update dmatrix

Author: vaaaaanquish

examples/basic/Cargo.toml

@@ -6,6 +6,6 @@ publish = false
 
 [dependencies]
 xgboost = { path = "../../" }
-sprs = "0.6"
+sprs = "0.11"
 log = "0.4"
 env_logger = "0.5"

examples/basic/src/main.rs

@@ -97,7 +97,7 @@ fn main() {
     let mut data = Vec::new();
 
     let reader = BufReader::new(File::open("../../xgboost-sys/xgboost/demo/data/agaricus.txt.train").unwrap());
-    let mut current_row: u64 = 0;
+    let mut current_row = 0;
     for line in reader.lines() {
         let line = line.unwrap();
         let sample: Vec<&str> = line.split_whitespace().collect();
@@ -106,7 +106,7 @@ fn main() {
         for entry in &sample[1..] {
             let pair: Vec<&str> = entry.split(':').collect();
             rows.push(current_row);
-            cols.push(pair[0].parse::<u64>().unwrap());
+            cols.push(pair[0].parse::<usize>().unwrap());
             data.push(pair[1].parse::<f32>().unwrap());
         }
 
@@ -116,11 +116,12 @@ fn main() {
     // work out size of sparse matrix from max row/col values
     let shape = ((*rows.iter().max().unwrap() + 1) as usize,
                  (*cols.iter().max().unwrap() + 1) as usize);
+    let num_col = Some((*cols.iter().max().unwrap() + 1) as usize);
     let triplet_mat = sprs::TriMatBase::from_triplets(shape, rows, cols, data);
     let csr_mat = triplet_mat.to_csr();
 
     let indices: Vec<usize> = csr_mat.indices().into_iter().map(|i| *i as usize).collect();
-    let mut dtrain = DMatrix::from_csr(csr_mat.indptr(), &indices, csr_mat.data(), None).unwrap();
+    let mut dtrain = DMatrix::from_csr(csr_mat.indptr().raw_storage(), &indices, csr_mat.data(), num_col).unwrap();
     dtrain.set_labels(&labels).unwrap();
 
     let training_params = parameters::TrainingParametersBuilder::default().dtrain(&dtrain).build().unwrap();

src/dmatrix.rs

@@ -124,9 +124,10 @@ impl DMatrix {
     /// `data[indptr[i]:indptr[i+1]`.
     ///
     /// If `num_cols` is set to None, number of columns will be inferred from given data.
-    pub fn from_csr(indptr: &[u64], indices: &[usize], data: &[f32], num_cols: Option<usize>) -> XGBResult<Self> {
+    pub fn from_csr(indptr: &[usize], indices: &[usize], data: &[f32], num_cols: Option<usize>) -> XGBResult<Self> {
         assert_eq!(indices.len(), data.len());
         let mut handle = ptr::null_mut();
+        let indptr: Vec<u64> = indptr.iter().map(|x| *x as u64).collect();
         let indices: Vec<u32> = indices.iter().map(|x| *x as u32).collect();
         let num_cols = num_cols.unwrap_or(0); // infer from data if 0
         xgb_call!(xgboost_sys::XGDMatrixCreateFromCSREx(indptr.as_ptr(),
@@ -147,9 +148,10 @@ impl DMatrix {
     /// `data[indptr[i]:indptr[i+1]`.
     ///
     /// If `num_rows` is set to None, number of rows will be inferred from given data.
-    pub fn from_csc(indptr: &[u64], indices: &[usize], data: &[f32], num_rows: Option<usize>) -> XGBResult<Self> {
+    pub fn from_csc(indptr: &[usize], indices: &[usize], data: &[f32], num_rows: Option<usize>) -> XGBResult<Self> {
         assert_eq!(indices.len(), data.len());
         let mut handle = ptr::null_mut();
+        let indptr: Vec<u64> = indptr.iter().map(|x| *x as u64).collect();
         let indices: Vec<u32> = indices.iter().map(|x| *x as u32).collect();
         let num_rows = num_rows.unwrap_or(0); // infer from data if 0
         xgb_call!(xgboost_sys::XGDMatrixCreateFromCSCEx(indptr.as_ptr(),