|
| 1 | +package sk.ainet.models.apertus |
| 2 | + |
| 3 | +/** |
| 4 | + * Parses HuggingFace `config.json` into [ApertusModelMetadata]. |
| 5 | + * |
| 6 | + * Uses lightweight manual JSON parsing to avoid external dependencies. |
| 7 | + */ |
| 8 | +public object ApertusConfigParser { |
| 9 | + |
| 10 | + /** |
| 11 | + * Parse a HuggingFace config.json string into ApertusModelMetadata. |
| 12 | + * |
| 13 | + * Required fields: hidden_size, num_hidden_layers, num_attention_heads, |
| 14 | + * num_key_value_heads, intermediate_size, vocab_size. |
| 15 | + * Optional: max_position_embeddings, head_dim, rope_theta, hidden_act. |
| 16 | + */ |
| 17 | + public fun parse(json: String): ApertusModelMetadata { |
| 18 | + val map = parseJsonObject(json.trim()) |
| 19 | + |
| 20 | + val hiddenSize = map.requireInt("hidden_size") |
| 21 | + val numLayers = map.requireInt("num_hidden_layers") |
| 22 | + val numHeads = map.requireInt("num_attention_heads") |
| 23 | + val numKvHeads = map.intOrNull("num_key_value_heads") ?: numHeads |
| 24 | + val intermediateSize = map.requireInt("intermediate_size") |
| 25 | + val vocabSize = map.requireInt("vocab_size") |
| 26 | + val contextLength = map.intOrNull("max_position_embeddings") ?: 2048 |
| 27 | + val headDim = map.intOrNull("head_dim") ?: (hiddenSize / numHeads) |
| 28 | + val architecture = map.stringOrNull("model_type") ?: "apertus" |
| 29 | + val ropeTheta = map.floatOrNull("rope_theta") ?: 12000000f |
| 30 | + val qkNorm = map["qk_norm"]?.lowercase()?.let { it == "true" || it == "per_head" } ?: true |
| 31 | + val hiddenAct = map.stringOrNull("hidden_act") ?: "xielu" |
| 32 | + val tiedEmbeddings = map["tie_word_embeddings"] == "true" |
| 33 | + |
| 34 | + return ApertusModelMetadata( |
| 35 | + architecture = architecture, |
| 36 | + embeddingLength = hiddenSize, |
| 37 | + contextLength = contextLength, |
| 38 | + blockCount = numLayers, |
| 39 | + headCount = numHeads, |
| 40 | + kvHeadCount = numKvHeads, |
| 41 | + feedForwardLength = intermediateSize, |
| 42 | + ropeDimensionCount = headDim, |
| 43 | + vocabSize = vocabSize, |
| 44 | + ropeTheta = ropeTheta, |
| 45 | + qkNorm = qkNorm, |
| 46 | + hiddenAct = hiddenAct, |
| 47 | + tiedEmbeddings = tiedEmbeddings |
| 48 | + ) |
| 49 | + } |
| 50 | + |
| 51 | + /** |
| 52 | + * Check if config.json indicates tied word embeddings. |
| 53 | + */ |
| 54 | + public fun isTiedEmbeddings(json: String): Boolean { |
| 55 | + val map = parseJsonObject(json.trim()) |
| 56 | + return map["tie_word_embeddings"] == "true" |
| 57 | + } |
| 58 | + |
| 59 | + // ========== Lightweight JSON parsing ========== |
| 60 | + |
| 61 | + private fun parseJsonObject(json: String): Map<String, String> { |
| 62 | + if (!json.startsWith("{") || !json.endsWith("}")) { |
| 63 | + error("config.json: expected JSON object") |
| 64 | + } |
| 65 | + val content = json.substring(1, json.length - 1) |
| 66 | + val result = mutableMapOf<String, String>() |
| 67 | + |
| 68 | + var i = 0 |
| 69 | + while (i < content.length) { |
| 70 | + while (i < content.length && content[i].isWhitespace()) i++ |
| 71 | + if (i >= content.length) break |
| 72 | + |
| 73 | + if (content[i] != '"') { i++; continue } |
| 74 | + val keyEnd = findStringEnd(content, i) |
| 75 | + val key = content.substring(i + 1, keyEnd) |
| 76 | + i = keyEnd + 1 |
| 77 | + |
| 78 | + while (i < content.length && (content[i].isWhitespace() || content[i] == ':')) i++ |
| 79 | + |
| 80 | + val valueStart = i |
| 81 | + i = skipValue(content, i) |
| 82 | + var value = content.substring(valueStart, i).trim() |
| 83 | + if (value.startsWith("\"") && value.endsWith("\"")) { |
| 84 | + value = value.substring(1, value.length - 1) |
| 85 | + } |
| 86 | + result[key] = value |
| 87 | + |
| 88 | + while (i < content.length && (content[i].isWhitespace() || content[i] == ',')) i++ |
| 89 | + } |
| 90 | + |
| 91 | + return result |
| 92 | + } |
| 93 | + |
| 94 | + private fun findStringEnd(s: String, start: Int): Int { |
| 95 | + var i = start + 1 |
| 96 | + while (i < s.length) { |
| 97 | + when (s[i]) { |
| 98 | + '"' -> return i |
| 99 | + '\\' -> i += 2 |
| 100 | + else -> i++ |
| 101 | + } |
| 102 | + } |
| 103 | + return s.length |
| 104 | + } |
| 105 | + |
| 106 | + private fun skipValue(s: String, start: Int): Int { |
| 107 | + if (start >= s.length) return start |
| 108 | + return when (s[start]) { |
| 109 | + '"' -> findStringEnd(s, start) + 1 |
| 110 | + '{' -> findMatching(s, start, '{', '}') |
| 111 | + '[' -> findMatching(s, start, '[', ']') |
| 112 | + else -> { |
| 113 | + var i = start |
| 114 | + while (i < s.length && s[i] != ',' && s[i] != '}' && s[i] != ']') i++ |
| 115 | + i |
| 116 | + } |
| 117 | + } |
| 118 | + } |
| 119 | + |
| 120 | + private fun findMatching(s: String, start: Int, open: Char, close: Char): Int { |
| 121 | + var depth = 0 |
| 122 | + var i = start |
| 123 | + var inString = false |
| 124 | + while (i < s.length) { |
| 125 | + val c = s[i] |
| 126 | + when { |
| 127 | + inString -> { |
| 128 | + if (c == '"') inString = false |
| 129 | + else if (c == '\\') i++ |
| 130 | + } |
| 131 | + c == '"' -> inString = true |
| 132 | + c == open -> depth++ |
| 133 | + c == close -> { depth--; if (depth == 0) return i + 1 } |
| 134 | + } |
| 135 | + i++ |
| 136 | + } |
| 137 | + return s.length |
| 138 | + } |
| 139 | + |
| 140 | + private fun Map<String, String>.requireInt(key: String): Int = |
| 141 | + this[key]?.toIntOrNull() ?: error("config.json: missing or invalid '$key'") |
| 142 | + |
| 143 | + private fun Map<String, String>.intOrNull(key: String): Int? = |
| 144 | + this[key]?.toIntOrNull() |
| 145 | + |
| 146 | + private fun Map<String, String>.floatOrNull(key: String): Float? = |
| 147 | + this[key]?.toFloatOrNull() |
| 148 | + |
| 149 | + private fun Map<String, String>.stringOrNull(key: String): String? = |
| 150 | + this[key]?.takeIf { it.isNotBlank() } |
| 151 | +} |
0 commit comments