@mevdragon/vidfarm-devcli 0.18.0 → 0.19.0

package/dist/src/config.js

@@ -101,6 +101,13 @@ const schema = z.object({
     RAPIDAPI_REMOVE_BACKGROUND_URL: z.string().url().default("https://remove-background18.p.rapidapi.com/public/remove-background/url"),
     RAPIDAPI_REMOVE_BACKGROUND_HOST: z.string().default("remove-background18.p.rapidapi.com"),
     RAPIDAPI_REMOVE_BACKGROUND_COST_USD: z.coerce.number().min(0).default(0.005),
+    // Local sharp chroma-key background removal (no third-party API) — a small
+    // compute/convenience wallet fee, far cheaper than the RapidAPI matting above.
+    GREENSCREEN_CHROMA_KEY_COST_USD: z.coerce.number().min(0).default(0.002),
+    // Two-step "create media overlay": BYOK AI image gen (provider cost rides the
+    // caller's own key) + local chroma key. This fee bills only the platform
+    // convenience/compute leg of the combined primitive.
+    MEDIA_OVERLAY_COST_USD: z.coerce.number().min(0).default(0.008),
     SUPERAGENCY_KEY: z.string().optional(),
     ARTIFACT_SIGNING_SECRET: z.string().optional(),
     HYPERFRAMES_LAMBDA_STACK_NAME: z.string().default("hyperframes-vidfarm-prod"),

package/dist/src/devcli/clip-store.js

@@ -130,7 +130,12 @@ export class ClipStore {
             this.attachEmbeddings([clip]);
         return clip;
     }
-    /** All clips (metadata only), newest first, optionally scoped to a source video. */
+    /**
+     * All clips (metadata only), newest first, optionally scoped to a source
+     * video. `contentTypes` filters by the shot-KIND tag (tags.content_type,
+     * exact match, case-insensitive); when present the limit/offset are applied
+     * AFTER the JS filter (the tag lives in tags_json, not a column).
+     */
     listClips(opts = {}) {
         let sql = "SELECT * FROM clips";
         const args = [];
@@ -139,9 +144,30 @@ export class ClipStore {
             args.push(opts.sourceVideoId);
         }
         sql += " ORDER BY created_at DESC";
-        if (opts.limit) {
+        const wanted = (opts.contentTypes ?? []).map((s) => s.trim().toLowerCase()).filter(Boolean);
+        if (wanted.length) {
+            // Content-type filter can't ride SQL (tag is in JSON), so filter every
+            // matching row in JS, then window with offset/limit.
+            let clips = this.db.prepare(sql).all(...args)
+                .map(rowToClip)
+                .filter((c) => (c.tags.content_type ?? []).some((t) => wanted.includes(String(t).toLowerCase())));
+            if (opts.offset)
+                clips = clips.slice(Math.max(0, Math.floor(opts.offset)));
+            if (opts.limit)
+                clips = clips.slice(0, opts.limit);
+            return clips;
+        }
+        if (opts.limit != null) {
             sql += " LIMIT ?";
             args.push(opts.limit);
+            if (opts.offset) {
+                sql += " OFFSET ?";
+                args.push(Math.max(0, Math.floor(opts.offset)));
+            }
+        }
+        else if (opts.offset) {
+            sql += " LIMIT -1 OFFSET ?";
+            args.push(Math.max(0, Math.floor(opts.offset)));
         }
         return this.db.prepare(sql).all(...args).map(rowToClip);
     }
@@ -313,6 +339,7 @@ function rowToClip(row) {
 }
 function emptyTags() {
     return {
+        content_type: [],
         subject: [],
         action: [],
         emotion: [],

package/dist/src/devcli/clips.js

@@ -12,7 +12,7 @@ import { homedir } from "node:os";
 import { parseArgs } from "node:util";
 import { createInterface } from "node:readline/promises";
 import { createIdV7 } from "../lib/ids.js";
-import { buildEffectiveGuidance, ClipModelClient, detectLocalAgent, detectScenes, estimateScanCostFromScenes, fitScenesToDurationBand, formatCostEstimate, hasFfmpeg, LocalAgentClipClient, normalizeAspect, normalizeCropFocus, normalizeWindows, parseClipHuntPrompt, parseTimeRanges, probeVideo, resolveDurationBand, scanVideo, searchClips } from "../services/clip-curation/index.js";
+import { buildEffectiveGuidance, ClipModelClient, DEFAULT_CLIP_ASPECT, DEFAULT_CLIP_DURATION_BAND, detectLocalAgent, detectScenes, effectiveHuntDurationSec, estimateScanCostFromScenes, fitScenesToDurationBand, formatCostEstimate, hasFfmpeg, LocalAgentClipClient, normalizeAspect, normalizeCropFocus, normalizeWindows, parseClipHuntPrompt, parseTimeRanges, probeVideo, resolveDurationBand, resolveTargetClipCount, scanVideo, searchClips } from "../services/clip-curation/index.js";
 import { ClipStore } from "./clip-store.js";
 export const CLIPS_HELP = `vidfarm raws — build & search a local raws library (local ffmpeg + local agent by default)
 
@@ -54,9 +54,14 @@ export const CLIPS_HELP = `vidfarm raws — build & search a local raws library
 
   raws list                        List raws in the library
     --source <id>                  Only raws from one source video
+    --content-type <a,b>           Only raws whose shot-KIND tag matches (comma-separated,
+                                   exact): talking_head, b_roll, product_shot, screen_recording,
+                                   demo, reaction, interview, establishing, lifestyle, text_graphic
+    --offset <n>                   Skip the first N (paginate with --limit)
     --limit <n> | --json
 
   raws search "<natural language>"   Hybrid structured + semantic search
+    --content-type <a,b>           Post-filter hits by shot-KIND tag (see \`raws list\`)
     --limit <n> | --json | --gemini-key <key>
 
   raws match "<text>"              Top library raws to REPLACE a scene (vector search)
@@ -125,10 +130,13 @@ async function runScan(argv) {
             aspect: { type: "string" },
             "crop-focus": { type: "string" },
             "no-text": { type: "boolean" },
+            "allow-text": { type: "boolean" },
+            "max-clips": { type: "string" },
             "no-refine": { type: "boolean" },
             cloud: { type: "boolean" },
             url: { type: "string" },
             tracer: { type: "string" },
+            folder: { type: "string" },
             "api-url": { type: "string" },
             "api-key": { type: "string" },
             "dry-run": { type: "boolean" },
@@ -153,16 +161,22 @@ async function runScan(argv) {
     const rawWindows = values.range?.length
         ? parseTimeRanges(values.range)
         : parsedPrompt.windows;
-    const durationBand = values.duration
+    // Explicit flags / prompt hints win; otherwise fall back to the same
+    // unspecified-prompt product defaults (5–15s clips, vertical 9:16, avoid text,
+    // ~10 clips / 10 min). Opt out with `--aspect original` / `--allow-text`.
+    const explicitBand = values.duration
         ? resolveDurationBand(Number(values.duration))
         : parsedPrompt.target_duration_sec != null
             ? resolveDurationBand(parsedPrompt.target_duration_sec)
             : null;
-    const aspect = values.aspect
-        ? (normalizeAspect(values.aspect) ?? (() => { throw new Error(`--aspect must be 9:16 | 16:9 | 4:3 | 1:1 (or vertical/horizontal/square), got "${values.aspect}"`); })())
+    const durationBand = explicitBand ?? { ...DEFAULT_CLIP_DURATION_BAND };
+    const explicitAspect = values.aspect
+        ? (normalizeAspect(values.aspect) ?? (() => { throw new Error(`--aspect must be 9:16 | 16:9 | 4:3 | 1:1 | original (or vertical/horizontal/square), got "${values.aspect}"`); })())
         : parsedPrompt.aspect;
+    const aspect = explicitAspect ?? DEFAULT_CLIP_ASPECT;
     const cropFocus = normalizeCropFocus(values["crop-focus"]);
-    const avoidText = Boolean(values["no-text"]) || parsedPrompt.avoid_text;
+    const avoidText = values["allow-text"] ? false : (Boolean(values["no-text"]) || parsedPrompt.avoid_text || true);
+    const maxClips = values["max-clips"] ? Math.max(1, Math.round(Number(values["max-clips"]))) : parsedPrompt.max_clips ?? undefined;
     const guidance = buildEffectiveGuidance({ contentPrompt: guidancePromptRaw, avoidText }) || undefined;
     // ── BACKUP path: run the hunt on the deployed pipeline (--cloud) ──────────
     if (values.cloud) {
@@ -175,13 +189,15 @@ async function runScan(argv) {
             apiUrl: values["api-url"],
             apiKey: values["api-key"],
             tracer: values.tracer,
+            folder: values.folder,
             prompt: guidancePromptRaw,
             provider: apiProvider,
             windows: rawWindows,
             durationBand,
             aspect,
             cropFocus,
-            avoidText
+            avoidText,
+            maxClips
         });
     }
     const videoArg = positionals[0];
@@ -224,6 +240,8 @@ async function runScan(argv) {
     if (rawWindows.length && !windows.length) {
         throw new Error("--range produced no usable window inside the video's duration.");
     }
+    // Density/uniqueness target: ~10 clips / 10 min unless --max-clips / "N clips".
+    const targetClipCount = resolveTargetClipCount(effectiveHuntDurationSec(probe.duration_sec, windows), { max_clips: maxClips });
     if (windows.length) {
         console.log(`[raws] hunting only ${windows.map((w) => `${fmtClock(w.start_sec)}–${fmtClock(w.end_sec)}`).join(", ")} (${fmtDuration(windows.reduce((a, w) => a + w.end_sec - w.start_sec, 0))} of ${fmtDuration(probe.duration_sec)}).`);
     }
@@ -234,6 +252,7 @@ async function runScan(argv) {
         console.log(`[raws] cropping clips to ${aspect} (${cropFocus}).`);
     if (avoidText)
         console.log(`[raws] preferring scenes without on-screen text/captions (selection only — GhostCut is never run on the source).`);
+    console.log(`[raws] aiming for ~${targetClipCount} distinct raw${targetClipCount === 1 ? "" : "s"}${maxClips ? " (--max-clips)" : " (~10 / 10 min)"}.`);
     console.log(`[raws] detecting scenes …`);
     const detected = await detectScenes(videoPath, { durationSec: probe.duration_sec, windows, ...sceneOptions });
     const scenes = fitScenesToDurationBand(detected, durationBand);
@@ -329,6 +348,7 @@ async function runScan(argv) {
         probe,
         windows,
         durationBand,
+        targetClipCount,
         sourceFilename: path.basename(videoPath),
         concurrency,
         onRefine: (r) => {
@@ -437,12 +457,16 @@ async function runScanCloud(input) {
             prompt: input.prompt ?? "",
             ...(input.provider ? { provider: input.provider } : {}),
             ...(input.tracer ? { tracer: input.tracer } : {}),
+            ...(input.folder ? { folder_path: input.folder } : {}),
             hunt_spec: {
                 ...(input.windows.length ? { windows: input.windows } : {}),
                 ...(input.durationBand ? { duration_band: input.durationBand } : {}),
                 ...(input.aspect ? { aspect: input.aspect } : {}),
                 crop_focus: input.cropFocus,
-                ...(input.avoidText ? { avoid_text: true } : {})
+                // Send explicitly (incl. false) so --allow-text can override the
+                // server-side avoid-text default.
+                avoid_text: input.avoidText,
+                ...(input.maxClips ? { max_clips: input.maxClips } : {})
             }
         })
     });
@@ -510,13 +534,20 @@ async function runList(argv) {
         args: argv,
         options: {
             source: { type: "string" },
+            "content-type": { type: "string" },
+            offset: { type: "string" },
             limit: { type: "string" },
             json: { type: "boolean" },
             home: { type: "string" }
         }
     });
     const store = new ClipStore(values.home);
-    const clips = store.listClips({ sourceVideoId: values.source, limit: clampInt(values.limit, 50, 1, 5000) });
+    const clips = store.listClips({
+        sourceVideoId: values.source,
+        contentTypes: parseContentTypes(values["content-type"]),
+        offset: values.offset ? clampInt(values.offset, 0, 0, 1_000_000) : undefined,
+        limit: clampInt(values.limit, 50, 1, 5000)
+    });
     store.close();
     if (values.json) {
         console.log(JSON.stringify(clips, null, 2));
@@ -538,6 +569,7 @@ async function runSearch(argv) {
         allowPositionals: true,
         options: {
             provider: { type: "string" },
+            "content-type": { type: "string" },
             limit: { type: "string" },
             json: { type: "boolean" },
             "gemini-key": { type: "string" },
@@ -549,6 +581,7 @@ async function runSearch(argv) {
     const query = positionals.join(" ").trim();
     if (!query)
         throw new Error('Usage: vidfarm raws search "<natural language>"');
+    const contentTypes = parseContentTypes(values["content-type"]);
     const limit = clampInt(values.limit, 10, 1, 200);
     const store = new ClipStore(values.home);
     if (store.stats().clips === 0) {
@@ -582,7 +615,13 @@ async function runSearch(argv) {
         console.warn(`[raws] no ${provider} key or local agent CLI — structured/keyword search only.`);
     }
     const t0 = timestamp();
-    const hits = store.search({ criteria, queryEmbedding, limit });
+    // Over-fetch when a content-type filter is on so the post-filter still fills
+    // the requested limit (the tag lives in tags_json, not the search index).
+    const searchLimit = contentTypes.length ? Math.min(1000, limit * 5) : limit;
+    let hits = store.search({ criteria, queryEmbedding, limit: searchLimit });
+    if (contentTypes.length) {
+        hits = hits.filter((h) => (h.clip.tags.content_type ?? []).some((t) => contentTypes.includes(String(t).toLowerCase()))).slice(0, limit);
+    }
     const ms = elapsedMs(t0);
     store.close();
     if (values.json) {
@@ -895,6 +934,13 @@ function normalizeProvider(value) {
         return value;
     throw new Error(`--provider must be gemini | openai | openrouter (got "${value}")`);
 }
+// Parse a comma-separated --content-type flag into lowercase shot-KIND tags for
+// the tags.content_type filter (talking_head, b_roll, product_shot, …).
+function parseContentTypes(value) {
+    if (!value)
+        return [];
+    return value.split(",").map((s) => s.trim().toLowerCase()).filter(Boolean);
+}
 function normalizeTier(value) {
     if (value === "flash")
         return "flash";