2 borrow::Cow, cell::RefCell, cmp::Ordering, collections::HashMap, iter::repeat, ops::Range,
7 dictionary::{self, Dictionary, VarWidth},
8 encoding::{default_encoding, get_encoding, Error as EncodingError},
10 format::{Error as FormatError, Spec, UncheckedSpec},
11 identifier::{Error as IdError, Identifier},
13 self, LongStringMissingValueRecord, MissingValues, ProductInfoRecord, RawDocumentLine,
14 RawStr, RawString, VarDisplayRecord, VarType, DecodedRecord,
17 use chrono::{NaiveDate, NaiveDateTime, NaiveTime};
18 use encoding_rs::{DecoderResult, Encoding};
19 use num::integer::div_ceil;
20 use ordered_float::OrderedFloat;
21 use thiserror::Error as ThisError;
23 pub use crate::raw::{CategoryLabels, Compression};
25 #[derive(ThisError, Debug)]
27 // XXX this is really an internal error and maybe we should change the
28 // interfaces to make it impossible
29 #[error("Missing header record")]
33 EncodingError(EncodingError),
35 #[error("Using default encoding {0}.")]
36 UsingDefaultEncoding(String),
38 #[error("Variable record from offset {:x} to {:x} specifies width {width} not in valid range [-1,255).", offsets.start, offsets.end)]
39 InvalidVariableWidth { offsets: Range<u64>, width: i32 },
41 #[error("This file has corrupted metadata written by a buggy version of PSPP. To ensure that other software can read it correctly, save a new copy of the file.")]
42 InvalidLongMissingValueFormat,
44 #[error("File creation date {creation_date} is not in the expected format \"DD MMM YY\" format. Using 01 Jan 1970.")]
45 InvalidCreationDate { creation_date: String },
47 #[error("File creation time {creation_time} is not in the expected format \"HH:MM:SS\" format. Using midnight.")]
48 InvalidCreationTime { creation_time: String },
50 #[error("{id_error} Renaming variable to {new_name}.")]
57 "Substituting {new_spec} for invalid print format on variable {variable}. {format_error}"
62 format_error: FormatError,
66 "Substituting {new_spec} for invalid write format on variable {variable}. {format_error}"
71 format_error: FormatError,
74 #[error("Renaming variable with duplicate name {duplicate_name} to {new_name}.")]
75 DuplicateVariableName {
76 duplicate_name: Identifier,
80 #[error("Dictionary index {dict_index} is outside valid range [1,{max_index}].")]
81 InvalidDictIndex { dict_index: usize, max_index: usize },
83 #[error("Dictionary index {0} refers to a long string continuation.")]
84 DictIndexIsContinuation(usize),
86 #[error("Variables associated with value label are not all of identical type. Variable {numeric_var} is numeric, but variable {string_var} is string.")]
87 ValueLabelsDifferentTypes {
88 numeric_var: Identifier,
89 string_var: Identifier,
93 "Value labels may not be added to long string variable {0} using record types 3 or 4."
95 InvalidLongStringValueLabel(Identifier),
97 #[error("Invalid multiple response set name. {0}")]
98 InvalidMrSetName(IdError),
100 #[error("Multiple response set {mr_set} includes unknown variable {short_name}.")]
101 UnknownMrSetVariable {
103 short_name: Identifier,
106 #[error("Multiple response set {0} has no variables.")]
107 EmptyMrSet(Identifier),
109 #[error("Multiple response set {0} has only one variable.")]
110 OneVarMrSet(Identifier),
112 #[error("Multiple response set {0} contains both string and numeric variables.")]
113 MixedMrSet(Identifier),
116 "Invalid numeric format for counted value {number} in multiple response set {mr_set}."
118 InvalidMDGroupCountedValue { mr_set: Identifier, number: String },
120 #[error("Counted value {value} has width {width}, but it must be no wider than {max_width}, the width of the narrowest variable in multiple response set {mr_set}.")]
121 TooWideMDGroupCountedValue {
128 #[error("Long string value label for variable {name} has width {width}, which is not in the valid range [{min_width},{max_width}].")]
129 InvalidLongValueLabelWidth {
136 #[error("Invalid attribute name. {0}")]
137 InvalidAttributeName(IdError),
139 #[error("Invalid short name in long variable name record. {0}")]
140 InvalidShortName(IdError),
142 #[error("Invalid name in long variable name record. {0}")]
143 InvalidLongName(IdError),
145 #[error("Invalid variable name in very long string record. {0}")]
146 InvalidLongStringName(IdError),
148 #[error("Invalid variable name in long string value label record. {0}")]
149 InvalidLongStringValueLabelName(IdError),
151 #[error("Invalid variable name in attribute record. {0}")]
152 InvalidAttributeVariableName(IdError),
154 // XXX This is risky because `text` might be arbitarily long.
155 #[error("Text string contains invalid bytes for {encoding} encoding: {text}")]
156 MalformedString { encoding: String, text: String },
158 #[error("Details TBD")]
162 #[derive(Clone, Debug)]
164 Header(HeaderRecord),
165 Variable(VariableRecord),
166 ValueLabel(ValueLabelRecord),
167 Document(DocumentRecord),
168 IntegerInfo(IntegerInfoRecord),
169 FloatInfo(FloatInfoRecord),
170 VariableSets(VariableSetRecord),
171 VarDisplay(VarDisplayRecord),
172 MultipleResponse(MultipleResponseRecord),
173 LongStringMissingValues(LongStringMissingValueRecord<String, String>),
174 LongStringValueLabels(LongStringValueLabelRecord),
175 Encoding(EncodingRecord),
176 NumberOfCases(NumberOfCasesRecord),
177 ProductInfo(ProductInfoRecord),
178 LongNames(LongNameRecord),
179 VeryLongStrings(VeryLongStringRecord),
180 FileAttributes(FileAttributeRecord),
181 VariableAttributes(VariableAttributeRecord),
182 OtherExtension(Extension),
186 pub use crate::raw::EncodingRecord;
187 pub use crate::raw::Extension;
188 pub use crate::raw::FloatInfoRecord;
189 pub use crate::raw::IntegerInfoRecord;
190 pub use crate::raw::NumberOfCasesRecord;
192 type DictIndex = usize;
194 pub struct Variable {
195 pub dict_index: DictIndex,
196 pub short_name: Identifier,
197 pub long_name: Option<Identifier>,
202 pub raw: raw::Decoder,
203 pub encoding: &'static Encoding,
204 pub variables: HashMap<DictIndex, Variable>,
205 pub var_names: HashMap<Identifier, DictIndex>,
206 pub dictionary: Dictionary,
207 n_dict_indexes: usize,
208 n_generated_names: usize,
213 header: Option<raw::HeaderRecord<Cow<'a, str>>>,
214 variables: Vec<raw::VariableRecord<Cow<'a, str>, String>>,
215 value_labels: Vec<&'a raw::ValueLabelRecord<RawStr<8>, RawString>>,
216 documents: Vec<raw::DocumentRecord<Cow<'a, str>>>,
217 integer_info: Option<&'a raw::IntegerInfoRecord>,
218 float_info: Option<&'a raw::FloatInfoRecord>,
219 variable_sets: Vec<&'a raw::VariableSetRecord>,
220 var_display: Option<&'a raw::VarDisplayRecord>,
221 multiple_response: Vec<&'a raw::MultipleResponseRecord<RawString, RawString>>,
222 long_string_value_labels: Vec<&'a raw::LongStringValueLabelRecord<RawString>>,
223 long_string_missing_values: Vec<raw::LongStringMissingValueRecord<Identifier, String>>,
224 encoding: Option<&'a raw::EncodingRecord>,
225 number_of_cases: Option<&'a raw::NumberOfCasesRecord>,
226 product_info: Option<&'a raw::ProductInfoRecord>,
227 long_names: Option<&'a raw::LongNamesRecord>,
228 very_long_strings: Vec<&'a raw::VeryLongStringsRecord>,
229 file_attributes: Vec<&'a raw::FileAttributeRecord>,
230 variable_attributes: Vec<&'a raw::VariableAttributeRecord>,
231 other_extensions: Vec<&'a raw::Extension>,
232 cases: Option<&'a Rc<RefCell<raw::Cases>>>,
235 fn set_or_warn<T>(option: &mut Option<T>, value: T, warn: &impl Fn(Error)) {
236 if option.is_none() {
237 let _ = option.insert(value);
243 impl<'a> Headers<'a> {
244 fn new(headers: &'a Vec<raw::Record>, decoder: &Decoder, warn: &impl Fn(Error)) -> Headers<'a> {
245 let mut h = Headers::default();
246 for header in headers {
248 raw::Record::Header(record) => {
249 set_or_warn(&mut h.header, record.decode(&decoder.raw), warn)
251 raw::Record::Variable(record) => h.variables.push(record.decode(&decoder.raw)),
252 raw::Record::ValueLabel(record) => h.value_labels.push(record),
253 raw::Record::Document(record) => h.documents.push(record.decode(&decoder.raw)),
254 raw::Record::IntegerInfo(record) => set_or_warn(&mut h.integer_info, record, warn),
255 raw::Record::FloatInfo(record) => set_or_warn(&mut h.float_info, record, warn),
256 raw::Record::VariableSets(record) => h.variable_sets.push(record),
257 raw::Record::VarDisplay(record) => set_or_warn(&mut h.var_display, record, warn),
258 raw::Record::MultipleResponse(record) => h.multiple_response.push(record),
259 raw::Record::LongStringValueLabels(record) => {
260 h.long_string_value_labels.push(record)
262 raw::Record::LongStringMissingValues(record) => h
263 .long_string_missing_values
264 .push(record.decode(&decoder.raw)),
265 raw::Record::Encoding(record) => set_or_warn(&mut h.encoding, record, warn),
266 raw::Record::NumberOfCases(record) => {
267 set_or_warn(&mut h.number_of_cases, record, warn)
269 raw::Record::ProductInfo(record) => set_or_warn(&mut h.product_info, record, warn),
270 raw::Record::LongNames(record) => set_or_warn(&mut h.long_names, record, warn),
271 raw::Record::VeryLongStrings(record) => h.very_long_strings.push(record),
272 raw::Record::FileAttributes(record) => h.file_attributes.push(record),
273 raw::Record::VariableAttributes(record) => h.variable_attributes.push(record),
274 raw::Record::OtherExtension(record) => h.other_extensions.push(record),
275 raw::Record::EndOfHeaders(_) => (),
276 raw::Record::ZHeader(_) => (),
277 raw::Record::ZTrailer(_) => (),
278 raw::Record::Cases(record) => set_or_warn(&mut h.cases, record, warn),
279 raw::Record::Text(_) => todo!(),
288 headers: Vec<DecodedRecord>,
289 decoder: raw::Decoder,
290 ) -> Result<(Vec<Record>, Metadata), Error> {
291 let dictionary = Dictionary::new(decoder.encoding);
292 let mut decoder = Decoder {
294 variables: HashMap::new(),
295 var_names: HashMap::new(),
298 n_generated_names: 0,
301 let h = Headers::new(&headers, &decoder);
302 let Some(header) = h.header else {
303 return Err(Error::MissingHeaderRecord);
306 let mut output = Vec::with_capacity(headers.len());
308 // Decode the records that don't use variables at all.
309 if let Some(header) = HeaderRecord::try_decode(&mut decoder, &header, warn)? {
310 output.push(Record::Header(header))
312 for document in h.documents {
313 for line in &document.lines {
314 decoder.dictionary.documents.push(line.to_string())
318 for &raw in &h.file_attributes {
319 let s = decoder.decode_string_cow(&raw.text.0, warn);
320 output.push(Record::FileAttributes(FileAttributeRecord::parse(
324 for &raw in &h.other_extensions {
325 output.push(Record::OtherExtension(raw.clone()));
328 // Decode the variable records, which are the basis of almost everything
330 for raw in &h.variables {
331 parse_variable_record(&mut decoder, raw, warn)?;
334 // Decode value labels and weight variable. These use indexes into the
335 // variable records, so we need to parse them before those indexes become
336 // invalidated by very long string variables.
337 for &raw in &h.value_labels {
338 if let Some(value_label) = ValueLabelRecord::try_decode(&mut decoder, raw, warn)? {
339 output.push(Record::ValueLabel(value_label));
343 if let Some(raw) = h.var_display {
344 output.push(Record::VarDisplay(raw.clone()));
347 // Decode records that use short names.
348 for &raw in &h.multiple_response {
349 if let Some(mrr) = MultipleResponseRecord::try_decode(&mut decoder, raw, warn)? {
350 output.push(Record::MultipleResponse(mrr))
353 for &raw in &h.very_long_strings {
354 let s = decoder.decode_string_cow(&raw.text.0, warn);
355 output.push(Record::VeryLongStrings(VeryLongStringRecord::parse(
360 // Rename variables to their long names.
361 for &raw in &h.long_names {
362 let s = decoder.decode_string_cow(&raw.text.0, warn);
363 output.push(Record::LongNames(LongNameRecord::parse(
370 // Decode recods that use long names.
371 for &raw in &h.variable_attributes {
372 let s = decoder.decode_string_cow(&raw.text.0, warn);
373 output.push(Record::VariableAttributes(VariableAttributeRecord::parse(
377 for &raw in &h.long_string_value_labels {
378 if let Some(mrr) = LongStringValueLabelRecord::try_decode(&mut decoder, raw, warn)? {
379 output.push(Record::LongStringValueLabels(mrr))
382 for &raw in &h.long_string_missing_values {
383 if let Some(mrr) = LongStringMissingValuesRecord::try_decode(&mut decoder, raw, warn)? {
384 output.push(Record::LongStringMissingValues(mrr))
387 for &raw in &h.variable_sets {
388 let s = decoder.decode_string_cow(&raw.text.0, warn);
389 output.push(Record::VariableSets(VariableSetRecord::parse(&s, warn)?));
392 let metadata = Metadata::decode(&header, h.integer_info, h.product_info, warn);
393 Ok((output, metadata))
397 fn generate_name(&mut self) -> Identifier {
399 self.n_generated_names += 1;
400 let name = Identifier::new(&format!("VAR{:03}", self.n_generated_names), self.encoding)
402 if !self.var_names.contains_key(&name) {
405 assert!(self.n_generated_names < usize::MAX);
408 fn decode_string_cow<'a>(&self, input: &'a [u8], warn: &impl Fn(Error)) -> Cow<'a, str> {
409 let (output, malformed) = self.encoding.decode_without_bom_handling(input);
411 warn(Error::MalformedString {
412 encoding: self.encoding.name().into(),
413 text: output.clone().into(),
418 fn decode_string(&self, input: &[u8], warn: &impl Fn(Error)) -> String {
419 self.decode_string_cow(input, warn).into()
421 pub fn decode_identifier(
424 warn: &impl Fn(Error),
425 ) -> Result<Identifier, IdError> {
426 let s = self.decode_string_cow(input, warn);
427 Identifier::new(&s, self.encoding)
429 fn get_var_by_index(&self, dict_index: usize) -> Result<&Variable, Error> {
430 let max_index = self.n_dict_indexes;
431 if dict_index == 0 || dict_index > max_index {
432 return Err(Error::InvalidDictIndex {
437 let Some(variable) = self.variables.get(&(dict_index - 1)) else {
438 return Err(Error::DictIndexIsContinuation(dict_index));
443 /// Returns `input` decoded from `self.encoding` into UTF-8 such that
444 /// re-encoding the result back into `self.encoding` will have exactly the
445 /// same length in bytes.
447 /// XXX warn about errors?
448 fn decode_exact_length<'a>(&self, input: &'a [u8]) -> Cow<'a, str> {
449 if let (s, false) = self.encoding.decode_without_bom_handling(input) {
450 // This is the common case. Usually there will be no errors.
453 // Unusual case. Don't bother to optimize it much.
454 let mut decoder = self.encoding.new_decoder_without_bom_handling();
455 let mut output = String::with_capacity(
457 .max_utf8_buffer_length_without_replacement(input.len())
460 let mut rest = input;
461 while !rest.is_empty() {
462 match decoder.decode_to_string_without_replacement(rest, &mut output, true) {
463 (DecoderResult::InputEmpty, _) => break,
464 (DecoderResult::OutputFull, _) => unreachable!(),
465 (DecoderResult::Malformed(a, b), consumed) => {
466 let skipped = a as usize + b as usize;
467 output.extend(repeat('?').take(skipped));
468 rest = &rest[consumed..];
472 assert_eq!(self.encoding.encode(&output).0.len(), input.len());
478 pub trait TryDecode: Sized {
481 decoder: &mut Decoder,
482 input: &Self::Input<'_>,
483 warn: impl Fn(Error),
484 ) -> Result<Option<Self>, Error>;
487 pub trait Decode<Input>: Sized {
488 fn decode(decoder: &Decoder, input: &Input, warn: impl Fn(Error)) -> Self;
491 impl<const N: usize> Decode<RawStr<N>> for String {
492 fn decode(decoder: &Decoder, input: &RawStr<N>, warn: impl Fn(Error)) -> Self {
493 decoder.decode_string(&input.0, &warn)
497 #[derive(Clone, Debug)]
498 pub struct HeaderRecord {
499 pub eye_catcher: String,
500 pub weight_index: Option<usize>,
501 pub n_cases: Option<u64>,
502 pub creation: NaiveDateTime,
503 pub file_label: String,
506 fn trim_end_spaces(mut s: String) -> String {
507 s.truncate(s.trim_end_matches(' ').len());
511 /// Data file info that doesn't fit in [Dictionary].
512 pub struct Metadata {
513 creation: NaiveDateTime,
515 compression: Option<Compression>,
516 n_cases: Option<u64>,
518 product_ext: Option<String>,
519 version: Option<(i32, i32, i32)>,
524 header: &crate::raw::HeaderRecord<Cow<str>>,
525 integer_info: Option<&IntegerInfoRecord>,
526 product_ext: Option<&ProductInfoRecord>,
527 warn: impl Fn(Error),
529 let creation_date = NaiveDate::parse_from_str(&header.creation_date, "%e %b %Y")
530 .unwrap_or_else(|_| {
531 warn(Error::InvalidCreationDate {
532 creation_date: header.creation_date.to_string(),
536 let creation_time = NaiveTime::parse_from_str(&header.creation_time, "%H:%M:%S")
537 .unwrap_or_else(|_| {
538 warn(Error::InvalidCreationTime {
539 creation_time: header.creation_time.to_string(),
543 let creation = NaiveDateTime::new(creation_date, creation_time);
547 .trim_start_matches("@(#) SPSS DATA FILE")
553 endian: header.endian,
554 compression: header.compression,
555 n_cases: header.n_cases.map(|n| n as u64),
557 product_ext: product_ext.map(|pe| pe.0.clone()),
558 version: integer_info.map(|ii| ii.version),
563 impl TryDecode for HeaderRecord {
564 type Input<'a> = crate::raw::HeaderRecord<Cow<'a, str>>;
567 _decoder: &mut Decoder,
568 input: &Self::Input<'_>,
569 warn: impl Fn(Error),
570 ) -> Result<Option<Self>, Error> {
571 let eye_catcher = trim_end_spaces(input.eye_catcher.to_string());
572 let file_label = trim_end_spaces(input.file_label.to_string());
573 let creation_date = NaiveDate::parse_from_str(&input.creation_date, "%e %b %Y")
574 .unwrap_or_else(|_| {
575 warn(Error::InvalidCreationDate {
576 creation_date: input.creation_date.to_string(),
580 let creation_time = NaiveTime::parse_from_str(&input.creation_time, "%H:%M:%S")
581 .unwrap_or_else(|_| {
582 warn(Error::InvalidCreationTime {
583 creation_time: input.creation_time.to_string(),
587 Ok(Some(HeaderRecord {
589 weight_index: input.weight_index.map(|n| n as usize),
590 n_cases: input.n_cases.map(|n| n as u64),
591 creation: NaiveDateTime::new(creation_date, creation_time),
597 #[derive(Clone, Debug)]
598 pub struct VariableRecord {
600 pub name: Identifier,
601 pub print_format: Spec,
602 pub write_format: Spec,
603 pub missing_values: MissingValues<String>,
604 pub label: Option<String>,
607 fn decode_format(raw: raw::Spec, width: VarWidth, warn: impl Fn(Spec, FormatError)) -> Spec {
608 UncheckedSpec::try_from(raw)
609 .and_then(Spec::try_from)
610 .and_then(|x| x.check_width_compatibility(width))
611 .unwrap_or_else(|error| {
612 let new_format = Spec::default_for_width(width);
613 warn(new_format, error);
618 fn parse_variable_record(
619 decoder: &mut Decoder,
620 input: &raw::VariableRecord<Cow<str>, String>,
621 warn: impl Fn(Error),
622 ) -> Result<(), Error> {
623 let width = match input.width {
624 0 => VarWidth::Numeric,
625 w @ 1..=255 => VarWidth::String(w as u16),
628 return Err(Error::InvalidVariableWidth {
629 offsets: input.offsets.clone(),
634 let name = trim_end_spaces(input.name.to_string());
635 let name = match Identifier::new(&name, decoder.encoding) {
637 if !decoder.var_names.contains_key(&name) {
640 let new_name = decoder.generate_name();
641 warn(Error::DuplicateVariableName {
642 duplicate_name: name.clone(),
643 new_name: new_name.clone(),
649 let new_name = decoder.generate_name();
650 warn(Error::InvalidVariableName {
652 new_name: new_name.clone(),
657 let variable = Variable {
658 dict_index: decoder.n_dict_indexes,
659 short_name: name.clone(),
663 decoder.n_dict_indexes += width.n_dict_indexes();
666 .insert(name.clone(), variable.dict_index)
670 .insert(variable.dict_index, variable)
673 let print_format = decode_format(input.print_format, width, |new_spec, format_error| {
674 warn(Error::InvalidPrintFormat {
676 variable: name.clone(),
680 let write_format = decode_format(input.write_format, width, |new_spec, format_error| {
681 warn(Error::InvalidWriteFormat {
683 variable: name.clone(),
687 let mut variable = dictionary::Variable::new(name, width);
688 variable.print_format = print_format;
689 variable.write_format = write_format;
690 variable.missing_values = input.missing_values.clone();
691 if let Some(ref label) = input.label {
692 variable.label = Some(label.to_string());
694 decoder.dictionary.add_var(variable).unwrap();
698 #[derive(Clone, Debug)]
699 pub struct DocumentRecord(Vec<String>);
701 impl TryDecode for DocumentRecord {
702 type Input<'a> = crate::raw::DocumentRecord<RawDocumentLine>;
705 decoder: &mut Decoder,
706 input: &Self::Input<'_>,
707 warn: impl Fn(Error),
708 ) -> Result<Option<Self>, Error> {
709 Ok(Some(DocumentRecord(
713 .map(|s| trim_end_spaces(decoder.decode_string(&s.0, &warn)))
723 const NAME: &'static str;
724 fn parse(input: &str, warn: impl Fn(Error)) -> Result<Self, Error>;
727 #[derive(Clone, Debug)]
728 pub struct VariableSet {
730 pub vars: Vec<String>,
734 fn parse(input: &str) -> Result<Self, Error> {
735 let (name, input) = input.split_once('=').ok_or(Error::TBD)?;
736 let vars = input.split_ascii_whitespace().map(String::from).collect();
744 trait WarnOnError<T> {
745 fn warn_on_error<F: Fn(Error)>(self, warn: &F) -> Option<T>;
747 impl<T> WarnOnError<T> for Result<T, Error> {
748 fn warn_on_error<F: Fn(Error)>(self, warn: &F) -> Option<T> {
750 Ok(result) => Some(result),
759 #[derive(Clone, Debug)]
760 pub struct ValueLabel {
765 #[derive(Clone, Debug)]
766 pub struct ValueLabelRecord {
767 pub var_type: VarType,
768 pub labels: Vec<ValueLabel>,
769 pub variables: Vec<Identifier>,
772 impl TryDecode for ValueLabelRecord {
773 type Input<'a> = crate::raw::ValueLabelRecord<RawStr<8>, RawString>;
775 decoder: &mut Decoder,
776 input: &Self::Input<'_>,
777 warn: impl Fn(Error),
778 ) -> Result<Option<ValueLabelRecord>, Error> {
779 let variables: Vec<&Variable> = input
782 .filter_map(|&dict_index| {
784 .get_var_by_index(dict_index as usize)
785 .warn_on_error(&warn)
787 .filter(|&variable| match variable.width {
788 VarWidth::String(width) if width > 8 => {
789 warn(Error::InvalidLongStringValueLabel(
790 variable.short_name.clone(),
797 let mut i = variables.iter();
798 let Some(&first_var) = i.next() else {
801 let var_type: VarType = first_var.width.into();
803 let this_type: VarType = variable.width.into();
804 if var_type != this_type {
805 let (numeric_var, string_var) = match var_type {
806 VarType::Numeric => (first_var, variable),
807 VarType::String => (variable, first_var),
809 warn(Error::ValueLabelsDifferentTypes {
810 numeric_var: numeric_var.short_name.clone(),
811 string_var: string_var.short_name.clone(),
819 .map(|raw::ValueLabel { value, label }| {
820 let label = decoder.decode_string(&label.0, &warn);
821 let value = Value::decode(value, decoder);
822 ValueLabel { value, label }
825 let variables = variables
827 .map(|&variable| variable.short_name.clone())
829 Ok(Some(ValueLabelRecord {
837 #[derive(Clone, Debug)]
838 pub struct VariableSetRecord(Vec<VariableSet>);
840 impl TextRecord for VariableSetRecord {
841 const NAME: &'static str = "variable set";
842 fn parse(input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
843 let mut sets = Vec::new();
844 for line in input.lines() {
845 if let Some(set) = VariableSet::parse(line).warn_on_error(&warn) {
849 Ok(VariableSetRecord(sets))
853 #[derive(Clone, Debug)]
854 pub struct LongName {
855 pub short_name: Identifier,
856 pub long_name: Identifier,
860 fn new(decoder: &mut Decoder, short_name: &str, long_name: &str) -> Result<LongName, Error> {
862 Identifier::new(short_name, decoder.encoding).map_err(Error::InvalidShortName)?;
864 Identifier::new(long_name, decoder.encoding).map_err(Error::InvalidLongName)?;
872 #[derive(Clone, Debug)]
873 pub struct LongNameRecord(Vec<LongName>);
875 impl LongNameRecord {
876 pub fn parse(decoder: &mut Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
877 let mut names = Vec::new();
878 for pair in input.split('\t').filter(|s| !s.is_empty()) {
879 if let Some((short_name, long_name)) = pair.split_once('=') {
880 if let Some(long_name) =
881 LongName::new(decoder, short_name, long_name).warn_on_error(&warn)
883 names.push(long_name);
889 Ok(LongNameRecord(names))
893 #[derive(Clone, Debug)]
894 pub struct VeryLongString {
895 pub short_name: Identifier,
899 impl VeryLongString {
900 fn parse(decoder: &Decoder, input: &str) -> Result<VeryLongString, Error> {
901 let Some((short_name, length)) = input.split_once('=') else {
902 return Err(Error::TBD);
905 Identifier::new(short_name, decoder.encoding).map_err(Error::InvalidLongStringName)?;
906 let length: u16 = length.parse().map_err(|_| Error::TBD)?;
907 if length > VarWidth::MAX_STRING {
908 return Err(Error::TBD);
910 Ok(VeryLongString { short_name, length })
914 #[derive(Clone, Debug)]
915 pub struct VeryLongStringRecord(Vec<VeryLongString>);
917 impl VeryLongStringRecord {
918 pub fn parse(decoder: &Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
919 let mut very_long_strings = Vec::new();
922 .map(|s| s.trim_end_matches('\t'))
923 .filter(|s| !s.is_empty())
925 if let Some(vls) = VeryLongString::parse(decoder, tuple).warn_on_error(&warn) {
926 very_long_strings.push(vls)
929 Ok(VeryLongStringRecord(very_long_strings))
933 #[derive(Clone, Debug)]
934 pub struct Attribute {
935 pub name: Identifier,
936 pub values: Vec<String>,
943 warn: &impl Fn(Error),
944 ) -> Result<(Option<Attribute>, &'a str), Error> {
945 let Some((name, mut input)) = input.split_once('(') else {
946 return Err(Error::TBD);
948 let mut values = Vec::new();
950 let Some((value, rest)) = input.split_once('\n') else {
951 return Err(Error::TBD);
953 if let Some(stripped) = value
955 .and_then(|value| value.strip_suffix('\''))
957 values.push(stripped.into());
960 values.push(value.into());
962 if let Some(rest) = rest.strip_prefix(')') {
963 let attribute = Identifier::new(name, decoder.encoding)
964 .map_err(Error::InvalidAttributeName)
966 .map(|name| Attribute { name, values });
967 return Ok((attribute, rest));
974 #[derive(Clone, Debug)]
975 pub struct AttributeSet(pub Vec<Attribute>);
981 sentinel: Option<char>,
982 warn: &impl Fn(Error),
983 ) -> Result<(AttributeSet, &'a str), Error> {
984 let mut attributes = Vec::new();
986 match input.chars().next() {
988 c if c == sentinel => break &input[1..],
990 let (attribute, rest) = Attribute::parse(decoder, input, &warn)?;
991 if let Some(attribute) = attribute {
992 attributes.push(attribute);
998 Ok((AttributeSet(attributes), rest))
1002 #[derive(Clone, Debug)]
1003 pub struct FileAttributeRecord(AttributeSet);
1005 impl FileAttributeRecord {
1006 pub fn parse(decoder: &Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1007 let (set, rest) = AttributeSet::parse(decoder, input, None, &warn)?;
1008 if !rest.is_empty() {
1011 Ok(FileAttributeRecord(set))
1015 #[derive(Clone, Debug)]
1016 pub struct VarAttributeSet {
1017 pub long_var_name: Identifier,
1018 pub attributes: AttributeSet,
1021 impl VarAttributeSet {
1025 warn: &impl Fn(Error),
1026 ) -> Result<(Option<VarAttributeSet>, &'a str), Error> {
1027 let Some((long_var_name, rest)) = input.split_once(':') else {
1028 return Err(Error::TBD);
1030 let (attributes, rest) = AttributeSet::parse(decoder, rest, Some('/'), warn)?;
1031 let var_attribute = Identifier::new(long_var_name, decoder.encoding)
1032 .map_err(Error::InvalidAttributeVariableName)
1033 .warn_on_error(warn)
1034 .map(|name| VarAttributeSet {
1035 long_var_name: name,
1038 Ok((var_attribute, rest))
1042 #[derive(Clone, Debug)]
1043 pub struct VariableAttributeRecord(Vec<VarAttributeSet>);
1045 impl VariableAttributeRecord {
1046 pub fn parse(decoder: &Decoder, mut input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1047 let mut var_attribute_sets = Vec::new();
1048 while !input.is_empty() {
1049 let Some((var_attribute, rest)) =
1050 VarAttributeSet::parse(decoder, input, &warn).warn_on_error(&warn)
1054 if let Some(var_attribute) = var_attribute {
1055 var_attribute_sets.push(var_attribute);
1059 Ok(VariableAttributeRecord(var_attribute_sets))
1063 #[derive(Clone, Debug)]
1064 pub enum MultipleResponseType {
1067 labels: CategoryLabels,
1072 impl MultipleResponseType {
1075 mr_set: &Identifier,
1076 input: &raw::MultipleResponseType,
1077 min_width: VarWidth,
1078 warn: &impl Fn(Error),
1079 ) -> Result<Self, Error> {
1080 let mr_type = match input {
1081 raw::MultipleResponseType::MultipleDichotomy { value, labels } => {
1082 let value = decoder.decode_string_cow(&value.0, warn);
1083 let value = match min_width {
1084 VarWidth::Numeric => {
1085 let number: f64 = value.trim().parse().map_err(|_| {
1086 Error::InvalidMDGroupCountedValue {
1087 mr_set: mr_set.clone(),
1088 number: value.into(),
1091 Value::Number(Some(number.into()))
1093 VarWidth::String(max_width) => {
1094 let value = value.trim_end_matches(' ');
1095 let width = value.len();
1096 if width > max_width as usize {
1097 return Err(Error::TooWideMDGroupCountedValue {
1098 mr_set: mr_set.clone(),
1099 value: value.into(),
1104 Value::String(value.into())
1107 MultipleResponseType::MultipleDichotomy {
1112 raw::MultipleResponseType::MultipleCategory => MultipleResponseType::MultipleCategory,
1118 #[derive(Clone, Debug)]
1119 pub struct MultipleResponseSet {
1120 pub name: Identifier,
1121 pub min_width: VarWidth,
1122 pub max_width: VarWidth,
1124 pub mr_type: MultipleResponseType,
1125 pub dict_indexes: Vec<DictIndex>,
1128 impl MultipleResponseSet {
1131 input: &raw::MultipleResponseSet<Identifier, Cow<str>>,
1132 warn: &impl Fn(Error),
1133 ) -> Result<Self, Error> {
1134 let mr_set_name = input.name.clone();
1135 let mut dict_indexes = Vec::with_capacity(input.short_names.len());
1136 for short_name in input.short_names.iter() {
1137 let Some(&dict_index) = decoder.var_names.get(&short_name) else {
1138 warn(Error::UnknownMrSetVariable {
1139 mr_set: mr_set_name.clone(),
1140 short_name: short_name.clone(),
1144 dict_indexes.push(dict_index);
1147 match dict_indexes.len() {
1148 0 => return Err(Error::EmptyMrSet(mr_set_name)),
1149 1 => return Err(Error::OneVarMrSet(mr_set_name)),
1153 let Some((Some(min_width), Some(max_width))) = dict_indexes
1155 .map(|dict_index| decoder.variables[dict_index].width)
1156 .map(|w| (Some(w), Some(w)))
1157 .reduce(|(na, wa), (nb, wb)| (VarWidth::narrower(na, nb), VarWidth::wider(wa, wb)))
1159 return Err(Error::MixedMrSet(mr_set_name));
1163 MultipleResponseType::decode(decoder, &mr_set_name, &input.mr_type, min_width, warn)?;
1165 Ok(MultipleResponseSet {
1169 label: input.label.to_string(),
1176 #[derive(Clone, Debug)]
1177 pub struct MultipleResponseRecord(pub Vec<MultipleResponseSet>);
1179 impl TryDecode for MultipleResponseRecord {
1180 type Input<'a> = raw::MultipleResponseRecord<Identifier, Cow<'a, str>>;
1183 decoder: &mut Decoder,
1184 input: &Self::Input<'_>,
1185 warn: impl Fn(Error),
1186 ) -> Result<Option<Self>, Error> {
1187 let mut sets = Vec::with_capacity(input.0.len());
1188 for set in &input.0 {
1189 match MultipleResponseSet::decode(decoder, set, &warn) {
1190 Ok(set) => sets.push(set),
1191 Err(error) => warn(error),
1194 Ok(Some(MultipleResponseRecord(sets)))
1198 #[derive(Clone, Debug)]
1199 pub struct LongStringValueLabels {
1200 pub var_name: Identifier,
1201 pub width: VarWidth,
1202 pub labels: Vec<ValueLabel>,
1205 impl LongStringValueLabels {
1208 input: &raw::LongStringValueLabels<RawString>,
1209 warn: &impl Fn(Error),
1210 ) -> Result<Self, Error> {
1211 let var_name = decoder.decode_string(&input.var_name.0, warn);
1212 let var_name = Identifier::new(var_name.trim_end(), decoder.encoding)
1213 .map_err(Error::InvalidLongStringValueLabelName)?;
1216 let max_width = VarWidth::MAX_STRING;
1217 if input.width < 9 || input.width > max_width as u32 {
1218 return Err(Error::InvalidLongValueLabelWidth {
1225 let width = input.width as u16;
1227 let mut labels = Vec::with_capacity(input.labels.len());
1228 for (value, label) in input.labels.iter() {
1229 let value = Value::String(decoder.decode_exact_length(&value.0).into());
1230 let label = decoder.decode_string(&label.0, warn);
1231 labels.push(ValueLabel { value, label });
1234 Ok(LongStringValueLabels {
1236 width: VarWidth::String(width),
1242 #[derive(Clone, Debug)]
1243 pub struct LongStringValueLabelRecord(pub Vec<LongStringValueLabels>);
1245 impl TryDecode for LongStringValueLabelRecord {
1246 type Input<'a> = raw::LongStringValueLabelRecord<RawString>;
1249 decoder: &mut Decoder,
1250 input: &Self::Input<'_>,
1251 warn: impl Fn(Error),
1252 ) -> Result<Option<Self>, Error> {
1253 let mut labels = Vec::with_capacity(input.0.len());
1254 for label in &input.0 {
1255 match LongStringValueLabels::decode(decoder, label, &warn) {
1256 Ok(set) => labels.push(set),
1257 Err(error) => warn(error),
1260 Ok(Some(LongStringValueLabelRecord(labels)))
1266 use encoding_rs::WINDOWS_1252;
1270 let mut s = String::new();
1271 s.push(char::REPLACEMENT_CHARACTER);
1272 let encoded = WINDOWS_1252.encode(&s).0;
1273 let decoded = WINDOWS_1252.decode(&encoded[..]).0;
1274 println!("{:?}", decoded);
1279 let charset: Vec<u8> = (0..=255).collect();
1280 println!("{}", charset.len());
1281 let decoded = WINDOWS_1252.decode(&charset[..]).0;
1282 println!("{}", decoded.len());
1283 let encoded = WINDOWS_1252.encode(&decoded[..]).0;
1284 println!("{}", encoded.len());
1285 assert_eq!(&charset[..], &encoded[..]);