[add] pydantic for parser result

README.md

@@ -5,7 +5,7 @@ colorFrom: green
 colorTo: indigo
 sdk: gradio
 sdk_version: 5.3.0
-app_file: app/main.py
+app_file: main.py
 pinned: false
 license: mit
 short_description: This project is a GUI for the gpustack/gguf-parser-go

app/models.py

@@ -0,0 +1,104 @@
+from typing import Optional
+
+from pydantic import BaseModel, Field
+
+
+class GgufParser(BaseModel):
+    metadata: "Metadata"
+    architecture: "Architecture"
+    tokenizer: "Tokenizer"
+    estimate: "Estimate"
+
+
+class Metadata(BaseModel):
+    type_: str = Field(alias="type")
+    architecture: str
+    quantization_version: int = Field(alias="quantizationVersion")
+    alignment: int
+    name: str
+    file_type: int = Field(alias="fileType")
+    little_endian: bool = Field(alias="littleEndian")
+    file_size: int = Field(alias="fileSize")
+    size: int
+    parameters: int
+    bits_per_weight: float = Field(alias="bitsPerWeight")
+
+
+class Architecture(BaseModel):
+    type_: str = Field(alias="type")
+    architecture: str
+    maximum_context_length: int = Field(alias="maximumContextLength")
+    embedding_length: int = Field(alias="embeddingLength")
+    block_count: int = Field(alias="blockCount")
+    feed_forward_length: int = Field(alias="feedForwardLength")
+    attention_head_count: int = Field(alias="attentionHeadCount")
+    attention_head_count_kv: int = Field(alias="attentionHeadCountKV")
+    attention_layer_norm_rmse_epsilon: float = Field(
+        alias="attentionLayerNormRMSEpsilon"
+    )
+    attention_key_length: int = Field(alias="attentionKeyLength")
+    attention_value_length: int = Field(alias="attentionValueLength")
+    attention_causal: bool = Field(alias="attentionCausal")
+    rope_dimension_count: int = Field(alias="ropeDimensionCount")
+    rope_frequency_base: int = Field(alias="ropeFrequencyBase")
+    vocabulary_length: int = Field(alias="vocabularyLength")
+    embedding_gqa: int = Field(alias="embeddingGQA")
+    embedding_key_gqa: int = Field(alias="embeddingKeyGQA")
+    embedding_value_gqa: int = Field(alias="embeddingValueGQA")
+
+
+class Tokenizer(BaseModel):
+    model: str
+    tokens_length: int = Field(alias="tokensLength")
+    merges_length: int = Field(alias="mergesLength")
+    added_token_length: int = Field(alias="addedTokenLength")
+    bos_token_id: int = Field(alias="bosTokenID")
+    eos_token_id: int = Field(alias="eosTokenID")
+    eot_token_id: int = Field(alias="eotTokenID")
+    eom_token_id: int = Field(alias="eomTokenID")
+    unknown_token_id: int = Field(alias="unknownTokenID")
+    separator_token_id: int = Field(alias="separatorTokenID")
+    padding_token_id: int = Field(alias="paddingTokenID")
+    tokens_size: int = Field(alias="tokensSize")
+    merges_size: int = Field(alias="mergesSize")
+
+
+class Ram(BaseModel):
+    handle_layers: int = Field(alias="handleLayers")
+    handle_last_layer: int = Field(alias="handleLastLayer")
+    handle_output_layer: bool = Field(alias="handleOutputLayer")
+    remote: bool
+    position: int
+    uma: int
+    nonuma: int
+
+
+class Item(BaseModel):
+    offload_layers: int = Field(alias="offloadLayers")
+    full_offloaded: bool = Field(alias="fullOffloaded")
+    ram: "Ram"
+    vrams: list["Ram"]
+
+
+class Estimate(BaseModel):
+    items: list["Item"]
+    type_: str = Field(alias="type")
+    architecture: str
+    context_size: int = Field(alias="contextSize")
+    flash_attention: bool = Field(alias="flashAttention")
+    no_mmap: bool = Field(alias="noMMap")
+    embedding_only: bool = Field(alias="embeddingOnly")
+    reranking: bool
+    distributable: bool
+    logical_batch_size: int = Field(alias="logicalBatchSize")
+    physical_batch_size: int = Field(alias="physicalBatchSize")
+    type_: str = Field(alias="type")
+    architecture: str
+    context_size: int = Field(alias="contextSize")
+    flash_attention: bool = Field(alias="flashAttention")
+    no_mmap: bool = Field(alias="noMMap")
+    embedding_only: bool = Field(alias="embeddingOnly")
+    reranking: bool
+    distributable: bool
+    logical_batch_size: int = Field(alias="logicalBatchSize")
+    physical_batch_size: int = Field(alias="physicalBatchSize")

app/main.py → main.py

@@ -5,6 +5,8 @@ from pathlib import Path
 import gradio as gr
 import pandas as pd
 
+from app.models import GgufParser
+
 GGUF_PARSER_VERSION = os.getenv("GGUF_PARSER_VERSION", "v0.12.0")
 gguf_parser = Path("gguf-parser-linux-amd64")
 gguf_parser_url = f"https://github.com/gpustack/gguf-parser-go/releases/download/{GGUF_PARSER_VERSION}/{gguf_parser}"
@@ -16,27 +18,17 @@ def process_url(url, context_length):
         res = os.popen(
             f"./{gguf_parser} --ctx-size={context_length} -url {url} --json"
         ).read()
-        data = json.loads(res)
+        parser_result = GgufParser.model_validate_json(res)
+        # data = json.loads(res)
 
-        metadata_df = pd.DataFrame([data["metadata"]])
+        metadata_df = pd.DataFrame([parser_result.metadata.model_dump()])
 
-        architecture_df = pd.DataFrame([data["architecture"]])
+        architecture_df = pd.DataFrame([parser_result.architecture.model_dump()])
 
-        tokenizer_df = pd.DataFrame([data["tokenizer"]])
+        tokenizer_df = pd.DataFrame([parser_result.tokenizer.model_dump()])
 
         estimate_df = pd.DataFrame(
-            [
-                {
-                    # "maximumTokensPerSecond": data["estimate"]["items"][0][
-                    #     "maximumTokensPerSecond"
-                    # ],
-                    "offloadLayers": data["estimate"]["items"][0]["offloadLayers"],
-                    "fullOffloaded": data["estimate"]["items"][0]["fullOffloaded"],
-                    "contextSize": data["estimate"]["contextSize"],
-                    "flashAttention": data["estimate"]["flashAttention"],
-                    "distributable": data["estimate"]["distributable"],
-                }
-            ]
+            [parser_result.estimate.model_dump(exclude_none=True)]
         )
 
         return metadata_df, architecture_df, tokenizer_df, estimate_df

pyproject.toml

@@ -8,7 +8,13 @@ dependencies = [
     "gradio==5.3.0",
     "orjson==3.10.10",
     "pandas>=2.2.3",
+    "pydantic>=2.9.2",
 ]
 
 [tool.pylint]
 max-line-length = 180
+
+[tool.uv]
+dev-dependencies = [
+    "pytest>=8.3.3",
+]

requirements.txt

@@ -0,0 +1 @@
+pydantic>=2.9.2

uv.lock

@@ -197,6 +197,12 @@ dependencies = [
     { name = "gradio" },
     { name = "orjson" },
     { name = "pandas" },
+    { name = "pydantic" },
+]
+
+[package.dev-dependencies]
+dev = [
+    { name = "pytest" },
 ]
 
 [package.metadata]
@@ -204,8 +210,12 @@ requires-dist = [
     { name = "gradio", specifier = "==5.3.0" },
     { name = "orjson", specifier = "==3.10.10" },
     { name = "pandas", specifier = ">=2.2.3" },
+    { name = "pydantic", specifier = ">=2.9.2" },
 ]
 
+[package.metadata.requires-dev]
+dev = [{ name = "pytest", specifier = ">=8.3.3" }]
+
 [[package]]
 name = "gradio"
 version = "5.3.0"
@@ -324,6 +334,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/76/c6/c88e154df9c4e1a2a66ccf0005a88dfb2650c1dffb6f5ce603dfbd452ce3/idna-3.10-py3-none-any.whl", hash = "sha256:946d195a0d259cbba61165e88e65941f16e9b36ea6ddb97f00452bae8b1287d3", size = 70442 },
 ]
 
+[[package]]
+name = "iniconfig"
+version = "2.0.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/d7/4b/cbd8e699e64a6f16ca3a8220661b5f83792b3017d0f79807cb8708d33913/iniconfig-2.0.0.tar.gz", hash = "sha256:2d91e135bf72d31a410b17c16da610a82cb55f6b0477d1a902134b24a455b8b3", size = 4646 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/ef/a6/62565a6e1cf69e10f5727360368e451d4b7f58beeac6173dc9db836a5b46/iniconfig-2.0.0-py3-none-any.whl", hash = "sha256:b6a85871a79d2e3b22d2d1b94ac2824226a63c6b741c88f7ae975f18b6778374", size = 5892 },
+]
+
 [[package]]
 name = "jinja2"
 version = "3.1.4"
@@ -616,6 +635,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/52/3b/ce7a01026a7cf46e5452afa86f97a5e88ca97f562cafa76570178ab56d8d/pillow-10.4.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:0755ffd4a0c6f267cccbae2e9903d95477ca2f77c4fcf3a3a09570001856c8a5", size = 2554661 },
 ]
 
+[[package]]
+name = "pluggy"
+version = "1.5.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/96/2d/02d4312c973c6050a18b314a5ad0b3210edb65a906f868e31c111dede4a6/pluggy-1.5.0.tar.gz", hash = "sha256:2cffa88e94fdc978c4c574f15f9e59b7f4201d439195c3715ca9e2486f1d0cf1", size = 67955 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/88/5f/e351af9a41f866ac3f1fac4ca0613908d9a41741cfcf2228f4ad853b697d/pluggy-1.5.0-py3-none-any.whl", hash = "sha256:44e1ad92c8ca002de6377e165f3e0f1be63266ab4d554740532335b9d75ea669", size = 20556 },
+]
+
 [[package]]
 name = "pydantic"
 version = "2.9.2"
@@ -715,6 +743,23 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/f7/3f/01c8b82017c199075f8f788d0d906b9ffbbc5a47dc9918a945e13d5a2bda/pygments-2.18.0-py3-none-any.whl", hash = "sha256:b8e6aca0523f3ab76fee51799c488e38782ac06eafcf95e7ba832985c8e7b13a", size = 1205513 },
 ]
 
+[[package]]
+name = "pytest"
+version = "8.3.3"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "colorama", marker = "sys_platform == 'win32'" },
+    { name = "exceptiongroup", marker = "python_full_version < '3.11'" },
+    { name = "iniconfig" },
+    { name = "packaging" },
+    { name = "pluggy" },
+    { name = "tomli", marker = "python_full_version < '3.11'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/8b/6c/62bbd536103af674e227c41a8f3dcd022d591f6eed5facb5a0f31ee33bbc/pytest-8.3.3.tar.gz", hash = "sha256:70b98107bd648308a7952b06e6ca9a50bc660be218d53c257cc1fc94fda10181", size = 1442487 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/6b/77/7440a06a8ead44c7757a64362dd22df5760f9b12dc5f11b6188cd2fc27a0/pytest-8.3.3-py3-none-any.whl", hash = "sha256:a6853c7375b2663155079443d2e45de913a911a11d669df02a50814944db57b2", size = 342341 },
+]
+
 [[package]]
 name = "python-dateutil"
 version = "2.9.0.post0"
@@ -891,6 +936,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/54/43/f185bfd0ca1d213beb4293bed51d92254df23d8ceaf6c0e17146d508a776/starlette-0.41.2-py3-none-any.whl", hash = "sha256:fbc189474b4731cf30fcef52f18a8d070e3f3b46c6a04c97579e85e6ffca942d", size = 73259 },
 ]
 
+[[package]]
+name = "tomli"
+version = "2.0.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/35/b9/de2a5c0144d7d75a57ff355c0c24054f965b2dc3036456ae03a51ea6264b/tomli-2.0.2.tar.gz", hash = "sha256:d46d457a85337051c36524bc5349dd91b1877838e2979ac5ced3e710ed8a60ed", size = 16096 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/cf/db/ce8eda256fa131af12e0a76d481711abe4681b6923c27efb9a255c9e4594/tomli-2.0.2-py3-none-any.whl", hash = "sha256:2ebe24485c53d303f690b0ec092806a085f07af5a5aa1464f3931eec36caaa38", size = 13237 },
+]
+
 [[package]]
 name = "tomlkit"
 version = "0.12.0"