2 borrow::Cow, cell::RefCell, cmp::Ordering, collections::HashMap, iter::repeat, ops::Range,
7 encoding::{default_encoding, get_encoding, Error as EncodingError},
9 format::{Error as FormatError, Spec, UncheckedSpec},
10 identifier::{Error as IdError, Identifier},
11 raw::{self, RawDocumentLine, RawStr, RawString, VarDisplayRecord, VarType},
13 use chrono::{NaiveDate, NaiveDateTime, NaiveTime};
14 use encoding_rs::{DecoderResult, Encoding};
15 use num::integer::div_ceil;
16 use ordered_float::OrderedFloat;
17 use thiserror::Error as ThisError;
19 pub use crate::raw::{CategoryLabels, Compression};
21 #[derive(ThisError, Debug)]
23 // XXX this is really an internal error and maybe we should change the
24 // interfaces to make it impossible
25 #[error("Missing header record")]
29 EncodingError(EncodingError),
31 #[error("Using default encoding {0}.")]
32 UsingDefaultEncoding(String),
34 #[error("Variable record from offset {:x} to {:x} specifies width {width} not in valid range [-1,255).", offsets.start, offsets.end)]
35 InvalidVariableWidth { offsets: Range<u64>, width: i32 },
37 #[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.")]
38 InvalidLongMissingValueFormat,
40 #[error("File creation date {creation_date} is not in the expected format \"DD MMM YY\" format. Using 01 Jan 1970.")]
41 InvalidCreationDate { creation_date: String },
43 #[error("File creation time {creation_time} is not in the expected format \"HH:MM:SS\" format. Using midnight.")]
44 InvalidCreationTime { creation_time: String },
46 #[error("{id_error} Renaming variable to {new_name}.")]
53 "Substituting {new_spec} for invalid print format on variable {variable}. {format_error}"
58 format_error: FormatError,
62 "Substituting {new_spec} for invalid write format on variable {variable}. {format_error}"
67 format_error: FormatError,
70 #[error("Renaming variable with duplicate name {duplicate_name} to {new_name}.")]
71 DuplicateVariableName {
72 duplicate_name: Identifier,
76 #[error("Dictionary index {dict_index} is outside valid range [1,{max_index}].")]
77 InvalidDictIndex { dict_index: usize, max_index: usize },
79 #[error("Dictionary index {0} refers to a long string continuation.")]
80 DictIndexIsContinuation(usize),
82 #[error("Variables associated with value label are not all of identical type. Variable {numeric_var} is numeric, but variable {string_var} is string.")]
83 ValueLabelsDifferentTypes {
84 numeric_var: Identifier,
85 string_var: Identifier,
89 "Value labels may not be added to long string variable {0} using record types 3 or 4."
91 InvalidLongStringValueLabel(Identifier),
93 #[error("Invalid multiple response set name. {0}")]
94 InvalidMrSetName(IdError),
96 #[error("Multiple response set {mr_set} includes unknown variable {short_name}.")]
97 UnknownMrSetVariable {
99 short_name: Identifier,
102 #[error("Multiple response set {0} has no variables.")]
103 EmptyMrSet(Identifier),
105 #[error("Multiple response set {0} has only one variable.")]
106 OneVarMrSet(Identifier),
108 #[error("Multiple response set {0} contains both string and numeric variables.")]
109 MixedMrSet(Identifier),
112 "Invalid numeric format for counted value {number} in multiple response set {mr_set}."
114 InvalidMDGroupCountedValue { mr_set: Identifier, number: String },
116 #[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}.")]
117 TooWideMDGroupCountedValue {
124 #[error("Long string value label for variable {name} has width {width}, which is not in the valid range [{min_width},{max_width}].")]
125 InvalidLongValueLabelWidth {
132 #[error("Invalid attribute name. {0}")]
133 InvalidAttributeName(IdError),
135 #[error("Invalid short name in long variable name record. {0}")]
136 InvalidShortName(IdError),
138 #[error("Invalid name in long variable name record. {0}")]
139 InvalidLongName(IdError),
141 #[error("Invalid variable name in very long string record. {0}")]
142 InvalidLongStringName(IdError),
144 #[error("Invalid variable name in long string value label record. {0}")]
145 InvalidLongStringValueLabelName(IdError),
147 #[error("Invalid variable name in attribute record. {0}")]
148 InvalidAttributeVariableName(IdError),
150 // XXX This is risky because `text` might be arbitarily long.
151 #[error("Text string contains invalid bytes for {encoding} encoding: {text}")]
152 MalformedString { encoding: String, text: String },
154 #[error("Details TBD")]
158 #[derive(Clone, Debug)]
160 Header(HeaderRecord),
161 Variable(VariableRecord),
162 ValueLabel(ValueLabelRecord),
163 Document(DocumentRecord),
164 IntegerInfo(IntegerInfoRecord),
165 FloatInfo(FloatInfoRecord),
166 VariableSets(VariableSetRecord),
167 VarDisplay(VarDisplayRecord),
168 MultipleResponse(MultipleResponseRecord),
169 LongStringMissingValues(LongStringMissingValuesRecord),
170 LongStringValueLabels(LongStringValueLabelRecord),
171 Encoding(EncodingRecord),
172 NumberOfCases(NumberOfCasesRecord),
173 ProductInfo(ProductInfoRecord),
174 LongNames(LongNameRecord),
175 VeryLongStrings(VeryLongStringRecord),
176 FileAttributes(FileAttributeRecord),
177 VariableAttributes(VariableAttributeRecord),
178 OtherExtension(Extension),
182 pub use crate::raw::EncodingRecord;
183 pub use crate::raw::Extension;
184 pub use crate::raw::FloatInfoRecord;
185 pub use crate::raw::IntegerInfoRecord;
186 pub use crate::raw::NumberOfCasesRecord;
188 type DictIndex = usize;
190 pub struct Variable {
191 pub dict_index: DictIndex,
192 pub short_name: Identifier,
193 pub long_name: Option<Identifier>,
198 pub compression: Option<Compression>,
200 pub encoding: &'static Encoding,
201 pub variables: HashMap<DictIndex, Variable>,
202 pub var_names: HashMap<Identifier, DictIndex>,
203 n_dict_indexes: usize,
204 n_generated_names: usize,
209 header: Option<&'a raw::HeaderRecord<RawString>>,
210 variables: Vec<&'a raw::VariableRecord<RawString, RawStr<8>>>,
211 value_labels: Vec<&'a raw::ValueLabelRecord<RawStr<8>, RawString>>,
212 document: Option<&'a raw::DocumentRecord<RawDocumentLine>>,
213 integer_info: Option<&'a raw::IntegerInfoRecord>,
214 float_info: Option<&'a raw::FloatInfoRecord>,
215 variable_sets: Vec<&'a raw::TextRecord>,
216 var_display: Option<&'a raw::VarDisplayRecord>,
217 multiple_response: Vec<&'a raw::MultipleResponseRecord<RawString>>,
218 long_string_value_labels: Vec<&'a raw::LongStringValueLabelRecord>,
219 long_string_missing_values: Vec<&'a raw::LongStringMissingValueRecord>,
220 encoding: Option<&'a raw::EncodingRecord>,
221 number_of_cases: Option<&'a raw::NumberOfCasesRecord>,
222 product_info: Option<&'a raw::TextRecord>,
223 long_names: Option<&'a raw::TextRecord>,
224 very_long_strings: Vec<&'a raw::TextRecord>,
225 file_attributes: Vec<&'a raw::TextRecord>,
226 variable_attributes: Vec<&'a raw::TextRecord>,
227 other_extensions: Vec<&'a raw::Extension>,
228 cases: Option<&'a Rc<RefCell<raw::Cases>>>,
231 fn set_or_warn<T>(option: &mut Option<T>, value: T, warn: &impl Fn(Error)) {
232 if option.is_none() {
233 let _ = option.insert(value);
239 impl<'a> Headers<'a> {
240 fn new(headers: &'a Vec<raw::Record>, warn: &impl Fn(Error)) -> Headers<'a> {
241 let mut h = Headers::default();
242 for header in headers {
244 raw::Record::Header(record) => set_or_warn(&mut h.header, record, warn),
245 raw::Record::Variable(record) => h.variables.push(record),
246 raw::Record::ValueLabel(record) => h.value_labels.push(record),
247 raw::Record::Document(record) => set_or_warn(&mut h.document, record, warn),
248 raw::Record::IntegerInfo(record) => set_or_warn(&mut h.integer_info, record, warn),
249 raw::Record::FloatInfo(record) => set_or_warn(&mut h.float_info, record, warn),
250 raw::Record::VariableSets(record) => h.variable_sets.push(record),
251 raw::Record::VarDisplay(record) => set_or_warn(&mut h.var_display, record, warn),
252 raw::Record::MultipleResponse(record) => h.multiple_response.push(record),
253 raw::Record::LongStringValueLabels(record) => {
254 h.long_string_value_labels.push(record)
256 raw::Record::LongStringMissingValues(record) => {
257 h.long_string_missing_values.push(record)
259 raw::Record::Encoding(record) => set_or_warn(&mut h.encoding, record, warn),
260 raw::Record::NumberOfCases(record) => {
261 set_or_warn(&mut h.number_of_cases, record, warn)
263 raw::Record::ProductInfo(record) => set_or_warn(&mut h.product_info, record, warn),
264 raw::Record::LongNames(record) => set_or_warn(&mut h.long_names, record, warn),
265 raw::Record::VeryLongStrings(record) => h.very_long_strings.push(record),
266 raw::Record::FileAttributes(record) => h.file_attributes.push(record),
267 raw::Record::VariableAttributes(record) => h.variable_attributes.push(record),
268 raw::Record::OtherExtension(record) => h.other_extensions.push(record),
269 raw::Record::EndOfHeaders(_) => (),
270 raw::Record::ZHeader(_) => (),
271 raw::Record::ZTrailer(_) => (),
272 raw::Record::Cases(record) => set_or_warn(&mut h.cases, record, warn),
280 headers: Vec<raw::Record>,
281 encoding: Option<&'static Encoding>,
282 warn: &impl Fn(Error),
283 ) -> Result<Vec<Record>, Error> {
284 let h = Headers::new(&headers, warn);
285 let Some(header) = h.header else {
286 return Err(Error::MissingHeaderRecord);
288 let encoding = match encoding {
289 Some(encoding) => encoding,
291 let encoding = h.encoding.map(|record| record.0.as_str());
292 let character_code = h.integer_info.map(|record| record.character_code);
293 match get_encoding(encoding, character_code) {
294 Ok(encoding) => encoding,
295 Err(err @ EncodingError::Ebcdic) => return Err(Error::EncodingError(err)),
297 warn(Error::EncodingError(err));
298 // Warn that we're using the default encoding.
305 //let mut dictionary = Dictionary::new(encoding);
307 let mut decoder = Decoder {
308 compression: header.compression,
309 endian: header.endian,
311 variables: HashMap::new(),
312 var_names: HashMap::new(),
314 n_generated_names: 0,
317 let mut output = Vec::with_capacity(headers.len());
319 // Decode the records that don't use variables at all.
320 if let Some(header) = HeaderRecord::try_decode(&mut decoder, header, warn)? {
321 output.push(Record::Header(header))
323 if let Some(raw) = h.document {
324 if let Some(document) = DocumentRecord::try_decode(&mut decoder, raw, warn)? {
325 output.push(Record::Document(document))
328 if let Some(raw) = h.integer_info {
329 output.push(Record::IntegerInfo(raw.clone()));
331 if let Some(raw) = h.float_info {
332 output.push(Record::FloatInfo(raw.clone()));
334 if let Some(raw) = h.product_info {
335 let s = decoder.decode_string_cow(&raw.text.0, warn);
336 output.push(Record::ProductInfo(ProductInfoRecord::parse(&s, warn)?));
338 if let Some(raw) = h.number_of_cases {
339 output.push(Record::NumberOfCases(raw.clone()))
341 for &raw in &h.file_attributes {
342 let s = decoder.decode_string_cow(&raw.text.0, warn);
343 output.push(Record::FileAttributes(FileAttributeRecord::parse(
347 for &raw in &h.other_extensions {
348 output.push(Record::OtherExtension(raw.clone()));
351 // Decode the variable records, which are the basis of almost everything
353 for &raw in &h.variables {
354 if let Some(variable) = VariableRecord::try_decode(&mut decoder, raw, warn)? {
355 output.push(Record::Variable(variable));
359 // Decode value labels and weight variable. These use indexes into the
360 // variable records, so we need to parse them before those indexes become
361 // invalidated by very long string variables.
362 for &raw in &h.value_labels {
363 if let Some(value_label) = ValueLabelRecord::try_decode(&mut decoder, raw, warn)? {
364 output.push(Record::ValueLabel(value_label));
368 if let Some(raw) = h.var_display {
369 output.push(Record::VarDisplay(raw.clone()));
372 // Decode records that use short names.
374 for &raw in &h.multiple_response {
375 if let Some(mrr) = MultipleResponseRecord::try_decode(&mut decoder, raw, warn)? {
376 output.push(Record::MultipleResponse(mrr))
380 for &raw in &h.very_long_strings {
381 let s = decoder.decode_string_cow(&raw.text.0, warn);
382 output.push(Record::VeryLongStrings(VeryLongStringRecord::parse(
387 // Rename variables to their long names.
388 for &raw in &h.long_names {
389 let s = decoder.decode_string_cow(&raw.text.0, warn);
390 output.push(Record::LongNames(LongNameRecord::parse(
397 // Decode recods that use long names.
398 for &raw in &h.variable_attributes {
399 let s = decoder.decode_string_cow(&raw.text.0, warn);
400 output.push(Record::VariableAttributes(VariableAttributeRecord::parse(
404 for &raw in &h.long_string_value_labels {
405 if let Some(mrr) = LongStringValueLabelRecord::try_decode(&mut decoder, raw, warn)? {
406 output.push(Record::LongStringValueLabels(mrr))
409 for &raw in &h.long_string_missing_values {
410 if let Some(mrr) = LongStringMissingValuesRecord::try_decode(&mut decoder, raw, warn)? {
411 output.push(Record::LongStringMissingValues(mrr))
414 for &raw in &h.variable_sets {
415 let s = decoder.decode_string_cow(&raw.text.0, warn);
416 output.push(Record::VariableSets(VariableSetRecord::parse(&s, warn)?));
422 fn generate_name(&mut self) -> Identifier {
424 self.n_generated_names += 1;
425 let name = Identifier::new(&format!("VAR{:03}", self.n_generated_names), self.encoding)
427 if !self.var_names.contains_key(&name) {
430 assert!(self.n_generated_names < usize::MAX);
433 fn decode_string_cow<'a>(&self, input: &'a [u8], warn: &impl Fn(Error)) -> Cow<'a, str> {
434 let (output, malformed) = self.encoding.decode_without_bom_handling(input);
436 warn(Error::MalformedString {
437 encoding: self.encoding.name().into(),
438 text: output.clone().into(),
443 fn decode_string(&self, input: &[u8], warn: &impl Fn(Error)) -> String {
444 self.decode_string_cow(input, warn).into()
446 pub fn decode_identifier(
449 warn: &impl Fn(Error),
450 ) -> Result<Identifier, IdError> {
451 let s = self.decode_string_cow(input, warn);
452 Identifier::new(&s, self.encoding)
454 fn get_var_by_index(&self, dict_index: usize) -> Result<&Variable, Error> {
455 let max_index = self.n_dict_indexes;
456 if dict_index == 0 || dict_index > max_index {
457 return Err(Error::InvalidDictIndex {
462 let Some(variable) = self.variables.get(&(dict_index - 1)) else {
463 return Err(Error::DictIndexIsContinuation(dict_index));
468 /// Returns `input` decoded from `self.encoding` into UTF-8 such that
469 /// re-encoding the result back into `self.encoding` will have exactly the
470 /// same length in bytes.
472 /// XXX warn about errors?
473 fn decode_exact_length<'a>(&self, input: &'a [u8]) -> Cow<'a, str> {
474 if let (s, false) = self.encoding.decode_without_bom_handling(input) {
475 // This is the common case. Usually there will be no errors.
478 // Unusual case. Don't bother to optimize it much.
479 let mut decoder = self.encoding.new_decoder_without_bom_handling();
480 let mut output = String::with_capacity(
482 .max_utf8_buffer_length_without_replacement(input.len())
485 let mut rest = input;
486 while !rest.is_empty() {
487 match decoder.decode_to_string_without_replacement(rest, &mut output, true) {
488 (DecoderResult::InputEmpty, _) => break,
489 (DecoderResult::OutputFull, _) => unreachable!(),
490 (DecoderResult::Malformed(a, b), consumed) => {
491 let skipped = a as usize + b as usize;
492 output.extend(repeat('?').take(skipped));
493 rest = &rest[consumed..];
497 assert_eq!(self.encoding.encode(&output).0.len(), input.len());
503 pub trait TryDecode: Sized {
506 decoder: &mut Decoder,
507 input: &Self::Input<'_>,
508 warn: impl Fn(Error),
509 ) -> Result<Option<Self>, Error>;
512 pub trait Decode<Input>: Sized {
513 fn decode(decoder: &Decoder, input: &Input, warn: impl Fn(Error)) -> Self;
516 impl<const N: usize> Decode<RawStr<N>> for String {
517 fn decode(decoder: &Decoder, input: &RawStr<N>, warn: impl Fn(Error)) -> Self {
518 decoder.decode_string(&input.0, &warn)
522 #[derive(Clone, Debug)]
523 pub struct HeaderRecord {
524 pub eye_catcher: String,
525 pub weight_index: Option<usize>,
526 pub n_cases: Option<u64>,
527 pub creation: NaiveDateTime,
528 pub file_label: String,
531 fn trim_end_spaces(mut s: String) -> String {
532 s.truncate(s.trim_end_matches(' ').len());
536 impl TryDecode for HeaderRecord {
537 type Input<'a> = crate::raw::HeaderRecord<RawString>;
540 decoder: &mut Decoder,
541 input: &Self::Input<'_>,
542 warn: impl Fn(Error),
543 ) -> Result<Option<Self>, Error> {
544 let eye_catcher = trim_end_spaces(decoder.decode_string(&input.eye_catcher.0, &warn));
545 let file_label = trim_end_spaces(decoder.decode_string(&input.file_label.0, &warn));
546 let creation_date = decoder.decode_string_cow(&input.creation_date.0, &warn);
548 NaiveDate::parse_from_str(&creation_date, "%e %b %Y").unwrap_or_else(|_| {
549 warn(Error::InvalidCreationDate {
550 creation_date: creation_date.into(),
554 let creation_time = decoder.decode_string_cow(&input.creation_time.0, &warn);
556 NaiveTime::parse_from_str(&creation_time, "%H:%M:%S").unwrap_or_else(|_| {
557 warn(Error::InvalidCreationTime {
558 creation_time: creation_time.into(),
562 Ok(Some(HeaderRecord {
564 weight_index: input.weight_index.map(|n| n as usize),
565 n_cases: input.n_cases.map(|n| n as u64),
566 creation: NaiveDateTime::new(creation_date, creation_time),
572 #[derive(Copy, Clone, Debug, PartialEq, Eq)]
578 impl PartialOrd for VarWidth {
579 fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
580 match (self, other) {
581 (VarWidth::Numeric, VarWidth::Numeric) => Some(Ordering::Equal),
582 (VarWidth::String(a), VarWidth::String(b)) => Some(a.cmp(b)),
589 const MAX_STRING: u16 = 32767;
591 fn n_dict_indexes(self) -> usize {
593 VarWidth::Numeric => 1,
594 VarWidth::String(w) => div_ceil(w as usize, 8),
601 f: impl Fn(u16, u16) -> u16,
602 ) -> Option<VarWidth> {
604 (Some(VarWidth::Numeric), Some(VarWidth::Numeric)) => Some(VarWidth::Numeric),
605 (Some(VarWidth::String(a)), Some(VarWidth::String(b))) => {
606 Some(VarWidth::String(f(a, b)))
612 /// Returns the wider of `self` and `other`:
613 /// - Numerical variable widths are equally wide.
614 /// - Longer strings are wider than shorter strings.
615 /// - Numerical and string types are incomparable, so result in `None`.
616 /// - Any `None` in the input yields `None` in the output.
617 pub fn wider(a: Option<VarWidth>, b: Option<VarWidth>) -> Option<VarWidth> {
618 Self::width_predicate(a, b, |a, b| a.max(b))
621 /// Returns the narrower of `self` and `other` (see [`Self::wider`]).
622 pub fn narrower(a: Option<VarWidth>, b: Option<VarWidth>) -> Option<VarWidth> {
623 Self::width_predicate(a, b, |a, b| a.min(b))
627 impl From<VarWidth> for VarType {
628 fn from(source: VarWidth) -> Self {
630 VarWidth::Numeric => VarType::Numeric,
631 VarWidth::String(_) => VarType::String,
636 #[derive(Clone, Debug)]
637 pub struct VariableRecord {
639 pub name: Identifier,
640 pub print_format: Spec,
641 pub write_format: Spec,
642 pub missing_values: MissingValues,
643 pub label: Option<String>,
646 #[derive(Clone, Debug)]
647 pub struct MissingValues {
648 /// Individual missing values, up to 3 of them.
649 pub values: Vec<Value>,
651 /// Optional range of missing values.
652 pub range: Option<(Value, Value)>,
655 impl Decode<raw::MissingValues<RawStr<8>>> for MissingValues {
658 input: &raw::MissingValues<RawStr<8>>,
659 _warn: impl Fn(Error),
665 .map(|value| Value::decode(value, decoder))
670 .map(|(low, high)| (Value::decode(low, decoder), Value::decode(high, decoder))),
675 fn decode_format(raw: raw::Spec, width: VarWidth, warn: impl Fn(Spec, FormatError)) -> Spec {
676 UncheckedSpec::try_from(raw)
677 .and_then(Spec::try_from)
678 .and_then(|x| x.check_width_compatibility(width))
679 .unwrap_or_else(|error| {
680 let new_format = Spec::default_for_width(width);
681 warn(new_format, error);
686 impl TryDecode for VariableRecord {
687 type Input<'a> = raw::VariableRecord<RawString, RawStr<8>>;
690 decoder: &mut Decoder,
691 input: &Self::Input<'_>,
692 warn: impl Fn(Error),
693 ) -> Result<Option<VariableRecord>, Error> {
694 let width = match input.width {
695 0 => VarWidth::Numeric,
696 w @ 1..=255 => VarWidth::String(w as u16),
697 -1 => return Ok(None),
699 return Err(Error::InvalidVariableWidth {
700 offsets: input.offsets.clone(),
705 let name = trim_end_spaces(decoder.decode_string(&input.name.0, &warn));
706 let name = match Identifier::new(&name, decoder.encoding) {
708 if !decoder.var_names.contains_key(&name) {
711 let new_name = decoder.generate_name();
712 warn(Error::DuplicateVariableName {
713 duplicate_name: name.clone(),
714 new_name: new_name.clone(),
720 let new_name = decoder.generate_name();
721 warn(Error::InvalidVariableName {
723 new_name: new_name.clone(),
728 let variable = Variable {
729 dict_index: decoder.n_dict_indexes,
730 short_name: name.clone(),
734 decoder.n_dict_indexes += width.n_dict_indexes();
737 .insert(name.clone(), variable.dict_index)
741 .insert(variable.dict_index, variable)
744 let print_format = decode_format(input.print_format, width, |new_spec, format_error| {
745 warn(Error::InvalidPrintFormat {
747 variable: name.clone(),
751 let write_format = decode_format(input.write_format, width, |new_spec, format_error| {
752 warn(Error::InvalidWriteFormat {
754 variable: name.clone(),
761 .map(|label| decoder.decode_string(&label.0, &warn));
762 Ok(Some(VariableRecord {
767 missing_values: MissingValues::decode(decoder, &input.missing_values, warn),
773 #[derive(Clone, Debug)]
774 pub struct DocumentRecord(Vec<String>);
776 impl TryDecode for DocumentRecord {
777 type Input<'a> = crate::raw::DocumentRecord<RawDocumentLine>;
780 decoder: &mut Decoder,
781 input: &Self::Input<'_>,
782 warn: impl Fn(Error),
783 ) -> Result<Option<Self>, Error> {
784 Ok(Some(DocumentRecord(
788 .map(|s| trim_end_spaces(decoder.decode_string(&s.0, &warn)))
798 const NAME: &'static str;
799 fn parse(input: &str, warn: impl Fn(Error)) -> Result<Self, Error>;
802 #[derive(Clone, Debug)]
803 pub struct VariableSet {
805 pub vars: Vec<String>,
809 fn parse(input: &str) -> Result<Self, Error> {
810 let (name, input) = input.split_once('=').ok_or(Error::TBD)?;
811 let vars = input.split_ascii_whitespace().map(String::from).collect();
819 trait WarnOnError<T> {
820 fn warn_on_error<F: Fn(Error)>(self, warn: &F) -> Option<T>;
822 impl<T> WarnOnError<T> for Result<T, Error> {
823 fn warn_on_error<F: Fn(Error)>(self, warn: &F) -> Option<T> {
825 Ok(result) => Some(result),
834 #[derive(Clone, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)]
836 Number(Option<OrderedFloat<f64>>),
841 pub fn decode(raw: &raw::Value<RawStr<8>>, decoder: &Decoder) -> Self {
843 raw::Value::Number(x) => Value::Number(x.map(|x| x.into())),
844 raw::Value::String(s) => Value::String(decoder.decode_exact_length(&s.0).into()),
849 #[derive(Clone, Debug)]
850 pub struct ValueLabel {
855 #[derive(Clone, Debug)]
856 pub struct ValueLabelRecord {
857 pub var_type: VarType,
858 pub labels: Vec<ValueLabel>,
859 pub variables: Vec<Identifier>,
862 impl TryDecode for ValueLabelRecord {
863 type Input<'a> = crate::raw::ValueLabelRecord<RawStr<8>, RawString>;
865 decoder: &mut Decoder,
866 input: &Self::Input<'_>,
867 warn: impl Fn(Error),
868 ) -> Result<Option<ValueLabelRecord>, Error> {
869 let variables: Vec<&Variable> = input
872 .filter_map(|&dict_index| {
874 .get_var_by_index(dict_index as usize)
875 .warn_on_error(&warn)
877 .filter(|&variable| match variable.width {
878 VarWidth::String(width) if width > 8 => {
879 warn(Error::InvalidLongStringValueLabel(
880 variable.short_name.clone(),
887 let mut i = variables.iter();
888 let Some(&first_var) = i.next() else {
891 let var_type: VarType = first_var.width.into();
893 let this_type: VarType = variable.width.into();
894 if var_type != this_type {
895 let (numeric_var, string_var) = match var_type {
896 VarType::Numeric => (first_var, variable),
897 VarType::String => (variable, first_var),
899 warn(Error::ValueLabelsDifferentTypes {
900 numeric_var: numeric_var.short_name.clone(),
901 string_var: string_var.short_name.clone(),
909 .map(|raw::ValueLabel { value, label }| {
910 let label = decoder.decode_string(&label.0, &warn);
911 let value = Value::decode(value, decoder);
912 ValueLabel { value, label }
915 let variables = variables
917 .map(|&variable| variable.short_name.clone())
919 Ok(Some(ValueLabelRecord {
927 #[derive(Clone, Debug)]
928 pub struct VariableSetRecord(Vec<VariableSet>);
930 impl TextRecord for VariableSetRecord {
931 const NAME: &'static str = "variable set";
932 fn parse(input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
933 let mut sets = Vec::new();
934 for line in input.lines() {
935 if let Some(set) = VariableSet::parse(line).warn_on_error(&warn) {
939 Ok(VariableSetRecord(sets))
943 #[derive(Clone, Debug)]
944 pub struct ProductInfoRecord(pub String);
946 impl TextRecord for ProductInfoRecord {
947 const NAME: &'static str = "extra product info";
948 fn parse(input: &str, _warn: impl Fn(Error)) -> Result<Self, Error> {
949 Ok(ProductInfoRecord(input.into()))
953 #[derive(Clone, Debug)]
954 pub struct LongName {
955 pub short_name: Identifier,
956 pub long_name: Identifier,
960 fn new(decoder: &mut Decoder, short_name: &str, long_name: &str) -> Result<LongName, Error> {
962 Identifier::new(short_name, decoder.encoding).map_err(Error::InvalidShortName)?;
964 Identifier::new(long_name, decoder.encoding).map_err(Error::InvalidLongName)?;
972 #[derive(Clone, Debug)]
973 pub struct LongNameRecord(Vec<LongName>);
975 impl LongNameRecord {
976 pub fn parse(decoder: &mut Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
977 let mut names = Vec::new();
978 for pair in input.split('\t').filter(|s| !s.is_empty()) {
979 if let Some((short_name, long_name)) = pair.split_once('=') {
980 if let Some(long_name) =
981 LongName::new(decoder, short_name, long_name).warn_on_error(&warn)
983 names.push(long_name);
989 Ok(LongNameRecord(names))
993 #[derive(Clone, Debug)]
994 pub struct VeryLongString {
995 pub short_name: Identifier,
999 impl VeryLongString {
1000 fn parse(decoder: &Decoder, input: &str) -> Result<VeryLongString, Error> {
1001 let Some((short_name, length)) = input.split_once('=') else {
1002 return Err(Error::TBD);
1005 Identifier::new(short_name, decoder.encoding).map_err(Error::InvalidLongStringName)?;
1006 let length: u16 = length.parse().map_err(|_| Error::TBD)?;
1007 if length > VarWidth::MAX_STRING {
1008 return Err(Error::TBD);
1010 Ok(VeryLongString { short_name, length })
1014 #[derive(Clone, Debug)]
1015 pub struct VeryLongStringRecord(Vec<VeryLongString>);
1017 impl VeryLongStringRecord {
1018 pub fn parse(decoder: &Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1019 let mut very_long_strings = Vec::new();
1022 .map(|s| s.trim_end_matches('\t'))
1023 .filter(|s| !s.is_empty())
1025 if let Some(vls) = VeryLongString::parse(decoder, tuple).warn_on_error(&warn) {
1026 very_long_strings.push(vls)
1029 Ok(VeryLongStringRecord(very_long_strings))
1033 #[derive(Clone, Debug)]
1034 pub struct Attribute {
1035 pub name: Identifier,
1036 pub values: Vec<String>,
1043 warn: &impl Fn(Error),
1044 ) -> Result<(Option<Attribute>, &'a str), Error> {
1045 let Some((name, mut input)) = input.split_once('(') else {
1046 return Err(Error::TBD);
1048 let mut values = Vec::new();
1050 let Some((value, rest)) = input.split_once('\n') else {
1051 return Err(Error::TBD);
1053 if let Some(stripped) = value
1055 .and_then(|value| value.strip_suffix('\''))
1057 values.push(stripped.into());
1060 values.push(value.into());
1062 if let Some(rest) = rest.strip_prefix(')') {
1063 let attribute = Identifier::new(name, decoder.encoding)
1064 .map_err(Error::InvalidAttributeName)
1065 .warn_on_error(warn)
1066 .map(|name| Attribute { name, values });
1067 return Ok((attribute, rest));
1074 #[derive(Clone, Debug)]
1075 pub struct AttributeSet(pub Vec<Attribute>);
1081 sentinel: Option<char>,
1082 warn: &impl Fn(Error),
1083 ) -> Result<(AttributeSet, &'a str), Error> {
1084 let mut attributes = Vec::new();
1086 match input.chars().next() {
1087 None => break input,
1088 c if c == sentinel => break &input[1..],
1090 let (attribute, rest) = Attribute::parse(decoder, input, &warn)?;
1091 if let Some(attribute) = attribute {
1092 attributes.push(attribute);
1098 Ok((AttributeSet(attributes), rest))
1102 #[derive(Clone, Debug)]
1103 pub struct FileAttributeRecord(AttributeSet);
1105 impl FileAttributeRecord {
1106 pub fn parse(decoder: &Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1107 let (set, rest) = AttributeSet::parse(decoder, input, None, &warn)?;
1108 if !rest.is_empty() {
1111 Ok(FileAttributeRecord(set))
1115 #[derive(Clone, Debug)]
1116 pub struct VarAttributeSet {
1117 pub long_var_name: Identifier,
1118 pub attributes: AttributeSet,
1121 impl VarAttributeSet {
1125 warn: &impl Fn(Error),
1126 ) -> Result<(Option<VarAttributeSet>, &'a str), Error> {
1127 let Some((long_var_name, rest)) = input.split_once(':') else {
1128 return Err(Error::TBD);
1130 let (attributes, rest) = AttributeSet::parse(decoder, rest, Some('/'), warn)?;
1131 let var_attribute = Identifier::new(long_var_name, decoder.encoding)
1132 .map_err(Error::InvalidAttributeVariableName)
1133 .warn_on_error(warn)
1134 .map(|name| VarAttributeSet {
1135 long_var_name: name,
1138 Ok((var_attribute, rest))
1142 #[derive(Clone, Debug)]
1143 pub struct VariableAttributeRecord(Vec<VarAttributeSet>);
1145 impl VariableAttributeRecord {
1146 pub fn parse(decoder: &Decoder, mut input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1147 let mut var_attribute_sets = Vec::new();
1148 while !input.is_empty() {
1149 let Some((var_attribute, rest)) =
1150 VarAttributeSet::parse(decoder, input, &warn).warn_on_error(&warn)
1154 if let Some(var_attribute) = var_attribute {
1155 var_attribute_sets.push(var_attribute);
1159 Ok(VariableAttributeRecord(var_attribute_sets))
1163 #[derive(Clone, Debug)]
1164 pub enum MultipleResponseType {
1167 labels: CategoryLabels,
1172 impl MultipleResponseType {
1175 mr_set: &Identifier,
1176 input: &raw::MultipleResponseType,
1177 min_width: VarWidth,
1178 warn: &impl Fn(Error),
1179 ) -> Result<Self, Error> {
1180 let mr_type = match input {
1181 raw::MultipleResponseType::MultipleDichotomy { value, labels } => {
1182 let value = decoder.decode_string_cow(&value.0, warn);
1183 let value = match min_width {
1184 VarWidth::Numeric => {
1185 let number: f64 = value.trim().parse().map_err(|_| {
1186 Error::InvalidMDGroupCountedValue {
1187 mr_set: mr_set.clone(),
1188 number: value.into(),
1191 Value::Number(Some(number.into()))
1193 VarWidth::String(max_width) => {
1194 let value = value.trim_end_matches(' ');
1195 let width = value.len();
1196 if width > max_width as usize {
1197 return Err(Error::TooWideMDGroupCountedValue {
1198 mr_set: mr_set.clone(),
1199 value: value.into(),
1204 Value::String(value.into())
1207 MultipleResponseType::MultipleDichotomy {
1212 raw::MultipleResponseType::MultipleCategory => MultipleResponseType::MultipleCategory,
1218 #[derive(Clone, Debug)]
1219 pub struct MultipleResponseSet {
1220 pub name: Identifier,
1221 pub min_width: VarWidth,
1222 pub max_width: VarWidth,
1224 pub mr_type: MultipleResponseType,
1225 pub dict_indexes: Vec<DictIndex>,
1228 impl MultipleResponseSet {
1231 input: &raw::MultipleResponseSet<Cow<str>>,
1232 warn: &impl Fn(Error),
1233 ) -> Result<Self, Error> {
1235 Identifier::new(&input.name, decoder.encoding).map_err(Error::InvalidMrSetName)?;
1237 let mut dict_indexes = Vec::with_capacity(input.short_names.len());
1238 for short_name in input.short_names.iter() {
1239 let short_name = match Identifier::new(&short_name, decoder.encoding) {
1242 warn(Error::InvalidMrSetName(error));
1246 let Some(&dict_index) = decoder.var_names.get(&short_name) else {
1247 warn(Error::UnknownMrSetVariable {
1248 mr_set: mr_set_name.clone(),
1249 short_name: short_name.clone(),
1253 dict_indexes.push(dict_index);
1256 match dict_indexes.len() {
1257 0 => return Err(Error::EmptyMrSet(mr_set_name)),
1258 1 => return Err(Error::OneVarMrSet(mr_set_name)),
1262 let Some((Some(min_width), Some(max_width))) = dict_indexes
1264 .map(|dict_index| decoder.variables[dict_index].width)
1265 .map(|w| (Some(w), Some(w)))
1266 .reduce(|(na, wa), (nb, wb)| (VarWidth::narrower(na, nb), VarWidth::wider(wa, wb)))
1268 return Err(Error::MixedMrSet(mr_set_name));
1272 MultipleResponseType::decode(decoder, &mr_set_name, &input.mr_type, min_width, warn)?;
1274 Ok(MultipleResponseSet {
1278 label: input.label.to_string(),
1285 #[derive(Clone, Debug)]
1286 pub struct MultipleResponseRecord(pub Vec<MultipleResponseSet>);
1288 impl TryDecode for MultipleResponseRecord {
1289 type Input<'a> = raw::MultipleResponseRecord<Cow<'a, str>>;
1292 decoder: &mut Decoder,
1293 input: &Self::Input<'_>,
1294 warn: impl Fn(Error),
1295 ) -> Result<Option<Self>, Error> {
1296 let mut sets = Vec::with_capacity(input.0.len());
1297 for set in &input.0 {
1298 match MultipleResponseSet::decode(decoder, set, &warn) {
1299 Ok(set) => sets.push(set),
1300 Err(error) => warn(error),
1303 Ok(Some(MultipleResponseRecord(sets)))
1307 #[derive(Clone, Debug)]
1308 pub struct LongStringMissingValues {
1310 pub var_name: Identifier,
1313 pub missing_values: MissingValues,
1316 impl LongStringMissingValues {
1319 input: &raw::LongStringMissingValues,
1320 warn: &impl Fn(Error),
1321 ) -> Result<Self, Error> {
1322 let var_name = decoder.decode_string(&input.var_name.0, warn);
1323 let var_name = Identifier::new(var_name.trim_end(), decoder.encoding)
1324 .map_err(Error::InvalidLongStringValueLabelName)?;
1326 let missing_values = MissingValues::decode(decoder, &input.missing_values, warn);
1328 Ok(LongStringMissingValues {
1335 #[derive(Clone, Debug)]
1336 pub struct LongStringMissingValuesRecord(Vec<LongStringMissingValues>);
1338 impl TryDecode for LongStringMissingValuesRecord {
1339 type Input<'a> = raw::LongStringMissingValueRecord;
1342 decoder: &mut Decoder,
1343 input: &Self::Input<'_>,
1344 warn: impl Fn(Error),
1345 ) -> Result<Option<Self>, Error> {
1346 let mut labels = Vec::with_capacity(input.0.len());
1347 for label in &input.0 {
1348 match LongStringMissingValues::decode(decoder, label, &warn) {
1349 Ok(set) => labels.push(set),
1350 Err(error) => warn(error),
1353 Ok(Some(LongStringMissingValuesRecord(labels)))
1357 #[derive(Clone, Debug)]
1358 pub struct LongStringValueLabels {
1359 pub var_name: Identifier,
1360 pub width: VarWidth,
1361 pub labels: Vec<ValueLabel>,
1364 impl LongStringValueLabels {
1367 input: &raw::LongStringValueLabels,
1368 warn: &impl Fn(Error),
1369 ) -> Result<Self, Error> {
1370 let var_name = decoder.decode_string(&input.var_name.0, warn);
1371 let var_name = Identifier::new(var_name.trim_end(), decoder.encoding)
1372 .map_err(Error::InvalidLongStringValueLabelName)?;
1375 let max_width = VarWidth::MAX_STRING;
1376 if input.width < 9 || input.width > max_width as u32 {
1377 return Err(Error::InvalidLongValueLabelWidth {
1384 let width = input.width as u16;
1386 let mut labels = Vec::with_capacity(input.labels.len());
1387 for (value, label) in input.labels.iter() {
1388 let value = Value::String(decoder.decode_exact_length(&value.0).into());
1389 let label = decoder.decode_string(&label.0, warn);
1390 labels.push(ValueLabel { value, label });
1393 Ok(LongStringValueLabels {
1395 width: VarWidth::String(width),
1401 #[derive(Clone, Debug)]
1402 pub struct LongStringValueLabelRecord(pub Vec<LongStringValueLabels>);
1404 impl TryDecode for LongStringValueLabelRecord {
1405 type Input<'a> = raw::LongStringValueLabelRecord;
1408 decoder: &mut Decoder,
1409 input: &Self::Input<'_>,
1410 warn: impl Fn(Error),
1411 ) -> Result<Option<Self>, Error> {
1412 let mut labels = Vec::with_capacity(input.0.len());
1413 for label in &input.0 {
1414 match LongStringValueLabels::decode(decoder, label, &warn) {
1415 Ok(set) => labels.push(set),
1416 Err(error) => warn(error),
1419 Ok(Some(LongStringValueLabelRecord(labels)))
1425 use encoding_rs::WINDOWS_1252;
1429 let mut s = String::new();
1430 s.push(char::REPLACEMENT_CHARACTER);
1431 let encoded = WINDOWS_1252.encode(&s).0;
1432 let decoded = WINDOWS_1252.decode(&encoded[..]).0;
1433 println!("{:?}", decoded);
1438 let charset: Vec<u8> = (0..=255).collect();
1439 println!("{}", charset.len());
1440 let decoded = WINDOWS_1252.decode(&charset[..]).0;
1441 println!("{}", decoded.len());
1442 let encoded = WINDOWS_1252.encode(&decoded[..]).0;
1443 println!("{}", encoded.len());
1444 assert_eq!(&charset[..], &encoded[..]);