2 borrow::Cow, cell::RefCell, cmp::Ordering, collections::HashMap, iter::repeat, ops::Range,
7 dictionary::{self, Dictionary},
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,
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<raw::Record>,
289 encoding: &'static Encoding,
290 warn: &impl Fn(Error),
291 ) -> Result<(Vec<Record>, Metadata), Error> {
292 let mut decoder = Decoder {
295 warn: Box::new(|error| println!("{error}")),
298 variables: HashMap::new(),
299 var_names: HashMap::new(),
300 dictionary: Dictionary::new(encoding),
302 n_generated_names: 0,
305 let h = Headers::new(&headers, &decoder, warn);
306 let Some(header) = h.header else {
307 return Err(Error::MissingHeaderRecord);
310 let mut output = Vec::with_capacity(headers.len());
312 // Decode the records that don't use variables at all.
313 if let Some(header) = HeaderRecord::try_decode(&mut decoder, &header, warn)? {
314 output.push(Record::Header(header))
316 for document in h.documents {
317 for line in &document.lines {
318 decoder.dictionary.documents.push(line.to_string())
322 for &raw in &h.file_attributes {
323 let s = decoder.decode_string_cow(&raw.text.0, warn);
324 output.push(Record::FileAttributes(FileAttributeRecord::parse(
328 for &raw in &h.other_extensions {
329 output.push(Record::OtherExtension(raw.clone()));
332 // Decode the variable records, which are the basis of almost everything
334 for raw in &h.variables {
335 parse_variable_record(&mut decoder, raw, warn)?;
338 // Decode value labels and weight variable. These use indexes into the
339 // variable records, so we need to parse them before those indexes become
340 // invalidated by very long string variables.
341 for &raw in &h.value_labels {
342 if let Some(value_label) = ValueLabelRecord::try_decode(&mut decoder, raw, warn)? {
343 output.push(Record::ValueLabel(value_label));
347 if let Some(raw) = h.var_display {
348 output.push(Record::VarDisplay(raw.clone()));
351 // Decode records that use short names.
352 for &raw in &h.multiple_response {
353 if let Some(mrr) = MultipleResponseRecord::try_decode(&mut decoder, raw, warn)? {
354 output.push(Record::MultipleResponse(mrr))
357 for &raw in &h.very_long_strings {
358 let s = decoder.decode_string_cow(&raw.text.0, warn);
359 output.push(Record::VeryLongStrings(VeryLongStringRecord::parse(
364 // Rename variables to their long names.
365 for &raw in &h.long_names {
366 let s = decoder.decode_string_cow(&raw.text.0, warn);
367 output.push(Record::LongNames(LongNameRecord::parse(
374 // Decode recods that use long names.
375 for &raw in &h.variable_attributes {
376 let s = decoder.decode_string_cow(&raw.text.0, warn);
377 output.push(Record::VariableAttributes(VariableAttributeRecord::parse(
381 for &raw in &h.long_string_value_labels {
382 if let Some(mrr) = LongStringValueLabelRecord::try_decode(&mut decoder, raw, warn)? {
383 output.push(Record::LongStringValueLabels(mrr))
386 for &raw in &h.long_string_missing_values {
387 if let Some(mrr) = LongStringMissingValuesRecord::try_decode(&mut decoder, raw, warn)? {
388 output.push(Record::LongStringMissingValues(mrr))
391 for &raw in &h.variable_sets {
392 let s = decoder.decode_string_cow(&raw.text.0, warn);
393 output.push(Record::VariableSets(VariableSetRecord::parse(&s, warn)?));
396 let metadata = Metadata::decode(&header, h.integer_info, h.product_info, warn);
397 Ok((output, metadata))
401 fn generate_name(&mut self) -> Identifier {
403 self.n_generated_names += 1;
404 let name = Identifier::new(&format!("VAR{:03}", self.n_generated_names), self.encoding)
406 if !self.var_names.contains_key(&name) {
409 assert!(self.n_generated_names < usize::MAX);
412 fn decode_string_cow<'a>(&self, input: &'a [u8], warn: &impl Fn(Error)) -> Cow<'a, str> {
413 let (output, malformed) = self.encoding.decode_without_bom_handling(input);
415 warn(Error::MalformedString {
416 encoding: self.encoding.name().into(),
417 text: output.clone().into(),
422 fn decode_string(&self, input: &[u8], warn: &impl Fn(Error)) -> String {
423 self.decode_string_cow(input, warn).into()
425 pub fn decode_identifier(
428 warn: &impl Fn(Error),
429 ) -> Result<Identifier, IdError> {
430 let s = self.decode_string_cow(input, warn);
431 Identifier::new(&s, self.encoding)
433 fn get_var_by_index(&self, dict_index: usize) -> Result<&Variable, Error> {
434 let max_index = self.n_dict_indexes;
435 if dict_index == 0 || dict_index > max_index {
436 return Err(Error::InvalidDictIndex {
441 let Some(variable) = self.variables.get(&(dict_index - 1)) else {
442 return Err(Error::DictIndexIsContinuation(dict_index));
447 /// Returns `input` decoded from `self.encoding` into UTF-8 such that
448 /// re-encoding the result back into `self.encoding` will have exactly the
449 /// same length in bytes.
451 /// XXX warn about errors?
452 fn decode_exact_length<'a>(&self, input: &'a [u8]) -> Cow<'a, str> {
453 if let (s, false) = self.encoding.decode_without_bom_handling(input) {
454 // This is the common case. Usually there will be no errors.
457 // Unusual case. Don't bother to optimize it much.
458 let mut decoder = self.encoding.new_decoder_without_bom_handling();
459 let mut output = String::with_capacity(
461 .max_utf8_buffer_length_without_replacement(input.len())
464 let mut rest = input;
465 while !rest.is_empty() {
466 match decoder.decode_to_string_without_replacement(rest, &mut output, true) {
467 (DecoderResult::InputEmpty, _) => break,
468 (DecoderResult::OutputFull, _) => unreachable!(),
469 (DecoderResult::Malformed(a, b), consumed) => {
470 let skipped = a as usize + b as usize;
471 output.extend(repeat('?').take(skipped));
472 rest = &rest[consumed..];
476 assert_eq!(self.encoding.encode(&output).0.len(), input.len());
482 pub trait TryDecode: Sized {
485 decoder: &mut Decoder,
486 input: &Self::Input<'_>,
487 warn: impl Fn(Error),
488 ) -> Result<Option<Self>, Error>;
491 pub trait Decode<Input>: Sized {
492 fn decode(decoder: &Decoder, input: &Input, warn: impl Fn(Error)) -> Self;
495 impl<const N: usize> Decode<RawStr<N>> for String {
496 fn decode(decoder: &Decoder, input: &RawStr<N>, warn: impl Fn(Error)) -> Self {
497 decoder.decode_string(&input.0, &warn)
501 #[derive(Clone, Debug)]
502 pub struct HeaderRecord {
503 pub eye_catcher: String,
504 pub weight_index: Option<usize>,
505 pub n_cases: Option<u64>,
506 pub creation: NaiveDateTime,
507 pub file_label: String,
510 fn trim_end_spaces(mut s: String) -> String {
511 s.truncate(s.trim_end_matches(' ').len());
515 /// Data file info that doesn't fit in [Dictionary].
516 pub struct Metadata {
517 creation: NaiveDateTime,
519 compression: Option<Compression>,
520 n_cases: Option<u64>,
522 product_ext: Option<String>,
523 version: Option<(i32, i32, i32)>,
528 header: &crate::raw::HeaderRecord<Cow<str>>,
529 integer_info: Option<&IntegerInfoRecord>,
530 product_ext: Option<&ProductInfoRecord>,
531 warn: impl Fn(Error),
533 let creation_date = NaiveDate::parse_from_str(&header.creation_date, "%e %b %Y")
534 .unwrap_or_else(|_| {
535 warn(Error::InvalidCreationDate {
536 creation_date: header.creation_date.to_string(),
540 let creation_time = NaiveTime::parse_from_str(&header.creation_time, "%H:%M:%S")
541 .unwrap_or_else(|_| {
542 warn(Error::InvalidCreationTime {
543 creation_time: header.creation_time.to_string(),
547 let creation = NaiveDateTime::new(creation_date, creation_time);
551 .trim_start_matches("@(#) SPSS DATA FILE")
557 endian: header.endian,
558 compression: header.compression,
559 n_cases: header.n_cases.map(|n| n as u64),
561 product_ext: product_ext.map(|pe| pe.0.clone()),
562 version: integer_info.map(|ii| ii.version),
567 impl TryDecode for HeaderRecord {
568 type Input<'a> = crate::raw::HeaderRecord<Cow<'a, str>>;
571 _decoder: &mut Decoder,
572 input: &Self::Input<'_>,
573 warn: impl Fn(Error),
574 ) -> Result<Option<Self>, Error> {
575 let eye_catcher = trim_end_spaces(input.eye_catcher.to_string());
576 let file_label = trim_end_spaces(input.file_label.to_string());
577 let creation_date = NaiveDate::parse_from_str(&input.creation_date, "%e %b %Y")
578 .unwrap_or_else(|_| {
579 warn(Error::InvalidCreationDate {
580 creation_date: input.creation_date.to_string(),
584 let creation_time = NaiveTime::parse_from_str(&input.creation_time, "%H:%M:%S")
585 .unwrap_or_else(|_| {
586 warn(Error::InvalidCreationTime {
587 creation_time: input.creation_time.to_string(),
591 Ok(Some(HeaderRecord {
593 weight_index: input.weight_index.map(|n| n as usize),
594 n_cases: input.n_cases.map(|n| n as u64),
595 creation: NaiveDateTime::new(creation_date, creation_time),
601 #[derive(Clone, Debug)]
602 pub struct VariableRecord {
604 pub name: Identifier,
605 pub print_format: Spec,
606 pub write_format: Spec,
607 pub missing_values: MissingValues<String>,
608 pub label: Option<String>,
611 fn decode_format(raw: raw::Spec, width: VarWidth, warn: impl Fn(Spec, FormatError)) -> Spec {
612 UncheckedSpec::try_from(raw)
613 .and_then(Spec::try_from)
614 .and_then(|x| x.check_width_compatibility(width))
615 .unwrap_or_else(|error| {
616 let new_format = Spec::default_for_width(width);
617 warn(new_format, error);
622 fn parse_variable_record(
623 decoder: &mut Decoder,
624 input: &raw::VariableRecord<Cow<str>, String>,
625 warn: impl Fn(Error),
626 ) -> Result<(), Error> {
627 let width = match input.width {
628 0 => VarWidth::Numeric,
629 w @ 1..=255 => VarWidth::String(w as u16),
632 return Err(Error::InvalidVariableWidth {
633 offsets: input.offsets.clone(),
638 let name = trim_end_spaces(input.name.to_string());
639 let name = match Identifier::new(&name, decoder.encoding) {
641 if !decoder.var_names.contains_key(&name) {
644 let new_name = decoder.generate_name();
645 warn(Error::DuplicateVariableName {
646 duplicate_name: name.clone(),
647 new_name: new_name.clone(),
653 let new_name = decoder.generate_name();
654 warn(Error::InvalidVariableName {
656 new_name: new_name.clone(),
661 let variable = Variable {
662 dict_index: decoder.n_dict_indexes,
663 short_name: name.clone(),
667 decoder.n_dict_indexes += width.n_dict_indexes();
670 .insert(name.clone(), variable.dict_index)
674 .insert(variable.dict_index, variable)
677 let print_format = decode_format(input.print_format, width, |new_spec, format_error| {
678 warn(Error::InvalidPrintFormat {
680 variable: name.clone(),
684 let write_format = decode_format(input.write_format, width, |new_spec, format_error| {
685 warn(Error::InvalidWriteFormat {
687 variable: name.clone(),
691 let mut variable = dictionary::Variable::new(name, width);
692 variable.print_format = print_format;
693 variable.write_format = write_format;
694 variable.missing_values = input.missing_values.clone();
695 if let Some(ref label) = input.label {
696 variable.label = Some(label.to_string());
698 decoder.dictionary.add_var(variable).unwrap();
702 #[derive(Clone, Debug)]
703 pub struct DocumentRecord(Vec<String>);
705 impl TryDecode for DocumentRecord {
706 type Input<'a> = crate::raw::DocumentRecord<RawDocumentLine>;
709 decoder: &mut Decoder,
710 input: &Self::Input<'_>,
711 warn: impl Fn(Error),
712 ) -> Result<Option<Self>, Error> {
713 Ok(Some(DocumentRecord(
717 .map(|s| trim_end_spaces(decoder.decode_string(&s.0, &warn)))
727 const NAME: &'static str;
728 fn parse(input: &str, warn: impl Fn(Error)) -> Result<Self, Error>;
731 #[derive(Clone, Debug)]
732 pub struct VariableSet {
734 pub vars: Vec<String>,
738 fn parse(input: &str) -> Result<Self, Error> {
739 let (name, input) = input.split_once('=').ok_or(Error::TBD)?;
740 let vars = input.split_ascii_whitespace().map(String::from).collect();
748 trait WarnOnError<T> {
749 fn warn_on_error<F: Fn(Error)>(self, warn: &F) -> Option<T>;
751 impl<T> WarnOnError<T> for Result<T, Error> {
752 fn warn_on_error<F: Fn(Error)>(self, warn: &F) -> Option<T> {
754 Ok(result) => Some(result),
763 #[derive(Clone, Debug)]
764 pub struct ValueLabel {
769 #[derive(Clone, Debug)]
770 pub struct ValueLabelRecord {
771 pub var_type: VarType,
772 pub labels: Vec<ValueLabel>,
773 pub variables: Vec<Identifier>,
776 impl TryDecode for ValueLabelRecord {
777 type Input<'a> = crate::raw::ValueLabelRecord<RawStr<8>, RawString>;
779 decoder: &mut Decoder,
780 input: &Self::Input<'_>,
781 warn: impl Fn(Error),
782 ) -> Result<Option<ValueLabelRecord>, Error> {
783 let variables: Vec<&Variable> = input
786 .filter_map(|&dict_index| {
788 .get_var_by_index(dict_index as usize)
789 .warn_on_error(&warn)
791 .filter(|&variable| match variable.width {
792 VarWidth::String(width) if width > 8 => {
793 warn(Error::InvalidLongStringValueLabel(
794 variable.short_name.clone(),
801 let mut i = variables.iter();
802 let Some(&first_var) = i.next() else {
805 let var_type: VarType = first_var.width.into();
807 let this_type: VarType = variable.width.into();
808 if var_type != this_type {
809 let (numeric_var, string_var) = match var_type {
810 VarType::Numeric => (first_var, variable),
811 VarType::String => (variable, first_var),
813 warn(Error::ValueLabelsDifferentTypes {
814 numeric_var: numeric_var.short_name.clone(),
815 string_var: string_var.short_name.clone(),
823 .map(|raw::ValueLabel { value, label }| {
824 let label = decoder.decode_string(&label.0, &warn);
825 let value = Value::decode(value, decoder);
826 ValueLabel { value, label }
829 let variables = variables
831 .map(|&variable| variable.short_name.clone())
833 Ok(Some(ValueLabelRecord {
841 #[derive(Clone, Debug)]
842 pub struct VariableSetRecord(Vec<VariableSet>);
844 impl TextRecord for VariableSetRecord {
845 const NAME: &'static str = "variable set";
846 fn parse(input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
847 let mut sets = Vec::new();
848 for line in input.lines() {
849 if let Some(set) = VariableSet::parse(line).warn_on_error(&warn) {
853 Ok(VariableSetRecord(sets))
857 #[derive(Clone, Debug)]
858 pub struct LongName {
859 pub short_name: Identifier,
860 pub long_name: Identifier,
864 fn new(decoder: &mut Decoder, short_name: &str, long_name: &str) -> Result<LongName, Error> {
866 Identifier::new(short_name, decoder.encoding).map_err(Error::InvalidShortName)?;
868 Identifier::new(long_name, decoder.encoding).map_err(Error::InvalidLongName)?;
876 #[derive(Clone, Debug)]
877 pub struct LongNameRecord(Vec<LongName>);
879 impl LongNameRecord {
880 pub fn parse(decoder: &mut Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
881 let mut names = Vec::new();
882 for pair in input.split('\t').filter(|s| !s.is_empty()) {
883 if let Some((short_name, long_name)) = pair.split_once('=') {
884 if let Some(long_name) =
885 LongName::new(decoder, short_name, long_name).warn_on_error(&warn)
887 names.push(long_name);
893 Ok(LongNameRecord(names))
897 #[derive(Clone, Debug)]
898 pub struct VeryLongString {
899 pub short_name: Identifier,
903 impl VeryLongString {
904 fn parse(decoder: &Decoder, input: &str) -> Result<VeryLongString, Error> {
905 let Some((short_name, length)) = input.split_once('=') else {
906 return Err(Error::TBD);
909 Identifier::new(short_name, decoder.encoding).map_err(Error::InvalidLongStringName)?;
910 let length: u16 = length.parse().map_err(|_| Error::TBD)?;
911 if length > VarWidth::MAX_STRING {
912 return Err(Error::TBD);
914 Ok(VeryLongString { short_name, length })
918 #[derive(Clone, Debug)]
919 pub struct VeryLongStringRecord(Vec<VeryLongString>);
921 impl VeryLongStringRecord {
922 pub fn parse(decoder: &Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
923 let mut very_long_strings = Vec::new();
926 .map(|s| s.trim_end_matches('\t'))
927 .filter(|s| !s.is_empty())
929 if let Some(vls) = VeryLongString::parse(decoder, tuple).warn_on_error(&warn) {
930 very_long_strings.push(vls)
933 Ok(VeryLongStringRecord(very_long_strings))
937 #[derive(Clone, Debug)]
938 pub struct Attribute {
939 pub name: Identifier,
940 pub values: Vec<String>,
947 warn: &impl Fn(Error),
948 ) -> Result<(Option<Attribute>, &'a str), Error> {
949 let Some((name, mut input)) = input.split_once('(') else {
950 return Err(Error::TBD);
952 let mut values = Vec::new();
954 let Some((value, rest)) = input.split_once('\n') else {
955 return Err(Error::TBD);
957 if let Some(stripped) = value
959 .and_then(|value| value.strip_suffix('\''))
961 values.push(stripped.into());
964 values.push(value.into());
966 if let Some(rest) = rest.strip_prefix(')') {
967 let attribute = Identifier::new(name, decoder.encoding)
968 .map_err(Error::InvalidAttributeName)
970 .map(|name| Attribute { name, values });
971 return Ok((attribute, rest));
978 #[derive(Clone, Debug)]
979 pub struct AttributeSet(pub Vec<Attribute>);
985 sentinel: Option<char>,
986 warn: &impl Fn(Error),
987 ) -> Result<(AttributeSet, &'a str), Error> {
988 let mut attributes = Vec::new();
990 match input.chars().next() {
992 c if c == sentinel => break &input[1..],
994 let (attribute, rest) = Attribute::parse(decoder, input, &warn)?;
995 if let Some(attribute) = attribute {
996 attributes.push(attribute);
1002 Ok((AttributeSet(attributes), rest))
1006 #[derive(Clone, Debug)]
1007 pub struct FileAttributeRecord(AttributeSet);
1009 impl FileAttributeRecord {
1010 pub fn parse(decoder: &Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1011 let (set, rest) = AttributeSet::parse(decoder, input, None, &warn)?;
1012 if !rest.is_empty() {
1015 Ok(FileAttributeRecord(set))
1019 #[derive(Clone, Debug)]
1020 pub struct VarAttributeSet {
1021 pub long_var_name: Identifier,
1022 pub attributes: AttributeSet,
1025 impl VarAttributeSet {
1029 warn: &impl Fn(Error),
1030 ) -> Result<(Option<VarAttributeSet>, &'a str), Error> {
1031 let Some((long_var_name, rest)) = input.split_once(':') else {
1032 return Err(Error::TBD);
1034 let (attributes, rest) = AttributeSet::parse(decoder, rest, Some('/'), warn)?;
1035 let var_attribute = Identifier::new(long_var_name, decoder.encoding)
1036 .map_err(Error::InvalidAttributeVariableName)
1037 .warn_on_error(warn)
1038 .map(|name| VarAttributeSet {
1039 long_var_name: name,
1042 Ok((var_attribute, rest))
1046 #[derive(Clone, Debug)]
1047 pub struct VariableAttributeRecord(Vec<VarAttributeSet>);
1049 impl VariableAttributeRecord {
1050 pub fn parse(decoder: &Decoder, mut input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1051 let mut var_attribute_sets = Vec::new();
1052 while !input.is_empty() {
1053 let Some((var_attribute, rest)) =
1054 VarAttributeSet::parse(decoder, input, &warn).warn_on_error(&warn)
1058 if let Some(var_attribute) = var_attribute {
1059 var_attribute_sets.push(var_attribute);
1063 Ok(VariableAttributeRecord(var_attribute_sets))
1067 #[derive(Clone, Debug)]
1068 pub enum MultipleResponseType {
1071 labels: CategoryLabels,
1076 impl MultipleResponseType {
1079 mr_set: &Identifier,
1080 input: &raw::MultipleResponseType,
1081 min_width: VarWidth,
1082 warn: &impl Fn(Error),
1083 ) -> Result<Self, Error> {
1084 let mr_type = match input {
1085 raw::MultipleResponseType::MultipleDichotomy { value, labels } => {
1086 let value = decoder.decode_string_cow(&value.0, warn);
1087 let value = match min_width {
1088 VarWidth::Numeric => {
1089 let number: f64 = value.trim().parse().map_err(|_| {
1090 Error::InvalidMDGroupCountedValue {
1091 mr_set: mr_set.clone(),
1092 number: value.into(),
1095 Value::Number(Some(number.into()))
1097 VarWidth::String(max_width) => {
1098 let value = value.trim_end_matches(' ');
1099 let width = value.len();
1100 if width > max_width as usize {
1101 return Err(Error::TooWideMDGroupCountedValue {
1102 mr_set: mr_set.clone(),
1103 value: value.into(),
1108 Value::String(value.into())
1111 MultipleResponseType::MultipleDichotomy {
1116 raw::MultipleResponseType::MultipleCategory => MultipleResponseType::MultipleCategory,
1122 #[derive(Clone, Debug)]
1123 pub struct MultipleResponseSet {
1124 pub name: Identifier,
1125 pub min_width: VarWidth,
1126 pub max_width: VarWidth,
1128 pub mr_type: MultipleResponseType,
1129 pub dict_indexes: Vec<DictIndex>,
1132 impl MultipleResponseSet {
1135 input: &raw::MultipleResponseSet<Identifier, Cow<str>>,
1136 warn: &impl Fn(Error),
1137 ) -> Result<Self, Error> {
1138 let mr_set_name = input.name.clone();
1139 let mut dict_indexes = Vec::with_capacity(input.short_names.len());
1140 for short_name in input.short_names.iter() {
1141 let Some(&dict_index) = decoder.var_names.get(&short_name) else {
1142 warn(Error::UnknownMrSetVariable {
1143 mr_set: mr_set_name.clone(),
1144 short_name: short_name.clone(),
1148 dict_indexes.push(dict_index);
1151 match dict_indexes.len() {
1152 0 => return Err(Error::EmptyMrSet(mr_set_name)),
1153 1 => return Err(Error::OneVarMrSet(mr_set_name)),
1157 let Some((Some(min_width), Some(max_width))) = dict_indexes
1159 .map(|dict_index| decoder.variables[dict_index].width)
1160 .map(|w| (Some(w), Some(w)))
1161 .reduce(|(na, wa), (nb, wb)| (VarWidth::narrower(na, nb), VarWidth::wider(wa, wb)))
1163 return Err(Error::MixedMrSet(mr_set_name));
1167 MultipleResponseType::decode(decoder, &mr_set_name, &input.mr_type, min_width, warn)?;
1169 Ok(MultipleResponseSet {
1173 label: input.label.to_string(),
1180 #[derive(Clone, Debug)]
1181 pub struct MultipleResponseRecord(pub Vec<MultipleResponseSet>);
1183 impl TryDecode for MultipleResponseRecord {
1184 type Input<'a> = raw::MultipleResponseRecord<Identifier, Cow<'a, str>>;
1187 decoder: &mut Decoder,
1188 input: &Self::Input<'_>,
1189 warn: impl Fn(Error),
1190 ) -> Result<Option<Self>, Error> {
1191 let mut sets = Vec::with_capacity(input.0.len());
1192 for set in &input.0 {
1193 match MultipleResponseSet::decode(decoder, set, &warn) {
1194 Ok(set) => sets.push(set),
1195 Err(error) => warn(error),
1198 Ok(Some(MultipleResponseRecord(sets)))
1202 #[derive(Clone, Debug)]
1203 pub struct LongStringValueLabels {
1204 pub var_name: Identifier,
1205 pub width: VarWidth,
1206 pub labels: Vec<ValueLabel>,
1209 impl LongStringValueLabels {
1212 input: &raw::LongStringValueLabels<RawString>,
1213 warn: &impl Fn(Error),
1214 ) -> Result<Self, Error> {
1215 let var_name = decoder.decode_string(&input.var_name.0, warn);
1216 let var_name = Identifier::new(var_name.trim_end(), decoder.encoding)
1217 .map_err(Error::InvalidLongStringValueLabelName)?;
1220 let max_width = VarWidth::MAX_STRING;
1221 if input.width < 9 || input.width > max_width as u32 {
1222 return Err(Error::InvalidLongValueLabelWidth {
1229 let width = input.width as u16;
1231 let mut labels = Vec::with_capacity(input.labels.len());
1232 for (value, label) in input.labels.iter() {
1233 let value = Value::String(decoder.decode_exact_length(&value.0).into());
1234 let label = decoder.decode_string(&label.0, warn);
1235 labels.push(ValueLabel { value, label });
1238 Ok(LongStringValueLabels {
1240 width: VarWidth::String(width),
1246 #[derive(Clone, Debug)]
1247 pub struct LongStringValueLabelRecord(pub Vec<LongStringValueLabels>);
1249 impl TryDecode for LongStringValueLabelRecord {
1250 type Input<'a> = raw::LongStringValueLabelRecord<RawString>;
1253 decoder: &mut Decoder,
1254 input: &Self::Input<'_>,
1255 warn: impl Fn(Error),
1256 ) -> Result<Option<Self>, Error> {
1257 let mut labels = Vec::with_capacity(input.0.len());
1258 for label in &input.0 {
1259 match LongStringValueLabels::decode(decoder, label, &warn) {
1260 Ok(set) => labels.push(set),
1261 Err(error) => warn(error),
1264 Ok(Some(LongStringValueLabelRecord(labels)))
1270 use encoding_rs::WINDOWS_1252;
1274 let mut s = String::new();
1275 s.push(char::REPLACEMENT_CHARACTER);
1276 let encoded = WINDOWS_1252.encode(&s).0;
1277 let decoded = WINDOWS_1252.decode(&encoded[..]).0;
1278 println!("{:?}", decoded);
1283 let charset: Vec<u8> = (0..=255).collect();
1284 println!("{}", charset.len());
1285 let decoded = WINDOWS_1252.decode(&charset[..]).0;
1286 println!("{}", decoded.len());
1287 let encoded = WINDOWS_1252.encode(&decoded[..]).0;
1288 println!("{}", encoded.len());
1289 assert_eq!(&charset[..], &encoded[..]);