|
| 1 | +""" |
| 2 | +Basic compression library for CFGPP configurations |
| 3 | +Phase 1: Standalone compression that doesn't integrate with parsing yet |
| 4 | +""" |
| 5 | + |
| 6 | +import gzip |
| 7 | +import logging |
| 8 | +import time |
| 9 | +from typing import Dict, Optional, Union |
| 10 | + |
| 11 | +from .features import FeatureFlags |
| 12 | + |
| 13 | + |
| 14 | +class CFGPPCompressor: |
| 15 | + """ |
| 16 | + Basic compression for configuration files |
| 17 | +
|
| 18 | + Phase 1 Implementation: |
| 19 | + - Compress/decompress content but don't integrate with parser yet |
| 20 | + - Support multiple algorithms for different use cases |
| 21 | + - All functionality disabled by default via feature flags |
| 22 | + """ |
| 23 | + |
| 24 | + def __init__(self): |
| 25 | + self.enabled = FeatureFlags.is_enabled("COMPRESSION") |
| 26 | + self.logger = logging.getLogger(__name__) |
| 27 | + |
| 28 | + # Algorithm preferences based on use case |
| 29 | + self.algorithm_map = { |
| 30 | + "ai-communication": "gzip", # Fast and widely supported |
| 31 | + "storage": "gzip", # Good compression ratio |
| 32 | + "network": "gzip", # Standard network compression |
| 33 | + "iot": "gzip", # Broad compatibility |
| 34 | + "default": "gzip", # Safe fallback |
| 35 | + } |
| 36 | + |
| 37 | + def compress( |
| 38 | + self, content: str, algorithm: str = "gzip", target: str = "default" |
| 39 | + ) -> bytes: |
| 40 | + """ |
| 41 | + Compress configuration content |
| 42 | +
|
| 43 | + Args: |
| 44 | + content: Configuration content to compress |
| 45 | + algorithm: Compression algorithm (gzip, lz4, zstd) |
| 46 | + target: Target use case (ai-communication, storage, network, iot) |
| 47 | +
|
| 48 | + Returns: |
| 49 | + Compressed bytes or original content encoded if disabled |
| 50 | + """ |
| 51 | + if not self.enabled: |
| 52 | + return content.encode("utf-8") # Pass-through when disabled |
| 53 | + |
| 54 | + # Auto-select algorithm based on target if not specified explicitly |
| 55 | + if algorithm == "auto": |
| 56 | + algorithm = self.algorithm_map.get(target, "gzip") |
| 57 | + |
| 58 | + try: |
| 59 | + content_bytes = content.encode("utf-8") |
| 60 | + |
| 61 | + if algorithm == "gzip": |
| 62 | + return gzip.compress(content_bytes, compresslevel=6) |
| 63 | + elif algorithm == "lz4": |
| 64 | + # Try to import lz4, fallback to gzip if not available |
| 65 | + try: |
| 66 | + import lz4.frame |
| 67 | + |
| 68 | + return lz4.frame.compress(content_bytes) |
| 69 | + except ImportError: |
| 70 | + self.logger.warning("LZ4 not available, falling back to gzip") |
| 71 | + return gzip.compress(content_bytes, compresslevel=1) |
| 72 | + elif algorithm == "zstd": |
| 73 | + # Try to import zstd, fallback to gzip if not available |
| 74 | + try: |
| 75 | + import zstandard as zstd |
| 76 | + |
| 77 | + compressor = zstd.ZstdCompressor(level=3) |
| 78 | + return compressor.compress(content_bytes) |
| 79 | + except ImportError: |
| 80 | + self.logger.warning("ZSTD not available, falling back to gzip") |
| 81 | + return gzip.compress(content_bytes, compresslevel=6) |
| 82 | + else: |
| 83 | + raise ValueError(f"Unsupported compression algorithm: {algorithm}") |
| 84 | + |
| 85 | + except Exception as e: |
| 86 | + self.logger.error(f"Compression failed: {e}") |
| 87 | + return content.encode("utf-8") # Return uncompressed on error |
| 88 | + |
| 89 | + def decompress(self, compressed_data: bytes, algorithm: str = "gzip") -> str: |
| 90 | + """ |
| 91 | + Decompress configuration content |
| 92 | +
|
| 93 | + Args: |
| 94 | + compressed_data: Compressed bytes to decompress |
| 95 | + algorithm: Algorithm used for compression |
| 96 | +
|
| 97 | + Returns: |
| 98 | + Decompressed content or original data decoded if disabled |
| 99 | + """ |
| 100 | + if not self.enabled: |
| 101 | + try: |
| 102 | + return compressed_data.decode("utf-8") |
| 103 | + except UnicodeDecodeError: |
| 104 | + # If it's actually compressed but feature is disabled, try gzip anyway |
| 105 | + try: |
| 106 | + return gzip.decompress(compressed_data).decode("utf-8") |
| 107 | + except: |
| 108 | + return "DECOMPRESSION_FAILED" |
| 109 | + |
| 110 | + try: |
| 111 | + if algorithm == "gzip": |
| 112 | + return gzip.decompress(compressed_data).decode("utf-8") |
| 113 | + elif algorithm == "lz4": |
| 114 | + try: |
| 115 | + import lz4.frame |
| 116 | + |
| 117 | + return lz4.frame.decompress(compressed_data).decode("utf-8") |
| 118 | + except ImportError: |
| 119 | + # Try gzip as fallback |
| 120 | + return gzip.decompress(compressed_data).decode("utf-8") |
| 121 | + elif algorithm == "zstd": |
| 122 | + try: |
| 123 | + import zstandard as zstd |
| 124 | + |
| 125 | + decompressor = zstd.ZstdDecompressor() |
| 126 | + return decompressor.decompress(compressed_data).decode("utf-8") |
| 127 | + except ImportError: |
| 128 | + # Try gzip as fallback |
| 129 | + return gzip.decompress(compressed_data).decode("utf-8") |
| 130 | + else: |
| 131 | + raise ValueError(f"Unsupported compression algorithm: {algorithm}") |
| 132 | + |
| 133 | + except Exception as e: |
| 134 | + self.logger.error(f"Decompression failed: {e}") |
| 135 | + return "DECOMPRESSION_ERROR" |
| 136 | + |
| 137 | + def compress_with_metadata(self, content: str, target: str = "default") -> Dict: |
| 138 | + """ |
| 139 | + Compress content and return with metadata for AI systems |
| 140 | +
|
| 141 | + Args: |
| 142 | + content: Configuration content to compress |
| 143 | + target: Target use case for algorithm selection |
| 144 | +
|
| 145 | + Returns: |
| 146 | + Dictionary with compressed data and metadata |
| 147 | + """ |
| 148 | + if not self.enabled: |
| 149 | + return { |
| 150 | + "compressed": False, |
| 151 | + "algorithm": "none", |
| 152 | + "original_size": len(content), |
| 153 | + "compressed_size": len(content), |
| 154 | + "compression_ratio": 1.0, |
| 155 | + "data": content.encode("utf-8"), |
| 156 | + } |
| 157 | + |
| 158 | + algorithm = self.algorithm_map.get(target, "gzip") |
| 159 | + |
| 160 | + start_time = time.time() |
| 161 | + compressed_data = self.compress(content, algorithm, target) |
| 162 | + compression_time = time.time() - start_time |
| 163 | + |
| 164 | + original_size = len(content.encode("utf-8")) |
| 165 | + compressed_size = len(compressed_data) |
| 166 | + compression_ratio = compressed_size / original_size if original_size > 0 else 1.0 |
| 167 | + |
| 168 | + return { |
| 169 | + "compressed": True, |
| 170 | + "algorithm": algorithm, |
| 171 | + "target": target, |
| 172 | + "original_size": original_size, |
| 173 | + "compressed_size": compressed_size, |
| 174 | + "compression_ratio": compression_ratio, |
| 175 | + "compression_time": compression_time, |
| 176 | + "data": compressed_data, |
| 177 | + } |
| 178 | + |
| 179 | + def get_algorithm_info(self) -> Dict[str, Dict]: |
| 180 | + """ |
| 181 | + Get information about available compression algorithms |
| 182 | +
|
| 183 | + Returns: |
| 184 | + Dictionary with algorithm availability and characteristics |
| 185 | + """ |
| 186 | + info = { |
| 187 | + "gzip": { |
| 188 | + "available": True, |
| 189 | + "use_case": "General purpose, widely supported", |
| 190 | + "speed": "Medium", |
| 191 | + "compression": "Good", |
| 192 | + } |
| 193 | + } |
| 194 | + |
| 195 | + # Check LZ4 availability |
| 196 | + try: |
| 197 | + import lz4.frame |
| 198 | + |
| 199 | + info["lz4"] = { |
| 200 | + "available": True, |
| 201 | + "use_case": "Real-time AI communication", |
| 202 | + "speed": "Very Fast", |
| 203 | + "compression": "Moderate", |
| 204 | + } |
| 205 | + except ImportError: |
| 206 | + info["lz4"] = { |
| 207 | + "available": False, |
| 208 | + "install": "pip install lz4", |
| 209 | + "use_case": "Real-time AI communication", |
| 210 | + "speed": "Very Fast", |
| 211 | + "compression": "Moderate", |
| 212 | + } |
| 213 | + |
| 214 | + # Check ZSTD availability |
| 215 | + try: |
| 216 | + import zstandard |
| 217 | + |
| 218 | + info["zstd"] = { |
| 219 | + "available": True, |
| 220 | + "use_case": "Storage and balanced performance", |
| 221 | + "speed": "Fast", |
| 222 | + "compression": "Excellent", |
| 223 | + } |
| 224 | + except ImportError: |
| 225 | + info["zstd"] = { |
| 226 | + "available": False, |
| 227 | + "install": "pip install zstandard", |
| 228 | + "use_case": "Storage and balanced performance", |
| 229 | + "speed": "Fast", |
| 230 | + "compression": "Excellent", |
| 231 | + } |
| 232 | + |
| 233 | + return info |
0 commit comments