 0, w: 1, h: 1 } };
         const w = /^(?:(left|right|top|bottom)(?::|$))?(.*)$/.exec(where.trim());
-        const region = partRegion({ half: w[1], crop: w[2] || undefined }, tree.get(media));
+        const img = tree.get(media);
+        const region = partRegion({ half: w[1], crop: w[2] || undefined }, img, mediaOrient(shared, img));
         return { media, region };
     });
 }
@@ -180,7 +183,7 @@ register({
             requireRecord(tree, opts.repo, "repository");
         if (opts.input)
             requireRecord(tree, opts.input, "input");
-        const clips = clipRefs(tree, listOpt(opts.clip));
+        const clips = clipRefs(tree, listOpt(opts.clip), ctx.settings.shared()?.value);
         if (clips.length > MAX_CLIPS)
             throw new UsageError(`at most ${MAX_CLIPS} clips: an entry, and its continuation over a page break`);
         const media = [...new Set([...mediaRefs(tree, csvOpt(opts.media)), ...clips.map((c) => c.media)])];
@@ -243,6 +246,7 @@ register({
     args: [{ name: "source", description: "source ID (S0001)", required: true }],
     run(ctx, { args }) {
         const tree = ctx.tree();
+        const shared = ctx.settings.shared()?.value;
         const s = requireRecord(tree, args[0], "source");
         const citing = [];
         const at = (c) => (c.locator ? ` (${c.locator})` : "");
@@ -265,7 +269,9 @@ register({
                     if (c.source === s.id)
                         citing.push(`${f.id} ${e.id} ${e.kind}${at(c)}`);
         }
-        const text = lines(`${s.id} ${s.title}`, `kind ${s.kind} · ${s.form} · information ${s.information}${s.language ? ` · language ${s.language}` : ""}${s.date ? ` · dated ${s.date}` : ""}`, s.recordset || s.repository ? `in     ${[s.recordset, s.repository].filter(Boolean).join(" · ")}${s.locator ? ` · ${s.locator}` : ""}` : s.locator ? `at     ${s.locator}` : undefined, s.url ? `url    ${s.url}` : undefined, s.media?.length ? `images ${s.media.map((id) => { const m = tree.get(id); return m ? `${id}${m.recordset ? ` (${m.recordset}:${m.image ?? "?"})` : ""}` : id; }).join(" · ")} — strom media view <M…>` : undefined, s.clips?.length ? `clips  ${s.clips.map((c, i) => `${i + 1}. ${clipText(c)}`).join(" · ")} — where the entry is on its image` : undefined, s.transcript ? `\ntranscript\n${s.transcript}` : undefined, s.translation ? `\ntranslation\n${s.translation}` : undefined, citing.length ? `\ncited by\n${citing.map((c) => `  ${c}`).join("\n")}` : "\ncited by nothing yet", ...s.notes.map((n) => `note   ${n.text}`));
+        const text = lines(`${s.id} ${s.title}`, `kind ${s.kind} · ${s.form} · information ${s.information}${s.language ? ` · language ${s.language}` : ""}${s.date ? ` · dated ${s.date}` : ""}`, s.recordset || s.repository ? `in     ${[s.recordset, s.repository].filter(Boolean).join(" · ")}${s.locator ? ` · ${s.locator}` : ""}` : s.locator ? `at     ${s.locator}` : undefined, s.url ? `url    ${s.url}` : undefined, s.media?.length ? `images ${s.media.map((id) => { const m = tree.get(id); return m ? `${id}${m.recordset ? ` (${m.recordset}:${m.image ?? "?"})` : ""}` : id; }).join(" · ")} — strom media view <M…>` : undefined, 
+        // as --clip takes them: of the picture as its view shows it
+        s.clips?.length ? `clips  ${s.clips.map((c, i) => `${i + 1}. ${clipText({ media: c.media, region: fileToShown(c.region, mediaOrient(shared, tree.get(c.media))) })}`).join(" · ")} — where the entry is on its image` : undefined, s.transcript ? `\ntranscript\n${s.transcript}` : undefined, s.translation ? `\ntranslation\n${s.translation}` : undefined, citing.length ? `\ncited by\n${citing.map((c) => `  ${c}`).join("\n")}` : "\ncited by nothing yet", ...s.notes.map((n) => `note   ${n.text}`));
         return { text, data: { source: s, citedBy: citing } };
     },
 });
@@ -725,7 +731,7 @@ register({
             requireRecord(tree, opts.recordset, "recordset");
         if (opts.repo)
             requireRecord(tree, opts.repo, "repository");
-        const clips = clipRefs(tree, listOpt(opts.clip));
+        const clips = clipRefs(tree, listOpt(opts.clip), ctx.settings.shared()?.value);
         const drop = str(opts["clip-remove"]);
         const media = [...new Set([...mediaRefs(tree, csvOpt(opts.media)), ...clips.map((c) => c.media)])];
         const s = editFields(tree, args[0], "source", {