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<RawString>>,
219 long_string_missing_values: Vec<&'a raw::LongStringMissingValueRecord<RawString, RawStr<8>>>,
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();
243 for header in headers {
245 raw::Record::Header(record) => set_or_warn(&mut h.header, record, warn),
246 raw::Record::Variable(record) => h.variables.push(record),
247 raw::Record::ValueLabel(record) => h.value_labels.push(record),
248 raw::Record::Document(record) => set_or_warn(&mut h.document, record, warn),
249 raw::Record::IntegerInfo(record) => set_or_warn(&mut h.integer_info, record, warn),
250 raw::Record::FloatInfo(record) => set_or_warn(&mut h.float_info, record, warn),
251 raw::Record::VariableSets(record) => h.variable_sets.push(record),
252 raw::Record::VarDisplay(record) => set_or_warn(&mut h.var_display, record, warn),
253 raw::Record::MultipleResponse(record) => h.multiple_response.push(record),
254 raw::Record::LongStringValueLabels(record) => {
255 h.long_string_value_labels.push(record)
257 raw::Record::LongStringMissingValues(record) => {
258 h.long_string_missing_values.push(record)
260 raw::Record::Encoding(record) => set_or_warn(&mut h.encoding, record, warn),
261 raw::Record::NumberOfCases(record) => {
262 set_or_warn(&mut h.number_of_cases, record, warn)
264 raw::Record::ProductInfo(record) => set_or_warn(&mut h.product_info, record, warn),
265 raw::Record::LongNames(record) => set_or_warn(&mut h.long_names, record, warn),
266 raw::Record::VeryLongStrings(record) => h.very_long_strings.push(record),
267 raw::Record::FileAttributes(record) => h.file_attributes.push(record),
268 raw::Record::VariableAttributes(record) => h.variable_attributes.push(record),
269 raw::Record::OtherExtension(record) => h.other_extensions.push(record),
270 raw::Record::EndOfHeaders(_) => (),
271 raw::Record::ZHeader(_) => (),
272 raw::Record::ZTrailer(_) => (),
273 raw::Record::Cases(record) => set_or_warn(&mut h.cases, record, warn),
282 headers: Vec<raw::Record>,
283 encoding: Option<&'static Encoding>,
284 warn: &impl Fn(Error),
285 ) -> Result<Vec<Record>, Error> {
286 let h = Headers::new(&headers, warn);
287 let Some(header) = h.header else {
288 return Err(Error::MissingHeaderRecord);
290 let encoding = match encoding {
291 Some(encoding) => encoding,
293 let encoding = h.encoding.map(|record| record.0.as_str());
294 let character_code = h.integer_info.map(|record| record.character_code);
295 match get_encoding(encoding, character_code) {
296 Ok(encoding) => encoding,
297 Err(err @ EncodingError::Ebcdic) => return Err(Error::EncodingError(err)),
299 warn(Error::EncodingError(err));
300 // Warn that we're using the default encoding.
307 //let mut dictionary = Dictionary::new(encoding);
309 let mut decoder = Decoder {
310 compression: header.compression,
311 endian: header.endian,
313 variables: HashMap::new(),
314 var_names: HashMap::new(),
316 n_generated_names: 0,
319 let mut output = Vec::with_capacity(headers.len());
321 // Decode the records that don't use variables at all.
322 if let Some(header) = HeaderRecord::try_decode(&mut decoder, header, warn)? {
323 output.push(Record::Header(header))
325 if let Some(raw) = h.document {
326 if let Some(document) = DocumentRecord::try_decode(&mut decoder, raw, warn)? {
327 output.push(Record::Document(document))
330 if let Some(raw) = h.integer_info {
331 output.push(Record::IntegerInfo(raw.clone()));
333 if let Some(raw) = h.float_info {
334 output.push(Record::FloatInfo(raw.clone()));
336 if let Some(raw) = h.product_info {
337 let s = decoder.decode_string_cow(&raw.text.0, warn);
338 output.push(Record::ProductInfo(ProductInfoRecord::parse(&s, warn)?));
340 if let Some(raw) = h.number_of_cases {
341 output.push(Record::NumberOfCases(raw.clone()))
343 for &raw in &h.file_attributes {
344 let s = decoder.decode_string_cow(&raw.text.0, warn);
345 output.push(Record::FileAttributes(FileAttributeRecord::parse(
349 for &raw in &h.other_extensions {
350 output.push(Record::OtherExtension(raw.clone()));
353 // Decode the variable records, which are the basis of almost everything
355 for &raw in &h.variables {
356 if let Some(variable) = VariableRecord::try_decode(&mut decoder, raw, warn)? {
357 output.push(Record::Variable(variable));
361 // Decode value labels and weight variable. These use indexes into the
362 // variable records, so we need to parse them before those indexes become
363 // invalidated by very long string variables.
364 for &raw in &h.value_labels {
365 if let Some(value_label) = ValueLabelRecord::try_decode(&mut decoder, raw, warn)? {
366 output.push(Record::ValueLabel(value_label));
370 if let Some(raw) = h.var_display {
371 output.push(Record::VarDisplay(raw.clone()));
374 // Decode records that use short names.
376 for &raw in &h.multiple_response {
377 if let Some(mrr) = MultipleResponseRecord::try_decode(&mut decoder, raw, warn)? {
378 output.push(Record::MultipleResponse(mrr))
382 for &raw in &h.very_long_strings {
383 let s = decoder.decode_string_cow(&raw.text.0, warn);
384 output.push(Record::VeryLongStrings(VeryLongStringRecord::parse(
389 // Rename variables to their long names.
390 for &raw in &h.long_names {
391 let s = decoder.decode_string_cow(&raw.text.0, warn);
392 output.push(Record::LongNames(LongNameRecord::parse(
399 // Decode recods that use long names.
400 for &raw in &h.variable_attributes {
401 let s = decoder.decode_string_cow(&raw.text.0, warn);
402 output.push(Record::VariableAttributes(VariableAttributeRecord::parse(
406 for &raw in &h.long_string_value_labels {
407 if let Some(mrr) = LongStringValueLabelRecord::try_decode(&mut decoder, raw, warn)? {
408 output.push(Record::LongStringValueLabels(mrr))
411 for &raw in &h.long_string_missing_values {
412 if let Some(mrr) = LongStringMissingValuesRecord::try_decode(&mut decoder, raw, warn)? {
413 output.push(Record::LongStringMissingValues(mrr))
416 for &raw in &h.variable_sets {
417 let s = decoder.decode_string_cow(&raw.text.0, warn);
418 output.push(Record::VariableSets(VariableSetRecord::parse(&s, warn)?));
424 fn generate_name(&mut self) -> Identifier {
426 self.n_generated_names += 1;
427 let name = Identifier::new(&format!("VAR{:03}", self.n_generated_names), self.encoding)
429 if !self.var_names.contains_key(&name) {
432 assert!(self.n_generated_names < usize::MAX);
435 fn decode_string_cow<'a>(&self, input: &'a [u8], warn: &impl Fn(Error)) -> Cow<'a, str> {
436 let (output, malformed) = self.encoding.decode_without_bom_handling(input);
438 warn(Error::MalformedString {
439 encoding: self.encoding.name().into(),
440 text: output.clone().into(),
445 fn decode_string(&self, input: &[u8], warn: &impl Fn(Error)) -> String {
446 self.decode_string_cow(input, warn).into()
448 pub fn decode_identifier(
451 warn: &impl Fn(Error),
452 ) -> Result<Identifier, IdError> {
453 let s = self.decode_string_cow(input, warn);
454 Identifier::new(&s, self.encoding)
456 fn get_var_by_index(&self, dict_index: usize) -> Result<&Variable, Error> {
457 let max_index = self.n_dict_indexes;
458 if dict_index == 0 || dict_index > max_index {
459 return Err(Error::InvalidDictIndex {
464 let Some(variable) = self.variables.get(&(dict_index - 1)) else {
465 return Err(Error::DictIndexIsContinuation(dict_index));
470 /// Returns `input` decoded from `self.encoding` into UTF-8 such that
471 /// re-encoding the result back into `self.encoding` will have exactly the
472 /// same length in bytes.
474 /// XXX warn about errors?
475 fn decode_exact_length<'a>(&self, input: &'a [u8]) -> Cow<'a, str> {
476 if let (s, false) = self.encoding.decode_without_bom_handling(input) {
477 // This is the common case. Usually there will be no errors.
480 // Unusual case. Don't bother to optimize it much.
481 let mut decoder = self.encoding.new_decoder_without_bom_handling();
482 let mut output = String::with_capacity(
484 .max_utf8_buffer_length_without_replacement(input.len())
487 let mut rest = input;
488 while !rest.is_empty() {
489 match decoder.decode_to_string_without_replacement(rest, &mut output, true) {
490 (DecoderResult::InputEmpty, _) => break,
491 (DecoderResult::OutputFull, _) => unreachable!(),
492 (DecoderResult::Malformed(a, b), consumed) => {
493 let skipped = a as usize + b as usize;
494 output.extend(repeat('?').take(skipped));
495 rest = &rest[consumed..];
499 assert_eq!(self.encoding.encode(&output).0.len(), input.len());
505 pub trait TryDecode: Sized {
508 decoder: &mut Decoder,
509 input: &Self::Input<'_>,
510 warn: impl Fn(Error),
511 ) -> Result<Option<Self>, Error>;
514 pub trait Decode<Input>: Sized {
515 fn decode(decoder: &Decoder, input: &Input, warn: impl Fn(Error)) -> Self;
518 impl<const N: usize> Decode<RawStr<N>> for String {
519 fn decode(decoder: &Decoder, input: &RawStr<N>, warn: impl Fn(Error)) -> Self {
520 decoder.decode_string(&input.0, &warn)
524 #[derive(Clone, Debug)]
525 pub struct HeaderRecord {
526 pub eye_catcher: String,
527 pub weight_index: Option<usize>,
528 pub n_cases: Option<u64>,
529 pub creation: NaiveDateTime,
530 pub file_label: String,
533 fn trim_end_spaces(mut s: String) -> String {
534 s.truncate(s.trim_end_matches(' ').len());
538 impl TryDecode for HeaderRecord {
539 type Input<'a> = crate::raw::HeaderRecord<RawString>;
542 decoder: &mut Decoder,
543 input: &Self::Input<'_>,
544 warn: impl Fn(Error),
545 ) -> Result<Option<Self>, Error> {
546 let eye_catcher = trim_end_spaces(decoder.decode_string(&input.eye_catcher.0, &warn));
547 let file_label = trim_end_spaces(decoder.decode_string(&input.file_label.0, &warn));
548 let creation_date = decoder.decode_string_cow(&input.creation_date.0, &warn);
550 NaiveDate::parse_from_str(&creation_date, "%e %b %Y").unwrap_or_else(|_| {
551 warn(Error::InvalidCreationDate {
552 creation_date: creation_date.into(),
556 let creation_time = decoder.decode_string_cow(&input.creation_time.0, &warn);
558 NaiveTime::parse_from_str(&creation_time, "%H:%M:%S").unwrap_or_else(|_| {
559 warn(Error::InvalidCreationTime {
560 creation_time: creation_time.into(),
564 Ok(Some(HeaderRecord {
566 weight_index: input.weight_index.map(|n| n as usize),
567 n_cases: input.n_cases.map(|n| n as u64),
568 creation: NaiveDateTime::new(creation_date, creation_time),
574 #[derive(Copy, Clone, Debug, PartialEq, Eq)]
580 impl PartialOrd for VarWidth {
581 fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
582 match (self, other) {
583 (VarWidth::Numeric, VarWidth::Numeric) => Some(Ordering::Equal),
584 (VarWidth::String(a), VarWidth::String(b)) => Some(a.cmp(b)),
591 const MAX_STRING: u16 = 32767;
593 fn n_dict_indexes(self) -> usize {
595 VarWidth::Numeric => 1,
596 VarWidth::String(w) => div_ceil(w as usize, 8),
603 f: impl Fn(u16, u16) -> u16,
604 ) -> Option<VarWidth> {
606 (Some(VarWidth::Numeric), Some(VarWidth::Numeric)) => Some(VarWidth::Numeric),
607 (Some(VarWidth::String(a)), Some(VarWidth::String(b))) => {
608 Some(VarWidth::String(f(a, b)))
614 /// Returns the wider of `self` and `other`:
615 /// - Numerical variable widths are equally wide.
616 /// - Longer strings are wider than shorter strings.
617 /// - Numerical and string types are incomparable, so result in `None`.
618 /// - Any `None` in the input yields `None` in the output.
619 pub fn wider(a: Option<VarWidth>, b: Option<VarWidth>) -> Option<VarWidth> {
620 Self::width_predicate(a, b, |a, b| a.max(b))
623 /// Returns the narrower of `self` and `other` (see [`Self::wider`]).
624 pub fn narrower(a: Option<VarWidth>, b: Option<VarWidth>) -> Option<VarWidth> {
625 Self::width_predicate(a, b, |a, b| a.min(b))
629 impl From<VarWidth> for VarType {
630 fn from(source: VarWidth) -> Self {
632 VarWidth::Numeric => VarType::Numeric,
633 VarWidth::String(_) => VarType::String,
638 #[derive(Clone, Debug)]
639 pub struct VariableRecord {
641 pub name: Identifier,
642 pub print_format: Spec,
643 pub write_format: Spec,
644 pub missing_values: MissingValues,
645 pub label: Option<String>,
648 #[derive(Clone, Debug)]
649 pub struct MissingValues {
650 /// Individual missing values, up to 3 of them.
651 pub values: Vec<Value>,
653 /// Optional range of missing values.
654 pub range: Option<(Value, Value)>,
657 impl Decode<raw::MissingValues<RawStr<8>>> for MissingValues {
660 input: &raw::MissingValues<RawStr<8>>,
661 _warn: impl Fn(Error),
667 .map(|value| Value::decode(value, decoder))
672 .map(|(low, high)| (Value::decode(low, decoder), Value::decode(high, decoder))),
677 fn decode_format(raw: raw::Spec, width: VarWidth, warn: impl Fn(Spec, FormatError)) -> Spec {
678 UncheckedSpec::try_from(raw)
679 .and_then(Spec::try_from)
680 .and_then(|x| x.check_width_compatibility(width))
681 .unwrap_or_else(|error| {
682 let new_format = Spec::default_for_width(width);
683 warn(new_format, error);
688 impl TryDecode for VariableRecord {
689 type Input<'a> = raw::VariableRecord<RawString, RawStr<8>>;
692 decoder: &mut Decoder,
693 input: &Self::Input<'_>,
694 warn: impl Fn(Error),
695 ) -> Result<Option<VariableRecord>, Error> {
696 let width = match input.width {
697 0 => VarWidth::Numeric,
698 w @ 1..=255 => VarWidth::String(w as u16),
699 -1 => return Ok(None),
701 return Err(Error::InvalidVariableWidth {
702 offsets: input.offsets.clone(),
707 let name = trim_end_spaces(decoder.decode_string(&input.name.0, &warn));
708 let name = match Identifier::new(&name, decoder.encoding) {
710 if !decoder.var_names.contains_key(&name) {
713 let new_name = decoder.generate_name();
714 warn(Error::DuplicateVariableName {
715 duplicate_name: name.clone(),
716 new_name: new_name.clone(),
722 let new_name = decoder.generate_name();
723 warn(Error::InvalidVariableName {
725 new_name: new_name.clone(),
730 let variable = Variable {
731 dict_index: decoder.n_dict_indexes,
732 short_name: name.clone(),
736 decoder.n_dict_indexes += width.n_dict_indexes();
739 .insert(name.clone(), variable.dict_index)
743 .insert(variable.dict_index, variable)
746 let print_format = decode_format(input.print_format, width, |new_spec, format_error| {
747 warn(Error::InvalidPrintFormat {
749 variable: name.clone(),
753 let write_format = decode_format(input.write_format, width, |new_spec, format_error| {
754 warn(Error::InvalidWriteFormat {
756 variable: name.clone(),
763 .map(|label| decoder.decode_string(&label.0, &warn));
764 Ok(Some(VariableRecord {
769 missing_values: MissingValues::decode(decoder, &input.missing_values, warn),
775 #[derive(Clone, Debug)]
776 pub struct DocumentRecord(Vec<String>);
778 impl TryDecode for DocumentRecord {
779 type Input<'a> = crate::raw::DocumentRecord<RawDocumentLine>;
782 decoder: &mut Decoder,
783 input: &Self::Input<'_>,
784 warn: impl Fn(Error),
785 ) -> Result<Option<Self>, Error> {
786 Ok(Some(DocumentRecord(
790 .map(|s| trim_end_spaces(decoder.decode_string(&s.0, &warn)))
800 const NAME: &'static str;
801 fn parse(input: &str, warn: impl Fn(Error)) -> Result<Self, Error>;
804 #[derive(Clone, Debug)]
805 pub struct VariableSet {
807 pub vars: Vec<String>,
811 fn parse(input: &str) -> Result<Self, Error> {
812 let (name, input) = input.split_once('=').ok_or(Error::TBD)?;
813 let vars = input.split_ascii_whitespace().map(String::from).collect();
821 trait WarnOnError<T> {
822 fn warn_on_error<F: Fn(Error)>(self, warn: &F) -> Option<T>;
824 impl<T> WarnOnError<T> for Result<T, Error> {
825 fn warn_on_error<F: Fn(Error)>(self, warn: &F) -> Option<T> {
827 Ok(result) => Some(result),
836 #[derive(Clone, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)]
838 Number(Option<OrderedFloat<f64>>),
843 pub fn decode(raw: &raw::Value<RawStr<8>>, decoder: &Decoder) -> Self {
845 raw::Value::Number(x) => Value::Number(x.map(|x| x.into())),
846 raw::Value::String(s) => Value::String(decoder.decode_exact_length(&s.0).into()),
851 #[derive(Clone, Debug)]
852 pub struct ValueLabel {
857 #[derive(Clone, Debug)]
858 pub struct ValueLabelRecord {
859 pub var_type: VarType,
860 pub labels: Vec<ValueLabel>,
861 pub variables: Vec<Identifier>,
864 impl TryDecode for ValueLabelRecord {
865 type Input<'a> = crate::raw::ValueLabelRecord<RawStr<8>, RawString>;
867 decoder: &mut Decoder,
868 input: &Self::Input<'_>,
869 warn: impl Fn(Error),
870 ) -> Result<Option<ValueLabelRecord>, Error> {
871 let variables: Vec<&Variable> = input
874 .filter_map(|&dict_index| {
876 .get_var_by_index(dict_index as usize)
877 .warn_on_error(&warn)
879 .filter(|&variable| match variable.width {
880 VarWidth::String(width) if width > 8 => {
881 warn(Error::InvalidLongStringValueLabel(
882 variable.short_name.clone(),
889 let mut i = variables.iter();
890 let Some(&first_var) = i.next() else {
893 let var_type: VarType = first_var.width.into();
895 let this_type: VarType = variable.width.into();
896 if var_type != this_type {
897 let (numeric_var, string_var) = match var_type {
898 VarType::Numeric => (first_var, variable),
899 VarType::String => (variable, first_var),
901 warn(Error::ValueLabelsDifferentTypes {
902 numeric_var: numeric_var.short_name.clone(),
903 string_var: string_var.short_name.clone(),
911 .map(|raw::ValueLabel { value, label }| {
912 let label = decoder.decode_string(&label.0, &warn);
913 let value = Value::decode(value, decoder);
914 ValueLabel { value, label }
917 let variables = variables
919 .map(|&variable| variable.short_name.clone())
921 Ok(Some(ValueLabelRecord {
929 #[derive(Clone, Debug)]
930 pub struct VariableSetRecord(Vec<VariableSet>);
932 impl TextRecord for VariableSetRecord {
933 const NAME: &'static str = "variable set";
934 fn parse(input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
935 let mut sets = Vec::new();
936 for line in input.lines() {
937 if let Some(set) = VariableSet::parse(line).warn_on_error(&warn) {
941 Ok(VariableSetRecord(sets))
945 #[derive(Clone, Debug)]
946 pub struct ProductInfoRecord(pub String);
948 impl TextRecord for ProductInfoRecord {
949 const NAME: &'static str = "extra product info";
950 fn parse(input: &str, _warn: impl Fn(Error)) -> Result<Self, Error> {
951 Ok(ProductInfoRecord(input.into()))
955 #[derive(Clone, Debug)]
956 pub struct LongName {
957 pub short_name: Identifier,
958 pub long_name: Identifier,
962 fn new(decoder: &mut Decoder, short_name: &str, long_name: &str) -> Result<LongName, Error> {
964 Identifier::new(short_name, decoder.encoding).map_err(Error::InvalidShortName)?;
966 Identifier::new(long_name, decoder.encoding).map_err(Error::InvalidLongName)?;
974 #[derive(Clone, Debug)]
975 pub struct LongNameRecord(Vec<LongName>);
977 impl LongNameRecord {
978 pub fn parse(decoder: &mut Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
979 let mut names = Vec::new();
980 for pair in input.split('\t').filter(|s| !s.is_empty()) {
981 if let Some((short_name, long_name)) = pair.split_once('=') {
982 if let Some(long_name) =
983 LongName::new(decoder, short_name, long_name).warn_on_error(&warn)
985 names.push(long_name);
991 Ok(LongNameRecord(names))
995 #[derive(Clone, Debug)]
996 pub struct VeryLongString {
997 pub short_name: Identifier,
1001 impl VeryLongString {
1002 fn parse(decoder: &Decoder, input: &str) -> Result<VeryLongString, Error> {
1003 let Some((short_name, length)) = input.split_once('=') else {
1004 return Err(Error::TBD);
1007 Identifier::new(short_name, decoder.encoding).map_err(Error::InvalidLongStringName)?;
1008 let length: u16 = length.parse().map_err(|_| Error::TBD)?;
1009 if length > VarWidth::MAX_STRING {
1010 return Err(Error::TBD);
1012 Ok(VeryLongString { short_name, length })
1016 #[derive(Clone, Debug)]
1017 pub struct VeryLongStringRecord(Vec<VeryLongString>);
1019 impl VeryLongStringRecord {
1020 pub fn parse(decoder: &Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1021 let mut very_long_strings = Vec::new();
1024 .map(|s| s.trim_end_matches('\t'))
1025 .filter(|s| !s.is_empty())
1027 if let Some(vls) = VeryLongString::parse(decoder, tuple).warn_on_error(&warn) {
1028 very_long_strings.push(vls)
1031 Ok(VeryLongStringRecord(very_long_strings))
1035 #[derive(Clone, Debug)]
1036 pub struct Attribute {
1037 pub name: Identifier,
1038 pub values: Vec<String>,
1045 warn: &impl Fn(Error),
1046 ) -> Result<(Option<Attribute>, &'a str), Error> {
1047 let Some((name, mut input)) = input.split_once('(') else {
1048 return Err(Error::TBD);
1050 let mut values = Vec::new();
1052 let Some((value, rest)) = input.split_once('\n') else {
1053 return Err(Error::TBD);
1055 if let Some(stripped) = value
1057 .and_then(|value| value.strip_suffix('\''))
1059 values.push(stripped.into());
1062 values.push(value.into());
1064 if let Some(rest) = rest.strip_prefix(')') {
1065 let attribute = Identifier::new(name, decoder.encoding)
1066 .map_err(Error::InvalidAttributeName)
1067 .warn_on_error(warn)
1068 .map(|name| Attribute { name, values });
1069 return Ok((attribute, rest));
1076 #[derive(Clone, Debug)]
1077 pub struct AttributeSet(pub Vec<Attribute>);
1083 sentinel: Option<char>,
1084 warn: &impl Fn(Error),
1085 ) -> Result<(AttributeSet, &'a str), Error> {
1086 let mut attributes = Vec::new();
1088 match input.chars().next() {
1089 None => break input,
1090 c if c == sentinel => break &input[1..],
1092 let (attribute, rest) = Attribute::parse(decoder, input, &warn)?;
1093 if let Some(attribute) = attribute {
1094 attributes.push(attribute);
1100 Ok((AttributeSet(attributes), rest))
1104 #[derive(Clone, Debug)]
1105 pub struct FileAttributeRecord(AttributeSet);
1107 impl FileAttributeRecord {
1108 pub fn parse(decoder: &Decoder, input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1109 let (set, rest) = AttributeSet::parse(decoder, input, None, &warn)?;
1110 if !rest.is_empty() {
1113 Ok(FileAttributeRecord(set))
1117 #[derive(Clone, Debug)]
1118 pub struct VarAttributeSet {
1119 pub long_var_name: Identifier,
1120 pub attributes: AttributeSet,
1123 impl VarAttributeSet {
1127 warn: &impl Fn(Error),
1128 ) -> Result<(Option<VarAttributeSet>, &'a str), Error> {
1129 let Some((long_var_name, rest)) = input.split_once(':') else {
1130 return Err(Error::TBD);
1132 let (attributes, rest) = AttributeSet::parse(decoder, rest, Some('/'), warn)?;
1133 let var_attribute = Identifier::new(long_var_name, decoder.encoding)
1134 .map_err(Error::InvalidAttributeVariableName)
1135 .warn_on_error(warn)
1136 .map(|name| VarAttributeSet {
1137 long_var_name: name,
1140 Ok((var_attribute, rest))
1144 #[derive(Clone, Debug)]
1145 pub struct VariableAttributeRecord(Vec<VarAttributeSet>);
1147 impl VariableAttributeRecord {
1148 pub fn parse(decoder: &Decoder, mut input: &str, warn: impl Fn(Error)) -> Result<Self, Error> {
1149 let mut var_attribute_sets = Vec::new();
1150 while !input.is_empty() {
1151 let Some((var_attribute, rest)) =
1152 VarAttributeSet::parse(decoder, input, &warn).warn_on_error(&warn)
1156 if let Some(var_attribute) = var_attribute {
1157 var_attribute_sets.push(var_attribute);
1161 Ok(VariableAttributeRecord(var_attribute_sets))
1165 #[derive(Clone, Debug)]
1166 pub enum MultipleResponseType {
1169 labels: CategoryLabels,
1174 impl MultipleResponseType {
1177 mr_set: &Identifier,
1178 input: &raw::MultipleResponseType,
1179 min_width: VarWidth,
1180 warn: &impl Fn(Error),
1181 ) -> Result<Self, Error> {
1182 let mr_type = match input {
1183 raw::MultipleResponseType::MultipleDichotomy { value, labels } => {
1184 let value = decoder.decode_string_cow(&value.0, warn);
1185 let value = match min_width {
1186 VarWidth::Numeric => {
1187 let number: f64 = value.trim().parse().map_err(|_| {
1188 Error::InvalidMDGroupCountedValue {
1189 mr_set: mr_set.clone(),
1190 number: value.into(),
1193 Value::Number(Some(number.into()))
1195 VarWidth::String(max_width) => {
1196 let value = value.trim_end_matches(' ');
1197 let width = value.len();
1198 if width > max_width as usize {
1199 return Err(Error::TooWideMDGroupCountedValue {
1200 mr_set: mr_set.clone(),
1201 value: value.into(),
1206 Value::String(value.into())
1209 MultipleResponseType::MultipleDichotomy {
1214 raw::MultipleResponseType::MultipleCategory => MultipleResponseType::MultipleCategory,
1220 #[derive(Clone, Debug)]
1221 pub struct MultipleResponseSet {
1222 pub name: Identifier,
1223 pub min_width: VarWidth,
1224 pub max_width: VarWidth,
1226 pub mr_type: MultipleResponseType,
1227 pub dict_indexes: Vec<DictIndex>,
1230 impl MultipleResponseSet {
1233 input: &raw::MultipleResponseSet<Cow<str>>,
1234 warn: &impl Fn(Error),
1235 ) -> Result<Self, Error> {
1237 Identifier::new(&input.name, decoder.encoding).map_err(Error::InvalidMrSetName)?;
1239 let mut dict_indexes = Vec::with_capacity(input.short_names.len());
1240 for short_name in input.short_names.iter() {
1241 let short_name = match Identifier::new(&short_name, decoder.encoding) {
1244 warn(Error::InvalidMrSetName(error));
1248 let Some(&dict_index) = decoder.var_names.get(&short_name) else {
1249 warn(Error::UnknownMrSetVariable {
1250 mr_set: mr_set_name.clone(),
1251 short_name: short_name.clone(),
1255 dict_indexes.push(dict_index);
1258 match dict_indexes.len() {
1259 0 => return Err(Error::EmptyMrSet(mr_set_name)),
1260 1 => return Err(Error::OneVarMrSet(mr_set_name)),
1264 let Some((Some(min_width), Some(max_width))) = dict_indexes
1266 .map(|dict_index| decoder.variables[dict_index].width)
1267 .map(|w| (Some(w), Some(w)))
1268 .reduce(|(na, wa), (nb, wb)| (VarWidth::narrower(na, nb), VarWidth::wider(wa, wb)))
1270 return Err(Error::MixedMrSet(mr_set_name));
1274 MultipleResponseType::decode(decoder, &mr_set_name, &input.mr_type, min_width, warn)?;
1276 Ok(MultipleResponseSet {
1280 label: input.label.to_string(),
1287 #[derive(Clone, Debug)]
1288 pub struct MultipleResponseRecord(pub Vec<MultipleResponseSet>);
1290 impl TryDecode for MultipleResponseRecord {
1291 type Input<'a> = raw::MultipleResponseRecord<Cow<'a, str>>;
1294 decoder: &mut Decoder,
1295 input: &Self::Input<'_>,
1296 warn: impl Fn(Error),
1297 ) -> Result<Option<Self>, Error> {
1298 let mut sets = Vec::with_capacity(input.0.len());
1299 for set in &input.0 {
1300 match MultipleResponseSet::decode(decoder, set, &warn) {
1301 Ok(set) => sets.push(set),
1302 Err(error) => warn(error),
1305 Ok(Some(MultipleResponseRecord(sets)))
1309 #[derive(Clone, Debug)]
1310 pub struct LongStringMissingValues {
1312 pub var_name: Identifier,
1315 pub missing_values: MissingValues,
1318 impl LongStringMissingValues {
1321 input: &raw::LongStringMissingValues<RawString, RawStr<8>>,
1322 warn: &impl Fn(Error),
1323 ) -> Result<Self, Error> {
1324 let var_name = decoder.decode_string(&input.var_name.0, warn);
1325 let var_name = Identifier::new(var_name.trim_end(), decoder.encoding)
1326 .map_err(Error::InvalidLongStringValueLabelName)?;
1328 let missing_values = MissingValues::decode(decoder, &input.missing_values, warn);
1330 Ok(LongStringMissingValues {
1337 #[derive(Clone, Debug)]
1338 pub struct LongStringMissingValuesRecord(Vec<LongStringMissingValues>);
1340 impl TryDecode for LongStringMissingValuesRecord {
1341 type Input<'a> = raw::LongStringMissingValueRecord<RawString, RawStr<8>>;
1344 decoder: &mut Decoder,
1345 input: &Self::Input<'_>,
1346 warn: impl Fn(Error),
1347 ) -> Result<Option<Self>, Error> {
1348 let mut labels = Vec::with_capacity(input.0.len());
1349 for label in &input.0 {
1350 match LongStringMissingValues::decode(decoder, label, &warn) {
1351 Ok(set) => labels.push(set),
1352 Err(error) => warn(error),
1355 Ok(Some(LongStringMissingValuesRecord(labels)))
1359 #[derive(Clone, Debug)]
1360 pub struct LongStringValueLabels {
1361 pub var_name: Identifier,
1362 pub width: VarWidth,
1363 pub labels: Vec<ValueLabel>,
1366 impl LongStringValueLabels {
1369 input: &raw::LongStringValueLabels<RawString>,
1370 warn: &impl Fn(Error),
1371 ) -> Result<Self, Error> {
1372 let var_name = decoder.decode_string(&input.var_name.0, warn);
1373 let var_name = Identifier::new(var_name.trim_end(), decoder.encoding)
1374 .map_err(Error::InvalidLongStringValueLabelName)?;
1377 let max_width = VarWidth::MAX_STRING;
1378 if input.width < 9 || input.width > max_width as u32 {
1379 return Err(Error::InvalidLongValueLabelWidth {
1386 let width = input.width as u16;
1388 let mut labels = Vec::with_capacity(input.labels.len());
1389 for (value, label) in input.labels.iter() {
1390 let value = Value::String(decoder.decode_exact_length(&value.0).into());
1391 let label = decoder.decode_string(&label.0, warn);
1392 labels.push(ValueLabel { value, label });
1395 Ok(LongStringValueLabels {
1397 width: VarWidth::String(width),
1403 #[derive(Clone, Debug)]
1404 pub struct LongStringValueLabelRecord(pub Vec<LongStringValueLabels>);
1406 impl TryDecode for LongStringValueLabelRecord {
1407 type Input<'a> = raw::LongStringValueLabelRecord<RawString>;
1410 decoder: &mut Decoder,
1411 input: &Self::Input<'_>,
1412 warn: impl Fn(Error),
1413 ) -> Result<Option<Self>, Error> {
1414 let mut labels = Vec::with_capacity(input.0.len());
1415 for label in &input.0 {
1416 match LongStringValueLabels::decode(decoder, label, &warn) {
1417 Ok(set) => labels.push(set),
1418 Err(error) => warn(error),
1421 Ok(Some(LongStringValueLabelRecord(labels)))
1427 use encoding_rs::WINDOWS_1252;
1431 let mut s = String::new();
1432 s.push(char::REPLACEMENT_CHARACTER);
1433 let encoded = WINDOWS_1252.encode(&s).0;
1434 let decoded = WINDOWS_1252.decode(&encoded[..]).0;
1435 println!("{:?}", decoded);
1440 let charset: Vec<u8> = (0..=255).collect();
1441 println!("{}", charset.len());
1442 let decoded = WINDOWS_1252.decode(&charset[..]).0;
1443 println!("{}", decoded.len());
1444 let encoded = WINDOWS_1252.encode(&decoded[..]).0;
1445 println!("{}", encoded.len());
1446 assert_eq!(&charset[..], &encoded[..]);