File size: 4,645 Bytes
345855e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 | import uuid
from datetime import datetime
# -------------------------------------------------
# UTILITIES
# -------------------------------------------------
def clamp(value, min_v=0, max_v=1):
return max(min_v, min(max_v, value))
def safe_len(x):
try:
return len(x)
except Exception:
return 0
# -------------------------------------------------
# CORE HIGHLIGHT DETECTION ENGINE
# -------------------------------------------------
def detect_peak_density(words, window=12):
"""
Simple sliding window density heuristic.
Returns indices where speech intensity peaks.
"""
if not words or len(words) < window:
return []
scores = []
for i in range(len(words) - window):
chunk = words[i:i + window]
# heuristic: repetition + punctuation + trigger words
unique_ratio = len(set(chunk)) / window
trigger_words = {"why", "how", "what", "secret", "hack", "never", "stop", "crazy", "insane"}
trigger_hits = sum(1 for w in chunk if str(w).lower() in trigger_words)
score = (1 - unique_ratio) + (trigger_hits * 0.15)
scores.append((i, score))
# sort by strongest signal
scores.sort(key=lambda x: x[1], reverse=True)
# take top peaks (non-overlapping)
selected = []
used = set()
for idx, _ in scores:
if any(abs(idx - u) < window for u in used):
continue
selected.append(idx)
used.add(idx)
if len(selected) >= 8: # max clips
break
return selected
def build_segments(words, indices, window=20):
"""
Convert peak indices into structured segments.
"""
segments = []
for idx in indices:
start = max(0, idx - window // 2)
end = min(len(words), idx + window // 2)
segment_words = words[start:end]
if not segment_words:
continue
segments.append({
"id": str(uuid.uuid4()),
"start_index": start,
"end_index": end,
"words": segment_words,
"length": len(segment_words),
})
return segments
# -------------------------------------------------
# FALLBACK MODE (NO SIGNAL DETECTED)
# -------------------------------------------------
def fallback_segments(words):
"""
Ensures highlights always exist even for weak input.
"""
if not words:
return []
chunk_size = max(25, len(words) // 5)
segments = []
for i in range(0, len(words), chunk_size):
chunk = words[i:i + chunk_size]
segments.append({
"id": str(uuid.uuid4()),
"start_index": i,
"end_index": i + len(chunk),
"words": chunk,
"length": len(chunk),
})
if len(segments) >= 5:
break
return segments
# -------------------------------------------------
# MAIN ENTRYPOINT (REGISTRY COMPATIBLE)
# -------------------------------------------------
def run(context):
"""
Expected input:
{
"words": [...]
}
"""
batch_id = str(uuid.uuid4())
started_at = datetime.utcnow().isoformat()
try:
words = []
if isinstance(context, dict):
words = context.get("words", [])
else:
words = getattr(context, "words", []) or []
# -------------------------------------------------
# DETECT PEAKS
# -------------------------------------------------
peak_indices = detect_peak_density(words)
if peak_indices:
segments = build_segments(words, peak_indices)
else:
segments = fallback_segments(words)
# -------------------------------------------------
# OUTPUT CONTRACT
# -------------------------------------------------
return {
"status": "success",
"task": "highlights",
"batch_id": batch_id,
"started_at": started_at,
"completed_at": datetime.utcnow().isoformat(),
# core output for downstream clipper
"highlights": segments,
# UI-friendly summary
"summary": {
"total_words": safe_len(words),
"segments_found": len(segments),
"peak_detection": bool(peak_indices),
}
}
except Exception as e:
return {
"status": "error",
"task": "highlights",
"batch_id": batch_id,
"message": str(e),
"stage": "highlight_detection_failed",
"highlights": []
} |