zenstory-ai/video-recap-skills / skills/video-recap
video-recap
Verified把一段或多集视频做成带画框字幕的中文解说成片或纯原声高光短片。用户提供 .mp4 / .mov / .mkv / .webm,要求做影视解说、短剧二创、加旁白配音、跨集混剪或导出剪映草稿时使用;串联完成看片识台词、定主线与留原声、按台词停顿切镜头、写词检查、TTS 配音与混音加字幕。触发词:视频解说、影视解说、视频旁白、生成解说、视频 recap、video recap、voiceover、narration、recap。
scripts/packaging_fonts.py
View on GitHub"""Cross-platform CJK and display font resolution plus @2x Pillow text helpers."""
from __future__ import annotations
import functools
import glob
import shutil
import subprocess
from pathlib import Path
try:
from PIL import Image, ImageDraw, ImageFilter, ImageFont
except ImportError: # pragma: no cover - isolated python -I without Pillow
Image = ImageDraw = ImageFilter = ImageFont = None # type: ignore[assignment]
_FONT_SEARCH_GLOBS = (
"/System/Library/AssetsV2/com_apple_MobileAsset_Font7/*/AssetData/*",
"/System/Library/Fonts/*",
"/System/Library/Fonts/Supplemental/*",
"/Library/Fonts/*",
)
_LINUX_CANDIDATES = {
"sans_heavy": (
"/usr/share/fonts/google-noto-cjk/NotoSansCJKsc-Black.otf",
"/usr/share/fonts/google-noto-cjk/NotoSansCJKsc-Bold.otf",
"/usr/share/fonts/opentype/noto/NotoSansCJK-Black.ttc",
"/usr/share/fonts/opentype/noto/NotoSansCJK-Bold.ttc",
"/usr/share/fonts/truetype/noto/NotoSansCJK-Bold.ttc",
"/usr/share/fonts/truetype/wqy/wqy-microhei.ttc",
"/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf",
),
"sans_medium": (
"/usr/share/fonts/google-noto-cjk/NotoSansCJKsc-Medium.otf",
"/usr/share/fonts/google-noto-cjk/NotoSansCJKsc-Regular.otf",
"/usr/share/fonts/opentype/noto/NotoSansCJK-Medium.ttc",
"/usr/share/fonts/opentype/noto/NotoSansCJK-Regular.ttc",
"/usr/share/fonts/truetype/wqy/wqy-microhei.ttc",
"/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf",
),
"calligraphy": (
"/usr/share/fonts/google-noto-cjk/NotoSerifCJKsc-Bold.otf",
"/usr/share/fonts/opentype/noto/NotoSerifCJK-Bold.ttc",
"/usr/share/fonts/google-noto-cjk/NotoSansCJKsc-Bold.otf",
"/usr/share/fonts/opentype/noto/NotoSansCJK-Bold.ttc",
),
"hand": (
"/usr/share/fonts/google-noto-cjk/NotoSerifCJKsc-SemiBold.otf",
"/usr/share/fonts/opentype/noto/NotoSerifCJK-SemiBold.ttc",
"/usr/share/fonts/google-noto-cjk/NotoSansCJKsc-Medium.otf",
"/usr/share/fonts/opentype/noto/NotoSansCJK-Medium.ttc",
),
"numeric": (
"/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf",
),
}
_WINDOWS_CANDIDATES = {
"sans_heavy": ("C:/Windows/Fonts/msyhbd.ttc", "C:/Windows/Fonts/simhei.ttf"),
"sans_medium": ("C:/Windows/Fonts/msyh.ttc", "C:/Windows/Fonts/simsun.ttc"),
"calligraphy": ("C:/Windows/Fonts/STXINGKA.TTF", "C:/Windows/Fonts/simkai.ttf"),
"hand": ("C:/Windows/Fonts/simkai.ttf", "C:/Windows/Fonts/msyhbd.ttc"),
"numeric": ("C:/Windows/Fonts/bahnschrift.ttf", "C:/Windows/Fonts/arialbd.ttf"),
}
_MACOS_ROLE_PREFS = {
"sans_heavy": ("Lantinghei.ttc", "PingFang.ttc", "Hiragino Sans GB.ttc", "STHeiti Medium.ttc"),
"sans_medium": ("PingFang.ttc", "Hiragino Sans GB.ttc", "STHeiti Medium.ttc"),
"calligraphy": ("Xingkai.ttc", "WeibeiSC-Bold.otf", "Kaiti.ttc", "Songti.ttc"),
"hand": ("Hannotate.ttc", "Kaiti.ttc", "PingFang.ttc"),
"numeric": ("DIN Alternate Bold.ttf", "Avenir Next Condensed.ttc", "Avenir Next.ttc"),
}
_WEIGHT_PREFS = {
"sans_heavy": ("heavy", "black", "semibold", "demibold", "bold", "w6", "medium"),
"sans_medium": ("medium", "regular", "w3", "demibold"),
"calligraphy": ("bold", "black", "semibold", "regular"),
"hand": ("bold", "regular", "medium"),
"numeric": ("bold", "heavy", "demi bold", "semibold"),
}
@functools.lru_cache(maxsize=1)
def _macos_font_index() -> dict[str, str]:
found: dict[str, str] = {}
home = Path.home()
patterns = (
*_FONT_SEARCH_GLOBS,
str(home / ".fonts/*"),
str(home / ".local/share/fonts/*"),
)
for pattern in patterns:
for raw in sorted(glob.glob(pattern)):
name = Path(raw).name
found.setdefault(name, raw)
return found
def _best_ttc_index(font_path: str, role: str) -> int:
if ImageFont is None or not font_path.lower().endswith((".ttc", ".otc")):
return 0
prefs = _WEIGHT_PREFS.get(role, ("bold", "regular"))
best_idx, best_score = 0, -100
for idx in range(16):
try:
face = ImageFont.truetype(font_path, 32, index=idx)
family, style = (s.lower() for s in face.getname())
except Exception: # noqa: BLE001
break
if family.startswith("."):
continue
score = 0
if " sc" in family or family.endswith("sc") or "gb" in family:
score += 40
elif " tc" in family or " hk" in family or " mo" in family:
score -= 15
if "italic" in style or "oblique" in style:
score -= 25
for rank, token in enumerate(prefs):
if token in style:
score += 30 - rank * 3
break
if score > best_score:
best_score = score
best_idx = idx
return best_idx
@functools.lru_cache(maxsize=32)
def resolve_font_spec(role: str = "sans_heavy", explicit_path: str = "") -> tuple[str | None, int]:
"""Return (font_path, face_index) for a typography role, or (None, 0) for PIL fallback."""
if explicit_path:
p = Path(explicit_path).expanduser()
if p.is_file():
return str(p.resolve()), _best_ttc_index(str(p.resolve()), role)
mac_map = _macos_font_index()
for fname in _MACOS_ROLE_PREFS.get(role, _MACOS_ROLE_PREFS["sans_heavy"]):
if fname in mac_map:
path = mac_map[fname]
return path, _best_ttc_index(path, role)
for cand in _LINUX_CANDIDATES.get(role, ()) + _WINDOWS_CANDIDATES.get(role, ()):
if Path(cand).is_file():
return cand, _best_ttc_index(cand, role)
if shutil.which("fc-match"):
query = {
"sans_heavy": "sans-serif:lang=zh-cn:weight=bold",
"sans_medium": "sans-serif:lang=zh-cn:weight=medium",
"calligraphy": "serif:lang=zh-cn:weight=bold",
"hand": "sans-serif:lang=zh-cn",
"numeric": "sans-serif:weight=bold",
}.get(role, "sans-serif:lang=zh-cn:weight=bold")
res = subprocess.run(
["fc-match", "-f", "%{file}", query],
capture_output=True, text=True, timeout=5, check=False,
)
matched = (res.stdout or "").strip()
if matched and Path(matched).is_file():
return matched, _best_ttc_index(matched, role)
if role != "sans_heavy":
return resolve_font_spec("sans_heavy", "")
return None, 0
@functools.lru_cache(maxsize=32)
def resolve_font_family(role: str = "sans_heavy", explicit_path: str = "") -> tuple[str | None, str | None]:
"""Return (family_name, font_path) suitable for ASS Fontname + libass fontsdir."""
path, index = resolve_font_spec(role, explicit_path)
if not path:
return None, None
if ImageFont is not None:
try:
face = ImageFont.truetype(path, 32, index=index)
family = str(face.getname()[0] or "").replace(",", " ").strip()
if family and not family.startswith("."):
return family, path
except Exception: # noqa: BLE001
pass
name = Path(path).name
for token, family in (
("NotoSansCJK", "Noto Sans CJK SC"),
("NotoSerifCJK", "Noto Serif CJK SC"),
("wqy-microhei", "WenQuanYi Micro Hei"),
("Lantinghei", "Lantinghei SC"),
("PingFang", "PingFang SC"),
("Hiragino", "Hiragino Sans GB"),
("STHeiti", "Heiti SC"),
("DejaVuSans", "DejaVu Sans"),
("msyh", "Microsoft YaHei"),
("simhei", "SimHei"),
):
if token in name:
return family, path
return None, path
@functools.lru_cache(maxsize=128)
def load_font(role: str, size_px: int, explicit_path: str = ""):
"""Load a Pillow font at exact pixel height."""
if ImageFont is None:
raise RuntimeError("生成文字贴片需要安装 Pillow (pip install Pillow)")
size_px = max(8, int(round(size_px)))
path, index = resolve_font_spec(role, explicit_path)
if path:
try:
return ImageFont.truetype(path, size_px, index=index)
except Exception: # noqa: BLE001
pass
try:
return ImageFont.load_default(size=size_px)
except TypeError:
return ImageFont.load_default()
def parse_rgba(color, default_alpha: int = 255) -> tuple[int, int, int, int]:
"""Parse '#RRGGBB', '#RRGGBBAA', or (r,g,b[,a]) into an RGBA tuple."""
if isinstance(color, (tuple, list)):
if len(color) == 4:
return tuple(int(c) for c in color) # type: ignore[return-value]
if len(color) == 3:
return (int(color[0]), int(color[1]), int(color[2]), int(default_alpha))
raw = str(color or "#FFFFFF").strip().lstrip("#")
if len(raw) == 6:
return (int(raw[0:2], 16), int(raw[2:4], 16), int(raw[4:6], 16), int(default_alpha))
if len(raw) == 8:
return (int(raw[0:2], 16), int(raw[2:4], 16), int(raw[4:6], 16), int(raw[6:8], 16))
raise ValueError(f"无法解析颜色值: {color!r}")
def split_highlight_spans(text: str, highlights: list[str] | tuple[str, ...] = ()) -> list[tuple[str, bool]]:
"""Split text into [(substring, is_highlighted)] for up to 2 non-overlapping phrases."""
clean_hl = [str(h).strip() for h in (highlights or ()) if str(h).strip()]
if not text or not clean_hl:
return [(text, False)]
chosen: list[tuple[int, int]] = []
for phrase in clean_hl:
if len(chosen) >= 2:
break
pos = text.find(phrase)
while pos != -1:
end = pos + len(phrase)
if all(end <= s or pos >= e for s, e in chosen):
chosen.append((pos, end))
break
pos = text.find(phrase, pos + 1)
if not chosen:
return [(text, False)]
chosen.sort()
spans: list[tuple[str, bool]] = []
cursor = 0
for s, e in chosen:
if s > cursor:
spans.append((text[cursor:s], False))
spans.append((text[s:e], True))
cursor = e
if cursor < len(text):
spans.append((text[cursor:], False))
return spans
def measure_text(draw: ImageDraw.ImageDraw, text: str, font, stroke_width: int = 0) -> tuple[int, int]:
"""Return (width, height) of single-line text."""
if not text:
return 0, 0
bbox = draw.textbbox((0, 0), text, font=font, stroke_width=stroke_width)
return max(0, bbox[2] - bbox[0]), max(0, bbox[3] - bbox[1])
def draw_styled_line(
base_img: Image.Image,
x: int,
y: int,
text: str,
font,
fill: tuple[int, int, int, int],
*,
highlights: list[str] | tuple[str, ...] = (),
highlight_fill: tuple[int, int, int, int] | None = None,
stroke_width: int = 0,
stroke_fill: tuple[int, int, int, int] = (0, 0, 0, 220),
shadow_blur: int = 0,
shadow_offset: tuple[int, int] = (0, 4),
shadow_color: tuple[int, int, int, int] = (0, 0, 0, 165),
) -> int:
"""Draw a crisp text line with optional optical soft shadow and inline highlight spans."""
spans = split_highlight_spans(text, highlights)
hl_fill = highlight_fill or fill
if shadow_blur > 0 or any(shadow_offset):
shadow_layer = Image.new("RGBA", base_img.size, (0, 0, 0, 0))
sdraw = ImageDraw.Draw(shadow_layer)
sx = x + shadow_offset[0]
sy = y + shadow_offset[1]
for chunk, _is_hl in spans:
sdraw.text(
(sx, sy), chunk, font=font, fill=shadow_color,
stroke_width=stroke_width, stroke_fill=shadow_color,
)
sx += int(round(sdraw.textlength(chunk, font=font)))
if shadow_blur > 0:
shadow_layer = shadow_layer.filter(ImageFilter.GaussianBlur(radius=shadow_blur))
base_img.alpha_composite(shadow_layer)
draw = ImageDraw.Draw(base_img)
cx = x
for chunk, is_hl in spans:
draw.text(
(cx, y),
chunk,
font=font,
fill=hl_fill if is_hl else fill,
stroke_width=stroke_width,
stroke_fill=stroke_fill if stroke_width > 0 else None,
)
cx += int(round(draw.textlength(chunk, font=font)))
return cx - x
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |