#!/usr/bin/env bash
# Chunked TTS for long text - splits into sentences, generates each, concatenates
# Usage: voice-long.sh "text" [voice] [format]

TTS_URL="https://hf4uwho-pocket-tts.hf.space/tts"
TEXT="${1:?Usage: voice-long.sh 'text' [voice] [format]}"
VOICE="${2:-af_alloy}"
FORMAT="${3:-ogg}"
OUTDIR="/tmp/tts"
mkdir -p "$OUTDIR"

HASH=$(echo -n "$TEXT$VOICE$(date +%s%N)" | md5sum | cut -c1-12)
OUTFILE="$OUTDIR/voice_${HASH}.${FORMAT}"
CHUNKDIR="$OUTDIR/chunks_${HASH}"
mkdir -p "$CHUNKDIR"

# Split text into sentences (rough but effective)
CHUNKS=$(python3 -c "
import re
text = '''$TEXT'''
# Split on sentence boundaries, group into chunks of ~200 chars
sentences = re.split(r'(?<=[.!?])\s+', text)
chunks = []
current = ''
for s in sentences:
    if len(current) + len(s) > 300 and current:
        chunks.append(current.strip())
        current = s
    else:
        current = (current + ' ' + s).strip()
if current.strip():
    chunks.append(current.strip())
for i, c in enumerate(chunks):
    print(f'CHUNK_{i}:{c}')
")

# Generate each chunk
INDEX=0
CHUNK_FILES=()
while IFS= read -r line; do
    if [[ "$line" == CHUNK_* ]]; then
        CHUNK_TEXT="${line#CHUNK_[0-9]*:}"
        # Strip the index prefix
        CHUNK_TEXT="${line#*:}"
        ENCODED=$(python3 -c "import urllib.parse; print(urllib.parse.quote('''$CHUNK_TEXT'''))")
        CHUNKFILE="$CHUNKDIR/chunk_${INDEX}.${FORMAT}"
        
        HTTP_CODE=$(curl -s -w "%{http_code}" -o "$CHUNKFILE" "${TTS_URL}?text=${ENCODED}&voice=${VOICE}&format=${FORMAT}" --max-time 60 2>/dev/null)
        
        if [ "$HTTP_CODE" != "200" ]; then
            echo "ERROR: Chunk $INDEX failed with HTTP $HTTP_CODE" >&2
            rm -rf "$CHUNKDIR"
            exit 1
        fi
        
        # Verify audio
        MIME=$(file -b --mime-type "$CHUNKFILE" 2>/dev/null)
        if [[ "$MIME" == audio/* ]] || [[ "$MIME" == application/ogg ]]; then
            CHUNK_FILES+=("$CHUNKFILE")
            INDEX=$((INDEX + 1))
        else
            echo "WARNING: Chunk $INDEX not audio ($MIME), skipping" >&2
        fi
    fi
done <<< "$CHUNKS"

if [ ${#CHUNK_FILES[@]} -eq 0 ]; then
    echo "ERROR: No chunks generated" >&2
    rm -rf "$CHUNKDIR"
    exit 1
fi

# If only one chunk, just use it
if [ ${#CHUNK_FILES[@]} -eq 1 ]; then
    cp "${CHUNK_FILES[0]}" "$OUTFILE"
else
    # Concatenate Ogg files using sox or ffmpeg, fallback to simple cat
    if command -v sox &>/dev/null; then
        sox "${CHUNK_FILES[@]}" "$OUTFILE" 2>/dev/null
    elif command -v ffmpeg &>/dev/null; then
        # Create concat list
        CONCFILE="$CHUNKDIR/concat.txt"
        > "$CONCFILE"
        for f in "${CHUNK_FILES[@]}"; do
            echo "file '$f'" >> "$CONCFILE"
        done
        ffmpeg -y -f concat -safe 0 -i "$CONCFILE" -c copy "$OUTFILE" 2>/dev/null
    else
        # Simple cat (works for raw Ogg but may have glitches)
        cat "${CHUNK_FILES[@]}" > "$OUTFILE"
    fi
fi

rm -rf "$CHUNKDIR"

if [ -f "$OUTFILE" ] && [ -s "$OUTFILE" ]; then
    echo "$OUTFILE"
else
    echo "ERROR: Final file empty or missing" >&2
    exit 1
fi