red-candle 1.4.0 → 1.4.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7a3ac57ccd28cb2eb647c5a71f42fdbdf84a9dcaabd906165daca3d57c4a8eb0
-  data.tar.gz: 283104e93802ac97f11525226c9e41dc3bebccb8706d5b69a6a648d62cf2ccad
+  metadata.gz: 56022d9cb3677fa07bedae0272b5855374e2a6a5af863e055b546e1a0a9ddd5d
+  data.tar.gz: b027603292ba34d1cf75460e74edf1d2472ea5651f8690d86adc50dc7000bdcc
 SHA512:
-  metadata.gz: d5be5ca76fe5441ee1fd87ea83bcd02f5a630c3410a4dccc347da28290ef408c4c2cae428134f2a8fba839fe5aacf2b7593fe45274c967600e6684d09f212a01
-  data.tar.gz: 72289aaf0c17dea679acfa4f92be5371f2543293e27dab0752b39c77a1c2be98c480c0ce7266bc07519551770abdf53fdbbf0f4d3c1fdf6ba928c2452e598060
+  metadata.gz: cbc00f433b91bddc9d117e836488f1991d42a4e7ca8c9968949adfe671277a22e36a2a65f26b7cd17036c24c6e24d6a45ce1f4684f6281811d5b25f5830b0558
+  data.tar.gz: 2970cd8e5d7d59c6d6be5e69219da7c518962e12b59f2c18f2d4fa806a50d264198b1c027f82263e185987e2f75d0fd4ee81194ed272a257fdd804575756b82c

data/Cargo.lock

@@ -300,9 +300,9 @@ checksum = "1fd0f2584146f6f2ef48085050886acf353beff7305ebd1ae69500e27c67f64b"
 
 [[package]]
 name = "bytes"
-version = "1.10.1"
+version = "1.11.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d71b6127be86fdcfddb610f7182ac57211d4b18a3e9c82eb2d17662f2227ad6a"
+checksum = "1e748733b7cbc798e1434b6ac524f0c1ff2ab456fe201501e6497c8417a4fc33"
 
 [[package]]
 name = "candle"
@@ -819,7 +819,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "39cab71617ae0d63f51a36d69f866391735b51691dbda63cf6f96d042b63efeb"
 dependencies = [
  "libc",
- "windows-sys 0.61.0",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -2468,9 +2468,9 @@ dependencies = [
 
 [[package]]
 name = "quinn-proto"
-version = "0.11.13"
+version = "0.11.14"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f1906b49b0c3bc04b5fe5d86a77925ae6524a19b816ae38ce1e426255f1d8a31"
+checksum = "434b42fec591c96ef50e21e886936e66d3cc3f737104fdb9b737c40ffb94c098"
 dependencies = [
  "bytes",
  "getrandom 0.3.3",
@@ -2498,7 +2498,7 @@ dependencies = [
  "once_cell",
  "socket2",
  "tracing",
- "windows-sys 0.60.2",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -2819,7 +2819,7 @@ dependencies = [
  "errno",
  "libc",
  "linux-raw-sys",
- "windows-sys 0.61.0",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -2850,9 +2850,9 @@ dependencies = [
 
 [[package]]
 name = "rustls-webpki"
-version = "0.103.5"
+version = "0.103.10"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b5a37813727b78798e53c2bec3f5e8fe12a6d6f8389bf9ca7802add4c9905ad8"
+checksum = "df33b2b81ac578cabaf06b89b0631153a3f416b0a886e8a7a1707fb51abbd1ef"
 dependencies = [
  "aws-lc-rs",
  "ring",
@@ -3195,9 +3195,9 @@ dependencies = [
 
 [[package]]
 name = "tar"
-version = "0.4.44"
+version = "0.4.45"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1d863878d212c87a19c1a610eb53bb01fe12951c0501cf5a0d65f724914a667a"
+checksum = "22692a6476a21fa75fdfc11d452fda482af402c008cdbaf3476414e122040973"
 dependencies = [
  "filetime",
  "libc",
@@ -3214,7 +3214,7 @@ dependencies = [
  "getrandom 0.3.3",
  "once_cell",
  "rustix",
- "windows-sys 0.61.0",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -3888,7 +3888,7 @@ version = "0.1.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c2a7b1c03c876122aa43f3020e6c3c3ee5c05081c9a00739faf7503aeba10d22"
 dependencies = [
- "windows-sys 0.61.0",
+ "windows-sys 0.48.0",
 ]
 
 [[package]]

data/ext/candle/src/lib.rs

@@ -1,6 +1,7 @@
 use magnus::{function, prelude::*, Ruby};
 
 use crate::ruby::candle_utils;
+use crate::ruby::utils::ensure_hf_cache_dir;
 use crate::ruby::Result;
 
 pub mod llm;
@@ -22,19 +23,20 @@ const DEFAULT_DEVICE: &str = "cpu";
 pub fn get_build_info() -> magnus::RHash {
     let ruby = magnus::Ruby::get().unwrap();
     let hash = ruby.hash_new();
-    
+
     let _ = hash.aset("default_device", DEFAULT_DEVICE);
     let _ = hash.aset("cuda_available", cfg!(feature = "cuda"));
     let _ = hash.aset("metal_available", cfg!(feature = "metal"));
     let _ = hash.aset("mkl_available", cfg!(feature = "mkl"));
     let _ = hash.aset("accelerate_available", cfg!(feature = "accelerate"));
     let _ = hash.aset("cudnn_available", cfg!(feature = "cudnn"));
-    
+
     hash
 }
 
 #[magnus::init]
 fn init(ruby: &Ruby) -> Result<()> {
+    ensure_hf_cache_dir();
     let rb_candle = ruby.define_module("Candle")?;
     
     // Export build info

data/ext/candle/src/llm/gemma.rs

@@ -34,7 +34,7 @@ impl Gemma {
     pub async fn from_pretrained_with_tokenizer(model_id: &str, device: Device, tokenizer_source: Option<&str>) -> CandleResult<Self> {
         let api = Api::new()
             .map_err(|e| candle_core::Error::Msg(format!("Failed to create HF API: {}", e)))?;
-        
+
         let repo = api.repo(Repo::model(model_id.to_string()));
         
         // Download model files

data/ext/candle/src/llm/llama.rs

@@ -41,7 +41,7 @@ impl Llama {
     pub async fn from_pretrained_with_tokenizer(model_id: &str, device: Device, tokenizer_source: Option<&str>) -> CandleResult<Self> {
         let api = Api::new()
             .map_err(|e| candle_core::Error::Msg(format!("Failed to create HF API: {}", e)))?;
-        
+
         let repo = api.repo(Repo::model(model_id.to_string()));
         
         // Download model files

data/ext/candle/src/llm/mistral.rs

@@ -34,7 +34,7 @@ impl Mistral {
     pub async fn from_pretrained_with_tokenizer(model_id: &str, device: Device, tokenizer_source: Option<&str>) -> CandleResult<Self> {
         let api = Api::new()
             .map_err(|e| candle_core::Error::Msg(format!("Failed to create HF API: {}", e)))?;
-        
+
         let repo = api.repo(Repo::model(model_id.to_string()));
         
         // Download model files

data/ext/candle/src/llm/phi.rs

@@ -42,7 +42,7 @@ impl Phi {
     pub async fn from_pretrained_with_tokenizer(model_id: &str, device: Device, tokenizer_source: Option<&str>) -> CandleResult<Self> {
         let api = Api::new()
             .map_err(|e| candle_core::Error::Msg(format!("Failed to create HF API: {}", e)))?;
-        
+
         let repo = api.model(model_id.to_string());
         
         // Download configuration

data/ext/candle/src/llm/quantized_gguf.rs

@@ -53,7 +53,7 @@ impl QuantizedGGUF {
 
         let api = Api::new()
             .map_err(|e| candle_core::Error::Msg(format!("Failed to create HF API: {}", e)))?;
-        
+
         let repo = api.model(actual_model_id.to_string());
         
         // Download GGUF file

data/ext/candle/src/llm/qwen.rs

@@ -34,7 +34,7 @@ impl Qwen {
     pub async fn from_pretrained_with_tokenizer(model_id: &str, device: Device, tokenizer_source: Option<&str>) -> CandleResult<Self> {
         let api = Api::new()
             .map_err(|e| candle_core::Error::Msg(format!("Failed to create HF API: {}", e)))?;
-        
+
         let repo = api.model(model_id.to_string());
         
         // Download configuration

data/ext/candle/src/ruby/utils.rs

@@ -4,6 +4,32 @@ use ::candle_core::Tensor as CoreTensor;
 
 use crate::ruby::Result;
 
+/// Ensures the HuggingFace cache directory exists before Api::new() is called.
+///
+/// The hf_hub crate stores downloaded models in a "hub" subdirectory under the
+/// cache root. When the parent directory doesn't exist, hf_hub may fail to
+/// create the full path or silently produce an empty cache. This function
+/// pre-creates the directory tree to avoid the race condition described in
+/// issue #72.
+///
+/// Resolution order for the cache root:
+///   1. $HF_HOME (if set)
+///   2. $XDG_CACHE_HOME/huggingface (if XDG_CACHE_HOME is set)
+///   3. ~/.cache/huggingface
+pub fn ensure_hf_cache_dir() {
+    let cache_root = if let Ok(hf_home) = std::env::var("HF_HOME") {
+        std::path::PathBuf::from(hf_home)
+    } else if let Ok(xdg) = std::env::var("XDG_CACHE_HOME") {
+        std::path::PathBuf::from(xdg).join("huggingface")
+    } else if let Ok(home) = std::env::var("HOME") {
+        std::path::PathBuf::from(home).join(".cache").join("huggingface")
+    } else {
+        return;
+    };
+    let hub_dir = cache_root.join("hub");
+    let _ = std::fs::create_dir_all(hub_dir);
+}
+
 pub fn actual_index(t: &CoreTensor, dim: usize, index: i64) -> candle_core::Result<usize> {
     let dim = t.dim(dim)?;
     if 0 <= index {

data/lib/candle/version.rb

@@ -1,5 +1,5 @@
 # :nocov:
 module Candle
-  VERSION = "1.4.0"
+  VERSION = "1.4.1"
 end
 # :nocov:

data/lib/candle.rb

@@ -1,5 +1,5 @@
 require_relative "candle/logger"
-require_relative "candle/candle"
+require "candle/candle"
 require_relative "candle/tensor"
 require_relative "candle/device_utils"
 require_relative "candle/embedding_model_type"

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: red-candle
 version: !ruby/object:Gem::Version
-  version: 1.4.0
+  version: 1.4.1
 platform: ruby
 authors:
 - Christopher Petersen
@@ -178,6 +178,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '3.13'
+- !ruby/object:Gem::Dependency
+  name: irb
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
 description: Ruby gem for running state-of-the-art language models locally. Access
   LLMs, embeddings, rerankers, and NER models directly from Ruby using Rust-powered
   Candle with Metal/CUDA acceleration.
@@ -267,7 +281,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '3.3'
 requirements:
 - Rust >= 1.85
-rubygems_version: 4.0.3
+rubygems_version: 4.0.6
 specification_version: 4
 summary: Ruby gem for running state-of-the-art language models locally. Access LLMs,
   embeddings, rerankers, and NER models directly from Ruby using Rust-powered Candle