reelforge 1.24.2 → 1.25.0

package/dist/commands/images.js

@@ -1,107 +0,0 @@
-import fs from "node:fs/promises";
-import fsSync from "node:fs";
-import path from "node:path";
-import { post } from "../client.js";
-import { uploadToOss } from "../utils/oss-upload.js";
-import { downloadTo } from "../utils/download.js";
-import { print, success, warn } from "../utils/output.js";
-const MAX_REF_BYTES = 15 * 1024 * 1024;
-async function resolveImageRef(input) {
-    const t = input.trim();
-    if (!t)
-        throw new Error("--image: empty value");
-    if (/^https?:\/\//i.test(t) || t.startsWith("data:"))
-        return { url: t };
-    const abs = path.resolve(t);
-    if (!fsSync.existsSync(abs)) {
-        throw new Error(`--image: local file not found: ${abs}`);
-    }
-    const ext = path.extname(abs).toLowerCase();
-    const mime = ext === ".jpg" || ext === ".jpeg" ? "image/jpeg" :
-        ext === ".webp" ? "image/webp" :
-            ext === ".png" ? "image/png" :
-                null;
-    if (!mime) {
-        throw new Error(`--image: unsupported extension ${ext} (use png/jpg/jpeg/webp)`);
-    }
-    try {
-        const key = await uploadToOss(abs);
-        return { key };
-    }
-    catch (e) {
-        const sz = (await fs.stat(abs)).size;
-        if (sz > MAX_REF_BYTES) {
-            throw new Error(`--image ${path.basename(abs)} 直传 OSS 失败 (${e instanceof Error ? e.message : e}),且文件超过内联回退上限 ${MAX_REF_BYTES} 字节。`);
-        }
-        warn(`--image ${path.basename(abs)}: OSS 直传不可用,回退内联上传 (${e instanceof Error ? e.message : e})`);
-        const buf = await fs.readFile(abs);
-        return { url: `data:${mime};base64,${buf.toString("base64")}` };
-    }
-}
-export function registerImages(program) {
-    const images = program
-        .command("images")
-        .description("图像生成: 文本 → 图片")
-        .helpOption("-h, --help", "show help");
-    images
-        .command("generate")
-        .description("Generate an image (text-to-image or ref-image, depending on model + --image)")
-        .helpOption("-h, --help", "show help")
-        .requiredOption("-p, --prompt <text>", "text prompt. With --image, reference each ref via 「图 N」/「Image N」.")
-        .option("-m, --model <id>", "image model id (rx-image-z | rx-image-flux | rx-image-qwen | rx-image-qwen-edit)")
-        .option("--width <n>", "image width", parseInt)
-        .option("--height <n>", "image height", parseInt)
-        .option("-i, --image <urlOrPath>", "reference image, repeatable up to 3 times. URL / data: URI / local png/jpg/webp path (auto-base64'd). Index N-1 maps to 「图 N」in the prompt. Only honored by edit-capable SKUs (rx-image-qwen-edit).", (val, prev) => [...prev, val], [])
-        .option("-o, --output <file>", "download first image to this local path")
-        .option("--all-output <dir>", "download ALL generated images into this directory")
-        .addHelpText("after", [
-        "",
-        "Examples:",
-        "  # plain text-to-image",
-        "  reelforge images generate -p 'a cat' -m rx-image-flux --width 1024 --height 1024 -o cat.png",
-        "",
-        "  # ref-image edit (Qwen-Image-Edit), 1 ref local file",
-        "  reelforge images generate -m rx-image-qwen-edit \\",
-        "    -p '保持图1人物，把背景改成雪山黄昏' \\",
-        "    -i ./hero.png -o out.png",
-        "",
-        "  # ref-image edit, 2 refs (URL + local), 「图1」+「图2」",
-        "  reelforge images generate -m rx-image-qwen-edit \\",
-        "    -p '让图1的人物穿上图2的衣服' \\",
-        "    -i https://example.com/person.jpg -i ./outfit.png -o composite.png",
-    ].join("\n"))
-        .action(async (opts) => {
-        const body = { prompt: opts.prompt };
-        if (opts.model)
-            body.model = opts.model;
-        if (opts.width !== undefined)
-            body.width = opts.width;
-        if (opts.height !== undefined)
-            body.height = opts.height;
-        if (opts.image && Array.isArray(opts.image) && opts.image.length > 0) {
-            if (opts.image.length > 3) {
-                throw new Error(`--image accepts at most 3 refs (got ${opts.image.length})`);
-            }
-            const resolved = await Promise.all(opts.image.map(resolveImageRef));
-            const urls = resolved.filter((r) => r.url).map((r) => r.url);
-            const keys = resolved.filter((r) => r.key).map((r) => r.key);
-            if (urls.length)
-                body.image_urls = urls;
-            if (keys.length)
-                body.image_keys = keys;
-        }
-        const r = await post("/api/v1/images/generate", body);
-        if (opts.output && r.images?.[0]) {
-            await downloadTo(r.images[0], opts.output);
-            success(`Saved → ${opts.output}`);
-        }
-        if (opts.allOutput) {
-            for (let i = 0; i < r.images.length; i++) {
-                const dest = `${opts.allOutput}/image_${i + 1}.png`;
-                await downloadTo(r.images[i], dest);
-                success(`Saved → ${dest}`);
-            }
-        }
-        print(r);
-    });
-}

package/dist/commands/llm.js

@@ -1,67 +0,0 @@
-import fs from "node:fs/promises";
-import { get, post } from "../client.js";
-import { print, table } from "../utils/output.js";
-export function registerLlm(program) {
-    const llm = program
-        .command("llm")
-        .description("LLM 工具: chat (单轮对话) / presets (模型预设清单)")
-        .helpOption("-h, --help", "show help");
-    llm
-        .command("chat")
-        .description("Send a single prompt to the configured LLM and print the response")
-        .helpOption("-h, --help", "show help")
-        .requiredOption("-p, --prompt <text>", "the prompt text (use @file to read from a file)")
-        .option("-m, --model <name>", "override model (default: from server config)")
-        .option("--base-url <url>", "override LLM base URL")
-        .option("--api-key <key>", "override LLM API key")
-        .option("-t, --temperature <n>", "sampling temperature (0..2)", parseFloat)
-        .option("--max-tokens <n>", "max tokens to generate", parseInt)
-        .option("--schema <file>", "JSON schema file for structured output (returns parsed object)")
-        .addHelpText("after", [
-        "",
-        "Examples:",
-        "  reelforge llm chat -p 'Hello'",
-        "  reelforge llm chat -p @prompt.txt -m deepseek-v4-flash -t 0.4",
-        "  reelforge llm chat -p 'movie review of Inception' --schema review.json --json",
-        "",
-        "Model IDs:",
-        "  Pass the bare model id (e.g. `model-name`).",
-        "  The provider/model form returned by `rf models` (e.g. `provider/model-name`)",
-        "  may need explicit provider access on your account and is more",
-        "  likely to return 'no permission'.",
-    ].join("\n"))
-        .action(async (opts) => {
-        let prompt = opts.prompt;
-        if (prompt.startsWith("@"))
-            prompt = await fs.readFile(prompt.slice(1), "utf-8");
-        const body = { prompt };
-        if (opts.model)
-            body.model = opts.model;
-        if (opts.baseUrl)
-            body.base_url = opts.baseUrl;
-        if (opts.apiKey)
-            body.api_key = opts.apiKey;
-        if (opts.temperature !== undefined)
-            body.temperature = opts.temperature;
-        if (opts.maxTokens !== undefined)
-            body.max_tokens = opts.maxTokens;
-        if (opts.schema) {
-            body.json_schema = JSON.parse(await fs.readFile(opts.schema, "utf-8"));
-        }
-        const r = await post("/api/v1/llm/chat", body);
-        print(r.output);
-    });
-    llm
-        .command("presets")
-        .description("List built-in model presets")
-        .helpOption("-h, --help", "show help")
-        .action(async () => {
-        const r = await get("/api/v1/llm/presets");
-        table(r.presets.map((p) => ({
-            id: p.id,
-            label: p.label,
-            base_url: p.baseUrl,
-            default_model: p.defaultModel,
-        })));
-    });
-}

package/dist/commands/media.js

@@ -1,136 +0,0 @@
-import fs from "node:fs/promises";
-import fsSync from "node:fs";
-import path from "node:path";
-import { post } from "../client.js";
-import { uploadToOss } from "../utils/oss-upload.js";
-import { info, isJson, print, warn } from "../utils/output.js";
-const MEDIA_EXT_TO_MIME = {
-    ".mp4": "video/mp4",
-    ".mov": "video/quicktime",
-    ".webm": "video/webm",
-    ".mkv": "video/x-matroska",
-    ".m4v": "video/mp4",
-    ".mp3": "audio/mpeg",
-    ".wav": "audio/wav",
-    ".m4a": "audio/mp4",
-    ".ogg": "audio/ogg",
-    ".flac": "audio/flac",
-    ".png": "image/png",
-    ".jpg": "image/jpeg",
-    ".jpeg": "image/jpeg",
-    ".webp": "image/webp",
-};
-function fmtSize(bytes) {
-    if (!bytes)
-        return "0";
-    if (bytes < 1024)
-        return `${bytes} B`;
-    if (bytes < 1024 * 1024)
-        return `${(bytes / 1024).toFixed(1)} KB`;
-    if (bytes < 1024 * 1024 * 1024)
-        return `${(bytes / 1024 / 1024).toFixed(1)} MB`;
-    return `${(bytes / 1024 / 1024 / 1024).toFixed(2)} GB`;
-}
-function formatHuman(r, label) {
-    const lines = [];
-    lines.push(`${label}  [${r.kind}]`);
-    if (r.formatName)
-        lines.push(`  format:       ${r.formatName}`);
-    if (r.durationSec) {
-        const min = Math.floor(r.durationSec / 60);
-        const sec = (r.durationSec % 60).toFixed(2);
-        lines.push(`  duration:     ${r.durationSec.toFixed(2)}s  (${min}:${sec.padStart(5, "0")})`);
-    }
-    if (r.sizeBytes)
-        lines.push(`  size:         ${fmtSize(r.sizeBytes)}`);
-    if (r.bitrate)
-        lines.push(`  bitrate:      ${(r.bitrate / 1000).toFixed(0)} kbps`);
-    if (r.video) {
-        lines.push(`  video:        ${r.video.codec}  ${r.video.width}×${r.video.height}  @ ${r.video.fps} fps` +
-            (r.video.pixFmt ? `  (${r.video.pixFmt})` : ""));
-    }
-    if (r.audio) {
-        lines.push(`  audio:        ${r.audio.codec}  ${r.audio.sampleRate} Hz  ${r.audio.channels}ch` +
-            (r.audio.channelLayout ? ` (${r.audio.channelLayout})` : ""));
-    }
-    const counts = r.streamCounts;
-    if (counts.subtitle > 0 || counts.other > 0) {
-        lines.push(`  streams:      ${counts.video}v + ${counts.audio}a + ${counts.subtitle}sub + ${counts.other}other`);
-    }
-    return lines;
-}
-export function registerMedia(program) {
-    const mediaCmd = program
-        .command("media")
-        .description("媒体处理原子能力。目前包含: probe (ffprobe 探针)")
-        .helpOption("-h, --help", "show help");
-    mediaCmd
-        .command("probe <input>")
-        .description("ffprobe 一个媒体文件 / URL → JSON metadata。常用于 agent 在决定下游参数前查媒体长度/编码/分辨率/帧率。")
-        .helpOption("-h, --help", "show help")
-        .addHelpText("after", [
-        "",
-        "Input 类型 (自动识别):",
-        "  本地路径    ./video.mp4              CLI base64 上传到服务端探针",
-        "  http(s) URL https://example.com/x.mp4  服务端 ffprobe 直读",
-        "",
-        "支持的扩展名:",
-        "  视频: .mp4 .mov .webm .mkv .m4v",
-        "  音频: .mp3 .wav .m4a .ogg .flac",
-        "  图片: .png .jpg .jpeg .webp",
-        "",
-        "Examples:",
-        "  # 本地 MP4",
-        "  rf media probe ./video.mp4",
-        "",
-        "  # 远程 URL (服务端直读, 不下载到本地)",
-        "  rf media probe 'https://example.com/clip.mp4'",
-        "",
-        "  # JSON 输出 (agent 解析)",
-        "  rf media probe ./video.mp4 --json",
-        "",
-        "  # 抽特定字段",
-        "  rf media probe ./video.mp4 --json | jq '.durationSec'",
-        "  rf media probe ./video.mp4 --json | jq '.video.fps'",
-        "",
-        "成本: $0 (纯 ffprobe, 无 LLM/ASR)",
-    ].join("\n"))
-        .action(async (input) => {
-        const isRemote = /^https?:\/\//i.test(input);
-        let body;
-        let label;
-        if (isRemote) {
-            body = { url: input };
-            label = input;
-        }
-        else {
-            const abs = path.resolve(input);
-            if (!fsSync.existsSync(abs))
-                throw new Error(`file not found: ${abs}`);
-            const ext = path.extname(abs).toLowerCase();
-            const mime = MEDIA_EXT_TO_MIME[ext];
-            if (!mime) {
-                throw new Error(`unsupported extension "${ext}". Supported: ${Object.keys(MEDIA_EXT_TO_MIME).join(" / ")}, ` +
-                    `or pass a URL.`);
-            }
-            try {
-                const key = await uploadToOss(abs);
-                body = { key };
-            }
-            catch (e) {
-                warn(`${path.basename(abs)}: OSS 直传不可用,回退内联上传 (${e instanceof Error ? e.message : e})`);
-                const buf = await fs.readFile(abs);
-                body = { data_uri: `data:${mime};base64,${buf.toString("base64")}` };
-            }
-            label = path.basename(abs);
-        }
-        const r = await post("/api/v1/media/probe", body);
-        if (isJson()) {
-            print(r);
-            return;
-        }
-        const lines = formatHuman(r, label);
-        for (const line of lines)
-            info(line);
-    });
-}

package/dist/commands/models.js

@@ -1,36 +0,0 @@
-import { get } from "../client.js";
-import { table } from "../utils/output.js";
-export function registerModels(program) {
-    program
-        .command("models")
-        .description("列出可用模型目录 (LLM / TTS / 图片 / ASR) 含价格")
-        .helpOption("-h, --help", "show help")
-        .option("--modality <m>", "filter by modality: llm | tts | image | asr")
-        .option("--refresh", "bypass the 5-minute server-side cache")
-        .addHelpText("after", [
-        "",
-        "Examples:",
-        "  reelforge models                       # all modalities",
-        "  reelforge models --modality llm        # only chat models",
-        "  reelforge models --modality tts        # only TTS models",
-        "  reelforge models --refresh             # force a re-fetch",
-    ].join("\n"))
-        .action(async (opts) => {
-        const qs = new URLSearchParams();
-        if (opts.modality)
-            qs.set("modality", opts.modality);
-        if (opts.refresh)
-            qs.set("refresh", "1");
-        const path = `/api/v1/models${qs.toString() ? `?${qs.toString()}` : ""}`;
-        const r = await get(path);
-        table(r.models.map((m) => ({
-            id: m.id,
-            modality: m.modality,
-            owned_by: m.owned_by,
-            context: m.context_length ?? "",
-            input_per_1m: m.pricing.input_per_1m ?? "",
-            output_per_1m: m.pricing.output_per_1m ?? "",
-            per_1m_chars: m.pricing.per_1m_chars ?? "",
-        })));
-    });
-}

package/dist/commands/pipelines.js

@@ -1,142 +0,0 @@
-import fs from "node:fs/promises";
-import { post } from "../client.js";
-import { waitForTask } from "../utils/task-waiter.js";
-import { downloadTo } from "../utils/download.js";
-import { print, success, warn } from "../utils/output.js";
-async function submitAndMaybeWait(endpoint, body, common) {
-    const { task_id, status } = await post(endpoint, body);
-    if (common.wait === false) {
-        print({ task_id, status });
-        return;
-    }
-    const t = await waitForTask(task_id, {
-        pollMs: common.pollMs ?? 1500,
-        timeoutMs: common.timeoutMs,
-    });
-    if (t.status !== "completed") {
-        throw new Error(t.error || `Task ended with status ${t.status}`);
-    }
-    const result = t.result;
-    if (common.output && result?.video_url) {
-        await downloadTo(result.video_url, common.output);
-        success(`Saved → ${common.output}`);
-    }
-    print({ task_id: t.id, status: t.status, ...result });
-}
-function commonOptions(cmd) {
-    return cmd
-        .option("--no-wait", "submit and return task_id immediately (do not poll)")
-        .option("-o, --output <file>", "save the final video to this path (only when waiting)")
-        .option("--poll-ms <ms>", "poll interval while waiting", (v) => parseInt(v, 10), 1500)
-        .option("--timeout-ms <ms>", "max wait time before aborting (default unlimited)", (v) => parseInt(v, 10));
-}
-export function registerPipelines(program) {
-    const pl = program
-        .command("pipelines")
-        .alias("pipeline")
-        .description("端到端视频流水线 (与 rf create 等价, 暴露更多底层参数)")
-        .helpOption("-h, --help", "show help");
-    commonOptions(pl
-        .command("standard")
-        .description("Audio-first pipeline: topic|script → master TTS → ASR → scene/subtitle layers → final MP4")
-        .helpOption("-h, --help", "show help")
-        .option("-t, --topic <text>", "video topic (mode=generate). Use @file to read from disk.")
-        .option("--script <text>", "your own master script text (mode=fixed). Use @file to read from disk.")
-        .option("--title <text>", "hard-override video title; LLM will NOT auto-summarize. Useful for compliance-sensitive content. Use @file to read from disk. Pass --title '' (empty) to explicitly suppress title rendering. Omit to keep LLM auto-title.")
-        .option("-d, --duration <sec>", "target video duration in seconds (generate mode; default 45)", (v) => parseInt(v, 10))
-        .option("-p, --pace <pace>", "visual rhythm hint: slow | normal | fast (default normal)")
-        .option("--motion <preset>", "per-scene image animation: off | lite (default) | max")
-        .option("--layout <preset>", "image layout: full (default) | blur-bg | letterbox. See below.")
-        .option("--layout-matte-color <css>", "letterbox matte color (CSS). Ignored unless --layout letterbox. Default: black.")
-        .option("--subtitle-style <preset>", "subtitle visual style: plate (default) | stroke | cinema")
-        .option("--preview-only", "只生成可预览的网页版,不出 MP4(快很多)。后续用 `rf render <task-id>` 再出 MP4。")
-        .option("--image-model <id>", "image model (rx-image-z | rx-image-flux | rx-image-qwen | rx-image-qwen-edit)")
-        .option("--prompt-prefix <text>", "style prefix prepended to every image prompt")
-        .option("--character-ref <urlOrPath>", "main character ref for cross-scene identity lock")
-        .option("--voice-id <id>", "TTS voice id (default 专业解说); see `rf tts voices`")
-        .option("--tts-speed <n>", "speech speed (0.5..2; default 1.0). Only honored by some TTS backends; the default cloud model ignores it.", parseFloat)
-        .option("--video-fps <n>", "output video fps (default 24 — cinema-standard, ~20% faster render vs 30; pass 30 if you want smoother motion)", (v) => parseInt(v, 10))
-        .option("--segment-min-chars <N>", "subtitle SEGMENT (分段) min chars (default 10). Segment = one on-screen subtitle unit, not a visual line.", (v) => parseInt(v, 10))
-        .option("--segment-max-chars <N>", "subtitle SEGMENT (分段) max chars (default 28 ≈ 2 visual lines at 820px safe width). HTML clamps to 2 lines + ellipsis.", (v) => parseInt(v, 10))
-        .addHelpText("after", [
-        "",
-        "Two content modes (exactly one required):",
-        "  generate    AI writes the script.   --topic / -t <text>  + optional --duration -d",
-        "  fixed       You supply the script.  --script <text-or-@file>",
-        "",
-        "Pace (LLM visual rhythm hint):  slow | normal | fast",
-        "Motion (per-scene animation):    off | lite | max",
-        "Subtitle style:                  plate | stroke | cinema",
-        "",
-        "Layout — how the AI image sits in the 1080×1920 canvas:",
-        "  full        image fills the whole canvas (default; punchy, 9:16-native content).",
-        "  blur-bg     image at 1080×1080 centered + same image scaled/blurred as background",
-        "              (小红书 / 抖音 style; best for charts, screenshots, non-9:16 source).",
-        "  letterbox   image at 1080×1080 centered + solid matte top/bottom (cinematic).",
-        "              tweak with --layout-matte-color (default 'black').",
-        "  · Image is generated at the actual on-screen size, so you don't pay for cropped pixels.",
-        "",
-        "Examples:",
-        " rf pipelines standard -t 'why we explore space' -d 60 -o space.mp4",
-        " rf pipelines standard --script @script.txt -p slow --motion max -o out.mp4",
-        " rf pipelines standard -t '财经日报' --layout blur-bg --subtitle-style plate -o out.mp4",
-        " rf pipelines standard -t '纪录片' --layout letterbox --motion max -o film.mp4",
-        "",
-        "Tip: `rf create` is a more ergonomic wrapper around the same endpoint.",
-    ].join("\n"))).action(async (opts) => {
-        const hasTopic = typeof opts.topic === "string" && opts.topic.length > 0;
-        const hasScript = typeof opts.script === "string" && opts.script.length > 0;
-        if (!hasTopic && !hasScript) {
-            throw new Error("either --topic / -t or --script is required");
-        }
-        if (hasTopic && hasScript) {
-            throw new Error("--topic and --script are mutually exclusive");
-        }
-        if (opts.pace && !["slow", "normal", "fast"].includes(opts.pace)) {
-            throw new Error(`--pace must be one of slow|normal|fast (got: ${opts.pace})`);
-        }
-        if (opts.motion && !["off", "lite", "max"].includes(opts.motion)) {
-            throw new Error(`--motion must be one of off|lite|max (got: ${opts.motion})`);
-        }
-        if (opts.layout && !["full", "blur-bg", "letterbox"].includes(opts.layout)) {
-            throw new Error(`--layout must be one of full|blur-bg|letterbox (got: ${opts.layout})`);
-        }
-        if (opts.subtitleStyle && !["plate", "stroke", "cinema"].includes(opts.subtitleStyle)) {
-            throw new Error(`--subtitle-style must be one of plate|stroke|cinema (got: ${opts.subtitleStyle})`);
-        }
-        if (opts.ttsSpeed !== undefined &&
-            (!opts.ttsModel || opts.ttsModel.startsWith("vox/"))) {
-            const m = opts.ttsModel || "default";
-            warn(`--tts-speed=${opts.ttsSpeed} will be ignored by TTS model "${m}". Speed control is only honored by Edge-TTS-backed models.`);
-        }
-        let topic = opts.topic;
-        let script = opts.script;
-        let title = opts.title;
-        if (topic?.startsWith("@"))
-            topic = await fs.readFile(topic.slice(1), "utf-8");
-        if (script?.startsWith("@"))
-            script = await fs.readFile(script.slice(1), "utf-8");
-        if (title?.startsWith("@"))
-            title = await fs.readFile(title.slice(1), "utf-8");
-        await submitAndMaybeWait("/api/v1/pipelines/standard", {
-            topic,
-            script,
-            title,
-            duration: opts.duration,
-            pace: opts.pace,
-            motion: opts.motion,
-            layout: opts.layout,
-            layout_matte_color: opts.layoutMatteColor,
-            subtitle_style: opts.subtitleStyle,
-            image_model: opts.imageModel,
-            prompt_prefix: opts.promptPrefix,
-            character_ref: opts.characterRef,
-            voice_id: opts.voiceId,
-            tts_speed: opts.ttsSpeed,
-            video_fps: opts.videoFps,
-            segment_min_chars: opts.segmentMinChars,
-            segment_max_chars: opts.segmentMaxChars,
-            preview_only: opts.previewOnly,
-        }, { wait: opts.wait, output: opts.output, pollMs: opts.pollMs, timeoutMs: opts.timeoutMs });
-    });
-}