parquet_derive/
parquet_field.rs

1// Licensed to the Apache Software Foundation (ASF) under one
2// or more contributor license agreements.  See the NOTICE file
3// distributed with this work for additional information
4// regarding copyright ownership.  The ASF licenses this file
5// to you under the Apache License, Version 2.0 (the
6// "License"); you may not use this file except in compliance
7// with the License.  You may obtain a copy of the License at
8//
9//   http://www.apache.org/licenses/LICENSE-2.0
10//
11// Unless required by applicable law or agreed to in writing,
12// software distributed under the License is distributed on an
13// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14// KIND, either express or implied.  See the License for the
15// specific language governing permissions and limitations
16// under the License.
17
18use syn::ext::IdentExt;
19
20/// Parquet physical types used while selecting the code emitted by the macro.
21/// Keeping this local avoids compiling `parquet` for the proc-macro host.
22#[derive(Debug, Clone, Copy, PartialEq, Eq)]
23enum PhysicalType {
24    Boolean,
25    Int32,
26    Int64,
27    Float,
28    Double,
29    ByteArray,
30    FixedLenByteArray,
31}
32
33#[derive(Debug, PartialEq)]
34pub struct Field {
35    ident: syn::Ident,
36    ty: Type,
37    is_a_byte_buf: bool,
38    third_party_type: Option<ThirdPartyType>,
39}
40
41/// Use third party libraries, detected
42/// at compile time. These libraries will
43/// be written to parquet as their preferred
44/// physical type.
45///
46///   ChronoNaiveDateTime is written as i64
47///   ChronoNaiveDate is written as i32
48#[derive(Debug, PartialEq)]
49enum ThirdPartyType {
50    ChronoNaiveDateTime,
51    ChronoNaiveDate,
52    Uuid,
53}
54
55impl Field {
56    pub fn from(f: &syn::Field) -> Self {
57        let ty = Type::from(f);
58        let is_a_byte_buf = ty.physical_type() == PhysicalType::ByteArray;
59
60        let third_party_type = match &ty.last_part()[..] {
61            "NaiveDateTime" => Some(ThirdPartyType::ChronoNaiveDateTime),
62            "NaiveDate" => Some(ThirdPartyType::ChronoNaiveDate),
63            "Uuid" => Some(ThirdPartyType::Uuid),
64            _ => None,
65        };
66
67        Field {
68            ident: f
69                .ident
70                .clone()
71                .expect("Only structs with named fields are currently supported"),
72            ty,
73            is_a_byte_buf,
74            third_party_type,
75        }
76    }
77
78    /// Takes the parsed field of the struct and emits a valid
79    /// column writer snippet. Should match exactly what you
80    /// would write by hand.
81    ///
82    /// Can only generate writers for basic structs, for example:
83    ///
84    /// struct Record {
85    ///   a_bool: bool,
86    ///   maybe_a_bool: `Option<bool>`
87    /// }
88    ///
89    /// but not
90    ///
91    /// struct UnsupportedNestedRecord {
92    ///   a_property: bool,
93    ///   nested_record: Record
94    /// }
95    ///
96    /// because this parsing logic is not sophisticated enough for definition
97    /// levels beyond 2.
98    pub fn writer_snippet(&self) -> proc_macro2::TokenStream {
99        let ident = &self.ident;
100        let column_writer = self.ty.column_writer();
101
102        let vals_builder = match &self.ty {
103            Type::TypePath(_) => self.copied_direct_vals(),
104            Type::Option(first_type) => match **first_type {
105                Type::TypePath(_) => self.option_into_vals(),
106                Type::Reference(_, ref second_type) => match **second_type {
107                    Type::TypePath(_) => self.option_into_vals(),
108                    _ => unimplemented!("Unsupported type encountered"),
109                },
110                Type::Vec(ref first_type) => match **first_type {
111                    Type::TypePath(_) => self.option_into_vals(),
112                    _ => unimplemented!("Unsupported type encountered"),
113                },
114                ref f => unimplemented!("Unsupported: {:#?}", f),
115            },
116            Type::Reference(_, first_type) => match **first_type {
117                Type::TypePath(_) => self.copied_direct_vals(),
118                Type::Option(ref second_type) => match **second_type {
119                    Type::TypePath(_) => self.option_into_vals(),
120                    Type::Reference(_, ref second_type) => match **second_type {
121                        Type::TypePath(_) => self.option_into_vals(),
122                        Type::Slice(ref second_type) => match **second_type {
123                            Type::TypePath(_) => self.option_into_vals(),
124                            ref f => unimplemented!("Unsupported: {:#?}", f),
125                        },
126                        _ => unimplemented!("Unsupported type encountered"),
127                    },
128                    Type::Vec(ref first_type) => match **first_type {
129                        Type::TypePath(_) => self.option_into_vals(),
130                        _ => unimplemented!("Unsupported type encountered"),
131                    },
132                    ref f => unimplemented!("Unsupported: {:#?}", f),
133                },
134                Type::Slice(ref second_type) => match **second_type {
135                    Type::TypePath(_) => self.copied_direct_vals(),
136                    ref f => unimplemented!("Unsupported: {:#?}", f),
137                },
138                ref f => unimplemented!("Unsupported: {:#?}", f),
139            },
140            Type::Vec(first_type) => match **first_type {
141                Type::TypePath(_) => self.copied_direct_vals(),
142                ref f => unimplemented!("Unsupported: {:#?}", f),
143            },
144            f => unimplemented!("Unsupported: {:#?}", f),
145        };
146
147        let definition_levels = match &self.ty {
148            Type::TypePath(_) => None,
149            Type::Option(first_type) => match **first_type {
150                Type::TypePath(_) => Some(self.optional_definition_levels()),
151                Type::Option(_) => unimplemented!("Unsupported nesting encountered"),
152                Type::Reference(_, ref second_type)
153                | Type::Vec(ref second_type)
154                | Type::Array(ref second_type, _)
155                | Type::Slice(ref second_type) => match **second_type {
156                    Type::TypePath(_) => Some(self.optional_definition_levels()),
157                    _ => unimplemented!("Unsupported nesting encountered"),
158                },
159            },
160            Type::Reference(_, first_type)
161            | Type::Vec(first_type)
162            | Type::Array(first_type, _)
163            | Type::Slice(first_type) => match **first_type {
164                Type::TypePath(_) => None,
165                Type::Vec(ref second_type)
166                | Type::Array(ref second_type, _)
167                | Type::Slice(ref second_type) => match **second_type {
168                    Type::TypePath(_) => None,
169                    Type::Reference(_, ref third_type) => match **third_type {
170                        Type::TypePath(_) => None,
171                        _ => unimplemented!("Unsupported definition encountered"),
172                    },
173                    _ => unimplemented!("Unsupported definition encountered"),
174                },
175                Type::Reference(_, ref second_type) | Type::Option(ref second_type) => {
176                    match **second_type {
177                        Type::TypePath(_) => Some(self.optional_definition_levels()),
178                        Type::Vec(ref third_type)
179                        | Type::Array(ref third_type, _)
180                        | Type::Slice(ref third_type) => match **third_type {
181                            Type::TypePath(_) => Some(self.optional_definition_levels()),
182                            Type::Reference(_, ref fourth_type) => match **fourth_type {
183                                Type::TypePath(_) => Some(self.optional_definition_levels()),
184                                _ => unimplemented!("Unsupported definition encountered"),
185                            },
186                            _ => unimplemented!("Unsupported definition encountered"),
187                        },
188                        Type::Reference(_, ref third_type) => match **third_type {
189                            Type::TypePath(_) => Some(self.optional_definition_levels()),
190                            Type::Slice(ref fourth_type) => match **fourth_type {
191                                Type::TypePath(_) => Some(self.optional_definition_levels()),
192                                _ => unimplemented!("Unsupported definition encountered"),
193                            },
194                            _ => unimplemented!("Unsupported definition encountered"),
195                        },
196                        _ => unimplemented!("Unsupported definition encountered"),
197                    }
198                }
199            },
200        };
201
202        // "vals" is the run of primitive data being written for the column
203        // "definition_levels" is a vector of bools which controls whether a value is missing or present
204        // this TokenStream is only one part of the code for writing a column and
205        // it relies on values calculated in prior code snippets, namely "definition_levels" and "vals_builder".
206        // All the context is put together in this functions final quote and
207        // this expression just switches between non-nullable and nullable write statements
208        let write_batch_expr = if definition_levels.is_some() {
209            quote! {
210                if let #column_writer(typed) = column_writer.untyped() {
211                    typed.write_batch(&vals[..], Some(&definition_levels[..]), None)?;
212                } else {
213                    panic!("Schema and struct disagree on type for {}", stringify!{#ident})
214                }
215            }
216        } else {
217            quote! {
218                if let #column_writer(typed) = column_writer.untyped() {
219                    typed.write_batch(&vals[..], None, None)?;
220                } else {
221                    panic!("Schema and struct disagree on type for {}", stringify!{#ident})
222                }
223            }
224        };
225
226        quote! {
227            {
228                #definition_levels
229
230                #vals_builder
231
232                #write_batch_expr
233            }
234        }
235    }
236
237    /// Takes the parsed field of the struct and emits a valid
238    /// column reader snippet. Should match exactly what you
239    /// would write by hand.
240    ///
241    /// Can only generate writers for basic structs, for example:
242    ///
243    /// struct Record {
244    ///   a_bool: bool
245    /// }
246    ///
247    /// but not
248    ///
249    /// struct UnsupportedNestedRecord {
250    ///   a_property: bool,
251    ///   nested_record: Record
252    /// }
253    ///
254    /// because this parsing logic is not sophisticated enough for definition
255    /// levels beyond 2.
256    ///
257    /// `Option` types and references not supported, but the column itself can be nullable
258    /// (i.e., def_level==1), as long as the values are all valid.
259    pub fn reader_snippet(&self) -> proc_macro2::TokenStream {
260        let ident = &self.ident;
261        let column_reader = self.ty.column_reader();
262
263        // generate the code to read the column into a vector `vals`
264        let write_batch_expr = quote! {
265            let mut vals = Vec::new();
266            if let #column_reader(mut typed) = column_reader {
267                let mut definition_levels = Vec::new();
268                let (total_num, valid_num, decoded_num) = typed.read_records(
269                    num_records, Some(&mut definition_levels), None, &mut vals)?;
270                if valid_num != decoded_num {
271                    panic!("Support only valid records, found {} null records in column type {}",
272                        decoded_num - valid_num, stringify!{#ident});
273                }
274            } else {
275                panic!("Schema and struct disagree on type for {}", stringify!{#ident});
276            }
277        };
278
279        // generate the code to convert each element of `vals` to the correct type and then write
280        // it to its field in the corresponding struct
281        let vals_writer = match &self.ty {
282            Type::TypePath(_) => self.copied_direct_fields(),
283            Type::Reference(_, first_type) => match **first_type {
284                Type::TypePath(_) => self.copied_direct_fields(),
285                Type::Slice(ref second_type) => match **second_type {
286                    Type::TypePath(_) => self.copied_direct_fields(),
287                    ref f => unimplemented!("Unsupported: {:#?}", f),
288                },
289                ref f => unimplemented!("Unsupported: {:#?}", f),
290            },
291            Type::Vec(first_type) => match **first_type {
292                Type::TypePath(_) => self.copied_direct_fields(),
293                ref f => unimplemented!("Unsupported: {:#?}", f),
294            },
295            f => unimplemented!("Unsupported: {:#?}", f),
296        };
297
298        quote! {
299            {
300                #write_batch_expr
301
302                #vals_writer
303            }
304        }
305    }
306
307    pub fn parquet_type(&self) -> proc_macro2::TokenStream {
308        // TODO: Support group types
309        // TODO: Add length if dealing with fixedlenbinary
310
311        // unraw the identifier, so a raw identifier like `r#type`
312        // becomes a column named `type` in the parquet schema
313        let field_name = self.ident.unraw().to_string();
314        let physical_type = match self.ty.physical_type() {
315            PhysicalType::Boolean => quote! {
316                ::parquet::basic::Type::BOOLEAN
317            },
318            PhysicalType::Int32 => quote! {
319                ::parquet::basic::Type::INT32
320            },
321            PhysicalType::Int64 => quote! {
322                ::parquet::basic::Type::INT64
323            },
324            PhysicalType::Float => quote! {
325                ::parquet::basic::Type::FLOAT
326            },
327            PhysicalType::Double => quote! {
328                ::parquet::basic::Type::DOUBLE
329            },
330            PhysicalType::ByteArray => quote! {
331                ::parquet::basic::Type::BYTE_ARRAY
332            },
333            PhysicalType::FixedLenByteArray => quote! {
334                ::parquet::basic::Type::FIXED_LEN_BYTE_ARRAY
335            },
336        };
337        let logical_type = self.ty.logical_type();
338        let repetition = self.ty.repetition();
339        let converted_type = self.ty.converted_type();
340        let length = self.ty.length();
341
342        let mut builder = quote! {
343            ParquetType::primitive_type_builder(#field_name, #physical_type)
344                .with_logical_type(#logical_type)
345                .with_repetition(#repetition)
346        };
347
348        if let Some(converted_type) = converted_type {
349            builder = quote! { #builder.with_converted_type(#converted_type) };
350        }
351
352        if let Some(length) = length {
353            builder = quote! { #builder.with_length(#length) };
354        }
355
356        quote! {  fields.push(#builder.build().unwrap().into()) }
357    }
358
359    fn option_into_vals(&self) -> proc_macro2::TokenStream {
360        let field_name = &self.ident;
361        let is_a_byte_buf = self.is_a_byte_buf;
362        let is_a_timestamp = self.third_party_type == Some(ThirdPartyType::ChronoNaiveDateTime);
363        let is_a_date = self.third_party_type == Some(ThirdPartyType::ChronoNaiveDate);
364        let is_a_uuid = self.third_party_type == Some(ThirdPartyType::Uuid);
365        let copy_to_vec = !matches!(
366            self.ty.physical_type(),
367            PhysicalType::ByteArray | PhysicalType::FixedLenByteArray
368        );
369
370        let binding = if copy_to_vec {
371            quote! { let Some(inner) = rec.#field_name }
372        } else {
373            quote! { let Some(inner) = &rec.#field_name }
374        };
375
376        let some = if is_a_timestamp {
377            quote! { Some(inner.timestamp_millis()) }
378        } else if is_a_date {
379            quote! { Some(inner.signed_duration_since(::chrono::NaiveDate::from_ymd(1970, 1, 1)).num_days() as i32)  }
380        } else if is_a_uuid {
381            quote! { Some((&inner.to_string()[..]).into()) }
382        } else if is_a_byte_buf {
383            quote! { Some((&inner[..]).into())}
384        } else {
385            // Type might need converting to a physical type
386            match self.ty.physical_type() {
387                PhysicalType::Int32 => quote! { Some(inner as i32) },
388                PhysicalType::Int64 => quote! { Some(inner as i64) },
389                _ => quote! { Some(inner) },
390            }
391        };
392
393        quote! {
394            let vals: Vec<_> = records.iter().filter_map(|rec| {
395                if #binding {
396                    #some
397                } else {
398                    None
399                }
400            }).collect();
401        }
402    }
403
404    // generates code to read `field_name` from each record into a vector `vals`
405    fn copied_direct_vals(&self) -> proc_macro2::TokenStream {
406        let field_name = &self.ident;
407
408        let access = match self.third_party_type {
409            Some(ThirdPartyType::ChronoNaiveDateTime) => {
410                quote! { rec.#field_name.timestamp_millis() }
411            }
412            Some(ThirdPartyType::ChronoNaiveDate) => {
413                quote! { rec.#field_name.signed_duration_since(::chrono::NaiveDate::from_ymd(1970, 1, 1)).num_days() as i32 }
414            }
415            Some(ThirdPartyType::Uuid) => {
416                quote! { rec.#field_name.as_bytes().to_vec().into() }
417            }
418            _ => {
419                if self.is_a_byte_buf {
420                    quote! { (&rec.#field_name[..]).into() }
421                } else {
422                    // Type might need converting to a physical type
423                    match self.ty.physical_type() {
424                        PhysicalType::Int32 => quote! { rec.#field_name as i32 },
425                        PhysicalType::Int64 => quote! { rec.#field_name as i64 },
426                        _ => quote! { rec.#field_name },
427                    }
428                }
429            }
430        };
431
432        quote! {
433            let vals: Vec<_> = records.iter().map(|rec| #access).collect();
434        }
435    }
436
437    // generates code to read a vector `records` into `field_name` for each record
438    fn copied_direct_fields(&self) -> proc_macro2::TokenStream {
439        let field_name = &self.ident;
440
441        let value = match self.third_party_type {
442            Some(ThirdPartyType::ChronoNaiveDateTime) => {
443                quote! { ::chrono::naive::NaiveDateTime::from_timestamp_millis(vals[i]).unwrap() }
444            }
445            Some(ThirdPartyType::ChronoNaiveDate) => {
446                // NaiveDateTime::UNIX_EPOCH.num_days_from_ce() == 719163
447                quote! {
448                    ::chrono::naive::NaiveDate::from_num_days_from_ce_opt(vals[i].saturating_add(719163)).unwrap()
449                }
450            }
451            Some(ThirdPartyType::Uuid) => {
452                quote! { ::uuid::Uuid::from_bytes(vals[i].data().try_into().unwrap()) }
453            }
454            _ => match &self.ty {
455                Type::TypePath(_) => match self.ty.last_part().as_str() {
456                    "String" => quote! { String::from(std::str::from_utf8(vals[i].data())
457                    .expect("invalid UTF-8 sequence")) },
458                    t => {
459                        let s: proc_macro2::TokenStream = t.parse().unwrap();
460                        quote! { vals[i] as #s }
461                    }
462                },
463                Type::Vec(_) => quote! { vals[i].data().to_vec() },
464                f => unimplemented!("Unsupported: {:#?}", f),
465            },
466        };
467
468        quote! {
469            for (i, r) in &mut records[..num_records].iter_mut().enumerate() {
470                r.#field_name = #value;
471            }
472        }
473    }
474
475    fn optional_definition_levels(&self) -> proc_macro2::TokenStream {
476        let field_name = &self.ident;
477
478        quote! {
479            let definition_levels: Vec<i16> = self
480              .iter()
481              .map(|rec| if rec.#field_name.is_some() { 1 } else { 0 })
482              .collect();
483        }
484    }
485}
486
487#[expect(clippy::enum_variant_names)]
488#[derive(Debug, PartialEq)]
489enum Type {
490    Array(Box<Type>, syn::Expr),
491    Option(Box<Type>),
492    Slice(Box<Type>),
493    Vec(Box<Type>),
494    TypePath(syn::Type),
495    Reference(Option<syn::Lifetime>, Box<Type>),
496}
497
498impl Type {
499    /// Takes a rust type and returns the appropriate
500    /// parquet-rs column writer
501    fn column_writer(&self) -> syn::TypePath {
502        match self.physical_type() {
503            PhysicalType::Boolean => {
504                syn::parse_quote!(ColumnWriter::BoolColumnWriter)
505            }
506            PhysicalType::Int32 => syn::parse_quote!(ColumnWriter::Int32ColumnWriter),
507            PhysicalType::Int64 => syn::parse_quote!(ColumnWriter::Int64ColumnWriter),
508            PhysicalType::Float => syn::parse_quote!(ColumnWriter::FloatColumnWriter),
509            PhysicalType::Double => syn::parse_quote!(ColumnWriter::DoubleColumnWriter),
510            PhysicalType::ByteArray => {
511                syn::parse_quote!(ColumnWriter::ByteArrayColumnWriter)
512            }
513            PhysicalType::FixedLenByteArray => {
514                syn::parse_quote!(ColumnWriter::FixedLenByteArrayColumnWriter)
515            }
516        }
517    }
518
519    /// Takes a rust type and returns the appropriate
520    /// parquet-rs column reader
521    fn column_reader(&self) -> syn::TypePath {
522        match self.physical_type() {
523            PhysicalType::Boolean => {
524                syn::parse_quote!(ColumnReader::BoolColumnReader)
525            }
526            PhysicalType::Int32 => syn::parse_quote!(ColumnReader::Int32ColumnReader),
527            PhysicalType::Int64 => syn::parse_quote!(ColumnReader::Int64ColumnReader),
528            PhysicalType::Float => syn::parse_quote!(ColumnReader::FloatColumnReader),
529            PhysicalType::Double => syn::parse_quote!(ColumnReader::DoubleColumnReader),
530            PhysicalType::ByteArray => {
531                syn::parse_quote!(ColumnReader::ByteArrayColumnReader)
532            }
533            PhysicalType::FixedLenByteArray => {
534                syn::parse_quote!(ColumnReader::FixedLenByteArrayColumnReader)
535            }
536        }
537    }
538
539    /// Helper to simplify a nested field definition to its leaf type
540    ///
541    /// Ex:
542    ///   `Option<&String>` => Type::TypePath(String)
543    ///   `&Option<i32>` => Type::TypePath(i32)
544    ///   `Vec<Vec<u8>>` => Type::Vec(u8)
545    ///
546    /// Useful in determining the physical type of a field and the
547    /// definition levels.
548    fn leaf_type_recursive(&self) -> &Type {
549        Type::leaf_type_recursive_helper(self, None)
550    }
551
552    fn leaf_type_recursive_helper<'a>(ty: &'a Type, parent_ty: Option<&'a Type>) -> &'a Type {
553        match ty {
554            Type::TypePath(_) => parent_ty.unwrap_or(ty),
555            Type::Option(first_type)
556            | Type::Vec(first_type)
557            | Type::Array(first_type, _)
558            | Type::Slice(first_type)
559            | Type::Reference(_, first_type) => {
560                Type::leaf_type_recursive_helper(first_type, Some(ty))
561            }
562        }
563    }
564
565    /// Helper method to further unwrap leaf_type() to get inner-most
566    /// type information, useful for determining the physical type
567    /// and normalizing the type paths.
568    fn inner_type(&self) -> &syn::Type {
569        let leaf_type = self.leaf_type_recursive();
570
571        match leaf_type {
572            Type::TypePath(type_) => type_,
573            Type::Option(first_type)
574            | Type::Vec(first_type)
575            | Type::Array(first_type, _)
576            | Type::Slice(first_type)
577            | Type::Reference(_, first_type) => match **first_type {
578                Type::TypePath(ref type_) => type_,
579                _ => unimplemented!("leaf_type() should only return shallow types"),
580            },
581        }
582    }
583
584    /// Helper to normalize a type path by extracting the
585    /// most identifiable part
586    ///
587    /// Ex:
588    ///   std::string::String => String
589    ///   `Vec<u8>` => `Vec<u8>`
590    ///   chrono::NaiveDateTime => NaiveDateTime
591    ///
592    /// Does run the risk of mis-identifying a type if import
593    /// rename is in play. Please note procedural macros always
594    /// run before type resolution so this is a risk the user
595    /// takes on when renaming imports.
596    fn last_part(&self) -> String {
597        let inner_type = self.inner_type();
598        let inner_type_str = (quote! { #inner_type }).to_string();
599
600        inner_type_str
601            .split("::")
602            .last()
603            .unwrap()
604            .trim()
605            .to_string()
606    }
607
608    /// Converts rust types to parquet physical types.
609    ///
610    /// Ex:
611    ///   [u8; 10] => FIXED_LEN_BYTE_ARRAY
612    ///   `Vec<u8>`  => BYTE_ARRAY
613    ///   String => BYTE_ARRAY
614    ///   i32 => INT32
615    fn physical_type(&self) -> PhysicalType {
616        let last_part = self.last_part();
617        let leaf_type = self.leaf_type_recursive();
618
619        match leaf_type {
620            Type::Array(first_type, _length) => {
621                if let Type::TypePath(_) = **first_type
622                    && last_part == "u8"
623                {
624                    return PhysicalType::FixedLenByteArray;
625                }
626            }
627            Type::Vec(first_type) | Type::Slice(first_type) => {
628                if let Type::TypePath(_) = **first_type
629                    && last_part == "u8"
630                {
631                    return PhysicalType::ByteArray;
632                }
633            }
634            _ => (),
635        }
636
637        match last_part.trim() {
638            "bool" => PhysicalType::Boolean,
639            "u8" | "u16" | "u32" => PhysicalType::Int32,
640            "i8" | "i16" | "i32" | "NaiveDate" => PhysicalType::Int32,
641            "u64" | "i64" | "NaiveDateTime" => PhysicalType::Int64,
642            "usize" | "isize" => {
643                if usize::BITS == 64 {
644                    PhysicalType::Int64
645                } else {
646                    PhysicalType::Int32
647                }
648            }
649            "f32" => PhysicalType::Float,
650            "f64" => PhysicalType::Double,
651            "String" | "str" | "Arc < str >" => PhysicalType::ByteArray,
652            "Uuid" => PhysicalType::FixedLenByteArray,
653            f => unimplemented!("{} currently is not supported", f),
654        }
655    }
656
657    fn length(&self) -> Option<syn::Expr> {
658        let last_part = self.last_part();
659        let leaf_type = self.leaf_type_recursive();
660
661        // `[u8; N]` => Some(N)
662        if let Type::Array(first_type, length) = leaf_type
663            && let Type::TypePath(_) = **first_type
664            && last_part == "u8"
665        {
666            return Some(length.clone());
667        }
668
669        match last_part.trim() {
670            // Uuid => [u8; 16] => Some(16)
671            "Uuid" => Some(syn::parse_quote!(16)),
672            _ => None,
673        }
674    }
675
676    fn logical_type(&self) -> proc_macro2::TokenStream {
677        let last_part = self.last_part();
678        let leaf_type = self.leaf_type_recursive();
679
680        match leaf_type {
681            Type::Array(first_type, _length) => {
682                if let Type::TypePath(_) = **first_type
683                    && last_part == "u8"
684                {
685                    return quote! { None };
686                }
687            }
688            Type::Vec(first_type) | Type::Slice(first_type) => {
689                if let Type::TypePath(_) = **first_type
690                    && last_part == "u8"
691                {
692                    return quote! { None };
693                }
694            }
695            _ => (),
696        }
697
698        match last_part.trim() {
699            "bool" => quote! { None },
700            "u8" => quote! { Some(LogicalType::integer(8, false)) },
701            "u16" => quote! { Some(LogicalType::integer(16, false)) },
702            "u32" => quote! { Some(LogicalType::integer(32, false)) },
703            "u64" => quote! { Some(LogicalType::integer(64, false)) },
704            "i8" => quote! { Some(LogicalType::integer(8, true)) },
705            "i16" => quote! { Some(LogicalType::integer(16, true)) },
706            "i32" | "i64" => quote! { None },
707            "usize" => {
708                quote! { Some(LogicalType::integer(usize::BITS as i8, false)) }
709            }
710            "isize" => {
711                quote! { Some(LogicalType::integer(usize::BITS as i8, true)) }
712            }
713            "NaiveDate" => quote! { Some(LogicalType::Date) },
714            "NaiveDateTime" => quote! { None },
715            "f32" | "f64" => quote! { None },
716            "String" | "str" | "Arc < str >" => quote! { Some(LogicalType::String) },
717            "Uuid" => quote! { Some(LogicalType::Uuid) },
718            f => unimplemented!("{} currently is not supported", f),
719        }
720    }
721
722    fn converted_type(&self) -> Option<proc_macro2::TokenStream> {
723        let last_part = self.last_part();
724
725        match last_part.trim() {
726            "NaiveDateTime" => Some(quote! { ::parquet::basic::ConvertedType::TIMESTAMP_MILLIS }),
727            _ => None,
728        }
729    }
730
731    fn repetition(&self) -> proc_macro2::TokenStream {
732        match self {
733            Type::Option(_) => quote! { ::parquet::basic::Repetition::OPTIONAL },
734            Type::Reference(_, ty) => ty.repetition(),
735            _ => quote! { ::parquet::basic::Repetition::REQUIRED },
736        }
737    }
738
739    /// Convert a parsed rust field AST in to a more easy to manipulate
740    /// parquet_derive::Field
741    fn from(f: &syn::Field) -> Self {
742        Type::from_type(f, &f.ty)
743    }
744
745    fn from_type(f: &syn::Field, ty: &syn::Type) -> Self {
746        match ty {
747            syn::Type::Path(p) => Type::from_type_path(f, p),
748            syn::Type::Reference(tr) => Type::from_type_reference(f, tr),
749            syn::Type::Array(ta) => Type::from_type_array(f, ta),
750            syn::Type::Slice(ts) => Type::from_type_slice(f, ts),
751            other => unimplemented!(
752                "Unable to derive {:?} - it is currently an unsupported type\n{:#?}",
753                f.ident.as_ref().unwrap(),
754                other
755            ),
756        }
757    }
758
759    fn from_type_path(f: &syn::Field, p: &syn::TypePath) -> Self {
760        let last_segment = p.path.segments.last().unwrap();
761
762        let is_vec = last_segment.ident == syn::Ident::new("Vec", proc_macro2::Span::call_site());
763        let is_option =
764            last_segment.ident == syn::Ident::new("Option", proc_macro2::Span::call_site());
765
766        if is_vec || is_option {
767            let generic_type = match &last_segment.arguments {
768                syn::PathArguments::AngleBracketed(angle_args) => {
769                    assert_eq!(angle_args.args.len(), 1);
770                    let first_arg = &angle_args.args[0];
771
772                    match first_arg {
773                        syn::GenericArgument::Type(typath) => typath.clone(),
774                        other => unimplemented!("Unsupported: {:#?}", other),
775                    }
776                }
777                other => unimplemented!("Unsupported: {:#?}", other),
778            };
779
780            if is_vec {
781                Type::Vec(Box::new(Type::from_type(f, &generic_type)))
782            } else {
783                Type::Option(Box::new(Type::from_type(f, &generic_type)))
784            }
785        } else {
786            Type::TypePath(syn::Type::Path(p.clone()))
787        }
788    }
789
790    fn from_type_reference(f: &syn::Field, tr: &syn::TypeReference) -> Self {
791        let lifetime = tr.lifetime.clone();
792        let inner_type = Type::from_type(f, tr.elem.as_ref());
793        Type::Reference(lifetime, Box::new(inner_type))
794    }
795
796    fn from_type_array(f: &syn::Field, ta: &syn::TypeArray) -> Self {
797        let inner_type = Type::from_type(f, ta.elem.as_ref());
798        Type::Array(Box::new(inner_type), ta.len.clone())
799    }
800
801    fn from_type_slice(f: &syn::Field, ts: &syn::TypeSlice) -> Self {
802        let inner_type = Type::from_type(f, ts.elem.as_ref());
803        Type::Slice(Box::new(inner_type))
804    }
805}
806
807#[cfg(test)]
808mod test {
809    use super::*;
810    use syn::{Data, DataStruct, DeriveInput};
811
812    fn extract_fields(input: proc_macro2::TokenStream) -> Vec<syn::Field> {
813        let input: DeriveInput = syn::parse2(input).unwrap();
814
815        let fields = match input.data {
816            Data::Struct(DataStruct { fields, .. }) => fields,
817            _ => panic!("Input must be a struct"),
818        };
819
820        fields.iter().map(|field| field.to_owned()).collect()
821    }
822
823    #[test]
824    fn test_generating_a_simple_writer_snippet() {
825        let snippet: proc_macro2::TokenStream = quote! {
826          struct ABoringStruct {
827            counter: usize,
828          }
829        };
830
831        let fields = extract_fields(snippet);
832        let counter = Field::from(&fields[0]);
833
834        let snippet = counter.writer_snippet().to_string();
835        assert_eq!(snippet,
836                   (quote!{
837                        {
838                            let vals : Vec < _ > = records . iter ( ) . map ( | rec | rec . counter as i64 ) . collect ( );
839
840                            if let ColumnWriter::Int64ColumnWriter ( typed ) = column_writer.untyped() {
841                                typed . write_batch ( & vals [ .. ] , None , None ) ?;
842                            }  else {
843                                panic!("Schema and struct disagree on type for {}" , stringify!{ counter } )
844                            }
845                        }
846                   }).to_string()
847        )
848    }
849
850    #[test]
851    fn test_generating_a_simple_reader_snippet() {
852        let snippet: proc_macro2::TokenStream = quote! {
853          struct ABoringStruct {
854            counter: usize,
855          }
856        };
857
858        let fields = extract_fields(snippet);
859        let counter = Field::from(&fields[0]);
860
861        let snippet = counter.reader_snippet().to_string();
862        assert_eq!(
863            snippet,
864            (quote! {
865                 {
866                    let mut vals = Vec::new();
867                    if let ColumnReader::Int64ColumnReader(mut typed) = column_reader {
868                        let mut definition_levels = Vec::new();
869                        let (total_num, valid_num, decoded_num) = typed.read_records(
870                            num_records, Some(&mut definition_levels), None, &mut vals)?;
871                        if valid_num != decoded_num {
872                            panic!("Support only valid records, found {} null records in column type {}",
873                                decoded_num - valid_num, stringify!{counter});
874                        }
875                    } else {
876                        panic!("Schema and struct disagree on type for {}", stringify!{counter});
877                    }
878                    for (i, r) in &mut records[..num_records].iter_mut().enumerate() {
879                        r.counter = vals[i] as usize;
880                    }
881                 }
882            })
883            .to_string()
884        )
885    }
886
887    #[test]
888    fn test_parquet_type_with_raw_identifier() {
889        let snippet: proc_macro2::TokenStream = quote! {
890          struct ABoringStruct {
891            r#type: i32,
892          }
893        };
894
895        let fields = extract_fields(snippet);
896        let r#type = Field::from(&fields[0]);
897
898        // the raw identifier `r#type` is named `type` in the parquet schema
899        let snippet = r#type.parquet_type().to_string();
900        assert!(
901            snippet.contains("primitive_type_builder (\"type\""),
902            "{snippet}"
903        );
904    }
905
906    #[test]
907    fn test_optional_to_writer_snippet() {
908        let struct_def: proc_macro2::TokenStream = quote! {
909          struct StringBorrower<'a> {
910            optional_str: Option<&'a str>,
911            optional_string: Option<&String>,
912            optional_dumb_int: Option<&i32>,
913          }
914        };
915
916        let fields = extract_fields(struct_def);
917
918        let optional = Field::from(&fields[0]);
919        let snippet = optional.writer_snippet();
920        assert_eq!(snippet.to_string(),
921          (quote! {
922          {
923                let definition_levels : Vec < i16 > = self . iter ( ) . map ( | rec | if rec . optional_str . is_some ( ) { 1 } else { 0 } ) . collect ( ) ;
924
925                let vals: Vec <_> = records.iter().filter_map( |rec| {
926                    if let Some ( inner ) = &rec . optional_str {
927                        Some ( (&inner[..]).into() )
928                    } else {
929                        None
930                    }
931                }).collect();
932
933                if let ColumnWriter::ByteArrayColumnWriter ( typed ) = column_writer.untyped() {
934                    typed . write_batch ( & vals [ .. ] , Some(&definition_levels[..]) , None ) ? ;
935                } else {
936                    panic!("Schema and struct disagree on type for {}" , stringify ! { optional_str } )
937                }
938           }
939            }
940          ).to_string());
941
942        let optional = Field::from(&fields[1]);
943        let snippet = optional.writer_snippet();
944        assert_eq!(snippet.to_string(),
945                   (quote!{
946                   {
947                        let definition_levels : Vec < i16 > = self . iter ( ) . map ( | rec | if rec . optional_string . is_some ( ) { 1 } else { 0 } ) . collect ( ) ;
948
949                        let vals: Vec <_> = records.iter().filter_map( |rec| {
950                            if let Some ( inner ) = &rec . optional_string {
951                                Some ( (&inner[..]).into() )
952                            } else {
953                                None
954                            }
955                        }).collect();
956
957                        if let ColumnWriter::ByteArrayColumnWriter ( typed ) = column_writer.untyped() {
958                            typed . write_batch ( & vals [ .. ] , Some(&definition_levels[..]) , None ) ? ;
959                        } else {
960                            panic!("Schema and struct disagree on type for {}" , stringify ! { optional_string } )
961                        }
962                    }
963        }).to_string());
964
965        let optional = Field::from(&fields[2]);
966        let snippet = optional.writer_snippet();
967        assert_eq!(snippet.to_string(),
968                   (quote!{
969                    {
970                        let definition_levels : Vec < i16 > = self . iter ( ) . map ( | rec | if rec . optional_dumb_int . is_some ( ) { 1 } else { 0 } ) . collect ( ) ;
971
972                        let vals: Vec <_> = records.iter().filter_map( |rec| {
973                            if let Some ( inner ) = rec . optional_dumb_int {
974                                Some ( inner as i32 )
975                            } else {
976                                None
977                            }
978                        }).collect();
979
980                        if let ColumnWriter::Int32ColumnWriter ( typed ) = column_writer.untyped() {
981                            typed . write_batch ( & vals [ .. ] , Some(&definition_levels[..]) , None ) ? ;
982                        }  else {
983                            panic!("Schema and struct disagree on type for {}" , stringify ! { optional_dumb_int } )
984                        }
985                    }
986        }).to_string());
987    }
988
989    #[test]
990    fn test_converting_to_column_writer_type() {
991        let snippet: proc_macro2::TokenStream = quote! {
992          struct ABasicStruct {
993            yes_no: bool,
994            name: String,
995          }
996        };
997
998        let fields = extract_fields(snippet);
999        let processed: Vec<_> = fields.iter().map(Field::from).collect();
1000
1001        let column_writers: Vec<_> = processed
1002            .iter()
1003            .map(|field| field.ty.column_writer())
1004            .collect();
1005
1006        assert_eq!(
1007            column_writers,
1008            vec![
1009                syn::parse_quote!(ColumnWriter::BoolColumnWriter),
1010                syn::parse_quote!(ColumnWriter::ByteArrayColumnWriter)
1011            ]
1012        );
1013    }
1014
1015    #[test]
1016    fn test_converting_to_column_reader_type() {
1017        let snippet: proc_macro2::TokenStream = quote! {
1018          struct ABasicStruct {
1019            yes_no: bool,
1020            name: String,
1021          }
1022        };
1023
1024        let fields = extract_fields(snippet);
1025        let processed: Vec<_> = fields.iter().map(Field::from).collect();
1026
1027        let column_readers: Vec<_> = processed
1028            .iter()
1029            .map(|field| field.ty.column_reader())
1030            .collect();
1031
1032        assert_eq!(
1033            column_readers,
1034            vec![
1035                syn::parse_quote!(ColumnReader::BoolColumnReader),
1036                syn::parse_quote!(ColumnReader::ByteArrayColumnReader)
1037            ]
1038        );
1039    }
1040
1041    #[test]
1042    fn convert_basic_struct() {
1043        let snippet: proc_macro2::TokenStream = quote! {
1044          struct ABasicStruct {
1045            yes_no: bool,
1046            name: String,
1047            length: usize
1048          }
1049        };
1050
1051        let fields = extract_fields(snippet);
1052        let processed: Vec<_> = fields.iter().map(Field::from).collect();
1053        assert_eq!(processed.len(), 3);
1054
1055        assert_eq!(
1056            processed,
1057            vec![
1058                Field {
1059                    ident: syn::Ident::new("yes_no", proc_macro2::Span::call_site()),
1060                    ty: Type::TypePath(syn::parse_quote!(bool)),
1061                    is_a_byte_buf: false,
1062                    third_party_type: None,
1063                },
1064                Field {
1065                    ident: syn::Ident::new("name", proc_macro2::Span::call_site()),
1066                    ty: Type::TypePath(syn::parse_quote!(String)),
1067                    is_a_byte_buf: true,
1068                    third_party_type: None,
1069                },
1070                Field {
1071                    ident: syn::Ident::new("length", proc_macro2::Span::call_site()),
1072                    ty: Type::TypePath(syn::parse_quote!(usize)),
1073                    is_a_byte_buf: false,
1074                    third_party_type: None,
1075                }
1076            ]
1077        )
1078    }
1079
1080    #[test]
1081    fn test_get_inner_type() {
1082        let snippet: proc_macro2::TokenStream = quote! {
1083          struct LotsOfInnerTypes {
1084            a_vec: Vec<u8>,
1085            a_option: ::std::option::Option<bool>,
1086            a_silly_string: ::std::string::String,
1087            a_complicated_thing: ::std::option::Option<::std::result::Result<(),()>>,
1088          }
1089        };
1090
1091        let fields = extract_fields(snippet);
1092        let converted_fields: Vec<_> = fields.iter().map(Type::from).collect();
1093        let inner_types: Vec<_> = converted_fields
1094            .iter()
1095            .map(|field| field.inner_type())
1096            .collect();
1097        let inner_types_strs: Vec<_> = inner_types
1098            .iter()
1099            .map(|ty| (quote! { #ty }).to_string())
1100            .collect();
1101
1102        assert_eq!(
1103            inner_types_strs,
1104            vec![
1105                "u8",
1106                "bool",
1107                ":: std :: string :: String",
1108                ":: std :: result :: Result < () , () >"
1109            ]
1110        )
1111    }
1112
1113    #[test]
1114    fn test_physical_type() {
1115        let snippet: proc_macro2::TokenStream = quote! {
1116          struct LotsOfInnerTypes {
1117            a_buf: ::std::vec::Vec<u8>,
1118            a_number: i32,
1119            a_verbose_option: ::std::option::Option<bool>,
1120            a_silly_string: String,
1121            a_fix_byte_buf: [u8; 10],
1122            a_complex_option: ::std::option::Option<&Vec<u8>>,
1123            a_complex_vec: &::std::vec::Vec<&Option<u8>>,
1124            a_uuid: ::uuid::Uuid,
1125          }
1126        };
1127
1128        let fields = extract_fields(snippet);
1129        let converted_fields: Vec<_> = fields.iter().map(Type::from).collect();
1130        let physical_types: Vec<_> = converted_fields
1131            .iter()
1132            .map(|ty| ty.physical_type())
1133            .collect();
1134
1135        assert_eq!(
1136            physical_types,
1137            vec![
1138                PhysicalType::ByteArray,
1139                PhysicalType::Int32,
1140                PhysicalType::Boolean,
1141                PhysicalType::ByteArray,
1142                PhysicalType::FixedLenByteArray,
1143                PhysicalType::ByteArray,
1144                PhysicalType::Int32,
1145                PhysicalType::FixedLenByteArray,
1146            ]
1147        )
1148    }
1149
1150    #[test]
1151    fn test_type_length() {
1152        let snippet: proc_macro2::TokenStream = quote! {
1153          struct LotsOfInnerTypes {
1154            a_buf: ::std::vec::Vec<u8>,
1155            a_number: i32,
1156            a_verbose_option: ::std::option::Option<bool>,
1157            a_silly_string: String,
1158            a_fix_byte_buf: [u8; 10],
1159            a_complex_option: ::std::option::Option<&Vec<u8>>,
1160            a_complex_vec: &::std::vec::Vec<&Option<u8>>,
1161            a_uuid: ::uuid::Uuid,
1162          }
1163        };
1164
1165        let fields = extract_fields(snippet);
1166        let converted_fields: Vec<_> = fields.iter().map(Type::from).collect();
1167        let lengths: Vec<_> = converted_fields.iter().map(|ty| ty.length()).collect();
1168
1169        assert_eq!(
1170            lengths,
1171            vec![
1172                None,
1173                None,
1174                None,
1175                None,
1176                Some(syn::parse_quote!(10)),
1177                None,
1178                None,
1179                Some(syn::parse_quote!(16)),
1180            ]
1181        )
1182    }
1183
1184    #[test]
1185    fn test_convert_comprehensive_owned_struct() {
1186        let snippet: proc_macro2::TokenStream = quote! {
1187          struct VecHolder {
1188            a_vec: ::std::vec::Vec<u8>,
1189            a_option: ::std::option::Option<bool>,
1190            a_silly_string: ::std::string::String,
1191            a_complicated_thing: ::std::option::Option<::std::result::Result<(),()>>,
1192          }
1193        };
1194
1195        let fields = extract_fields(snippet);
1196        let converted_fields: Vec<_> = fields.iter().map(Type::from).collect();
1197
1198        assert_eq!(
1199            converted_fields,
1200            vec![
1201                Type::Vec(Box::new(Type::TypePath(syn::parse_quote!(u8)))),
1202                Type::Option(Box::new(Type::TypePath(syn::parse_quote!(bool)))),
1203                Type::TypePath(syn::parse_quote!(::std::string::String)),
1204                Type::Option(Box::new(Type::TypePath(
1205                    syn::parse_quote!(::std::result::Result<(),()>)
1206                ))),
1207            ]
1208        );
1209    }
1210
1211    #[test]
1212    fn test_convert_borrowed_struct() {
1213        let snippet: proc_macro2::TokenStream = quote! {
1214          struct Borrower<'a> {
1215            a_str: &'a str,
1216            a_borrowed_option: &'a Option<bool>,
1217            so_many_borrows: &'a Option<&'a str>,
1218          }
1219        };
1220
1221        let fields = extract_fields(snippet);
1222        let types: Vec<_> = fields.iter().map(Type::from).collect();
1223
1224        assert_eq!(
1225            types,
1226            vec![
1227                Type::Reference(
1228                    Some(syn::Lifetime::new("'a", proc_macro2::Span::call_site())),
1229                    Box::new(Type::TypePath(syn::parse_quote!(str)))
1230                ),
1231                Type::Reference(
1232                    Some(syn::Lifetime::new("'a", proc_macro2::Span::call_site())),
1233                    Box::new(Type::Option(Box::new(Type::TypePath(syn::parse_quote!(
1234                        bool
1235                    )))))
1236                ),
1237                Type::Reference(
1238                    Some(syn::Lifetime::new("'a", proc_macro2::Span::call_site())),
1239                    Box::new(Type::Option(Box::new(Type::Reference(
1240                        Some(syn::Lifetime::new("'a", proc_macro2::Span::call_site())),
1241                        Box::new(Type::TypePath(syn::parse_quote!(str)))
1242                    ))))
1243                ),
1244            ]
1245        );
1246    }
1247
1248    #[test]
1249    fn test_chrono_timestamp_millis_write() {
1250        let snippet: proc_macro2::TokenStream = quote! {
1251          struct ATimestampStruct {
1252            henceforth: chrono::NaiveDateTime,
1253            maybe_happened: Option<&chrono::NaiveDateTime>,
1254          }
1255        };
1256
1257        let fields = extract_fields(snippet);
1258        let when = Field::from(&fields[0]);
1259        assert_eq!(when.writer_snippet().to_string(),(quote!{
1260            {
1261                let vals : Vec<_> = records.iter().map(|rec| rec.henceforth.timestamp_millis() ).collect();
1262                if let ColumnWriter::Int64ColumnWriter(typed) = column_writer.untyped() {
1263                    typed.write_batch(&vals[..], None, None) ?;
1264                } else {
1265                    panic!("Schema and struct disagree on type for {}" , stringify!{ henceforth })
1266                }
1267            }
1268        }).to_string());
1269
1270        let maybe_happened = Field::from(&fields[1]);
1271        assert_eq!(maybe_happened.writer_snippet().to_string(),(quote!{
1272            {
1273                let definition_levels : Vec<i16> = self.iter().map(|rec| if rec.maybe_happened.is_some() { 1 } else { 0 }).collect();
1274                let vals : Vec<_> = records.iter().filter_map(|rec| {
1275                    if let Some(inner) = rec.maybe_happened {
1276                        Some(inner.timestamp_millis())
1277                    } else {
1278                        None
1279                    }
1280                }).collect();
1281
1282                if let ColumnWriter::Int64ColumnWriter(typed) = column_writer.untyped() {
1283                    typed.write_batch(&vals[..], Some(&definition_levels[..]), None) ?;
1284                } else {
1285                    panic!("Schema and struct disagree on type for {}" , stringify!{ maybe_happened })
1286                }
1287            }
1288        }).to_string());
1289    }
1290
1291    #[test]
1292    fn test_chrono_timestamp_millis_read() {
1293        let snippet: proc_macro2::TokenStream = quote! {
1294          struct ATimestampStruct {
1295            henceforth: chrono::NaiveDateTime,
1296          }
1297        };
1298
1299        let fields = extract_fields(snippet);
1300        let when = Field::from(&fields[0]);
1301        assert_eq!(when.reader_snippet().to_string(),(quote!{
1302            {
1303                let mut vals = Vec::new();
1304                if let ColumnReader::Int64ColumnReader(mut typed) = column_reader {
1305                    let mut definition_levels = Vec::new();
1306                    let (total_num, valid_num, decoded_num) = typed.read_records(
1307                        num_records, Some(&mut definition_levels), None, &mut vals)?;
1308                    if valid_num != decoded_num {
1309                        panic!("Support only valid records, found {} null records in column type {}",
1310                            decoded_num - valid_num, stringify!{henceforth});
1311                    }
1312                } else {
1313                    panic!("Schema and struct disagree on type for {}", stringify!{ henceforth });
1314                }
1315                for (i, r) in &mut records[..num_records].iter_mut().enumerate() {
1316                    r.henceforth = ::chrono::naive::NaiveDateTime::from_timestamp_millis(vals[i]).unwrap();
1317                }
1318            }
1319        }).to_string());
1320    }
1321
1322    #[test]
1323    fn test_chrono_date_write() {
1324        let snippet: proc_macro2::TokenStream = quote! {
1325          struct ATimestampStruct {
1326            henceforth: chrono::NaiveDate,
1327            maybe_happened: Option<&chrono::NaiveDate>,
1328          }
1329        };
1330
1331        let fields = extract_fields(snippet);
1332        let when = Field::from(&fields[0]);
1333        assert_eq!(when.writer_snippet().to_string(),(quote!{
1334            {
1335                let vals : Vec<_> = records.iter().map(|rec| rec.henceforth.signed_duration_since(::chrono::NaiveDate::from_ymd(1970, 1, 1)).num_days() as i32).collect();
1336                if let ColumnWriter::Int32ColumnWriter(typed) = column_writer.untyped() {
1337                    typed.write_batch(&vals[..], None, None) ?;
1338                } else {
1339                    panic!("Schema and struct disagree on type for {}" , stringify!{ henceforth })
1340                }
1341            }
1342        }).to_string());
1343
1344        let maybe_happened = Field::from(&fields[1]);
1345        assert_eq!(maybe_happened.writer_snippet().to_string(),(quote!{
1346            {
1347                let definition_levels : Vec<i16> = self.iter().map(|rec| if rec.maybe_happened.is_some() { 1 } else { 0 }).collect();
1348                let vals : Vec<_> = records.iter().filter_map(|rec| {
1349                    if let Some(inner) = rec.maybe_happened {
1350                        Some(inner.signed_duration_since(::chrono::NaiveDate::from_ymd(1970, 1, 1)).num_days() as i32)
1351                    } else {
1352                        None
1353                    }
1354                }).collect();
1355
1356                if let ColumnWriter::Int32ColumnWriter(typed) = column_writer.untyped() {
1357                    typed.write_batch(&vals[..], Some(&definition_levels[..]), None) ?;
1358                } else {
1359                    panic!("Schema and struct disagree on type for {}" , stringify!{ maybe_happened })
1360                }
1361            }
1362        }).to_string());
1363    }
1364
1365    #[test]
1366    fn test_chrono_date_read() {
1367        let snippet: proc_macro2::TokenStream = quote! {
1368          struct ATimestampStruct {
1369            henceforth: chrono::NaiveDate,
1370          }
1371        };
1372
1373        let fields = extract_fields(snippet);
1374        let when = Field::from(&fields[0]);
1375        assert_eq!(when.reader_snippet().to_string(),(quote!{
1376            {
1377                let mut vals = Vec::new();
1378                if let ColumnReader::Int32ColumnReader(mut typed) = column_reader {
1379                    let mut definition_levels = Vec::new();
1380                    let (total_num, valid_num, decoded_num) = typed.read_records(
1381                        num_records, Some(&mut definition_levels), None, &mut vals)?;
1382                    if valid_num != decoded_num {
1383                        panic!("Support only valid records, found {} null records in column type {}",
1384                            decoded_num - valid_num, stringify!{henceforth});
1385                    }
1386                } else {
1387                    panic!("Schema and struct disagree on type for {}", stringify!{ henceforth });
1388                }
1389                for (i, r) in &mut records[..num_records].iter_mut().enumerate() {
1390                    r.henceforth = ::chrono::naive::NaiveDate::from_num_days_from_ce_opt(vals[i].saturating_add(719163)).unwrap();
1391                }
1392            }
1393        }).to_string());
1394    }
1395
1396    #[test]
1397    fn test_uuid_write() {
1398        let snippet: proc_macro2::TokenStream = quote! {
1399          struct AUuidStruct {
1400            unique_id: uuid::Uuid,
1401            maybe_unique_id: Option<&uuid::Uuid>,
1402          }
1403        };
1404
1405        let fields = extract_fields(snippet);
1406        let when = Field::from(&fields[0]);
1407        assert_eq!(when.writer_snippet().to_string(),(quote!{
1408            {
1409                let vals : Vec<_> = records.iter().map(|rec| rec.unique_id.as_bytes().to_vec().into() ).collect();
1410                if let ColumnWriter::FixedLenByteArrayColumnWriter(typed) = column_writer.untyped() {
1411                    typed.write_batch(&vals[..], None, None) ?;
1412                } else {
1413                    panic!("Schema and struct disagree on type for {}" , stringify!{ unique_id })
1414                }
1415            }
1416        }).to_string());
1417
1418        let maybe_happened = Field::from(&fields[1]);
1419        assert_eq!(maybe_happened.writer_snippet().to_string(),(quote!{
1420            {
1421                let definition_levels : Vec<i16> = self.iter().map(|rec| if rec.maybe_unique_id.is_some() { 1 } else { 0 }).collect();
1422                let vals : Vec<_> = records.iter().filter_map(|rec| {
1423                    if let Some(inner) = &rec.maybe_unique_id {
1424                        Some((&inner.to_string()[..]).into())
1425                    } else {
1426                        None
1427                    }
1428                }).collect();
1429
1430                if let ColumnWriter::FixedLenByteArrayColumnWriter(typed) = column_writer.untyped() {
1431                    typed.write_batch(&vals[..], Some(&definition_levels[..]), None) ?;
1432                } else {
1433                    panic!("Schema and struct disagree on type for {}" , stringify!{ maybe_unique_id })
1434                }
1435            }
1436        }).to_string());
1437    }
1438
1439    #[test]
1440    fn test_uuid_read() {
1441        let snippet: proc_macro2::TokenStream = quote! {
1442          struct AUuidStruct {
1443            unique_id: uuid::Uuid,
1444          }
1445        };
1446
1447        let fields = extract_fields(snippet);
1448        let when = Field::from(&fields[0]);
1449        assert_eq!(when.reader_snippet().to_string(),(quote!{
1450            {
1451                let mut vals = Vec::new();
1452                if let ColumnReader::FixedLenByteArrayColumnReader(mut typed) = column_reader {
1453                    let mut definition_levels = Vec::new();
1454                    let (total_num, valid_num, decoded_num) = typed.read_records(
1455                        num_records, Some(&mut definition_levels), None, &mut vals)?;
1456                    if valid_num != decoded_num {
1457                        panic!("Support only valid records, found {} null records in column type {}",
1458                            decoded_num - valid_num, stringify!{unique_id});
1459                    }
1460                } else {
1461                    panic!("Schema and struct disagree on type for {}", stringify!{ unique_id });
1462                }
1463                for (i, r) in &mut records[..num_records].iter_mut().enumerate() {
1464                    r.unique_id = ::uuid::Uuid::from_bytes(vals[i].data().try_into().unwrap());
1465                }
1466            }
1467        }).to_string());
1468    }
1469
1470    #[test]
1471    fn test_converted_type() {
1472        let snippet: proc_macro2::TokenStream = quote! {
1473          struct ATimeStruct {
1474            time: chrono::NaiveDateTime,
1475          }
1476        };
1477
1478        let fields = extract_fields(snippet);
1479
1480        let time = Field::from(&fields[0]);
1481
1482        let converted_type = time.ty.converted_type();
1483        assert_eq!(
1484            converted_type.unwrap().to_string(),
1485            quote! { ::parquet::basic::ConvertedType::TIMESTAMP_MILLIS }.to_string()
1486        );
1487    }
1488}
parquet_derive/parquet_field.rs

parquet_derive/
parquet_field.rs
