Skip to content

armgit / hypit

Public Imported from GitHub
0

Clone any viral video with AI agents. Not just a script, the whole workflow: swap the face, the words, the B-roll, ship 100 variants in one command, and get your 100M views.

docs: rebaseline unreleased language as v1

kashorin896d7b432195
diff --git a/README.md b/README.md
index 6fab20d7..d6c4d77c 100644
--- a/README.md
+++ b/README.md
@@ -24,14 +24,14 @@ not enter the source language.
The repository contains a strict TypeScript compiler and an intentionally small
stdlib. The deterministic slice is working end to end:
-- frozen Script Surface with dialogue, speech and caption projections;
+- an early Script parser with dialogue, speech and caption projections;
- closed, possibly disconnected Selection sets and zero-width Moments;
- `.svc` content modules and typed `.svs` parameter classes;
- stable Plan identities, separate execution digests and a verified `svml.lock`;
- isolated `.svk` component lowering with scoped CSS and a JSON-only ABI;
- typed `.svk` ports, parameters and recursive child-content schemas;
- manifest-enforced temporal `one` / `each` / `set` consumption;
-- a Speech Spine compatibility component, media/presentation mapping, ranking, B-roll,
+- a temporary Speech Spine prototype, media/presentation mapping, ranking, B-roll,
captions, text and audio;
- frame-exact Located IR, Canvas/Timeline views and HyperFrames HTML.
@@ -42,12 +42,13 @@ falls back to a provider call. Direct existing `Image`, `Video` and `Audio` valu
work as usual. This keeps the first end-to-end reference deterministic while
preserving the eventual capability boundary.
-The executable prototype still implements frozen Script Surface v1: one globally
-monotonic Speech Spine alignment with shared adjacent Segment cuts. The architecture
-draft and Script Surface v2 draft specify the next boundary: a replaceable Basis
-Component, an importable Locator Component that outputs SemanticMap, and `2M + 2N`
-independent Segment/word endpoint identities. The repository does not claim that
-migration is implemented.
+SVML has not been publicly released, so the repository does not preserve the early
+prototype as a compatibility contract. The sole v1 target is the architecture draft
+and Script Surface v1: a replaceable Basis Component, an imported Locator Component,
+and `2M + 2N` independent Segment/word endpoint identities. The executable prototype
+still uses shared adjacent Segment cuts and a temporary `speech-spine`; it is an
+incomplete implementation scheduled for direct replacement, not an older language
+version that v1 must support.
## Commands
@@ -76,7 +77,7 @@ pnpm svml render build/index.html --out build/video.mp4
```
`--artifacts` is needed only when the reachable Plan contains capability Components.
-`fmt` currently canonicalizes only the frozen Script body and refuses any rewrite
+`fmt` currently canonicalizes only the prototype Script body and refuses any rewrite
whose reparsed semantic IR differs.
`canvas` requires no timing or provider result. `estimate` deterministically
creates provisional syllable-based alignment; `timeline` and `compile` consume
@@ -88,7 +89,7 @@ SemanticMap digests are not all part of the executable lock path.
## Reference implementation
[`examples/regen-ranking/regen-ranking.svml`](examples/regen-ranking/regen-ranking.svml)
-reconstructs a real production video containing the Speech Spine compatibility component,
+reconstructs a real production video using the temporary Speech Spine prototype,
a five-item ranking track, five B-roll items, captions, title, music and sound effects. The media files
are intentionally gitignored; their production provenance and frozen assertions
are recorded in
@@ -100,11 +101,11 @@ speech program (0.46% duration error); it remains preview evidence, not measured
alignment.
[`examples/flat-track-launch/flat-track-launch.svml`](examples/flat-track-launch/flat-track-launch.svml)
-is the readable fixture for the next architecture boundary: separate Basis and
-Locator Components, Script Surface v2, continuous A-roll Present changes, B-roll
+is the readable fixture for the v1 architecture target: separate Basis and
+Locator Components, Script Surface v1, continuous A-roll Present changes, B-roll
transition audio, disconnected caption selections, absolute z and Film as the sole
-`Track[]` consumer. It is a draft fixture, not yet accepted by the executable v1
-compiler.
+`Track[]` consumer. It is a draft fixture not yet accepted by the early compiler
+prototype.
[`examples/composite-speech-program/minimal.svml`](examples/composite-speech-program/minimal.svml)
and its adjacent `speech-program.svk` define the proposed `composite-v1` author
@@ -115,15 +116,13 @@ data consumed by one assembler, and the Component has no ambient document access
## Specification
-- [Script Surface v1](spec/script-surface-v1.md) — the frozen, human-readable
- source language for narrative, text projections, selections, and moments.
-- [Script Surface v2 draft](spec/script-surface-v2-draft.md) — the syntax-preserving
- independent Segment endpoint and `2M + 2N` migration.
+- [Script Surface v1](spec/script-surface-v1.md) — the sole draft target for the
+ human-readable Script, including independent Segment endpoints and `2M + 2N`.
- [Source Architecture Draft](spec/source-architecture-draft.md) — the current
draft for `.svk` components, `.svs` parameter sheets, `.svc` content modules,
Script-dependent generation, five public output boundaries, `composite-v1`, flat
Tracks, and a Composition root.
-- [Compiler prototype record](docs/compiler-prototype-v1.md) — implemented
+- [Compiler prototype record](docs/compiler-prototype.md) — implemented
boundaries, real-video evidence and remaining work.
- [External Runtime Host Architecture Draft](docs/runtime-host-architecture-draft.md)
— portable runtime, Artifact, Effect/Receipt, Worker, queue, provider and
diff --git a/examples/composite-speech-program/README.md b/examples/composite-speech-program/README.md
index 890b7e58..b8e0a439 100644
--- a/examples/composite-speech-program/README.md
+++ b/examples/composite-speech-program/README.md
@@ -33,10 +33,10 @@ not select a language-level family or mode. `after` names a stable Segment id, a
the following Segment comes from source order. A join cannot declare both `gap`
and `overlap`.
-An implementation may retain `speech-spine` as a compatibility facade for V1
-authors. It remains an ordinary imported Composite, never a compiler keyword or a
-required singleton. Alternative Locators use explicit low-level wiring or another
-Composite instead of extending a closed `locator="..."` enum.
+The v1 stdlib publishes `speech-program` as its chosen author surface. It remains
+an ordinary imported Composite, never a compiler keyword or a required singleton.
+Alternative Locators use explicit low-level wiring or another Composite instead
+of extending a closed `locator="..."` enum.
The bracket expression in `ports.script.segment[item.id]` is a stable id-keyed
lookup, not a positional array lookup. The `for` iterates only over finite children
diff --git a/examples/composite-speech-program/minimal.svml b/examples/composite-speech-program/minimal.svml
index f2783a8e..02a13028 100644
--- a/examples/composite-speech-program/minimal.svml
+++ b/examples/composite-speech-program/minimal.svml
@@ -1,15 +1,15 @@
<!--
Proposed composite-v1 author surface. The current compiler does not implement
- this profile or Script Surface v2 yet.
+ this profile or the complete Script Surface v1 yet.
-->
-<svml version="2">
+<svml version="1">
<import from="./speech-program.svk"/>
<import from="@svml/std/media-track.svk"/>
<import from="@svml/std/film.svk"/>
<import as="house" from="./house.svs"/>
- <script version="2">
+ <script>
<segment id="intro">
<HOST> Hello, this is SVML.
</segment>
diff --git a/examples/flat-track-launch/README.md b/examples/flat-track-launch/README.md
index 4d0265b5..9940f1ec 100644
--- a/examples/flat-track-launch/README.md
+++ b/examples/flat-track-launch/README.md
@@ -1,10 +1,10 @@
# Flat Track Launch architecture fixture
This source is the readable fixture for the unimplemented Source Architecture
-Draft. It deliberately exercises contracts that the current v1 compiler does not
+Draft. It deliberately exercises v1 contracts that the early compiler does not
yet accept:
-- Script Surface v2 independent Segment endpoints (`2M + 2N`);
+- Script Surface v1 independent Segment endpoints (`2M + 2N`);
- a Basis Component that overlaps adjacent Segment media;
- an imported Locator Component that produces the selected ExactSemanticMap;
- one ProgramBasis and one absolute ProgramSpace;
diff --git a/examples/flat-track-launch/flat-track-launch.svml b/examples/flat-track-launch/flat-track-launch.svml
index 5aa021ad..22bccfbb 100644
--- a/examples/flat-track-launch/flat-track-launch.svml
+++ b/examples/flat-track-launch/flat-track-launch.svml
@@ -1,10 +1,10 @@
<!--
Architecture fixture for Source Architecture Draft.
- It is intentionally not accepted by the current Script Surface v1 prototype.
+ It is intentionally not accepted by the current early compiler prototype.
-->
-<svml version="2">
+<svml version="1">
- <import from="@svml/std/crossfade-speech.svk"/>
+ <import from="@svml/std/speech-assemble.svk"/>
<import from="@svml/std/speech-locator.svk"/>
<import from="@svml/std/media-track.svk"/>
<import from="@svml/std/broll-track.svk"/>
@@ -16,7 +16,7 @@
<import as="house" from="./launch.svs"/>
- <script version="2">
+ <script>
@whole
@prelude
@@ -51,7 +51,7 @@
source maps, program audio and a program-bound visual facet. The Locator
Component separately binds this Script's 2M + 2N identities to the basis.
-->
- <crossfade-speech id="voice" overlap="500ms">
+ <speech-assemble id="voice" defaultJoin="cut">
<segment
id="prelude"
script={script.segment.prelude}
@@ -70,7 +70,21 @@
source={host-long}
sourceWindow="30s .. 50s"
/>
- </crossfade-speech>
+
+ <join
+ after="prelude"
+ overlap="500ms"
+ audio="crossfade"
+ visual="dissolve"
+ />
+
+ <join
+ after="intro"
+ overlap="500ms"
+ audio="crossfade"
+ visual="dissolve"
+ />
+ </speech-assemble>
<!--
SemanticMap is the fifth SVK output role. A simple speech-locator may expose
diff --git a/spec/script-surface-v1.md b/spec/script-surface-v1.md
index 01179c9c..5f579bf7 100644
--- a/spec/script-surface-v1.md
+++ b/spec/script-surface-v1.md
@@ -1,16 +1,16 @@
# SVML Script Surface v1
-> **Surface frozen; implementation pending.**
+> **Draft; not frozen.**
>
-> 本规范冻结 SVML 稿子区的 v1 源语言。Script 是后端无关的语义源,不声明
-> 某个画布、节点注册表或渲染器已经实现这些规则。实现只有在通过本文的
+> SVML 尚未公开发布,不存在需要兼容的旧语言版本。本文直接定义首个公开目标
+> Script Surface v1;仓库中的早期原型不是规范合同。实现只有在通过本文的
> parser、projection、source-map 和 temporal golden fixtures 后,才可声称
-> 支持 Script Surface v1。
+> 支持 SVML v1。
-SVML 是语义视频源语言,不是像素渲染格式。本文只冻结它最高频、最需要保持
+SVML 是语义视频源语言,不是像素渲染格式。本文只定义它最高频、最需要保持
可读性的稿子区 `<script>`;画布、节点、样式、生成参数和其他 Program 区域的
-外层文档结构另行设计。冻结之后,新增语义应首先组合本文已有构造;确需改变
-正文语法时必须发布新的 Script Surface 版本,不能静默扩写 v1。
+外层文档结构另行设计。v1 发布后,新增语义应首先组合本文已有构造;确需改变
+正文语法时再发布新版本,不能因为早期原型行为而削弱当前 v1。
## 一页合同
@@ -18,7 +18,7 @@ SVML 是语义视频源语言,不是像素渲染格式。本文只冻结它最
| 构造 | 只负责 | 不负责 |
|---|---|---|
-| `Segment` | 有序的媒体/口播块和结构切点 | 自动 Selection、时长、静音 |
+| `Segment` | 有序的媒体/口播块和独立结构端点 | 自动 Selection、时长、静音 |
| `Role Cue` | `dialogue` 导出时的“谁说了什么”文本前缀 | speaker 实体、音色、字段路由 |
| `Dual Text` | 显示文字与实际读音不同 | TTS 厂商参数、样式、动作 |
| `Selection` | 一个或多个显式闭合的语义时间区间 | 消费者声明、绝对秒数 |
@@ -80,7 +80,7 @@ Meet Hypit.
```
`silence` 只是 Selection 的名字,不会自动让音频静音。它完整选择了空
-Segment 的两个结构切点;真正的静音或素材行为由外部 Program 决定。
+Segment 的两个独立结构端点;真正的静音或素材行为由外部 Program 决定。
`pop` 是一个默认吸右的 Moment,解析到 `laughed` 的起音;它同样不进入
任何文本投影。
@@ -107,8 +107,9 @@ Segment 的两个结构切点;真正的静音或素材行为由外部 Program
结果或 `<script>` 外的 Program。
- Segment 不自动声明同名 Selection。选择一个、三个或任意多个 Segment,
都必须显式写闭合 Selection。
-- 相邻 Segment 共享一个切点。因此 `m` 个 Segment 始终贡献 `m + 1` 个
- 具有稳定身份的结构切点。
+- 每个 Segment 独立拥有 `start` 与 `end` 两个稳定 identity;相邻 Segment
+ 不共享结构端点 identity。两个 identity 可以在硬切时落到同一 ProgramPoint,
+ 也可以因 overlap 或 gap 落到不同点,但不能合并身份。
例如选择三个连续 Segment:
@@ -228,10 +229,10 @@ Selection 是区间声明,不是供下游自行拼接的两个公开锚点:
| 端点写法 | 吸附方向 | 典型含义 |
|---|---|---|
-| `@id` | 右 | 从右侧词首或右侧结构切点开始 |
-| `~@id` | 左 | 向外扩到左侧词尾或左侧结构切点 |
-| `@/id` | 左 | 在左侧词尾或左侧结构切点结束 |
-| `@/id~` | 右 | 向外扩到右侧词首或右侧结构切点 |
+| `@id` | 右 | 从右侧词首或右侧结构端点开始 |
+| `~@id` | 左 | 向外扩到左侧词尾或左侧结构端点 |
+| `@/id` | 左 | 在左侧词尾或左侧结构端点结束 |
+| `@/id~` | 右 | 向外扩到右侧词首或右侧结构端点 |
例子:
@@ -296,24 +297,71 @@ Slot 绑定、Dual Text speech 投影和 NFC 归一化之后,v1 使用规范
边界,可以在新的 timing profile / Script Surface 版本中复用相同的
Selection/Moment 类型;v1 不应为尚不存在的精度引入新正文符号。
-若 speech 中有 `n` 个 token、Script 中有 `m` 个 Segment,Temporal Anchor
-Map 恰有:
+若 Script 有 `N` 个 Segment,第 `k` 个 Segment 有 `mₖ` 个 speech token,且
+`M = Σmₖ`,Semantic Anchor Index 恰有:
```text
-2n + m + 1
+Σ(2mₖ + 2) = 2M + 2N
```
-个稳定身份:每个 token 的词首/词尾各一个,加上 `m + 1` 个唯一结构切点。
-这些身份即使最终解析到同一帧也不能合并。
+个稳定身份。每个 Segment 内的局部顺序是:
-同一份 SelectionSet 可先在 Estimate Anchor Map 上得到预览区间,再在
-Evidence Anchor Map 上得到精确区间。Script 本身不含秒数、帧号或采样点。
+```text
+segment[k].start
+token[k,1].start
+token[k,1].end
+...
+token[k,mₖ].start
+token[k,mₖ].end
+segment[k].end
+```
+
+Program 起点和终点属于 ProgramBasis,不额外进入 Semantic Anchor Index。空
+Segment 仍有独立 start/end,即使二者最终重合;任意不同 identity 即使最终
+落到同一帧也不能合并。
+
+Locator 必须提交覆盖全部 `2M + 2N` identity 的总映射:
+
+```text
+SemanticAnchorIdentity → ProgramPoint(basisDigest)
+```
+
+完整性与精度正交:每个点都必须存在,同时可以标记为 `estimated`、`derived`
+或 `measured`。消费者不得补点、移动点或从相邻 occurrence 借点。每个 Segment
+内部必须非降序:
+
+```text
+segment.start ≤ token₁.start ≤ token₁.end ≤ ... ≤ segment.end
+```
+
+按源码相邻的 Segment `A`、`B` 还必须保持:
+
+```text
+A.start ≤ B.start
+A.end ≤ B.end
+```
+
+这允许硬切、重叠与留白,同时禁止一个 Locator 把后写的整个 Segment 静默排到
+前写 Segment 之前:
+
+```text
+hard cut A.end == B.start
+overlap B.start < A.end
+gap B.start > A.end
+```
+
+若未来需要真正重排或并行 speech,应发布显式的非线性叙事模型;v1 不让普通
+Range 在不同 Locator 下反向或消失。
+
+同一份 SelectionSet 可先在 `EstimatedSemanticMap` 上得到预览区间,再在
+`ExactSemanticMap` 上得到成片区间。两者使用完全相同的 identity、不同的精度
+类型。Script 本身不含秒数、帧号或采样点。
Script Surface 不绑定帧率、采样率或渲染器。后端一旦选择物理时钟,必须只
量化一次并让所有消费者复用同一整数边界;后端时钟变化不改变本语言表面。
-对齐、结构切点或帧量化可能使某个 occurrence 最终成为零长或反向区间。
-消费者先应用自己的 projection,再只丢弃该非正区间 occurrence;不得移动
-Selection 端点、借用相邻 occurrence 或替作者补范围。
+对齐、结构端点、projection 或帧量化可能使某个 occurrence 最终成为零长或
+反向区间。实现必须产生明确诊断,不能静默丢弃、移动 Selection 端点、借用
+相邻 occurrence 或替作者补范围。
## 5. Moment
@@ -328,8 +376,8 @@ Moment 是完整的点声明,不是省略了 close 的 Selection:
| 写法 | 吸附方向 | 典型含义 |
|---|---|---|
-| `@id!` | 右 | 右侧词首或右侧结构切点 |
-| `~@id!` | 左 | 左侧词尾或左侧结构切点 |
+| `@id!` | 右 | 右侧词首或右侧结构端点 |
+| `~@id!` | 左 | 左侧词尾或左侧结构端点 |
默认右吸覆盖“与某个词一起开始”的高频情况:
@@ -343,8 +391,9 @@ I just @pop! laughed my ass out.
I laughed ~@pop! and left.
```
-`pop` 解析到 `laughed` 的收音。位于两个 Segment 之间时,左右候选是对应
-方向上的共享结构切点。
+`pop` 解析到 `laughed` 的收音。位于两个 Segment 之间时,左候选是前一
+Segment 的 `end`,右候选是后一 Segment 的 `start`;两者可以同点,也可以因
+overlap 或 gap 落在不同点。
规范规则:
@@ -359,10 +408,10 @@ I laughed ~@pop! and left.
temporal syntax,不能退化为正文。
- Moment 与 Selection marker 可出现在相同位置,包括 Segment 之间和
Dual Text 的 speech 侧,但均不得切入 v1 speech token。
-- Moment 只选择 Temporal Anchor Map 中已有的一个候选点,不新增 anchor
- identity;因此 `2n + m + 1` 的精确计数不因 Moment 数量改变。
-- MomentSet 与 SelectionSet 复用同一份 Estimate/Evidence Anchor Map 和
- 一次性帧量化;区别只在最终载体是 `Point[]` 而不是 `Range[]`。
+- Moment 只选择 Semantic Anchor Index 中已有的一个候选点,不新增 anchor
+ identity;因此 `2M + 2N` 的精确计数不因 Moment 数量改变。
+- MomentSet 与 SelectionSet 复用同一份 Estimated/Exact SemanticMap 和一次性
+ 帧量化;区别只在最终载体是 `Point[]` 而不是 `Range[]`。
消费端保持两个互不相混的类型:
@@ -463,7 +512,7 @@ temporal marker、Role Cue 或结构标签内部。注释可出现在 atom 之
解析结果至少保留:
-- Segment 的顺序、id 和结构边界;
+- Segment 的顺序、id 和独立 start/end identity;
- 每个 Segment 的有序 spoken turn,以及各 turn 的可选 Role Cue;
- plain / Dual Text / Slot atom 及其源码范围;
- caption atom 到一个或多个 speech token 的显式映射;
@@ -533,7 +582,7 @@ Caption 的多样式、region、cue segmentation、annotation、mute 和 layout
一个实现只有同时满足以下证据,才能声明支持 Script Surface v1:
-1. 冻结本文全部正例、反例、投影和 source-map golden fixtures;
+1. 本文全部正例、反例、投影和 source-map golden fixtures;
2. parser、formatter、Narrative IR 与三投影不依赖 provider 或渲染器;
3. SelectionSet 支持闭合、交叉、非连通 occurrence 与左右 affinity;
4. MomentSet 支持 `@id!` / `~@id!`、单/多 occurrence 与 `at` 类型闸;
@@ -541,3 +590,6 @@ Caption 的多样式、region、cue segmentation、annotation、mute 和 layout
6. Slot 采用 parse-first、literal-only binding,不能注入语法;
7. 未知版本、未知保留语法、混合 temporal type 与 partial dual atom 全部
fail closed。
+8. `2M + 2N` identity 的规范序列化、digest、完整 Map 与相邻 Segment 双单调
+ 约束都有 golden fixtures;硬切、overlap、gap、空 Segment 和端点重合均被
+ 覆盖。