You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@iceberg.apache.org by gi...@apache.org on 2025/09/03 09:51:44 UTC

(iceberg-rust) branch gh-pages updated: deploy: bd228fe247d617c1507db200d6c0014c4eb5ca8f

This is an automated email from the ASF dual-hosted git repository.

github-bot pushed a commit to branch gh-pages
in repository https://gitbox.apache.org/repos/asf/iceberg-rust.git


The following commit(s) were added to refs/heads/gh-pages by this push:
     new d667cc5c5 deploy: bd228fe247d617c1507db200d6c0014c4eb5ca8f
d667cc5c5 is described below

commit d667cc5c5a418080e3c930c71c274b79325a31be
Author: liurenjie1024 <li...@users.noreply.github.com>
AuthorDate: Wed Sep 3 09:51:40 2025 +0000

    deploy: bd228fe247d617c1507db200d6c0014c4eb5ca8f
---
 api/iceberg/arrow/index.html                       |   2 +-
 api/src-files.js                                   |   4 +-
 api/src/iceberg/arrow/mod.rs.html                  |   3 +-
 .../arrow/record_batch_partition_splitter.rs.html  | 304 +++++++++++++++++++++
 4 files changed, 309 insertions(+), 4 deletions(-)

diff --git a/api/iceberg/arrow/index.html b/api/iceberg/arrow/index.html
index 90b41d837..8ccdf3ba9 100644
--- a/api/iceberg/arrow/index.html
+++ b/api/iceberg/arrow/index.html
@@ -1,3 +1,3 @@
-<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="Conversion between Iceberg and Arrow schema"><title>iceberg::arrow - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,FiraSans-Med [...]
+<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="Conversion between Iceberg and Arrow schema"><title>iceberg::arrow - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,FiraSans-Med [...]
 </div></details><h2 id="modules" class="section-header">Modules<a href="#modules" class="anchor">§</a></h2><dl class="item-table"><dt><a class="mod" href="delete_file_loader/index.html" title="mod iceberg::arrow::delete_file_loader">delete_<wbr>file_<wbr>loader</a></dt><dd>Delete File loader</dd></dl><h2 id="structs" class="section-header">Structs<a href="#structs" class="anchor">§</a></h2><dl class="item-table"><dt><a class="struct" href="struct.ArrowArrayAccessor.html" title="struct ic [...]
 This function will assume the schema of arrow struct array is the same as iceberg struct type.</dd><dt><a class="fn" href="fn.arrow_type_to_type.html" title="fn iceberg::arrow::arrow_type_to_type">arrow_<wbr>type_<wbr>to_<wbr>type</a></dt><dd>Convert Arrow type to iceberg type.</dd><dt><a class="fn" href="fn.schema_to_arrow_schema.html" title="fn iceberg::arrow::schema_to_arrow_schema">schema_<wbr>to_<wbr>arrow_<wbr>schema</a></dt><dd>Convert iceberg schema to an arrow schema.</dd><dt><a [...]
\ No newline at end of file
diff --git a/api/src-files.js b/api/src-files.js
index e06967384..43a784bed 100644
--- a/api/src-files.js
+++ b/api/src-files.js
@@ -1,3 +1,3 @@
-var srcIndex = new Map(JSON.parse('[["iceberg",["",[["arrow",[],["caching_delete_file_loader.rs","delete_file_loader.rs","delete_filter.rs","mod.rs","nan_val_cnt_visitor.rs","reader.rs","record_batch_projector.rs","record_batch_transformer.rs","schema.rs","value.rs"]],["avro",[],["mod.rs","schema.rs"]],["catalog",[["memory",[],["catalog.rs","mod.rs","namespace_state.rs"]]],["metadata_location.rs","mod.rs"]],["expr",[["visitors",[],["bound_predicate_visitor.rs","expression_evaluator.rs"," [...]
+var srcIndex = new Map(JSON.parse('[["iceberg",["",[["arrow",[],["caching_delete_file_loader.rs","delete_file_loader.rs","delete_filter.rs","mod.rs","nan_val_cnt_visitor.rs","reader.rs","record_batch_partition_splitter.rs","record_batch_projector.rs","record_batch_transformer.rs","schema.rs","value.rs"]],["avro",[],["mod.rs","schema.rs"]],["catalog",[["memory",[],["catalog.rs","mod.rs","namespace_state.rs"]]],["metadata_location.rs","mod.rs"]],["expr",[["visitors",[],["bound_predicate_vi [...]
 createSrcSidebar();
-//{"start":36,"fragment_lengths":[2283,42,91,90,46,80,72,67,256,49,55,109,63]}
\ No newline at end of file
+//{"start":36,"fragment_lengths":[2320,42,91,90,46,80,72,67,256,49,55,109,63]}
\ No newline at end of file
diff --git a/api/src/iceberg/arrow/mod.rs.html b/api/src/iceberg/arrow/mod.rs.html
index 8c232744b..4233b0395 100644
--- a/api/src/iceberg/arrow/mod.rs.html
+++ b/api/src/iceberg/arrow/mod.rs.html
@@ -33,4 +33,5 @@
 <a href=#33 id=33 data-nosnippet>33</a><span class="kw">mod </span>value;
 <a href=#34 id=34 data-nosnippet>34</a>
 <a href=#35 id=35 data-nosnippet>35</a><span class="kw">pub use </span>reader::<span class="kw-2">*</span>;
-<a href=#36 id=36 data-nosnippet>36</a><span class="kw">pub use </span>value::<span class="kw-2">*</span>;</code></pre></div></section></main></body></html>
\ No newline at end of file
+<a href=#36 id=36 data-nosnippet>36</a><span class="kw">pub use </span>value::<span class="kw-2">*</span>;
+<a href=#37 id=37 data-nosnippet>37</a><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">mod </span>record_batch_partition_splitter;</code></pre></div></section></main></body></html>
\ No newline at end of file
diff --git a/api/src/iceberg/arrow/record_batch_partition_splitter.rs.html b/api/src/iceberg/arrow/record_batch_partition_splitter.rs.html
new file mode 100644
index 000000000..f91b60034
--- /dev/null
+++ b/api/src/iceberg/arrow/record_batch_partition_splitter.rs.html
@@ -0,0 +1,304 @@
+<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="Source of the Rust file `crates/iceberg/src/arrow/record_batch_partition_splitter.rs`."><title>record_batch_partition_splitter.rs - source</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-I [...]
+<a href=#2 id=2 data-nosnippet>2</a>// or more contributor license agreements.  See the NOTICE file
+<a href=#3 id=3 data-nosnippet>3</a>// distributed with this work for additional information
+<a href=#4 id=4 data-nosnippet>4</a>// regarding copyright ownership.  The ASF licenses this file
+<a href=#5 id=5 data-nosnippet>5</a>// to you under the Apache License, Version 2.0 (the
+<a href=#6 id=6 data-nosnippet>6</a>// "License"); you may not use this file except in compliance
+<a href=#7 id=7 data-nosnippet>7</a>// with the License.  You may obtain a copy of the License at
+<a href=#8 id=8 data-nosnippet>8</a>//
+<a href=#9 id=9 data-nosnippet>9</a>//   http://www.apache.org/licenses/LICENSE-2.0
+<a href=#10 id=10 data-nosnippet>10</a>//
+<a href=#11 id=11 data-nosnippet>11</a>// Unless required by applicable law or agreed to in writing,
+<a href=#12 id=12 data-nosnippet>12</a>// software distributed under the License is distributed on an
+<a href=#13 id=13 data-nosnippet>13</a>// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+<a href=#14 id=14 data-nosnippet>14</a>// KIND, either express or implied.  See the License for the
+<a href=#15 id=15 data-nosnippet>15</a>// specific language governing permissions and limitations
+<a href=#16 id=16 data-nosnippet>16</a>// under the License.
+<a href=#17 id=17 data-nosnippet>17</a>
+<a href=#18 id=18 data-nosnippet>18</a></span><span class="kw">use </span>std::collections::HashMap;
+<a href=#19 id=19 data-nosnippet>19</a><span class="kw">use </span>std::sync::Arc;
+<a href=#20 id=20 data-nosnippet>20</a>
+<a href=#21 id=21 data-nosnippet>21</a><span class="kw">use </span>arrow_array::{ArrayRef, BooleanArray, RecordBatch, StructArray};
+<a href=#22 id=22 data-nosnippet>22</a><span class="kw">use </span>arrow_schema::{DataType, SchemaRef <span class="kw">as </span>ArrowSchemaRef};
+<a href=#23 id=23 data-nosnippet>23</a><span class="kw">use </span>arrow_select::filter::filter_record_batch;
+<a href=#24 id=24 data-nosnippet>24</a><span class="kw">use </span>itertools::Itertools;
+<a href=#25 id=25 data-nosnippet>25</a><span class="kw">use </span>parquet::arrow::PARQUET_FIELD_ID_META_KEY;
+<a href=#26 id=26 data-nosnippet>26</a>
+<a href=#27 id=27 data-nosnippet>27</a><span class="kw">use </span><span class="kw">super</span>::arrow_struct_to_literal;
+<a href=#28 id=28 data-nosnippet>28</a><span class="kw">use </span><span class="kw">super</span>::record_batch_projector::RecordBatchProjector;
+<a href=#29 id=29 data-nosnippet>29</a><span class="kw">use </span><span class="kw">crate</span>::arrow::type_to_arrow_type;
+<a href=#30 id=30 data-nosnippet>30</a><span class="kw">use </span><span class="kw">crate</span>::spec::{Literal, PartitionSpecRef, SchemaRef, Struct, StructType, Type};
+<a href=#31 id=31 data-nosnippet>31</a><span class="kw">use </span><span class="kw">crate</span>::transform::{BoxedTransformFunction, create_transform_function};
+<a href=#32 id=32 data-nosnippet>32</a><span class="kw">use crate</span>::{Error, ErrorKind, <span class="prelude-ty">Result</span>};
+<a href=#33 id=33 data-nosnippet>33</a>
+<a href=#34 id=34 data-nosnippet>34</a><span class="doccomment">/// The splitter used to split the record batch into multiple record batches by the partition spec.
+<a href=#35 id=35 data-nosnippet>35</a>/// 1. It will project and transform the input record batch based on the partition spec, get the partitioned record batch.
+<a href=#36 id=36 data-nosnippet>36</a>/// 2. Split the input record batch into multiple record batches based on the partitioned record batch.
+<a href=#37 id=37 data-nosnippet>37</a></span><span class="comment">// # TODO
+<a href=#38 id=38 data-nosnippet>38</a>// Remove this after partition writer supported.
+<a href=#39 id=39 data-nosnippet>39</a></span><span class="attr">#[allow(dead_code)]
+<a href=#40 id=40 data-nosnippet>40</a></span><span class="kw">pub struct </span>RecordBatchPartitionSplitter {
+<a href=#41 id=41 data-nosnippet>41</a>    schema: SchemaRef,
+<a href=#42 id=42 data-nosnippet>42</a>    partition_spec: PartitionSpecRef,
+<a href=#43 id=43 data-nosnippet>43</a>    projector: RecordBatchProjector,
+<a href=#44 id=44 data-nosnippet>44</a>    transform_functions: Vec&lt;BoxedTransformFunction&gt;,
+<a href=#45 id=45 data-nosnippet>45</a>
+<a href=#46 id=46 data-nosnippet>46</a>    partition_type: StructType,
+<a href=#47 id=47 data-nosnippet>47</a>    partition_arrow_type: DataType,
+<a href=#48 id=48 data-nosnippet>48</a>}
+<a href=#49 id=49 data-nosnippet>49</a>
+<a href=#50 id=50 data-nosnippet>50</a><span class="comment">// # TODO
+<a href=#51 id=51 data-nosnippet>51</a>// Remove this after partition writer supported.
+<a href=#52 id=52 data-nosnippet>52</a></span><span class="attr">#[allow(dead_code)]
+<a href=#53 id=53 data-nosnippet>53</a></span><span class="kw">impl </span>RecordBatchPartitionSplitter {
+<a href=#54 id=54 data-nosnippet>54</a>    <span class="kw">pub fn </span>new(
+<a href=#55 id=55 data-nosnippet>55</a>        input_schema: ArrowSchemaRef,
+<a href=#56 id=56 data-nosnippet>56</a>        iceberg_schema: SchemaRef,
+<a href=#57 id=57 data-nosnippet>57</a>        partition_spec: PartitionSpecRef,
+<a href=#58 id=58 data-nosnippet>58</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;<span class="self">Self</span>&gt; {
+<a href=#59 id=59 data-nosnippet>59</a>        <span class="kw">let </span>projector = RecordBatchProjector::new(
+<a href=#60 id=60 data-nosnippet>60</a>            input_schema,
+<a href=#61 id=61 data-nosnippet>61</a>            <span class="kw-2">&amp;</span>partition_spec
+<a href=#62 id=62 data-nosnippet>62</a>                .fields()
+<a href=#63 id=63 data-nosnippet>63</a>                .iter()
+<a href=#64 id=64 data-nosnippet>64</a>                .map(|field| field.source_id)
+<a href=#65 id=65 data-nosnippet>65</a>                .collect::&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;(),
+<a href=#66 id=66 data-nosnippet>66</a>            <span class="comment">// The source columns, selected by ids, must be a primitive type and cannot be contained in a map or list, but may be nested in a struct.
+<a href=#67 id=67 data-nosnippet>67</a>            // ref: https://iceberg.apache.org/spec/#partitioning
+<a href=#68 id=68 data-nosnippet>68</a>            </span>|field| {
+<a href=#69 id=69 data-nosnippet>69</a>                <span class="kw">if </span>!field.data_type().is_primitive() {
+<a href=#70 id=70 data-nosnippet>70</a>                    <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="prelude-val">None</span>);
+<a href=#71 id=71 data-nosnippet>71</a>                }
+<a href=#72 id=72 data-nosnippet>72</a>                field
+<a href=#73 id=73 data-nosnippet>73</a>                    .metadata()
+<a href=#74 id=74 data-nosnippet>74</a>                    .get(PARQUET_FIELD_ID_META_KEY)
+<a href=#75 id=75 data-nosnippet>75</a>                    .map(|s| {
+<a href=#76 id=76 data-nosnippet>76</a>                        s.parse::&lt;i64&gt;()
+<a href=#77 id=77 data-nosnippet>77</a>                            .map_err(|e| Error::new(ErrorKind::Unexpected, e.to_string()))
+<a href=#78 id=78 data-nosnippet>78</a>                    })
+<a href=#79 id=79 data-nosnippet>79</a>                    .transpose()
+<a href=#80 id=80 data-nosnippet>80</a>            },
+<a href=#81 id=81 data-nosnippet>81</a>            |<span class="kw">_</span>| <span class="bool-val">true</span>,
+<a href=#82 id=82 data-nosnippet>82</a>        )<span class="question-mark">?</span>;
+<a href=#83 id=83 data-nosnippet>83</a>        <span class="kw">let </span>transform_functions = partition_spec
+<a href=#84 id=84 data-nosnippet>84</a>            .fields()
+<a href=#85 id=85 data-nosnippet>85</a>            .iter()
+<a href=#86 id=86 data-nosnippet>86</a>            .map(|field| create_transform_function(<span class="kw-2">&amp;</span>field.transform))
+<a href=#87 id=87 data-nosnippet>87</a>            .collect::&lt;<span class="prelude-ty">Result</span>&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;&gt;()<span class="question-mark">?</span>;
+<a href=#88 id=88 data-nosnippet>88</a>
+<a href=#89 id=89 data-nosnippet>89</a>        <span class="kw">let </span>partition_type = partition_spec.partition_type(<span class="kw-2">&amp;</span>iceberg_schema)<span class="question-mark">?</span>;
+<a href=#90 id=90 data-nosnippet>90</a>        <span class="kw">let </span>partition_arrow_type = type_to_arrow_type(<span class="kw-2">&amp;</span>Type::Struct(partition_type.clone()))<span class="question-mark">?</span>;
+<a href=#91 id=91 data-nosnippet>91</a>
+<a href=#92 id=92 data-nosnippet>92</a>        <span class="prelude-val">Ok</span>(<span class="self">Self </span>{
+<a href=#93 id=93 data-nosnippet>93</a>            schema: iceberg_schema,
+<a href=#94 id=94 data-nosnippet>94</a>            partition_spec,
+<a href=#95 id=95 data-nosnippet>95</a>            projector,
+<a href=#96 id=96 data-nosnippet>96</a>            transform_functions,
+<a href=#97 id=97 data-nosnippet>97</a>            partition_type,
+<a href=#98 id=98 data-nosnippet>98</a>            partition_arrow_type,
+<a href=#99 id=99 data-nosnippet>99</a>        })
+<a href=#100 id=100 data-nosnippet>100</a>    }
+<a href=#101 id=101 data-nosnippet>101</a>
+<a href=#102 id=102 data-nosnippet>102</a>    <span class="kw">fn </span>partition_columns_to_struct(<span class="kw-2">&amp;</span><span class="self">self</span>, partition_columns: Vec&lt;ArrayRef&gt;) -&gt; <span class="prelude-ty">Result</span>&lt;Vec&lt;Struct&gt;&gt; {
+<a href=#103 id=103 data-nosnippet>103</a>        <span class="kw">let </span>arrow_struct_array = {
+<a href=#104 id=104 data-nosnippet>104</a>            <span class="kw">let </span>partition_arrow_fields = {
+<a href=#105 id=105 data-nosnippet>105</a>                <span class="kw">let </span>DataType::Struct(fields) = <span class="kw-2">&amp;</span><span class="self">self</span>.partition_arrow_type <span class="kw">else </span>{
+<a href=#106 id=106 data-nosnippet>106</a>                    <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#107 id=107 data-nosnippet>107</a>                        ErrorKind::DataInvalid,
+<a href=#108 id=108 data-nosnippet>108</a>                        <span class="string">"The partition arrow type is not a struct type"</span>,
+<a href=#109 id=109 data-nosnippet>109</a>                    ));
+<a href=#110 id=110 data-nosnippet>110</a>                };
+<a href=#111 id=111 data-nosnippet>111</a>                fields.clone()
+<a href=#112 id=112 data-nosnippet>112</a>            };
+<a href=#113 id=113 data-nosnippet>113</a>            Arc::new(StructArray::try_new(
+<a href=#114 id=114 data-nosnippet>114</a>                partition_arrow_fields,
+<a href=#115 id=115 data-nosnippet>115</a>                partition_columns,
+<a href=#116 id=116 data-nosnippet>116</a>                <span class="prelude-val">None</span>,
+<a href=#117 id=117 data-nosnippet>117</a>            )<span class="question-mark">?</span>) <span class="kw">as </span>ArrayRef
+<a href=#118 id=118 data-nosnippet>118</a>        };
+<a href=#119 id=119 data-nosnippet>119</a>        <span class="kw">let </span>struct_array = {
+<a href=#120 id=120 data-nosnippet>120</a>            <span class="kw">let </span>struct_array = arrow_struct_to_literal(<span class="kw-2">&amp;</span>arrow_struct_array, <span class="kw-2">&amp;</span><span class="self">self</span>.partition_type)<span class="question-mark">?</span>;
+<a href=#121 id=121 data-nosnippet>121</a>            struct_array
+<a href=#122 id=122 data-nosnippet>122</a>                .into_iter()
+<a href=#123 id=123 data-nosnippet>123</a>                .map(|s| {
+<a href=#124 id=124 data-nosnippet>124</a>                    <span class="kw">if let </span><span class="prelude-val">Some</span>(s) = s {
+<a href=#125 id=125 data-nosnippet>125</a>                        <span class="kw">if let </span>Literal::Struct(s) = s {
+<a href=#126 id=126 data-nosnippet>126</a>                            <span class="prelude-val">Ok</span>(s)
+<a href=#127 id=127 data-nosnippet>127</a>                        } <span class="kw">else </span>{
+<a href=#128 id=128 data-nosnippet>128</a>                            <span class="prelude-val">Err</span>(Error::new(
+<a href=#129 id=129 data-nosnippet>129</a>                                ErrorKind::DataInvalid,
+<a href=#130 id=130 data-nosnippet>130</a>                                <span class="string">"The struct is not a struct literal"</span>,
+<a href=#131 id=131 data-nosnippet>131</a>                            ))
+<a href=#132 id=132 data-nosnippet>132</a>                        }
+<a href=#133 id=133 data-nosnippet>133</a>                    } <span class="kw">else </span>{
+<a href=#134 id=134 data-nosnippet>134</a>                        <span class="prelude-val">Err</span>(Error::new(ErrorKind::DataInvalid, <span class="string">"The struct is null"</span>))
+<a href=#135 id=135 data-nosnippet>135</a>                    }
+<a href=#136 id=136 data-nosnippet>136</a>                })
+<a href=#137 id=137 data-nosnippet>137</a>                .collect::&lt;<span class="prelude-ty">Result</span>&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;&gt;()<span class="question-mark">?
+<a href=#138 id=138 data-nosnippet>138</a>        </span>};
+<a href=#139 id=139 data-nosnippet>139</a>
+<a href=#140 id=140 data-nosnippet>140</a>        <span class="prelude-val">Ok</span>(struct_array)
+<a href=#141 id=141 data-nosnippet>141</a>    }
+<a href=#142 id=142 data-nosnippet>142</a>
+<a href=#143 id=143 data-nosnippet>143</a>    <span class="doccomment">/// Split the record batch into multiple record batches based on the partition spec.
+<a href=#144 id=144 data-nosnippet>144</a>    </span><span class="kw">pub fn </span>split(<span class="kw-2">&amp;</span><span class="self">self</span>, batch: <span class="kw-2">&amp;</span>RecordBatch) -&gt; <span class="prelude-ty">Result</span>&lt;Vec&lt;(Struct, RecordBatch)&gt;&gt; {
+<a href=#145 id=145 data-nosnippet>145</a>        <span class="kw">let </span>source_columns = <span class="self">self</span>.projector.project_column(batch.columns())<span class="question-mark">?</span>;
+<a href=#146 id=146 data-nosnippet>146</a>        <span class="kw">let </span>partition_columns = source_columns
+<a href=#147 id=147 data-nosnippet>147</a>            .into_iter()
+<a href=#148 id=148 data-nosnippet>148</a>            .zip_eq(<span class="self">self</span>.transform_functions.iter())
+<a href=#149 id=149 data-nosnippet>149</a>            .map(|(source_column, transform_function)| transform_function.transform(source_column))
+<a href=#150 id=150 data-nosnippet>150</a>            .collect::&lt;<span class="prelude-ty">Result</span>&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;&gt;()<span class="question-mark">?</span>;
+<a href=#151 id=151 data-nosnippet>151</a>
+<a href=#152 id=152 data-nosnippet>152</a>        <span class="kw">let </span>partition_structs = <span class="self">self</span>.partition_columns_to_struct(partition_columns)<span class="question-mark">?</span>;
+<a href=#153 id=153 data-nosnippet>153</a>
+<a href=#154 id=154 data-nosnippet>154</a>        <span class="comment">// Group the batch by row value.
+<a href=#155 id=155 data-nosnippet>155</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>group_ids = HashMap::new();
+<a href=#156 id=156 data-nosnippet>156</a>        partition_structs
+<a href=#157 id=157 data-nosnippet>157</a>            .iter()
+<a href=#158 id=158 data-nosnippet>158</a>            .enumerate()
+<a href=#159 id=159 data-nosnippet>159</a>            .for_each(|(row_id, row)| {
+<a href=#160 id=160 data-nosnippet>160</a>                group_ids.entry(row.clone()).or_insert(<span class="macro">vec!</span>[]).push(row_id);
+<a href=#161 id=161 data-nosnippet>161</a>            });
+<a href=#162 id=162 data-nosnippet>162</a>
+<a href=#163 id=163 data-nosnippet>163</a>        <span class="comment">// Partition the batch with same partition partition_values
+<a href=#164 id=164 data-nosnippet>164</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>partition_batches = Vec::with_capacity(group_ids.len());
+<a href=#165 id=165 data-nosnippet>165</a>        <span class="kw">for </span>(row, row_ids) <span class="kw">in </span>group_ids.into_iter() {
+<a href=#166 id=166 data-nosnippet>166</a>            <span class="comment">// generate the bool filter array from column_ids
+<a href=#167 id=167 data-nosnippet>167</a>            </span><span class="kw">let </span>filter_array: BooleanArray = {
+<a href=#168 id=168 data-nosnippet>168</a>                <span class="kw">let </span><span class="kw-2">mut </span>filter = <span class="macro">vec!</span>[<span class="bool-val">false</span>; batch.num_rows()];
+<a href=#169 id=169 data-nosnippet>169</a>                row_ids.into_iter().for_each(|row_id| {
+<a href=#170 id=170 data-nosnippet>170</a>                    filter[row_id] = <span class="bool-val">true</span>;
+<a href=#171 id=171 data-nosnippet>171</a>                });
+<a href=#172 id=172 data-nosnippet>172</a>                filter.into()
+<a href=#173 id=173 data-nosnippet>173</a>            };
+<a href=#174 id=174 data-nosnippet>174</a>
+<a href=#175 id=175 data-nosnippet>175</a>            <span class="comment">// filter the RecordBatch
+<a href=#176 id=176 data-nosnippet>176</a>            </span>partition_batches.push((row, filter_record_batch(batch, <span class="kw-2">&amp;</span>filter_array)<span class="question-mark">?</span>));
+<a href=#177 id=177 data-nosnippet>177</a>        }
+<a href=#178 id=178 data-nosnippet>178</a>
+<a href=#179 id=179 data-nosnippet>179</a>        <span class="prelude-val">Ok</span>(partition_batches)
+<a href=#180 id=180 data-nosnippet>180</a>    }
+<a href=#181 id=181 data-nosnippet>181</a>}
+<a href=#182 id=182 data-nosnippet>182</a>
+<a href=#183 id=183 data-nosnippet>183</a><span class="attr">#[cfg(test)]
+<a href=#184 id=184 data-nosnippet>184</a></span><span class="kw">mod </span>tests {
+<a href=#185 id=185 data-nosnippet>185</a>    <span class="kw">use </span>std::sync::Arc;
+<a href=#186 id=186 data-nosnippet>186</a>
+<a href=#187 id=187 data-nosnippet>187</a>    <span class="kw">use </span>arrow_array::{Int32Array, RecordBatch, StringArray};
+<a href=#188 id=188 data-nosnippet>188</a>
+<a href=#189 id=189 data-nosnippet>189</a>    <span class="kw">use super</span>::<span class="kw-2">*</span>;
+<a href=#190 id=190 data-nosnippet>190</a>    <span class="kw">use </span><span class="kw">crate</span>::arrow::schema_to_arrow_schema;
+<a href=#191 id=191 data-nosnippet>191</a>    <span class="kw">use </span><span class="kw">crate</span>::spec::{
+<a href=#192 id=192 data-nosnippet>192</a>        NestedField, PartitionSpecBuilder, PrimitiveLiteral, Schema, Transform,
+<a href=#193 id=193 data-nosnippet>193</a>        UnboundPartitionField,
+<a href=#194 id=194 data-nosnippet>194</a>    };
+<a href=#195 id=195 data-nosnippet>195</a>
+<a href=#196 id=196 data-nosnippet>196</a>    <span class="attr">#[test]
+<a href=#197 id=197 data-nosnippet>197</a>    </span><span class="kw">fn </span>test_record_batch_partition_split() {
+<a href=#198 id=198 data-nosnippet>198</a>        <span class="kw">let </span>schema = Arc::new(
+<a href=#199 id=199 data-nosnippet>199</a>            Schema::builder()
+<a href=#200 id=200 data-nosnippet>200</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#201 id=201 data-nosnippet>201</a>                    NestedField::required(
+<a href=#202 id=202 data-nosnippet>202</a>                        <span class="number">1</span>,
+<a href=#203 id=203 data-nosnippet>203</a>                        <span class="string">"id"</span>,
+<a href=#204 id=204 data-nosnippet>204</a>                        Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::Int),
+<a href=#205 id=205 data-nosnippet>205</a>                    )
+<a href=#206 id=206 data-nosnippet>206</a>                    .into(),
+<a href=#207 id=207 data-nosnippet>207</a>                    NestedField::required(
+<a href=#208 id=208 data-nosnippet>208</a>                        <span class="number">2</span>,
+<a href=#209 id=209 data-nosnippet>209</a>                        <span class="string">"name"</span>,
+<a href=#210 id=210 data-nosnippet>210</a>                        Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::String),
+<a href=#211 id=211 data-nosnippet>211</a>                    )
+<a href=#212 id=212 data-nosnippet>212</a>                    .into(),
+<a href=#213 id=213 data-nosnippet>213</a>                ])
+<a href=#214 id=214 data-nosnippet>214</a>                .build()
+<a href=#215 id=215 data-nosnippet>215</a>                .unwrap(),
+<a href=#216 id=216 data-nosnippet>216</a>        );
+<a href=#217 id=217 data-nosnippet>217</a>        <span class="kw">let </span>partition_spec = Arc::new(
+<a href=#218 id=218 data-nosnippet>218</a>            PartitionSpecBuilder::new(schema.clone())
+<a href=#219 id=219 data-nosnippet>219</a>                .with_spec_id(<span class="number">1</span>)
+<a href=#220 id=220 data-nosnippet>220</a>                .add_unbound_field(UnboundPartitionField {
+<a href=#221 id=221 data-nosnippet>221</a>                    source_id: <span class="number">1</span>,
+<a href=#222 id=222 data-nosnippet>222</a>                    field_id: <span class="prelude-val">None</span>,
+<a href=#223 id=223 data-nosnippet>223</a>                    name: <span class="string">"id_bucket"</span>.to_string(),
+<a href=#224 id=224 data-nosnippet>224</a>                    transform: Transform::Identity,
+<a href=#225 id=225 data-nosnippet>225</a>                })
+<a href=#226 id=226 data-nosnippet>226</a>                .unwrap()
+<a href=#227 id=227 data-nosnippet>227</a>                .build()
+<a href=#228 id=228 data-nosnippet>228</a>                .unwrap(),
+<a href=#229 id=229 data-nosnippet>229</a>        );
+<a href=#230 id=230 data-nosnippet>230</a>        <span class="kw">let </span>input_schema = Arc::new(schema_to_arrow_schema(<span class="kw-2">&amp;</span>schema).unwrap());
+<a href=#231 id=231 data-nosnippet>231</a>        <span class="kw">let </span>partition_splitter =
+<a href=#232 id=232 data-nosnippet>232</a>            RecordBatchPartitionSplitter::new(input_schema.clone(), schema.clone(), partition_spec)
+<a href=#233 id=233 data-nosnippet>233</a>                .expect(<span class="string">"Failed to create splitter"</span>);
+<a href=#234 id=234 data-nosnippet>234</a>
+<a href=#235 id=235 data-nosnippet>235</a>        <span class="kw">let </span>id_array = Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">1</span>, <span class="number">3</span>, <span class="number">2</span>, <span class="number">3</span>, <span class="number">1</span>]);
+<a href=#236 id=236 data-nosnippet>236</a>        <span class="kw">let </span>data_array = StringArray::from(<span class="macro">vec!</span>[<span class="string">"a"</span>, <span class="string">"b"</span>, <span class="string">"c"</span>, <span class="string">"d"</span>, <span class="string">"e"</span>, <span class="string">"f"</span>, <span class="string">"g"</span>]);
+<a href=#237 id=237 data-nosnippet>237</a>        <span class="kw">let </span>batch = RecordBatch::try_new(input_schema.clone(), <span class="macro">vec!</span>[
+<a href=#238 id=238 data-nosnippet>238</a>            Arc::new(id_array),
+<a href=#239 id=239 data-nosnippet>239</a>            Arc::new(data_array),
+<a href=#240 id=240 data-nosnippet>240</a>        ])
+<a href=#241 id=241 data-nosnippet>241</a>        .expect(<span class="string">"Failed to create RecordBatch"</span>);
+<a href=#242 id=242 data-nosnippet>242</a>
+<a href=#243 id=243 data-nosnippet>243</a>        <span class="kw">let </span><span class="kw-2">mut </span>partitioned_batches = partition_splitter
+<a href=#244 id=244 data-nosnippet>244</a>            .split(<span class="kw-2">&amp;</span>batch)
+<a href=#245 id=245 data-nosnippet>245</a>            .expect(<span class="string">"Failed to split RecordBatch"</span>);
+<a href=#246 id=246 data-nosnippet>246</a>        partitioned_batches.sort_by_key(|(row, <span class="kw">_</span>)| {
+<a href=#247 id=247 data-nosnippet>247</a>            <span class="kw">if let </span>PrimitiveLiteral::Int(i) = row.fields()[<span class="number">0</span>]
+<a href=#248 id=248 data-nosnippet>248</a>                .as_ref()
+<a href=#249 id=249 data-nosnippet>249</a>                .unwrap()
+<a href=#250 id=250 data-nosnippet>250</a>                .as_primitive_literal()
+<a href=#251 id=251 data-nosnippet>251</a>                .unwrap()
+<a href=#252 id=252 data-nosnippet>252</a>            {
+<a href=#253 id=253 data-nosnippet>253</a>                i
+<a href=#254 id=254 data-nosnippet>254</a>            } <span class="kw">else </span>{
+<a href=#255 id=255 data-nosnippet>255</a>                <span class="macro">panic!</span>(<span class="string">"The partition value is not a int"</span>);
+<a href=#256 id=256 data-nosnippet>256</a>            }
+<a href=#257 id=257 data-nosnippet>257</a>        });
+<a href=#258 id=258 data-nosnippet>258</a>        <span class="macro">assert_eq!</span>(partitioned_batches.len(), <span class="number">3</span>);
+<a href=#259 id=259 data-nosnippet>259</a>        {
+<a href=#260 id=260 data-nosnippet>260</a>            <span class="comment">// check the first partition
+<a href=#261 id=261 data-nosnippet>261</a>            </span><span class="kw">let </span>expected_id_array = Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">1</span>, <span class="number">1</span>]);
+<a href=#262 id=262 data-nosnippet>262</a>            <span class="kw">let </span>expected_data_array = StringArray::from(<span class="macro">vec!</span>[<span class="string">"a"</span>, <span class="string">"c"</span>, <span class="string">"g"</span>]);
+<a href=#263 id=263 data-nosnippet>263</a>            <span class="kw">let </span>expected_batch = RecordBatch::try_new(input_schema.clone(), <span class="macro">vec!</span>[
+<a href=#264 id=264 data-nosnippet>264</a>                Arc::new(expected_id_array),
+<a href=#265 id=265 data-nosnippet>265</a>                Arc::new(expected_data_array),
+<a href=#266 id=266 data-nosnippet>266</a>            ])
+<a href=#267 id=267 data-nosnippet>267</a>            .expect(<span class="string">"Failed to create expected RecordBatch"</span>);
+<a href=#268 id=268 data-nosnippet>268</a>            <span class="macro">assert_eq!</span>(partitioned_batches[<span class="number">0</span>].<span class="number">1</span>, expected_batch);
+<a href=#269 id=269 data-nosnippet>269</a>        }
+<a href=#270 id=270 data-nosnippet>270</a>        {
+<a href=#271 id=271 data-nosnippet>271</a>            <span class="comment">// check the second partition
+<a href=#272 id=272 data-nosnippet>272</a>            </span><span class="kw">let </span>expected_id_array = Int32Array::from(<span class="macro">vec!</span>[<span class="number">2</span>, <span class="number">2</span>]);
+<a href=#273 id=273 data-nosnippet>273</a>            <span class="kw">let </span>expected_data_array = StringArray::from(<span class="macro">vec!</span>[<span class="string">"b"</span>, <span class="string">"e"</span>]);
+<a href=#274 id=274 data-nosnippet>274</a>            <span class="kw">let </span>expected_batch = RecordBatch::try_new(input_schema.clone(), <span class="macro">vec!</span>[
+<a href=#275 id=275 data-nosnippet>275</a>                Arc::new(expected_id_array),
+<a href=#276 id=276 data-nosnippet>276</a>                Arc::new(expected_data_array),
+<a href=#277 id=277 data-nosnippet>277</a>            ])
+<a href=#278 id=278 data-nosnippet>278</a>            .expect(<span class="string">"Failed to create expected RecordBatch"</span>);
+<a href=#279 id=279 data-nosnippet>279</a>            <span class="macro">assert_eq!</span>(partitioned_batches[<span class="number">1</span>].<span class="number">1</span>, expected_batch);
+<a href=#280 id=280 data-nosnippet>280</a>        }
+<a href=#281 id=281 data-nosnippet>281</a>        {
+<a href=#282 id=282 data-nosnippet>282</a>            <span class="comment">// check the third partition
+<a href=#283 id=283 data-nosnippet>283</a>            </span><span class="kw">let </span>expected_id_array = Int32Array::from(<span class="macro">vec!</span>[<span class="number">3</span>, <span class="number">3</span>]);
+<a href=#284 id=284 data-nosnippet>284</a>            <span class="kw">let </span>expected_data_array = StringArray::from(<span class="macro">vec!</span>[<span class="string">"d"</span>, <span class="string">"f"</span>]);
+<a href=#285 id=285 data-nosnippet>285</a>            <span class="kw">let </span>expected_batch = RecordBatch::try_new(input_schema.clone(), <span class="macro">vec!</span>[
+<a href=#286 id=286 data-nosnippet>286</a>                Arc::new(expected_id_array),
+<a href=#287 id=287 data-nosnippet>287</a>                Arc::new(expected_data_array),
+<a href=#288 id=288 data-nosnippet>288</a>            ])
+<a href=#289 id=289 data-nosnippet>289</a>            .expect(<span class="string">"Failed to create expected RecordBatch"</span>);
+<a href=#290 id=290 data-nosnippet>290</a>            <span class="macro">assert_eq!</span>(partitioned_batches[<span class="number">2</span>].<span class="number">1</span>, expected_batch);
+<a href=#291 id=291 data-nosnippet>291</a>        }
+<a href=#292 id=292 data-nosnippet>292</a>
+<a href=#293 id=293 data-nosnippet>293</a>        <span class="kw">let </span>partition_values = partitioned_batches
+<a href=#294 id=294 data-nosnippet>294</a>            .iter()
+<a href=#295 id=295 data-nosnippet>295</a>            .map(|(row, <span class="kw">_</span>)| row.clone())
+<a href=#296 id=296 data-nosnippet>296</a>            .collect::&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;();
+<a href=#297 id=297 data-nosnippet>297</a>        <span class="comment">// check partition value is struct(1), struct(2), struct(3)
+<a href=#298 id=298 data-nosnippet>298</a>        </span><span class="macro">assert_eq!</span>(partition_values, <span class="macro">vec!</span>[
+<a href=#299 id=299 data-nosnippet>299</a>            Struct::from_iter(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(Literal::int(<span class="number">1</span>))]),
+<a href=#300 id=300 data-nosnippet>300</a>            Struct::from_iter(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(Literal::int(<span class="number">2</span>))]),
+<a href=#301 id=301 data-nosnippet>301</a>            Struct::from_iter(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(Literal::int(<span class="number">3</span>))]),
+<a href=#302 id=302 data-nosnippet>302</a>        ]);
+<a href=#303 id=303 data-nosnippet>303</a>    }
+<a href=#304 id=304 data-nosnippet>304</a>}</code></pre></div></section></main></body></html>
\ No newline at end of file