@aspan-corporation/ac-shared 1.2.37 → 1.2.39

package/lib/utils/diary.d.ts

@@ -15,8 +15,26 @@ export declare const TAG_DIARY_TITLE = "ac:diary:title";
 export declare const TAG_DIARY_PREVIEW = "ac:diary:preview";
 /** One per embedded photo: value is the referenced media key. */
 export declare const TAG_DIARY_PHOTO = "ac:diary:photo";
+/** One per embedded audio recording: value is the referenced diary key. */
+export declare const TAG_DIARY_AUDIO = "ac:diary:audio";
+/** One per embedded video: value is the referenced media key. Unlike audio,
+ *  diary videos DO run through the full video pipeline (encode + thumbnail),
+ *  so they get their own tag to keep the photo count accurate rather than
+ *  reusing TAG_DIARY_PHOTO. */
+export declare const TAG_DIARY_VIDEO = "ac:diary:video";
 /** Namespace for body word tokens: `ac:text:<word>`. */
 export declare const TEXT_TOKEN_PREFIX = "ac:text:";
+/**
+ * Audio formats a browser MediaRecorder can produce: Safari records
+ * audio/mp4 (.m4a), Chromium/Firefox audio/webm or audio/ogg (opus).
+ * Deliberately kept SEPARATE from the image-only `isAllowedExtension`
+ * (thumbsKey.ts) — that allowlist is what keeps diary audio out of the
+ * meta-extractor/resizer dispatch on save, so recordings stay entry-only
+ * (no meta item, never in browse/search).
+ */
+export declare const AUDIO_EXTENSIONS: string[];
+/** True when a key's extension is a supported diary audio format. */
+export declare const isAllowedAudioExtension: (key: string) => boolean;
 /** Max body characters surfaced in the preview tag. */
 export declare const DIARY_PREVIEW_LENGTH = 160;
 /** Upper bound on distinct word tokens written per entry (bounds item size). */

package/lib/utils/diary.js

@@ -5,6 +5,7 @@
  * searchable by tokenising it into `ac:text:<word>` tags that ride the existing
  * inverted-index search table.
  */
+import { getKeyExtension } from "./thumbsKey.js";
 /** Key prefix for every diary object (and the `folder`/id namespace). */
 export const DIARY_PREFIX = "diary/";
 /** Marks a meta item as a diary entry (value "true"). */
@@ -15,8 +16,26 @@ export const TAG_DIARY_TITLE = "ac:diary:title";
 export const TAG_DIARY_PREVIEW = "ac:diary:preview";
 /** One per embedded photo: value is the referenced media key. */
 export const TAG_DIARY_PHOTO = "ac:diary:photo";
+/** One per embedded audio recording: value is the referenced diary key. */
+export const TAG_DIARY_AUDIO = "ac:diary:audio";
+/** One per embedded video: value is the referenced media key. Unlike audio,
+ *  diary videos DO run through the full video pipeline (encode + thumbnail),
+ *  so they get their own tag to keep the photo count accurate rather than
+ *  reusing TAG_DIARY_PHOTO. */
+export const TAG_DIARY_VIDEO = "ac:diary:video";
 /** Namespace for body word tokens: `ac:text:<word>`. */
 export const TEXT_TOKEN_PREFIX = "ac:text:";
+/**
+ * Audio formats a browser MediaRecorder can produce: Safari records
+ * audio/mp4 (.m4a), Chromium/Firefox audio/webm or audio/ogg (opus).
+ * Deliberately kept SEPARATE from the image-only `isAllowedExtension`
+ * (thumbsKey.ts) — that allowlist is what keeps diary audio out of the
+ * meta-extractor/resizer dispatch on save, so recordings stay entry-only
+ * (no meta item, never in browse/search).
+ */
+export const AUDIO_EXTENSIONS = ["m4a", "webm", "ogg"];
+/** True when a key's extension is a supported diary audio format. */
+export const isAllowedAudioExtension = (key) => AUDIO_EXTENSIONS.includes(getKeyExtension(key));
 /** Max body characters surfaced in the preview tag. */
 export const DIARY_PREVIEW_LENGTH = 160;
 /** Upper bound on distinct word tokens written per entry (bounds item size). */
@@ -77,14 +96,85 @@ export const parseDiaryKeyDate = (key) => {
 // A small English stop-word set — high-frequency words that add noise and bloat
 // the index without improving recall for a personal diary.
 const STOP_WORDS = new Set([
-    "the", "and", "for", "are", "but", "not", "you", "all", "any", "can", "had",
-    "her", "was", "one", "our", "out", "day", "get", "has", "him", "his", "how",
-    "its", "may", "new", "now", "old", "see", "two", "way", "who", "did", "yes",
-    "his", "she", "they", "them", "this", "that", "with", "have", "from", "your",
-    "were", "been", "their", "what", "when", "then", "than", "into", "just",
-    "like", "over", "also", "back", "after", "would", "could", "there", "here",
-    "about", "which", "while", "these", "those", "where", "very", "much", "some",
-    "such", "only", "more", "most", "will", "well", "went", "going", "got",
+    "the",
+    "and",
+    "for",
+    "are",
+    "but",
+    "not",
+    "you",
+    "all",
+    "any",
+    "can",
+    "had",
+    "her",
+    "was",
+    "one",
+    "our",
+    "out",
+    "day",
+    "get",
+    "has",
+    "him",
+    "his",
+    "how",
+    "its",
+    "may",
+    "new",
+    "now",
+    "old",
+    "see",
+    "two",
+    "way",
+    "who",
+    "did",
+    "yes",
+    "his",
+    "she",
+    "they",
+    "them",
+    "this",
+    "that",
+    "with",
+    "have",
+    "from",
+    "your",
+    "were",
+    "been",
+    "their",
+    "what",
+    "when",
+    "then",
+    "than",
+    "into",
+    "just",
+    "like",
+    "over",
+    "also",
+    "back",
+    "after",
+    "would",
+    "could",
+    "there",
+    "here",
+    "about",
+    "which",
+    "while",
+    "these",
+    "those",
+    "where",
+    "very",
+    "much",
+    "some",
+    "such",
+    "only",
+    "more",
+    "most",
+    "will",
+    "well",
+    "went",
+    "going",
+    "got",
 ]);
 /**
  * Tokenise diary markdown into a deduplicated set of searchable lowercase

package/lib/utils/hintFallback.d.ts

@@ -0,0 +1,14 @@
+export type TagInput = {
+    key: string;
+    value: string;
+};
+/**
+ * Fallback tags from device hints for fields the file's own metadata (EXIF
+ * for photos, ffprobe container tags for videos) didn't provide. The
+ * extracted metadata always wins: a hint is used only when the corresponding
+ * key is absent from `extractedTags`. Only `dateCreated` / `latitude` /
+ * `longitude` are pushed; processMeta derives year/month/day from dateCreated
+ * and reverse-geocodes lat/long into country/region tags exactly as it does
+ * for real metadata. Shared by the image and video meta-extractors.
+ */
+export default function hintFallbackTags(extractedTags: TagInput[], s3Metadata: Record<string, string> | undefined): TagInput[];

package/lib/utils/hintFallback.js

@@ -0,0 +1,51 @@
+// S3 user-metadata keys written by the diary upload-url route when the
+// browser supplied device hints (mobile OSes strip GPS EXIF/ffprobe metadata —
+// and sometimes the date — from files handed to web pages, so the client
+// sends what it knows: the file's own timestamp and, for just-taken photos or
+// videos, the device's current position). S3 lowercases metadata keys, so
+// these are the exact wire names as they come back from HeadObject.
+const HINT_DATE = "hint-date";
+const HINT_LATITUDE = "hint-latitude";
+const HINT_LONGITUDE = "hint-longitude";
+// Number() quirks: Number("") and Number("  ") are 0, not NaN — an empty
+// hint must not turn into a valid coordinate at (0, 0).
+const parseFinite = (value) => {
+    if (value === undefined || value.trim() === "")
+        return undefined;
+    const n = Number(value);
+    return Number.isFinite(n) ? n : undefined;
+};
+/**
+ * Fallback tags from device hints for fields the file's own metadata (EXIF
+ * for photos, ffprobe container tags for videos) didn't provide. The
+ * extracted metadata always wins: a hint is used only when the corresponding
+ * key is absent from `extractedTags`. Only `dateCreated` / `latitude` /
+ * `longitude` are pushed; processMeta derives year/month/day from dateCreated
+ * and reverse-geocodes lat/long into country/region tags exactly as it does
+ * for real metadata. Shared by the image and video meta-extractors.
+ */
+export default function hintFallbackTags(extractedTags, s3Metadata) {
+    if (!s3Metadata)
+        return [];
+    const has = (key) => extractedTags.some((t) => t.key === key);
+    const out = [];
+    if (!has("dateCreated") && s3Metadata[HINT_DATE]) {
+        const d = new Date(s3Metadata[HINT_DATE]);
+        if (!isNaN(d.getTime())) {
+            out.push({ key: "dateCreated", value: d.toISOString() });
+        }
+    }
+    // Coordinates only ever make sense as a pair; note 0 is a valid value
+    // (equator / prime meridian), so presence checks must not be falsy checks.
+    if (!has("latitude") && !has("longitude")) {
+        const lat = parseFinite(s3Metadata[HINT_LATITUDE]);
+        const lon = parseFinite(s3Metadata[HINT_LONGITUDE]);
+        if (lat !== undefined &&
+            lon !== undefined &&
+            Math.abs(lat) <= 90 &&
+            Math.abs(lon) <= 180) {
+            out.push({ key: "latitude", value: String(lat) }, { key: "longitude", value: String(lon) });
+        }
+    }
+    return out;
+}

package/lib/utils/index.d.ts

@@ -5,3 +5,4 @@ export * from "./helpers.js";
 export * from "./processMeta.js";
 export * from "./parseFolderDate.js";
 export * from "./diary.js";
+export { default as hintFallbackTags, type TagInput } from "./hintFallback.js";

package/lib/utils/index.js

@@ -9,3 +9,4 @@ export * from "./helpers.js";
 export * from "./processMeta.js";
 export * from "./parseFolderDate.js";
 export * from "./diary.js";
+export { default as hintFallbackTags } from "./hintFallback.js";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@aspan-corporation/ac-shared",
-  "version": "1.2.37",
+  "version": "1.2.39",
   "description": "",
   "keywords": [],
   "exports": {