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, UnencodedStr, 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("Invalid variable measurement level value {0}")]
155 InvalidMeasurement(u32),
157 #[error("Invalid variable display alignment value {0}")]
158 InvalidAlignment(u32),
160 #[error("Details TBD")]
164 #[derive(Clone, Debug)]
166 Header(HeaderRecord),
167 Variable(VariableRecord),
168 ValueLabel(ValueLabelRecord),
169 Document(DocumentRecord),
170 IntegerInfo(IntegerInfoRecord),
171 FloatInfo(FloatInfoRecord),
172 VariableSets(VariableSetRecord),
173 VarDisplay(VarDisplayRecord),
174 MultipleResponse(MultipleResponseRecord),
175 LongStringMissingValues(LongStringMissingValuesRecord),
176 LongStringValueLabels(LongStringValueLabelRecord),
177 Encoding(EncodingRecord),
178 NumberOfCases(NumberOfCasesRecord),
179 ProductInfo(ProductInfoRecord),
180 LongNames(LongNameRecord),
181 VeryLongStrings(VeryLongStringRecord),
182 FileAttributes(FileAttributeRecord),
183 VariableAttributes(VariableAttributeRecord),
184 OtherExtension(Extension),
188 pub use crate::raw::EncodingRecord;
189 pub use crate::raw::Extension;
190 pub use crate::raw::FloatInfoRecord;
191 pub use crate::raw::IntegerInfoRecord;
192 pub use crate::raw::NumberOfCasesRecord;
194 type DictIndex = usize;
196 pub struct Variable {
197 pub dict_index: DictIndex,
198 pub short_name: Identifier,
199 pub long_name: Option<Identifier>,
204 pub compression: Option<Compression>,
206 pub encoding: &'static Encoding,
207 pub variables: HashMap<DictIndex, Variable>,
208 pub var_names: HashMap<Identifier, DictIndex>,
209 n_dict_indexes: usize,
210 n_generated_names: usize,
215 header: Option<&'a raw::HeaderRecord>,
216 variables: Vec<&'a raw::VariableRecord>,
217 value_labels: Vec<&'a raw::ValueLabelRecord>,
218 document: Option<&'a raw::DocumentRecord>,
219 integer_info: Option<&'a raw::IntegerInfoRecord>,
220 float_info: Option<&'a raw::FloatInfoRecord>,
221 variable_sets: Vec<&'a raw::TextRecord>,
222 var_display: Option<&'a raw::VarDisplayRecord>,
223 multiple_response: Vec<&'a raw::MultipleResponseRecord>,
224 long_string_value_labels: Vec<&'a raw::LongStringValueLabelRecord>,
225 long_string_missing_values: Vec<&'a raw::LongStringMissingValueRecord>,
226 encoding: Option<&'a raw::EncodingRecord>,
227 number_of_cases: Option<&'a raw::NumberOfCasesRecord>,
228 product_info: Option<&'a raw::TextRecord>,
229 long_names: Option<&'a raw::TextRecord>,
230 very_long_strings: Vec<&'a raw::TextRecord>,
231 file_attributes: Vec<&'a raw::TextRecord>,
232 variable_attributes: Vec<&'a raw::TextRecord>,
233 other_extensions: Vec<&'a raw::Extension>,
234 cases: Option<&'a Rc<RefCell<raw::Cases>>>,
237 fn set_or_warn<T>(option: &mut Option<T>, value: T, warn: &impl Fn(Error)) {
238 if option.is_none() {
239 let _ = option.insert(value);
245 impl<'a> Headers<'a> {
246 fn new(headers: &'a Vec<raw::Record>, warn: &impl Fn(Error)) -> Headers<'a> {
247 let mut h = Headers::default();
248 for header in headers {
250 raw::Record::Header(record) => set_or_warn(&mut h.header, record, warn),
251 raw::Record::Variable(record) => h.variables.push(record),
252 raw::Record::ValueLabel(record) => h.value_labels.push(record),
253 raw::Record::Document(record) => set_or_warn(&mut h.document, record, warn),
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) => {
263 h.long_string_missing_values.push(record)
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),
286 headers: Vec<raw::Record>,
287 encoding: Option<&'static Encoding>,
288 warn: &impl Fn(Error),
289 ) -> Result<Vec<Record>, Error> {
290 let h = Headers::new(&headers, warn);
291 let Some(header) = h.header else {
292 return Err(Error::MissingHeaderRecord);
294 let encoding = match encoding {
295 Some(encoding) => encoding,
297 let encoding = h.encoding.map(|record| record.0.as_str());
298 let character_code = h.integer_info.map(|record| record.character_code);
299 match get_encoding(encoding, character_code) {
300 Ok(encoding) => encoding,
301 Err(err @ EncodingError::Ebcdic) => return Err(Error::EncodingError(err)),
303 warn(Error::EncodingError(err));
304 // Warn that we're using the default encoding.
311 let mut decoder = Decoder {
312 compression: header.compression,
313 endian: header.endian,
315 variables: HashMap::new(),
316 var_names: HashMap::new(),
318 n_generated_names: 0,
321 let mut output = Vec::with_capacity(headers.len());
323 // Decode the records that don't use variables at all.
324 if let Some(header) = HeaderRecord::try_decode(&mut decoder, header, warn)? {
325 output.push(Record::Header(header))
327 if let Some(raw) = h.document {
328 if let Some(document) = DocumentRecord::try_decode(&mut decoder, raw, warn)? {
329 output.push(Record::Document(document))
332 if let Some(raw) = h.integer_info {
333 output.push(Record::IntegerInfo(raw.clone()));
335 if let Some(raw) = h.float_info {
336 output.push(Record::FloatInfo(raw.clone()));
338 if let Some(raw) = h.product_info {
339 let s = decoder.decode_string_cow(&raw.text.0, warn);
340 output.push(Record::ProductInfo(ProductInfoRecord::parse(&s, warn)?));
342 if let Some(raw) = h.number_of_cases {
343 output.push(Record::NumberOfCases(raw.clone()))
345 for &raw in &h.file_attributes {
346 let s = decoder.decode_string_cow(&raw.text.0, warn);
347 output.push(Record::FileAttributes(FileAttributeRecord::parse(
351 for &raw in &h.other_extensions {
352 output.push(Record::OtherExtension(raw.clone()));
355 // Decode the variable records, which are the basis of almost everything
357 for &raw in &h.variables {
358 if let Some(variable) = VariableRecord::try_decode(&mut decoder, raw, warn)? {
359 output.push(Record::Variable(variable));
363 // Decode value labels and weight variable. These use indexes into the
364 // variable records, so we need to parse them before those indexes become
365 // invalidated by very long string variables.
366 for &raw in &h.value_labels {
367 if let Some(value_label) = ValueLabelRecord::try_decode(&mut decoder, raw, warn)? {
368 output.push(Record::ValueLabel(value_label));
372 if let Some(raw) = h.var_display {
373 if let Some(vdr) = VarDisplayRecord::try_decode(&mut decoder, raw, warn)? {
374 output.push(Record::VarDisplay(vdr))
378 // Decode records that use short names.
379 for &raw in &h.multiple_response {
380 if let Some(mrr) = MultipleResponseRecord::try_decode(&mut decoder, raw, warn)? {
381 output.push(Record::MultipleResponse(mrr))
384 for &raw in &h.very_long_strings {
385 let s = decoder.decode_string_cow(&raw.text.0, warn);
386 output.push(Record::VeryLongStrings(VeryLongStringRecord::parse(
391 // Rename variables to their long names.
392 for &raw in &h.long_names {
393 let s = decoder.decode_string_cow(&raw.text.0, warn);
394 output.push(Record::LongNames(LongNameRecord::parse(
401 // Decode recods that use long names.
402 for &raw in &h.variable_attributes {
403 let s = decoder.decode_string_cow(&raw.text.0, warn);
404 output.push(Record::VariableAttributes(VariableAttributeRecord::parse(
408 for &raw in &h.long_string_value_labels {
409 if let Some(mrr) = LongStringValueLabelRecord::try_decode(&mut decoder, raw, warn)? {
410 output.push(Record::LongStringValueLabels(mrr))
413 for &raw in &h.long_string_missing_values {
414 if let Some(mrr) = LongStringMissingValuesRecord::try_decode(&mut decoder, raw, warn)? {
415 output.push(Record::LongStringMissingValues(mrr))
418 for &raw in &h.variable_sets {
419 let s = decoder.decode_string_cow(&raw.text.0, warn);
420 output.push(Record::VariableSets(VariableSetRecord::parse(&s, warn)?));
426 fn generate_name(&mut self) -> Identifier {
428 self.n_generated_names += 1;
429 let name = Identifier::new(&format!("VAR{:03}", self.n_generated_names), self.encoding)
431 if !self.var_names.contains_key(&name) {
434 assert!(self.n_generated_names < usize::MAX);
437 fn decode_string_cow<'a>(&self, input: &'a [u8], warn: &impl Fn(Error)) -> Cow<'a, str> {
438 let (output, malformed) = self.encoding.decode_without_bom_handling(input);
440 warn(Error::MalformedString {
441 encoding: self.encoding.name().into(),
442 text: output.clone().into(),
447 fn decode_string(&self, input: &[u8], warn: &impl Fn(Error)) -> String {
448 self.decode_string_cow(input, warn).into()
450 pub fn decode_identifier(
453 warn: &impl Fn(Error),
454 ) -> Result<Identifier, IdError> {
455 let s = self.decode_string_cow(input, warn);
456 Identifier::new(&s, self.encoding)
458 fn get_var_by_index(&self, dict_index: usize) -> Result<&Variable, Error> {
459 let max_index = self.n_dict_indexes;
460 if dict_index == 0 || dict_index > max_index {
461 return Err(Error::InvalidDictIndex {
466 let Some(variable) = self.variables.get(&(dict_index - 1)) else {
467 return Err(Error::DictIndexIsContinuation(dict_index));
472 /// Returns `input` decoded from `self.encoding` into UTF-8 such that
473 /// re-encoding the result back into `self.encoding` will have exactly the
474 /// same length in bytes.
476 /// XXX warn about errors?
477 fn decode_exact_length<'a>(&self, input: &'a [u8]) -> Cow<'a, str> {
478 if let (s, false) = self.encoding.decode_without_bom_handling(input) {
479 // This is the common case. Usually there will be no errors.
482 // Unusual case. Don't bother to optimize it much.
483 let mut decoder = self.encoding.new_decoder_without_bom_handling();
484 let mut output = String::with_capacity(
486 .max_utf8_buffer_length_without_replacement(input.len())
489 let mut rest = input;
490 while !rest.is_empty() {
491 match decoder.decode_to_string_without_replacement(rest, &mut output, true) {
492 (DecoderResult::InputEmpty, _) => break,
493 (DecoderResult::OutputFull, _) => unreachable!(),
494 (DecoderResult::Malformed(a, b), consumed) => {
495 let skipped = a as usize + b as usize;
496 output.extend(repeat('?').take(skipped));
497 rest = &rest[consumed..];
501 assert_eq!(self.encoding.encode(&output).0.len(), input.len());
507 pub trait TryDecode: Sized {
510 decoder: &mut Decoder,
512 warn: impl Fn(Error),
513 ) -> Result<Option<Self>, Error>;
516 pub trait Decode<Input>: Sized {
517 fn decode(decoder: &Decoder, input: &Input, warn: impl Fn(Error)) -> Self;
520 impl<const N: usize> Decode<UnencodedStr<N>> for String {
521 fn decode(decoder: &Decoder, input: &UnencodedStr<N>, warn: impl Fn(Error)) -> Self {
522 decoder.decode_string(&input.0, &warn)
526 #[derive(Clone, Debug)]
527 pub struct HeaderRecord {
528 pub eye_catcher: String,
529 pub weight_index: Option<usize>,
530 pub n_cases: Option<u64>,
531 pub creation: NaiveDateTime,
532 pub file_label: String,
535 fn trim_end_spaces(mut s: String) -> String {
536 s.truncate(s.trim_end_matches(' ').len());
540 impl TryDecode for HeaderRecord {
541 type Input = crate::raw::HeaderRecord;
544 decoder: &mut Decoder,
546 warn: impl Fn(Error),
547 ) -> Result<Option<Self>, Error> {
548 let eye_catcher = trim_end_spaces(decoder.decode_string(&input.eye_catcher.0, &warn));
549 let file_label = trim_end_spaces(decoder.decode_string(&input.file_label.0, &warn));
550 let creation_date = decoder.decode_string_cow(&input.creation_date.0, &warn);
552 NaiveDate::parse_from_str(&creation_date, "%e %b %Y").unwrap_or_else(|_| {
553 warn(Error::InvalidCreationDate {
554 creation_date: creation_date.into(),
558 let creation_time = decoder.decode_string_cow(&input.creation_time.0, &warn);
560 NaiveTime::parse_from_str(&creation_time, "%H:%M:%S").unwrap_or_else(|_| {
561 warn(Error::InvalidCreationTime {
562 creation_time: creation_time.into(),
566 Ok(Some(HeaderRecord {
568 weight_index: input.weight_index.map(|n| n as usize),
569 n_cases: input.n_cases.map(|n| n as u64),
570 creation: NaiveDateTime::new(creation_date, creation_time),
576 #[derive(Copy, Clone, Debug, PartialEq, Eq)]
582 impl PartialOrd for VarWidth {
583 fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
584 match (self, other) {
585 (VarWidth::Numeric, VarWidth::Numeric) => Some(Ordering::Equal),
586 (VarWidth::String(a), VarWidth::String(b)) => Some(a.cmp(b)),
593 const MAX_STRING: u16 = 32767;
595 fn n_dict_indexes(self) -> usize {
597 VarWidth::Numeric => 1,
598 VarWidth::String(w) => div_ceil(w as usize, 8),
605 f: impl Fn(u16, u16) -> u16,
606 ) -> Option<VarWidth> {
608 (Some(VarWidth::Numeric), Some(VarWidth::Numeric)) => Some(VarWidth::Numeric),
609 (Some(VarWidth::String(a)), Some(VarWidth::String(b))) => {
610 Some(VarWidth::String(f(a, b)))
616 /// Returns the wider of `self` and `other`:
617 /// - Numerical variable widths are equally wide.
618 /// - Longer strings are wider than shorter strings.
619 /// - Numerical and string types are incomparable, so result in `None`.
620 /// - Any `None` in the input yields `None` in the output.
621 pub fn wider(a: Option<VarWidth>, b: Option<VarWidth>) -> Option<VarWidth> {
622 Self::width_predicate(a, b, |a, b| a.max(b))
625 /// Returns the narrower of `self` and `other` (see [`Self::wider`]).
626 pub fn narrower(a: Option<VarWidth>, b: Option<VarWidth>) -> Option<VarWidth> {
627 Self::width_predicate(a, b, |a, b| a.min(b))
631 impl From<VarWidth> for VarType {
632 fn from(source: VarWidth) -> Self {
634 VarWidth::Numeric => VarType::Numeric,
635 VarWidth::String(_) => VarType::String,
640 #[derive(Clone, Debug)]
641 pub struct VariableRecord {
643 pub name: Identifier,
644 pub print_format: Spec,
645 pub write_format: Spec,
646 pub missing_values: MissingValues,
647 pub label: Option<String>,
650 #[derive(Clone, Debug)]
651 pub struct MissingValues {
652 /// Individual missing values, up to 3 of them.
653 pub values: Vec<Value>,
655 /// Optional range of missing values.
656 pub range: Option<(Value, Value)>,
659 impl Decode<raw::MissingValues> for MissingValues {
660 fn decode(decoder: &Decoder, input: &raw::MissingValues, _warn: impl Fn(Error)) -> Self {
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 = raw::VariableRecord;
690 decoder: &mut Decoder,
691 input: &crate::raw::VariableRecord,
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 = crate::raw::DocumentRecord;
780 decoder: &mut Decoder,
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, 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 = crate::raw::ValueLabelRecord;
865 decoder: &mut Decoder,
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(|(value, label)| {
910 let label = decoder.decode_string(&label.0, &warn);
911 let value = Value::decode(
912 &raw::Value::from_raw(value, var_type, decoder.endian),
915 ValueLabel { value, label }
918 let variables = variables
920 .map(|&variable| variable.short_name.clone())
922 Ok(Some(ValueLabelRecord {
930 #[derive(Clone, Debug)]
931 pub struct VariableSetRecord(Vec<VariableSet>);
933 impl TextRecord for VariableSetRecord {
934 const NAME: &'static str = "variable set";
935 fn parse(input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
936 let mut sets = Vec::new();
937 for line in input.lines() {
938 if let Some(set) = VariableSet::parse(line).warn_on_error(&warn) {
942 Ok(VariableSetRecord(sets))
946 #[derive(Clone, Debug)]
947 pub struct ProductInfoRecord(pub String);
949 impl TextRecord for ProductInfoRecord {
950 const NAME: &'static str = "extra product info";
951 fn parse(input: &str, _warn: impl Fn(Error)) -> Result<Self, Error> {
952 Ok(ProductInfoRecord(input.into()))
956 #[derive(Clone, Debug)]
957 pub struct LongName {
958 pub short_name: Identifier,
959 pub long_name: Identifier,
963 fn new(decoder: &mut Decoder, short_name: &str, long_name: &str) -> Result<LongName, Error> {
965 Identifier::new(short_name, decoder.encoding).map_err(Error::InvalidShortName)?;
967 Identifier::new(long_name, decoder.encoding).map_err(Error::InvalidLongName)?;
975 #[derive(Clone, Debug)]
976 pub struct LongNameRecord(Vec<LongName>);
978 impl LongNameRecord {
979 pub fn parse(decoder: &mut Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
980 let mut names = Vec::new();
981 for pair in input.split('\t').filter(|s| !s.is_empty()) {
982 if let Some((short_name, long_name)) = pair.split_once('=') {
983 if let Some(long_name) =
984 LongName::new(decoder, short_name, long_name).warn_on_error(&warn)
986 names.push(long_name);
992 Ok(LongNameRecord(names))
996 #[derive(Clone, Debug)]
997 pub struct VeryLongString {
998 pub short_name: Identifier,
1002 impl VeryLongString {
1003 fn parse(decoder: &Decoder, input: &str) -> Result<VeryLongString, Error> {
1004 let Some((short_name, length)) = input.split_once('=') else {
1005 return Err(Error::TBD);
1008 Identifier::new(short_name, decoder.encoding).map_err(Error::InvalidLongStringName)?;
1009 let length: u16 = length.parse().map_err(|_| Error::TBD)?;
1010 if length > VarWidth::MAX_STRING {
1011 return Err(Error::TBD);
1013 Ok(VeryLongString { short_name, length })
1017 #[derive(Clone, Debug)]
1018 pub struct VeryLongStringRecord(Vec<VeryLongString>);
1020 impl VeryLongStringRecord {
1021 pub fn parse(decoder: &Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1022 let mut very_long_strings = Vec::new();
1025 .map(|s| s.trim_end_matches('\t'))
1026 .filter(|s| !s.is_empty())
1028 if let Some(vls) = VeryLongString::parse(decoder, tuple).warn_on_error(&warn) {
1029 very_long_strings.push(vls)
1032 Ok(VeryLongStringRecord(very_long_strings))
1036 #[derive(Clone, Debug)]
1037 pub struct Attribute {
1038 pub name: Identifier,
1039 pub values: Vec<String>,
1046 warn: &impl Fn(Error),
1047 ) -> Result<(Option<Attribute>, &'a str), Error> {
1048 let Some((name, mut input)) = input.split_once('(') else {
1049 return Err(Error::TBD);
1051 let mut values = Vec::new();
1053 let Some((value, rest)) = input.split_once('\n') else {
1054 return Err(Error::TBD);
1056 if let Some(stripped) = value
1058 .and_then(|value| value.strip_suffix('\''))
1060 values.push(stripped.into());
1063 values.push(value.into());
1065 if let Some(rest) = rest.strip_prefix(')') {
1066 let attribute = Identifier::new(name, decoder.encoding)
1067 .map_err(Error::InvalidAttributeName)
1068 .warn_on_error(warn)
1069 .map(|name| Attribute { name, values });
1070 return Ok((attribute, rest));
1077 #[derive(Clone, Debug)]
1078 pub struct AttributeSet(pub Vec<Attribute>);
1084 sentinel: Option<char>,
1085 warn: &impl Fn(Error),
1086 ) -> Result<(AttributeSet, &'a str), Error> {
1087 let mut attributes = Vec::new();
1089 match input.chars().next() {
1090 None => break input,
1091 c if c == sentinel => break &input[1..],
1093 let (attribute, rest) = Attribute::parse(decoder, input, &warn)?;
1094 if let Some(attribute) = attribute {
1095 attributes.push(attribute);
1101 Ok((AttributeSet(attributes), rest))
1105 #[derive(Clone, Debug)]
1106 pub struct FileAttributeRecord(AttributeSet);
1108 impl FileAttributeRecord {
1109 pub fn parse(decoder: &Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1110 let (set, rest) = AttributeSet::parse(decoder, input, None, &warn)?;
1111 if !rest.is_empty() {
1114 Ok(FileAttributeRecord(set))
1118 #[derive(Clone, Debug)]
1119 pub struct VarAttributeSet {
1120 pub long_var_name: Identifier,
1121 pub attributes: AttributeSet,
1124 impl VarAttributeSet {
1128 warn: &impl Fn(Error),
1129 ) -> Result<(Option<VarAttributeSet>, &'a str), Error> {
1130 let Some((long_var_name, rest)) = input.split_once(':') else {
1131 return Err(Error::TBD);
1133 let (attributes, rest) = AttributeSet::parse(decoder, rest, Some('/'), warn)?;
1134 let var_attribute = Identifier::new(long_var_name, decoder.encoding)
1135 .map_err(Error::InvalidAttributeVariableName)
1136 .warn_on_error(warn)
1137 .map(|name| VarAttributeSet {
1138 long_var_name: name,
1141 Ok((var_attribute, rest))
1145 #[derive(Clone, Debug)]
1146 pub struct VariableAttributeRecord(Vec<VarAttributeSet>);
1148 impl VariableAttributeRecord {
1149 pub fn parse(decoder: &Decoder, mut input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1150 let mut var_attribute_sets = Vec::new();
1151 while !input.is_empty() {
1152 let Some((var_attribute, rest)) =
1153 VarAttributeSet::parse(decoder, input, &warn).warn_on_error(&warn)
1157 if let Some(var_attribute) = var_attribute {
1158 var_attribute_sets.push(var_attribute);
1162 Ok(VariableAttributeRecord(var_attribute_sets))
1166 #[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)]
1174 fn try_decode(source: u32) -> Result<Option<Measure>, Error> {
1177 1 => Ok(Some(Measure::Nominal)),
1178 2 => Ok(Some(Measure::Ordinal)),
1179 3 => Ok(Some(Measure::Scale)),
1180 _ => Err(Error::InvalidMeasurement(source)),
1185 #[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)]
1186 pub enum Alignment {
1193 fn try_decode(source: u32) -> Result<Option<Alignment>, Error> {
1196 1 => Ok(Some(Alignment::Left)),
1197 2 => Ok(Some(Alignment::Right)),
1198 3 => Ok(Some(Alignment::Center)),
1199 _ => Err(Error::InvalidAlignment(source)),
1204 #[derive(Clone, Debug)]
1205 pub struct VarDisplay {
1206 pub measure: Option<Measure>,
1207 pub width: Option<u32>,
1208 pub alignment: Option<Alignment>,
1211 #[derive(Clone, Debug)]
1212 pub struct VarDisplayRecord(pub Vec<VarDisplay>);
1214 impl TryDecode for VarDisplayRecord {
1215 type Input = raw::VarDisplayRecord;
1217 decoder: &mut Decoder,
1218 input: &Self::Input,
1219 warn: impl Fn(Error),
1220 ) -> Result<Option<Self>, Error> {
1221 let n_vars = decoder.variables.len();
1222 let n_per_var = if input.0.len() == 3 * n_vars {
1224 } else if input.0.len() == 2 * n_vars {
1227 return Err(Error::TBD);
1230 let var_displays = input
1234 let (measure, width, alignment) = match n_per_var == 3 {
1235 true => (chunk[0], Some(chunk[1]), chunk[2]),
1236 false => (chunk[0], None, chunk[1]),
1238 let measure = Measure::try_decode(measure).warn_on_error(&warn).flatten();
1239 let alignment = Alignment::try_decode(alignment)
1240 .warn_on_error(&warn)
1249 Ok(Some(VarDisplayRecord(var_displays)))
1253 #[derive(Clone, Debug)]
1254 pub enum MultipleResponseType {
1257 labels: CategoryLabels,
1262 impl MultipleResponseType {
1265 mr_set: &Identifier,
1266 input: &raw::MultipleResponseType,
1267 min_width: VarWidth,
1268 warn: &impl Fn(Error),
1269 ) -> Result<Self, Error> {
1270 let mr_type = match input {
1271 raw::MultipleResponseType::MultipleDichotomy { value, labels } => {
1272 let value = decoder.decode_string_cow(&value.0, warn);
1273 let value = match min_width {
1274 VarWidth::Numeric => {
1275 let number: f64 = value.trim().parse().map_err(|_| {
1276 Error::InvalidMDGroupCountedValue {
1277 mr_set: mr_set.clone(),
1278 number: value.into(),
1281 Value::Number(Some(number.into()))
1283 VarWidth::String(max_width) => {
1284 let value = value.trim_end_matches(' ');
1285 let width = value.len();
1286 if width > max_width as usize {
1287 return Err(Error::TooWideMDGroupCountedValue {
1288 mr_set: mr_set.clone(),
1289 value: value.into(),
1294 Value::String(value.into())
1297 MultipleResponseType::MultipleDichotomy {
1302 raw::MultipleResponseType::MultipleCategory => MultipleResponseType::MultipleCategory,
1308 #[derive(Clone, Debug)]
1309 pub struct MultipleResponseSet {
1310 pub name: Identifier,
1311 pub min_width: VarWidth,
1312 pub max_width: VarWidth,
1314 pub mr_type: MultipleResponseType,
1315 pub dict_indexes: Vec<DictIndex>,
1318 impl MultipleResponseSet {
1321 input: &raw::MultipleResponseSet,
1322 warn: &impl Fn(Error),
1323 ) -> Result<Self, Error> {
1324 let mr_set_name = decoder
1325 .decode_identifier(&input.name.0, warn)
1326 .map_err(Error::InvalidMrSetName)?;
1328 let label = decoder.decode_string(&input.label.0, warn);
1330 let mut dict_indexes = Vec::with_capacity(input.short_names.len());
1331 for short_name in input.short_names.iter() {
1332 let short_name = match decoder.decode_identifier(&short_name.0, warn) {
1335 warn(Error::InvalidMrSetName(error));
1339 let Some(&dict_index) = decoder.var_names.get(&short_name) else {
1340 warn(Error::UnknownMrSetVariable {
1341 mr_set: mr_set_name.clone(),
1342 short_name: short_name.clone(),
1346 dict_indexes.push(dict_index);
1349 match dict_indexes.len() {
1350 0 => return Err(Error::EmptyMrSet(mr_set_name)),
1351 1 => return Err(Error::OneVarMrSet(mr_set_name)),
1355 let Some((Some(min_width), Some(max_width))) = dict_indexes
1357 .map(|dict_index| decoder.variables[dict_index].width)
1358 .map(|w| (Some(w), Some(w)))
1359 .reduce(|(na, wa), (nb, wb)| (VarWidth::narrower(na, nb), VarWidth::wider(wa, wb)))
1361 return Err(Error::MixedMrSet(mr_set_name));
1365 MultipleResponseType::decode(decoder, &mr_set_name, &input.mr_type, min_width, warn)?;
1367 Ok(MultipleResponseSet {
1378 #[derive(Clone, Debug)]
1379 pub struct MultipleResponseRecord(pub Vec<MultipleResponseSet>);
1381 impl TryDecode for MultipleResponseRecord {
1382 type Input = raw::MultipleResponseRecord;
1385 decoder: &mut Decoder,
1386 input: &Self::Input,
1387 warn: impl Fn(Error),
1388 ) -> Result<Option<Self>, Error> {
1389 let mut sets = Vec::with_capacity(input.0.len());
1390 for set in &input.0 {
1391 match MultipleResponseSet::decode(decoder, set, &warn) {
1392 Ok(set) => sets.push(set),
1393 Err(error) => warn(error),
1396 Ok(Some(MultipleResponseRecord(sets)))
1400 #[derive(Clone, Debug)]
1401 pub struct LongStringMissingValues {
1403 pub var_name: Identifier,
1406 pub missing_values: MissingValues,
1409 impl LongStringMissingValues {
1412 input: &raw::LongStringMissingValues,
1413 warn: &impl Fn(Error),
1414 ) -> Result<Self, Error> {
1415 let var_name = decoder.decode_string(&input.var_name.0, warn);
1416 let var_name = Identifier::new(var_name.trim_end(), decoder.encoding)
1417 .map_err(Error::InvalidLongStringValueLabelName)?;
1419 let missing_values = MissingValues::decode(decoder, &input.missing_values, warn);
1421 Ok(LongStringMissingValues {
1428 #[derive(Clone, Debug)]
1429 pub struct LongStringMissingValuesRecord(Vec<LongStringMissingValues>);
1431 impl TryDecode for LongStringMissingValuesRecord {
1432 type Input = raw::LongStringMissingValueRecord;
1435 decoder: &mut Decoder,
1436 input: &Self::Input,
1437 warn: impl Fn(Error),
1438 ) -> Result<Option<Self>, Error> {
1439 let mut labels = Vec::with_capacity(input.0.len());
1440 for label in &input.0 {
1441 match LongStringMissingValues::decode(decoder, label, &warn) {
1442 Ok(set) => labels.push(set),
1443 Err(error) => warn(error),
1446 Ok(Some(LongStringMissingValuesRecord(labels)))
1450 #[derive(Clone, Debug)]
1451 pub struct LongStringValueLabels {
1452 pub var_name: Identifier,
1453 pub width: VarWidth,
1454 pub labels: Vec<ValueLabel>,
1457 impl LongStringValueLabels {
1460 input: &raw::LongStringValueLabels,
1461 warn: &impl Fn(Error),
1462 ) -> Result<Self, Error> {
1463 let var_name = decoder.decode_string(&input.var_name.0, warn);
1464 let var_name = Identifier::new(var_name.trim_end(), decoder.encoding)
1465 .map_err(Error::InvalidLongStringValueLabelName)?;
1468 let max_width = VarWidth::MAX_STRING;
1469 if input.width < 9 || input.width > max_width as u32 {
1470 return Err(Error::InvalidLongValueLabelWidth {
1477 let width = input.width as u16;
1479 let mut labels = Vec::with_capacity(input.labels.len());
1480 for (value, label) in input.labels.iter() {
1481 let value = Value::String(decoder.decode_exact_length(&value.0).into());
1482 let label = decoder.decode_string(&label.0, warn);
1483 labels.push(ValueLabel { value, label });
1486 Ok(LongStringValueLabels {
1488 width: VarWidth::String(width),
1494 #[derive(Clone, Debug)]
1495 pub struct LongStringValueLabelRecord(pub Vec<LongStringValueLabels>);
1497 impl TryDecode for LongStringValueLabelRecord {
1498 type Input = raw::LongStringValueLabelRecord;
1501 decoder: &mut Decoder,
1502 input: &Self::Input,
1503 warn: impl Fn(Error),
1504 ) -> Result<Option<Self>, Error> {
1505 let mut labels = Vec::with_capacity(input.0.len());
1506 for label in &input.0 {
1507 match LongStringValueLabels::decode(decoder, label, &warn) {
1508 Ok(set) => labels.push(set),
1509 Err(error) => warn(error),
1512 Ok(Some(LongStringValueLabelRecord(labels)))
1518 use encoding_rs::WINDOWS_1252;
1522 let mut s = String::new();
1523 s.push(char::REPLACEMENT_CHARACTER);
1524 let encoded = WINDOWS_1252.encode(&s).0;
1525 let decoded = WINDOWS_1252.decode(&encoded[..]).0;
1526 println!("{:?}", decoded);
1531 let charset: Vec<u8> = (0..=255).collect();
1532 println!("{}", charset.len());
1533 let decoded = WINDOWS_1252.decode(&charset[..]).0;
1534 println!("{}", decoded.len());
1535 let encoded = WINDOWS_1252.encode(&decoded[..]).0;
1536 println!("{}", encoded.len());
1537 assert_eq!(&charset[..], &encoded[..]);