oxidize_pdf/parser/
lexer.rs

1//! PDF Lexer
2//!
3//! Tokenizes PDF syntax according to ISO 32000-1 Section 7.2
4
5use super::{ParseError, ParseOptions, ParseResult, ParseWarning};
6use std::io::{Read, Seek, SeekFrom};
7
8/// PDF Token types
9#[derive(Debug, Clone, PartialEq)]
10pub enum Token {
11    /// Boolean: true or false
12    Boolean(bool),
13
14    /// Integer number
15    Integer(i64),
16
17    /// Real number
18    Real(f64),
19
20    /// String (literal or hexadecimal)
21    String(Vec<u8>),
22
23    /// Name object (e.g., /Type)
24    Name(String),
25
26    /// Left square bracket [
27    ArrayStart,
28
29    /// Right square bracket ]
30    ArrayEnd,
31
32    /// Dictionary start <<
33    DictStart,
34
35    /// Dictionary end >>
36    DictEnd,
37
38    /// Stream keyword
39    Stream,
40
41    /// Endstream keyword
42    EndStream,
43
44    /// Obj keyword
45    Obj,
46
47    /// Endobj keyword
48    EndObj,
49
50    /// StartXRef keyword
51    StartXRef,
52
53    /// Reference (e.g., 1 0 R)
54    Reference(u32, u16),
55
56    /// Null object
57    Null,
58
59    /// Comment (usually ignored)
60    Comment(String),
61
62    /// End of file
63    Eof,
64}
65
66/// PDF Lexer for tokenizing PDF content
67pub struct Lexer<R> {
68    reader: std::io::BufReader<R>,
69    #[allow(dead_code)]
70    buffer: Vec<u8>,
71    position: usize,
72    peek_buffer: Option<u8>,
73    token_buffer: Vec<Token>,
74    options: ParseOptions,
75    warnings: Vec<ParseWarning>,
76}
77
78impl<R: Read> Lexer<R> {
79    /// Create a new lexer from a reader with default options
80    pub fn new(reader: R) -> Self {
81        Self::new_with_options(reader, ParseOptions::default())
82    }
83
84    /// Create a new lexer from a reader with custom options
85    pub fn new_with_options(reader: R, options: ParseOptions) -> Self {
86        Self {
87            reader: std::io::BufReader::new(reader),
88            buffer: Vec::with_capacity(1024),
89            position: 0,
90            peek_buffer: None,
91            token_buffer: Vec::new(),
92            options,
93            warnings: Vec::new(),
94        }
95    }
96
97    /// Get warnings collected during lexing (if enabled)
98    pub fn warnings(&self) -> &[ParseWarning] {
99        &self.warnings
100    }
101
102    /// Get the next token
103    pub fn next_token(&mut self) -> ParseResult<Token> {
104        // Check if we have a pushed-back token
105        if let Some(token) = self.token_buffer.pop() {
106            return Ok(token);
107        }
108
109        self.skip_whitespace()?;
110
111        let ch = match self.peek_char()? {
112            Some(ch) => ch,
113            None => return Ok(Token::Eof),
114        };
115
116        match ch {
117            b'%' => self.read_comment(),
118            b'/' => self.read_name(),
119            b'(' => self.read_literal_string(),
120            b'<' => self.read_angle_bracket(),
121            b'>' => {
122                self.consume_char()?;
123                if self.peek_char()? == Some(b'>') {
124                    self.consume_char()?;
125                    Ok(Token::DictEnd)
126                } else {
127                    Err(ParseError::SyntaxError {
128                        position: self.position,
129                        message: "Expected '>' after '>'".to_string(),
130                    })
131                }
132            }
133            b'[' => {
134                self.consume_char()?;
135                Ok(Token::ArrayStart)
136            }
137            b']' => {
138                self.consume_char()?;
139                Ok(Token::ArrayEnd)
140            }
141            b't' | b'f' => self.read_boolean(),
142            b'n' => self.read_null(),
143            b'+' | b'-' | b'0'..=b'9' | b'.' => self.read_number(),
144            b'R' => {
145                // R could be a keyword (for references)
146                self.consume_char()?;
147                Ok(Token::Name("R".to_string()))
148            }
149            _ if ch.is_ascii_alphabetic() => self.read_keyword(),
150            b';' => {
151                // Skip semicolons (corrupted PDF recovery)
152                self.consume_char()?;
153                self.next_token() // Recursively get next valid token
154            }
155            _ => {
156                // Check if this is a problematic encoding character
157                if self.is_problematic_encoding_char(ch) {
158                    self.handle_encoding_char_in_token_stream(ch)
159                } else if self.options.lenient_syntax {
160                    // In lenient mode, skip unexpected characters with a warning
161                    if self.options.collect_warnings {
162                        tracing::debug!(
163                            "Warning: Skipping unexpected character '{}' at position {}",
164                            ch as char,
165                            self.position
166                        );
167                    }
168                    self.consume_char()?;
169                    self.next_token() // Continue with next token
170                } else {
171                    Err(ParseError::SyntaxError {
172                        position: self.position,
173                        message: format!("Unexpected character: {}", ch as char),
174                    })
175                }
176            }
177        }
178    }
179
180    /// Peek at the next character without consuming it
181    fn peek_char(&mut self) -> ParseResult<Option<u8>> {
182        if let Some(ch) = self.peek_buffer {
183            return Ok(Some(ch));
184        }
185
186        let mut buf = [0u8; 1];
187        match self.reader.read_exact(&mut buf) {
188            Ok(_) => {
189                self.peek_buffer = Some(buf[0]);
190                Ok(Some(buf[0]))
191            }
192            Err(e) if e.kind() == std::io::ErrorKind::UnexpectedEof => Ok(None),
193            Err(e) => Err(e.into()),
194        }
195    }
196
197    /// Consume the next character
198    fn consume_char(&mut self) -> ParseResult<Option<u8>> {
199        let ch = self.peek_char()?;
200        if ch.is_some() {
201            self.peek_buffer = None;
202            self.position += 1;
203        }
204        Ok(ch)
205    }
206
207    /// Skip whitespace and return the number of bytes skipped
208    pub(crate) fn skip_whitespace(&mut self) -> ParseResult<usize> {
209        let mut count = 0;
210        while let Some(ch) = self.peek_char()? {
211            if ch.is_ascii_whitespace() {
212                self.consume_char()?;
213                count += 1;
214            } else {
215                break;
216            }
217        }
218        Ok(count)
219    }
220
221    /// Read a comment (from % to end of line)
222    fn read_comment(&mut self) -> ParseResult<Token> {
223        self.consume_char()?; // consume '%'
224        let mut comment = String::new();
225
226        while let Some(ch) = self.peek_char()? {
227            if ch == b'\n' || ch == b'\r' {
228                break;
229            }
230            self.consume_char()?;
231            comment.push(ch as char);
232        }
233
234        Ok(Token::Comment(comment))
235    }
236
237    /// Read a name object (e.g., /Type)
238    fn read_name(&mut self) -> ParseResult<Token> {
239        self.consume_char()?; // consume '/'
240        let mut name = String::new();
241
242        while let Some(ch) = self.peek_char()? {
243            if ch.is_ascii_whitespace()
244                || matches!(ch, b'/' | b'<' | b'>' | b'[' | b']' | b'(' | b')' | b'%')
245            {
246                break;
247            }
248            self.consume_char()?;
249
250            // Handle hex codes in names (e.g., /A#20B means /A B)
251            if ch == b'#' {
252                let hex1 = self
253                    .consume_char()?
254                    .ok_or_else(|| ParseError::SyntaxError {
255                        position: self.position,
256                        message: "Incomplete hex code in name".to_string(),
257                    })?;
258                let hex2 = self
259                    .consume_char()?
260                    .ok_or_else(|| ParseError::SyntaxError {
261                        position: self.position,
262                        message: "Incomplete hex code in name".to_string(),
263                    })?;
264
265                let value = u8::from_str_radix(&format!("{}{}", hex1 as char, hex2 as char), 16)
266                    .map_err(|_| ParseError::SyntaxError {
267                        position: self.position,
268                        message: "Invalid hex code in name".to_string(),
269                    })?;
270
271                name.push(value as char);
272            } else {
273                name.push(ch as char);
274            }
275        }
276
277        Ok(Token::Name(name))
278    }
279
280    /// Read a literal string (parentheses)
281    fn read_literal_string(&mut self) -> ParseResult<Token> {
282        self.consume_char()?; // consume '('
283        let mut string = Vec::new();
284        let mut paren_depth = 1;
285        let mut escape = false;
286
287        while paren_depth > 0 {
288            let ch = match self.consume_char()? {
289                Some(c) => c,
290                None => {
291                    if self.options.lenient_syntax {
292                        // In lenient mode, return what we have so far
293                        if self.options.collect_warnings {
294                            self.warnings.push(ParseWarning::SyntaxErrorRecovered {
295                                position: self.position,
296                                expected: "closing parenthesis".to_string(),
297                                found: "EOF".to_string(),
298                                recovery_action: "returned partial string content".to_string(),
299                            });
300                        }
301                        break;
302                    } else {
303                        return Err(ParseError::SyntaxError {
304                            position: self.position,
305                            message: "Unterminated string".to_string(),
306                        });
307                    }
308                }
309            };
310
311            if escape {
312                let escaped = match ch {
313                    b'n' => b'\n',
314                    b'r' => b'\r',
315                    b't' => b'\t',
316                    b'b' => b'\x08',
317                    b'f' => b'\x0C',
318                    b'(' => b'(',
319                    b')' => b')',
320                    b'\\' => b'\\',
321                    b'0'..=b'7' => {
322                        // Octal escape sequence.
323                        // Use u16 to avoid overflow panic on malformed octal (e.g. \777).
324                        // Per ISO 32000-1:2008 §7.3.4.2: "high-order overflow shall be ignored".
325                        let mut value = u16::from(ch - b'0');
326                        for _ in 0..2 {
327                            if let Some(next) = self.peek_char()? {
328                                if matches!(next, b'0'..=b'7') {
329                                    self.consume_char()?;
330                                    value = value * 8 + u16::from(next - b'0');
331                                } else {
332                                    break;
333                                }
334                            }
335                        }
336                        value as u8
337                    }
338                    _ => ch, // Unknown escape, use literal
339                };
340                string.push(escaped);
341                escape = false;
342            } else {
343                match ch {
344                    b'\\' => escape = true,
345                    b'(' => {
346                        string.push(ch);
347                        paren_depth += 1;
348                    }
349                    b')' => {
350                        paren_depth -= 1;
351                        if paren_depth > 0 {
352                            string.push(ch);
353                        }
354                    }
355                    _ => string.push(ch),
356                }
357            }
358        }
359
360        // Apply character encoding recovery if enabled
361        let processed_string = if self.options.lenient_encoding {
362            self.process_string_with_encoding_recovery(&string)?
363        } else {
364            string
365        };
366
367        Ok(Token::String(processed_string))
368    }
369
370    /// Read angle bracket tokens (hex strings or dict markers)
371    fn read_angle_bracket(&mut self) -> ParseResult<Token> {
372        self.consume_char()?; // consume '<'
373
374        if self.peek_char()? == Some(b'<') {
375            self.consume_char()?;
376            Ok(Token::DictStart)
377        } else {
378            // Hex string
379            let mut hex_chars = String::new();
380            let mut found_end = false;
381
382            while let Some(ch) = self.peek_char()? {
383                if ch == b'>' {
384                    self.consume_char()?;
385                    found_end = true;
386                    break;
387                }
388                self.consume_char()?;
389                if ch.is_ascii_hexdigit() {
390                    hex_chars.push(ch as char);
391                } else if !ch.is_ascii_whitespace() {
392                    if self.options.lenient_syntax {
393                        // In lenient mode, skip invalid characters
394                        if self.options.collect_warnings {
395                            self.warnings.push(ParseWarning::SyntaxErrorRecovered {
396                                position: self.position,
397                                expected: "hex digit".to_string(),
398                                found: format!("'{}'", ch as char),
399                                recovery_action: "skipped invalid character".to_string(),
400                            });
401                        }
402                    } else {
403                        return Err(ParseError::SyntaxError {
404                            position: self.position,
405                            message: "Invalid character in hex string".to_string(),
406                        });
407                    }
408                }
409            }
410
411            if !found_end {
412                if self.options.lenient_syntax {
413                    // In lenient mode, return what we have so far
414                    if self.options.collect_warnings {
415                        self.warnings.push(ParseWarning::SyntaxErrorRecovered {
416                            position: self.position,
417                            expected: ">".to_string(),
418                            found: "EOF".to_string(),
419                            recovery_action: "returned partial hex string".to_string(),
420                        });
421                    }
422                } else {
423                    return Err(ParseError::SyntaxError {
424                        position: self.position,
425                        message: "Unterminated hex string".to_string(),
426                    });
427                }
428            }
429
430            // Pad with 0 if odd number of digits
431            if hex_chars.len() % 2 != 0 {
432                hex_chars.push('0');
433            }
434
435            // Convert hex to bytes
436            let mut bytes = Vec::new();
437            for chunk in hex_chars.as_bytes().chunks(2) {
438                let hex_str = std::str::from_utf8(chunk).map_err(|_| ParseError::SyntaxError {
439                    position: self.position,
440                    message: "Invalid UTF-8 in hex string".to_string(),
441                })?;
442                let byte =
443                    u8::from_str_radix(hex_str, 16).map_err(|_| ParseError::SyntaxError {
444                        position: self.position,
445                        message: "Invalid hex string".to_string(),
446                    })?;
447                bytes.push(byte);
448            }
449
450            Ok(Token::String(bytes))
451        }
452    }
453
454    /// Read boolean (true/false)
455    fn read_boolean(&mut self) -> ParseResult<Token> {
456        let word = self.read_word()?;
457        match word.as_str() {
458            "true" => Ok(Token::Boolean(true)),
459            "false" => Ok(Token::Boolean(false)),
460            _ => {
461                // Not a boolean, might be a keyword
462                self.process_keyword(word)
463            }
464        }
465    }
466
467    /// Read null
468    fn read_null(&mut self) -> ParseResult<Token> {
469        let word = self.read_word()?;
470        if word == "null" {
471            Ok(Token::Null)
472        } else {
473            // Not null, might be a keyword
474            self.process_keyword(word)
475        }
476    }
477
478    /// Read a number (integer or real)
479    fn read_number(&mut self) -> ParseResult<Token> {
480        let mut number_str = String::new();
481        let mut has_dot = false;
482
483        // Handle sign - consume it first
484        if let Some(ch) = self.peek_char()? {
485            if ch == b'+' || ch == b'-' {
486                self.consume_char()?;
487                number_str.push(ch as char);
488
489                // After sign, we must have at least one digit
490                if let Some(next) = self.peek_char()? {
491                    if !next.is_ascii_digit() && next != b'.' {
492                        return Err(ParseError::SyntaxError {
493                            position: self.position,
494                            message: "Expected digit after sign".to_string(),
495                        });
496                    }
497                }
498            }
499        }
500
501        // Read digits and decimal point
502        while let Some(ch) = self.peek_char()? {
503            match ch {
504                b'0'..=b'9' => {
505                    self.consume_char()?;
506                    number_str.push(ch as char);
507                }
508                b'.' if !has_dot => {
509                    self.consume_char()?;
510                    number_str.push(ch as char);
511                    has_dot = true;
512                }
513                _ => break,
514            }
515        }
516
517        // Handle scientific notation (e/E)
518        if let Some(ch) = self.peek_char()? {
519            if ch == b'e' || ch == b'E' {
520                self.consume_char()?;
521                number_str.push(ch as char);
522
523                // Check for optional sign after e/E
524                if let Some(sign_ch) = self.peek_char()? {
525                    if sign_ch == b'+' || sign_ch == b'-' {
526                        self.consume_char()?;
527                        number_str.push(sign_ch as char);
528                    }
529                }
530
531                // Read exponent digits
532                while let Some(digit_ch) = self.peek_char()? {
533                    if digit_ch.is_ascii_digit() {
534                        self.consume_char()?;
535                        number_str.push(digit_ch as char);
536                    } else {
537                        break;
538                    }
539                }
540
541                // Scientific notation always results in a real number
542                has_dot = true;
543            }
544        }
545
546        // Don't try to parse references here - let the parser handle it
547        // References are just "num num R" and can be handled at a higher level
548
549        // Parse as number
550        if has_dot {
551            let value = number_str
552                .parse::<f64>()
553                .map_err(|_| ParseError::SyntaxError {
554                    position: self.position,
555                    message: format!("Invalid real number: '{number_str}'"),
556                })?;
557            Ok(Token::Real(value))
558        } else {
559            let value = number_str
560                .parse::<i64>()
561                .map_err(|_| ParseError::SyntaxError {
562                    position: self.position,
563                    message: format!("Invalid integer: '{number_str}'"),
564                })?;
565            Ok(Token::Integer(value))
566        }
567    }
568
569    /// Read a keyword
570    fn read_keyword(&mut self) -> ParseResult<Token> {
571        let word = self.read_word()?;
572        self.process_keyword(word)
573    }
574
575    /// Process a word as a keyword
576    fn process_keyword(&self, word: String) -> ParseResult<Token> {
577        match word.as_str() {
578            "stream" => Ok(Token::Stream),
579            "endstream" => Ok(Token::EndStream),
580            "obj" => Ok(Token::Obj),
581            "endobj" => Ok(Token::EndObj),
582            "startxref" => Ok(Token::StartXRef),
583            _ => Err(ParseError::SyntaxError {
584                position: self.position,
585                message: format!("Unknown keyword: {word}"),
586            }),
587        }
588    }
589
590    /// Read a word (sequence of non-delimiter characters)
591    fn read_word(&mut self) -> ParseResult<String> {
592        let mut word = String::new();
593
594        while let Some(ch) = self.peek_char()? {
595            if ch.is_ascii_whitespace()
596                || matches!(ch, b'/' | b'<' | b'>' | b'[' | b']' | b'(' | b')' | b'%')
597            {
598                break;
599            }
600            self.consume_char()?;
601            word.push(ch as char);
602        }
603
604        Ok(word)
605    }
606
607    /// Read a sequence of digits
608    #[allow(dead_code)]
609    fn read_digits(&mut self) -> ParseResult<String> {
610        let mut digits = String::new();
611
612        while let Some(ch) = self.peek_char()? {
613            if ch.is_ascii_digit() {
614                self.consume_char()?;
615                digits.push(ch as char);
616            } else {
617                break;
618            }
619        }
620
621        Ok(digits)
622    }
623
624    /// Read a newline sequence (CR, LF, or CRLF)
625    pub fn read_newline(&mut self) -> ParseResult<()> {
626        match self.peek_char()? {
627            Some(b'\r') => {
628                self.consume_char()?;
629                // Check for CRLF
630                if self.peek_char()? == Some(b'\n') {
631                    self.consume_char()?;
632                }
633                Ok(())
634            }
635            Some(b'\n') => {
636                self.consume_char()?;
637                Ok(())
638            }
639            _ => Err(ParseError::SyntaxError {
640                position: self.position,
641                message: "Expected newline".to_string(),
642            }),
643        }
644    }
645
646    /// Read exactly n bytes
647    /// Peek at the next byte without consuming it
648    pub fn peek_byte(&mut self) -> ParseResult<u8> {
649        match self.peek_char()? {
650            Some(b) => Ok(b),
651            None => Err(ParseError::UnexpectedToken {
652                expected: "byte".to_string(),
653                found: "EOF".to_string(),
654            }),
655        }
656    }
657
658    /// Read a single byte
659    pub fn read_byte(&mut self) -> ParseResult<u8> {
660        match self.consume_char()? {
661            Some(b) => Ok(b),
662            None => Err(ParseError::UnexpectedToken {
663                expected: "byte".to_string(),
664                found: "EOF".to_string(),
665            }),
666        }
667    }
668
669    /// Seek to a specific position
670    pub fn seek(&mut self, pos: u64) -> ParseResult<()>
671    where
672        R: Seek,
673    {
674        self.reader.seek(SeekFrom::Start(pos))?;
675        self.position = pos as usize;
676        Ok(())
677    }
678
679    pub fn read_bytes(&mut self, n: usize) -> ParseResult<Vec<u8>> {
680        let mut bytes = Vec::with_capacity(n);
681
682        // First consume any peeked byte to avoid duplication
683        if self.peek_buffer.is_some() && n > 0 {
684            if let Some(byte) = self.consume_char()? {
685                bytes.push(byte);
686            }
687        }
688
689        // Read remaining bytes directly
690        let remaining = n - bytes.len();
691        if remaining > 0 {
692            let mut rest = vec![0u8; remaining];
693            self.reader.read_exact(&mut rest)?;
694            self.position += remaining;
695            bytes.extend_from_slice(&rest);
696        }
697
698        Ok(bytes)
699    }
700
701    /// Read until a specific byte sequence is found
702    pub fn read_until_sequence(&mut self, sequence: &[u8]) -> ParseResult<Vec<u8>> {
703        let mut result = Vec::new();
704        let mut match_pos = 0;
705
706        while let Some(ch) = self.consume_char()? {
707            result.push(ch);
708
709            if ch == sequence[match_pos] {
710                match_pos += 1;
711                if match_pos == sequence.len() {
712                    // Found the sequence, remove it from result
713                    result.truncate(result.len() - sequence.len());
714                    break;
715                }
716            } else if ch == sequence[0] {
717                match_pos = 1;
718            } else {
719                match_pos = 0;
720            }
721        }
722
723        if match_pos < sequence.len() {
724            return Err(ParseError::SyntaxError {
725                position: self.position,
726                message: format!("Sequence {sequence:?} not found"),
727            });
728        }
729
730        Ok(result)
731    }
732
733    /// Get current position
734    pub fn position(&self) -> usize {
735        self.position
736    }
737
738    /// Push back a token to be returned by the next call to next_token
739    pub fn push_token(&mut self, token: Token) {
740        self.token_buffer.push(token);
741    }
742
743    /// Expect a specific keyword token
744    pub fn expect_keyword(&mut self, keyword: &str) -> ParseResult<()> {
745        let token = self.next_token()?;
746        match (keyword, &token) {
747            ("endstream", Token::EndStream) => Ok(()),
748            ("stream", Token::Stream) => Ok(()),
749            ("endobj", Token::EndObj) => Ok(()),
750            ("obj", Token::Obj) => Ok(()),
751            ("startxref", Token::StartXRef) => Ok(()),
752            _ => Err(ParseError::UnexpectedToken {
753                expected: format!("keyword '{keyword}'"),
754                found: format!("{token:?}"),
755            }),
756        }
757    }
758
759    /// Find a keyword ahead in the stream without consuming bytes
760    /// Returns the number of bytes until the keyword is found
761    pub fn find_keyword_ahead(
762        &mut self,
763        keyword: &str,
764        max_bytes: usize,
765    ) -> ParseResult<Option<usize>>
766    where
767        R: Seek,
768    {
769        use std::io::{Read, Seek, SeekFrom};
770
771        // Save current position
772        let current_pos = self.reader.stream_position()?;
773        let start_buffer_state = self.peek_buffer;
774
775        let keyword_bytes = keyword.as_bytes();
776        let mut bytes_read = 0;
777        let mut match_buffer = Vec::new();
778
779        // Search for the keyword
780        while bytes_read < max_bytes {
781            let mut byte = [0u8; 1];
782            match self.reader.read_exact(&mut byte) {
783                Ok(_) => {
784                    bytes_read += 1;
785                    match_buffer.push(byte[0]);
786
787                    // Keep only the last keyword.len() bytes in match_buffer
788                    if match_buffer.len() > keyword_bytes.len() {
789                        match_buffer.remove(0);
790                    }
791
792                    // Check if we found the keyword
793                    if match_buffer.len() == keyword_bytes.len() && match_buffer == keyword_bytes {
794                        // Restore position
795                        self.reader.seek(SeekFrom::Start(current_pos))?;
796                        self.peek_buffer = start_buffer_state;
797                        return Ok(Some(bytes_read - keyword_bytes.len()));
798                    }
799                }
800                Err(_) => break, // EOF or error
801            }
802        }
803
804        // Restore position
805        self.reader.seek(SeekFrom::Start(current_pos))?;
806        self.peek_buffer = start_buffer_state;
807        Ok(None)
808    }
809
810    /// Peek ahead n bytes without consuming them
811    pub fn peek_ahead(&mut self, n: usize) -> ParseResult<Vec<u8>>
812    where
813        R: Seek,
814    {
815        use std::io::{Read, Seek, SeekFrom};
816
817        // Save current position
818        let current_pos = self.reader.stream_position()?;
819        let start_buffer_state = self.peek_buffer;
820
821        // Read n bytes
822        let mut bytes = vec![0u8; n];
823        let bytes_read = self.reader.read(&mut bytes)?;
824        bytes.truncate(bytes_read);
825
826        // Restore position
827        self.reader.seek(SeekFrom::Start(current_pos))?;
828        self.peek_buffer = start_buffer_state;
829
830        Ok(bytes)
831    }
832
833    /// Save the current position for later restoration
834    pub fn save_position(&mut self) -> ParseResult<(u64, Option<u8>)>
835    where
836        R: Seek,
837    {
838        use std::io::Seek;
839        let pos = self.reader.stream_position()?;
840        Ok((pos, self.peek_buffer))
841    }
842
843    /// Restore a previously saved position
844    pub fn restore_position(&mut self, saved: (u64, Option<u8>)) -> ParseResult<()>
845    where
846        R: Seek,
847    {
848        use std::io::{Seek, SeekFrom};
849        self.reader.seek(SeekFrom::Start(saved.0))?;
850        self.peek_buffer = saved.1;
851        self.position = saved.0 as usize;
852        Ok(())
853    }
854
855    /// Peek the next token without consuming it
856    pub fn peek_token(&mut self) -> ParseResult<Token>
857    where
858        R: Seek,
859    {
860        let saved_pos = self.save_position()?;
861        let token = self.next_token()?;
862        self.restore_position(saved_pos)?;
863        Ok(token)
864    }
865
866    /// Process string bytes with enhanced character encoding recovery
867    fn process_string_with_encoding_recovery(
868        &mut self,
869        string_bytes: &[u8],
870    ) -> ParseResult<Vec<u8>> {
871        use super::encoding::{CharacterDecoder, EncodingOptions, EncodingType, EnhancedDecoder};
872
873        // First check for common problematic bytes that need special handling
874        let has_problematic_chars = string_bytes.iter().any(|&b| {
875            // Control characters and Latin-1 supplement range that often cause issues
876            (0x80..=0x9F).contains(&b)
877                || b == 0x07
878                || (b <= 0x1F && b != 0x09 && b != 0x0A && b != 0x0D)
879        });
880
881        let decoder = EnhancedDecoder::new();
882
883        // Use more aggressive encoding options if problematic characters detected
884        let encoding_options = if has_problematic_chars {
885            EncodingOptions {
886                lenient_mode: true, // Always use lenient mode for problematic chars
887                preferred_encoding: Some(EncodingType::Windows1252), // Try Windows-1252 first for control chars
888                max_replacements: std::cmp::max(100, string_bytes.len() / 10), // More generous replacement limit
889                log_issues: self.options.collect_warnings,
890            }
891        } else {
892            EncodingOptions {
893                lenient_mode: self.options.lenient_encoding,
894                preferred_encoding: self.options.preferred_encoding,
895                max_replacements: 50,
896                log_issues: self.options.collect_warnings,
897            }
898        };
899
900        match decoder.decode(string_bytes, &encoding_options) {
901            Ok(result) => {
902                // Log warning if replacements were made or problematic chars detected
903                if (result.replacement_count > 0 || has_problematic_chars)
904                    && self.options.collect_warnings
905                {
906                    self.warnings.push(ParseWarning::InvalidEncoding {
907                        position: self.position,
908                        recovered_text: if result.text.len() > 50 {
909                            // Ultra-safe character boundary truncation
910                            let truncate_at = result
911                                .text
912                                .char_indices()
913                                .map(|(i, _)| i)
914                                .nth(47)
915                                .unwrap_or_else(|| {
916                                    // Fallback: find last valid boundary within first 47 bytes
917                                    let limit = result.text.len().min(47);
918                                    let mut pos = limit;
919                                    while pos > 0 && !result.text.is_char_boundary(pos) {
920                                        pos -= 1;
921                                    }
922                                    pos
923                                });
924
925                            // Double-check the boundary before slicing
926                            let safe_text = if truncate_at <= result.text.len()
927                                && result.text.is_char_boundary(truncate_at)
928                            {
929                                result.text[..truncate_at].to_string()
930                            } else {
931                                // Emergency fallback: use chars().take() for absolute safety
932                                result.text.chars().take(47).collect::<String>()
933                            };
934
935                            format!(
936                                "{}... (truncated, {} chars total)",
937                                safe_text,
938                                result.text.chars().count()
939                            )
940                        } else {
941                            result.text.clone()
942                        },
943                        encoding_used: result.detected_encoding,
944                        replacement_count: result.replacement_count,
945                    });
946                }
947
948                // Convert back to bytes
949                Ok(result.text.into_bytes())
950            }
951            Err(encoding_error) => {
952                if self.options.lenient_encoding {
953                    // Enhanced fallback strategy
954                    let fallback_result = self.apply_fallback_encoding_strategy(string_bytes);
955
956                    if self.options.collect_warnings {
957                        self.warnings.push(ParseWarning::InvalidEncoding {
958                            position: self.position,
959                            recovered_text: format!(
960                                "Fallback strategy applied: {} -> {} chars",
961                                string_bytes.len(),
962                                fallback_result.len()
963                            ),
964                            encoding_used: None,
965                            replacement_count: string_bytes.len(),
966                        });
967                    }
968                    Ok(fallback_result)
969                } else {
970                    Err(ParseError::CharacterEncodingError {
971                        position: self.position,
972                        message: format!(
973                            "Failed to decode string with any supported encoding: {encoding_error}"
974                        ),
975                    })
976                }
977            }
978        }
979    }
980
981    /// Apply fallback encoding strategy for severely corrupted strings
982    fn apply_fallback_encoding_strategy(&self, string_bytes: &[u8]) -> Vec<u8> {
983        let mut result = Vec::with_capacity(string_bytes.len());
984
985        for &byte in string_bytes {
986            match byte {
987                // Replace common problematic control characters with safe alternatives
988                0x00..=0x08 | 0x0B | 0x0C | 0x0E..=0x1F => {
989                    result.push(b' '); // Replace control chars with space
990                }
991                0x80..=0x9F => {
992                    // Windows-1252 control character range - try to map to reasonable alternatives
993                    let replacement = match byte {
994                        0x80 => b'E',  // Euro sign -> E
995                        0x81 => b' ',  // Undefined -> space
996                        0x82 => b',',  // Single low-9 quotation mark -> comma
997                        0x83 => b'f',  // Latin small letter f with hook -> f
998                        0x84 => b'"',  // Double low-9 quotation mark -> quote
999                        0x85 => b'.',  // Horizontal ellipsis -> period
1000                        0x86 => b'+',  // Dagger -> plus
1001                        0x87 => b'+',  // Double dagger -> plus
1002                        0x88 => b'^',  // Modifier letter circumflex accent -> caret
1003                        0x89 => b'%',  // Per mille sign -> percent
1004                        0x8A => b'S',  // Latin capital letter S with caron -> S
1005                        0x8B => b'<',  // Single left-pointing angle quotation mark
1006                        0x8C => b'O',  // Latin capital ligature OE -> O
1007                        0x8D => b' ',  // Undefined -> space
1008                        0x8E => b'Z',  // Latin capital letter Z with caron -> Z
1009                        0x8F => b' ',  // Undefined -> space
1010                        0x90 => b' ',  // Undefined -> space
1011                        0x91 => b'\'', // Left single quotation mark
1012                        0x92 => b'\'', // Right single quotation mark
1013                        0x93 => b'"',  // Left double quotation mark
1014                        0x94 => b'"',  // Right double quotation mark
1015                        0x95 => b'*',  // Bullet -> asterisk
1016                        0x96 => b'-',  // En dash -> hyphen
1017                        0x97 => b'-',  // Em dash -> hyphen
1018                        0x98 => b'~',  // Small tilde
1019                        0x99 => b'T',  // Trade mark sign -> T
1020                        0x9A => b's',  // Latin small letter s with caron -> s
1021                        0x9B => b'>',  // Single right-pointing angle quotation mark
1022                        0x9C => b'o',  // Latin small ligature oe -> o
1023                        0x9D => b' ',  // Undefined -> space
1024                        0x9E => b'z',  // Latin small letter z with caron -> z
1025                        0x9F => b'Y',  // Latin capital letter Y with diaeresis -> Y
1026                        _ => b'?',     // Fallback
1027                    };
1028                    result.push(replacement);
1029                }
1030                _ => {
1031                    result.push(byte); // Keep valid bytes as-is
1032                }
1033            }
1034        }
1035
1036        result
1037    }
1038
1039    /// Check if a character is likely a problematic encoding character
1040    fn is_problematic_encoding_char(&self, ch: u8) -> bool {
1041        // Control characters and Latin-1 supplement range that often indicate encoding issues
1042        (0x80..=0x9F).contains(&ch) ||
1043        ch == 0x07 || // Bell character
1044        (ch <= 0x1F && ch != 0x09 && ch != 0x0A && ch != 0x0D) || // Control chars except tab, LF, CR
1045        // In lenient mode, also handle extended Latin-1 characters that may appear in corrupted streams
1046        (self.options.lenient_syntax && ch >= 0xA0) // Extended Latin-1 range (u8 max is 0xFF)
1047    }
1048
1049    /// Handle problematic encoding characters in the main token stream
1050    fn handle_encoding_char_in_token_stream(&mut self, ch: u8) -> ParseResult<Token> {
1051        if self.options.lenient_encoding {
1052            // Consume the problematic character and continue
1053            self.consume_char()?;
1054
1055            // Log warning about the character recovery
1056            if self.options.collect_warnings {
1057                let replacement_char = match ch {
1058                    0x07 => "bell",
1059                    0x00..=0x1F => "control",
1060                    0x80..=0x9F => "latin1-supplement",
1061                    _ => "unknown",
1062                };
1063
1064                self.warnings.push(ParseWarning::InvalidEncoding {
1065                    position: self.position,
1066                    recovered_text: format!(
1067                        "Skipped problematic {replacement_char} character (0x{ch:02X})"
1068                    ),
1069                    encoding_used: None,
1070                    replacement_count: 1,
1071                });
1072            }
1073
1074            // Skip this character and try to get the next token
1075            self.skip_whitespace()?;
1076            if let Ok(Some(_)) = self.peek_char() {
1077                self.next_token() // Recursively try next token
1078            } else {
1079                Err(ParseError::SyntaxError {
1080                    position: self.position,
1081                    message: "Unexpected end of file after problematic character".to_string(),
1082                })
1083            }
1084        } else {
1085            // In strict mode, generate a more descriptive error
1086            let char_description = match ch {
1087                0x07 => "Bell character (\\u{07})".to_string(),
1088                0x00..=0x1F => format!("Control character (\\u{{{ch:02X}}})"),
1089                0x80..=0x9F => format!("Latin-1 supplement character (\\u{{{ch:02X}}})"),
1090                _ => format!("Problematic character (\\u{{{ch:02X}}})"),
1091            };
1092
1093            Err(ParseError::CharacterEncodingError {
1094                position: self.position,
1095                message: format!(
1096                    "Unexpected character: {char_description} - Consider using lenient parsing mode"
1097                ),
1098            })
1099        }
1100    }
1101}
1102
1103#[cfg(test)]
1104mod tests {
1105    use super::*;
1106    use std::io::Cursor;
1107
1108    #[test]
1109    fn test_lexer_basic_tokens() {
1110        // Test positive and negative numbers
1111        let input = b"123 -456 3.14 true false null /Name";
1112        let mut lexer = Lexer::new(Cursor::new(input));
1113
1114        assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1115        assert_eq!(lexer.next_token().unwrap(), Token::Integer(-456));
1116        assert_eq!(lexer.next_token().unwrap(), Token::Real(3.14));
1117        assert_eq!(lexer.next_token().unwrap(), Token::Boolean(true));
1118        assert_eq!(lexer.next_token().unwrap(), Token::Boolean(false));
1119        assert_eq!(lexer.next_token().unwrap(), Token::Null);
1120        assert_eq!(lexer.next_token().unwrap(), Token::Name("Name".to_string()));
1121        assert_eq!(lexer.next_token().unwrap(), Token::Eof);
1122    }
1123
1124    #[test]
1125    fn test_lexer_negative_numbers() {
1126        // Test negative numbers without space
1127        let input = b"-123 -45.67";
1128        let mut lexer = Lexer::new(Cursor::new(input));
1129
1130        assert_eq!(lexer.next_token().unwrap(), Token::Integer(-123));
1131        assert_eq!(lexer.next_token().unwrap(), Token::Real(-45.67));
1132    }
1133
1134    #[test]
1135    fn test_lexer_strings() {
1136        let input = b"(Hello World) <48656C6C6F>";
1137        let mut lexer = Lexer::new(Cursor::new(input));
1138
1139        assert_eq!(
1140            lexer.next_token().unwrap(),
1141            Token::String(b"Hello World".to_vec())
1142        );
1143        assert_eq!(
1144            lexer.next_token().unwrap(),
1145            Token::String(b"Hello".to_vec())
1146        );
1147    }
1148
1149    #[test]
1150    fn test_lexer_dictionaries() {
1151        let input = b"<< /Type /Page >>";
1152        let mut lexer = Lexer::new(Cursor::new(input));
1153
1154        assert_eq!(lexer.next_token().unwrap(), Token::DictStart);
1155        assert_eq!(lexer.next_token().unwrap(), Token::Name("Type".to_string()));
1156        assert_eq!(lexer.next_token().unwrap(), Token::Name("Page".to_string()));
1157        assert_eq!(lexer.next_token().unwrap(), Token::DictEnd);
1158    }
1159
1160    #[test]
1161    fn test_lexer_arrays() {
1162        let input = b"[1 2 3]";
1163        let mut lexer = Lexer::new(Cursor::new(input));
1164
1165        assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1166        assert_eq!(lexer.next_token().unwrap(), Token::Integer(1));
1167        assert_eq!(lexer.next_token().unwrap(), Token::Integer(2));
1168        assert_eq!(lexer.next_token().unwrap(), Token::Integer(3));
1169        assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1170    }
1171
1172    #[test]
1173    fn test_lexer_references() {
1174        let input = b"1 0 R 25 1 R";
1175        let mut lexer = Lexer::new(Cursor::new(input));
1176
1177        // Now references are parsed as separate tokens
1178        assert_eq!(lexer.next_token().unwrap(), Token::Integer(1));
1179        assert_eq!(lexer.next_token().unwrap(), Token::Integer(0));
1180        // 'R' should be parsed as a keyword or name
1181        match lexer.next_token().unwrap() {
1182            Token::Name(s) if s == "R" => {} // Could be a name
1183            other => panic!("Expected R token, got {other:?}"),
1184        }
1185
1186        assert_eq!(lexer.next_token().unwrap(), Token::Integer(25));
1187        assert_eq!(lexer.next_token().unwrap(), Token::Integer(1));
1188        match lexer.next_token().unwrap() {
1189            Token::Name(s) if s == "R" => {} // Could be a name
1190            other => panic!("Expected R token, got {other:?}"),
1191        }
1192    }
1193
1194    #[test]
1195    fn test_lexer_comments() {
1196        let input = b"%PDF-1.7\n123";
1197        let mut lexer = Lexer::new(Cursor::new(input));
1198
1199        assert_eq!(
1200            lexer.next_token().unwrap(),
1201            Token::Comment("PDF-1.7".to_string())
1202        );
1203        assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1204    }
1205
1206    // Comprehensive tests for Lexer
1207    mod comprehensive_tests {
1208        use super::*;
1209        use std::io::Cursor;
1210
1211        #[test]
1212        fn test_token_debug_trait() {
1213            let token = Token::Integer(42);
1214            let debug_str = format!("{token:?}");
1215            assert!(debug_str.contains("Integer"));
1216            assert!(debug_str.contains("42"));
1217        }
1218
1219        #[test]
1220        fn test_token_clone() {
1221            let token = Token::String(b"hello".to_vec());
1222            let cloned = token.clone();
1223            assert_eq!(token, cloned);
1224        }
1225
1226        #[test]
1227        fn test_token_equality() {
1228            assert_eq!(Token::Integer(42), Token::Integer(42));
1229            assert_ne!(Token::Integer(42), Token::Integer(43));
1230            assert_eq!(Token::Boolean(true), Token::Boolean(true));
1231            assert_ne!(Token::Boolean(true), Token::Boolean(false));
1232            assert_eq!(Token::Null, Token::Null);
1233            assert_ne!(Token::Null, Token::Integer(0));
1234        }
1235
1236        #[test]
1237        fn test_lexer_empty_input() {
1238            let input = b"";
1239            let mut lexer = Lexer::new(Cursor::new(input));
1240            assert_eq!(lexer.next_token().unwrap(), Token::Eof);
1241        }
1242
1243        #[test]
1244        fn test_lexer_whitespace_only() {
1245            let input = b"   \t\n\r  ";
1246            let mut lexer = Lexer::new(Cursor::new(input));
1247            assert_eq!(lexer.next_token().unwrap(), Token::Eof);
1248        }
1249
1250        #[test]
1251        fn test_lexer_integer_edge_cases() {
1252            let input = b"0 +123 -0 9876543210";
1253            let mut lexer = Lexer::new(Cursor::new(input));
1254
1255            assert_eq!(lexer.next_token().unwrap(), Token::Integer(0));
1256            assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1257            assert_eq!(lexer.next_token().unwrap(), Token::Integer(0));
1258            assert_eq!(lexer.next_token().unwrap(), Token::Integer(9876543210));
1259        }
1260
1261        #[test]
1262        fn test_lexer_real_edge_cases() {
1263            let input = b"0.0 +3.14 -2.71828 .5 5. 123.456789";
1264            let mut lexer = Lexer::new(Cursor::new(input));
1265
1266            assert_eq!(lexer.next_token().unwrap(), Token::Real(0.0));
1267            assert_eq!(lexer.next_token().unwrap(), Token::Real(3.14));
1268            assert_eq!(lexer.next_token().unwrap(), Token::Real(-2.71828));
1269            assert_eq!(lexer.next_token().unwrap(), Token::Real(0.5));
1270            assert_eq!(lexer.next_token().unwrap(), Token::Real(5.0));
1271            assert_eq!(lexer.next_token().unwrap(), Token::Real(123.456789));
1272        }
1273
1274        #[test]
1275        fn test_lexer_scientific_notation() {
1276            let input = b"1.23e10 -4.56E-5 1e0 2E+3";
1277            let mut lexer = Lexer::new(Cursor::new(input));
1278
1279            assert_eq!(lexer.next_token().unwrap(), Token::Real(1.23e10));
1280            assert_eq!(lexer.next_token().unwrap(), Token::Real(-4.56e-5));
1281            assert_eq!(lexer.next_token().unwrap(), Token::Real(1e0));
1282            assert_eq!(lexer.next_token().unwrap(), Token::Real(2e3));
1283        }
1284
1285        #[test]
1286        fn test_lexer_string_literal_escapes() {
1287            let input = b"(Hello\\nWorld) (Tab\\tChar) (Quote\\\"Mark) (Backslash\\\\)";
1288            let mut lexer = Lexer::new(Cursor::new(input));
1289
1290            assert_eq!(
1291                lexer.next_token().unwrap(),
1292                Token::String(b"Hello\nWorld".to_vec())
1293            );
1294            assert_eq!(
1295                lexer.next_token().unwrap(),
1296                Token::String(b"Tab\tChar".to_vec())
1297            );
1298            assert_eq!(
1299                lexer.next_token().unwrap(),
1300                Token::String(b"Quote\"Mark".to_vec())
1301            );
1302            assert_eq!(
1303                lexer.next_token().unwrap(),
1304                Token::String(b"Backslash\\".to_vec())
1305            );
1306        }
1307
1308        #[test]
1309        fn test_lexer_string_literal_nested_parens() {
1310            let input = b"(Nested (parentheses) work)";
1311            let mut lexer = Lexer::new(Cursor::new(input));
1312
1313            assert_eq!(
1314                lexer.next_token().unwrap(),
1315                Token::String(b"Nested (parentheses) work".to_vec())
1316            );
1317        }
1318
1319        #[test]
1320        fn test_lexer_string_literal_empty() {
1321            let input = b"()";
1322            let mut lexer = Lexer::new(Cursor::new(input));
1323
1324            assert_eq!(lexer.next_token().unwrap(), Token::String(b"".to_vec()));
1325        }
1326
1327        #[test]
1328        fn test_lexer_hexadecimal_strings() {
1329            let input = b"<48656C6C6F> <20576F726C64> <>";
1330            let mut lexer = Lexer::new(Cursor::new(input));
1331
1332            assert_eq!(
1333                lexer.next_token().unwrap(),
1334                Token::String(b"Hello".to_vec())
1335            );
1336            assert_eq!(
1337                lexer.next_token().unwrap(),
1338                Token::String(b" World".to_vec())
1339            );
1340            assert_eq!(lexer.next_token().unwrap(), Token::String(b"".to_vec()));
1341        }
1342
1343        #[test]
1344        fn test_lexer_hexadecimal_strings_odd_length() {
1345            let input = b"<48656C6C6F2> <1> <ABC>";
1346            let mut lexer = Lexer::new(Cursor::new(input));
1347
1348            // Odd length hex strings should pad with 0
1349            assert_eq!(
1350                lexer.next_token().unwrap(),
1351                Token::String(b"Hello ".to_vec())
1352            );
1353            assert_eq!(lexer.next_token().unwrap(), Token::String(b"\x10".to_vec()));
1354            assert_eq!(
1355                lexer.next_token().unwrap(),
1356                Token::String(b"\xAB\xC0".to_vec())
1357            );
1358        }
1359
1360        #[test]
1361        fn test_lexer_hexadecimal_strings_whitespace() {
1362            let input = b"<48 65 6C 6C 6F>";
1363            let mut lexer = Lexer::new(Cursor::new(input));
1364
1365            assert_eq!(
1366                lexer.next_token().unwrap(),
1367                Token::String(b"Hello".to_vec())
1368            );
1369        }
1370
1371        #[test]
1372        fn test_lexer_names() {
1373            let input = b"/Type /Page /Root /Kids /Count /MediaBox";
1374            let mut lexer = Lexer::new(Cursor::new(input));
1375
1376            assert_eq!(lexer.next_token().unwrap(), Token::Name("Type".to_string()));
1377            assert_eq!(lexer.next_token().unwrap(), Token::Name("Page".to_string()));
1378            assert_eq!(lexer.next_token().unwrap(), Token::Name("Root".to_string()));
1379            assert_eq!(lexer.next_token().unwrap(), Token::Name("Kids".to_string()));
1380            assert_eq!(
1381                lexer.next_token().unwrap(),
1382                Token::Name("Count".to_string())
1383            );
1384            assert_eq!(
1385                lexer.next_token().unwrap(),
1386                Token::Name("MediaBox".to_string())
1387            );
1388        }
1389
1390        #[test]
1391        fn test_lexer_names_with_special_chars() {
1392            let input = b"/Name#20with#20spaces /Name#2Fwith#2Fslashes";
1393            let mut lexer = Lexer::new(Cursor::new(input));
1394
1395            assert_eq!(
1396                lexer.next_token().unwrap(),
1397                Token::Name("Name with spaces".to_string())
1398            );
1399            assert_eq!(
1400                lexer.next_token().unwrap(),
1401                Token::Name("Name/with/slashes".to_string())
1402            );
1403        }
1404
1405        #[test]
1406        fn test_lexer_names_edge_cases() {
1407            let input = b"/ /A /123 /true /false /null";
1408            let mut lexer = Lexer::new(Cursor::new(input));
1409
1410            assert_eq!(lexer.next_token().unwrap(), Token::Name("".to_string()));
1411            assert_eq!(lexer.next_token().unwrap(), Token::Name("A".to_string()));
1412            assert_eq!(lexer.next_token().unwrap(), Token::Name("123".to_string()));
1413            assert_eq!(lexer.next_token().unwrap(), Token::Name("true".to_string()));
1414            assert_eq!(
1415                lexer.next_token().unwrap(),
1416                Token::Name("false".to_string())
1417            );
1418            assert_eq!(lexer.next_token().unwrap(), Token::Name("null".to_string()));
1419        }
1420
1421        #[test]
1422        fn test_lexer_nested_dictionaries() {
1423            let input = b"<< /Type /Page /Resources << /Font << /F1 123 0 R >> >> >>";
1424            let mut lexer = Lexer::new(Cursor::new(input));
1425
1426            assert_eq!(lexer.next_token().unwrap(), Token::DictStart);
1427            assert_eq!(lexer.next_token().unwrap(), Token::Name("Type".to_string()));
1428            assert_eq!(lexer.next_token().unwrap(), Token::Name("Page".to_string()));
1429            assert_eq!(
1430                lexer.next_token().unwrap(),
1431                Token::Name("Resources".to_string())
1432            );
1433            assert_eq!(lexer.next_token().unwrap(), Token::DictStart);
1434            assert_eq!(lexer.next_token().unwrap(), Token::Name("Font".to_string()));
1435            assert_eq!(lexer.next_token().unwrap(), Token::DictStart);
1436            assert_eq!(lexer.next_token().unwrap(), Token::Name("F1".to_string()));
1437            assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1438            assert_eq!(lexer.next_token().unwrap(), Token::Integer(0));
1439            assert_eq!(lexer.next_token().unwrap(), Token::Name("R".to_string()));
1440            assert_eq!(lexer.next_token().unwrap(), Token::DictEnd);
1441            assert_eq!(lexer.next_token().unwrap(), Token::DictEnd);
1442            assert_eq!(lexer.next_token().unwrap(), Token::DictEnd);
1443        }
1444
1445        #[test]
1446        fn test_lexer_nested_arrays() {
1447            let input = b"[[1 2] [3 4] [5 [6 7]]]";
1448            let mut lexer = Lexer::new(Cursor::new(input));
1449
1450            assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1451            assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1452            assert_eq!(lexer.next_token().unwrap(), Token::Integer(1));
1453            assert_eq!(lexer.next_token().unwrap(), Token::Integer(2));
1454            assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1455            assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1456            assert_eq!(lexer.next_token().unwrap(), Token::Integer(3));
1457            assert_eq!(lexer.next_token().unwrap(), Token::Integer(4));
1458            assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1459            assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1460            assert_eq!(lexer.next_token().unwrap(), Token::Integer(5));
1461            assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1462            assert_eq!(lexer.next_token().unwrap(), Token::Integer(6));
1463            assert_eq!(lexer.next_token().unwrap(), Token::Integer(7));
1464            assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1465            assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1466            assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1467        }
1468
1469        #[test]
1470        fn test_lexer_mixed_content() {
1471            let input = b"<< /Type /Page /MediaBox [0 0 612 792] /Resources << /Font << /F1 << /Type /Font /Subtype /Type1 >> >> >> >>";
1472            let mut lexer = Lexer::new(Cursor::new(input));
1473
1474            // Just test that we can parse this without errors
1475            let mut tokens = Vec::new();
1476            loop {
1477                match lexer.next_token().unwrap() {
1478                    Token::Eof => break,
1479                    token => tokens.push(token),
1480                }
1481            }
1482            assert!(tokens.len() > 10);
1483        }
1484
1485        #[test]
1486        fn test_lexer_keywords() {
1487            let input = b"obj endobj stream endstream startxref";
1488            let mut lexer = Lexer::new(Cursor::new(input));
1489
1490            assert_eq!(lexer.next_token().unwrap(), Token::Obj);
1491            assert_eq!(lexer.next_token().unwrap(), Token::EndObj);
1492            assert_eq!(lexer.next_token().unwrap(), Token::Stream);
1493            assert_eq!(lexer.next_token().unwrap(), Token::EndStream);
1494            assert_eq!(lexer.next_token().unwrap(), Token::StartXRef);
1495        }
1496
1497        #[test]
1498        fn test_lexer_multiple_comments() {
1499            let input = b"%First comment\n%Second comment\n123";
1500            let mut lexer = Lexer::new(Cursor::new(input));
1501
1502            assert_eq!(
1503                lexer.next_token().unwrap(),
1504                Token::Comment("First comment".to_string())
1505            );
1506            assert_eq!(
1507                lexer.next_token().unwrap(),
1508                Token::Comment("Second comment".to_string())
1509            );
1510            assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1511        }
1512
1513        #[test]
1514        fn test_lexer_comment_without_newline() {
1515            let input = b"%Comment at end";
1516            let mut lexer = Lexer::new(Cursor::new(input));
1517
1518            assert_eq!(
1519                lexer.next_token().unwrap(),
1520                Token::Comment("Comment at end".to_string())
1521            );
1522            assert_eq!(lexer.next_token().unwrap(), Token::Eof);
1523        }
1524
1525        #[test]
1526        fn test_lexer_special_characters_in_streams() {
1527            let input = b"<< /Length 5 >> stream\nHello endstream";
1528            let mut lexer = Lexer::new(Cursor::new(input));
1529
1530            assert_eq!(lexer.next_token().unwrap(), Token::DictStart);
1531            assert_eq!(
1532                lexer.next_token().unwrap(),
1533                Token::Name("Length".to_string())
1534            );
1535            assert_eq!(lexer.next_token().unwrap(), Token::Integer(5));
1536            assert_eq!(lexer.next_token().unwrap(), Token::DictEnd);
1537            assert_eq!(lexer.next_token().unwrap(), Token::Stream);
1538            // The actual stream content would be handled by a higher-level parser
1539        }
1540
1541        #[test]
1542        fn test_lexer_push_token() {
1543            let input = b"123 456";
1544            let mut lexer = Lexer::new(Cursor::new(input));
1545
1546            let token1 = lexer.next_token().unwrap();
1547            assert_eq!(token1, Token::Integer(123));
1548
1549            let token2 = lexer.next_token().unwrap();
1550            assert_eq!(token2, Token::Integer(456));
1551
1552            // Push token2 back
1553            lexer.push_token(token2.clone());
1554
1555            // Should get token2 again
1556            let token3 = lexer.next_token().unwrap();
1557            assert_eq!(token3, token2);
1558
1559            // Should get EOF
1560            let token4 = lexer.next_token().unwrap();
1561            assert_eq!(token4, Token::Eof);
1562        }
1563
1564        #[test]
1565        fn test_lexer_push_multiple_tokens() {
1566            let input = b"123";
1567            let mut lexer = Lexer::new(Cursor::new(input));
1568
1569            let original_token = lexer.next_token().unwrap();
1570            assert_eq!(original_token, Token::Integer(123));
1571
1572            // Push multiple tokens
1573            lexer.push_token(Token::Boolean(true));
1574            lexer.push_token(Token::Boolean(false));
1575            lexer.push_token(Token::Null);
1576
1577            // Should get them back in reverse order (stack behavior)
1578            assert_eq!(lexer.next_token().unwrap(), Token::Null);
1579            assert_eq!(lexer.next_token().unwrap(), Token::Boolean(false));
1580            assert_eq!(lexer.next_token().unwrap(), Token::Boolean(true));
1581            assert_eq!(lexer.next_token().unwrap(), Token::Eof);
1582        }
1583
1584        #[test]
1585        fn test_lexer_read_newline() {
1586            let input = b"123\n456\r\n789";
1587            let mut lexer = Lexer::new(Cursor::new(input));
1588
1589            // Read first digits
1590            let digits1 = lexer.read_digits().unwrap();
1591            assert_eq!(digits1, "123");
1592            assert!(lexer.read_newline().is_ok());
1593
1594            // Read second digits
1595            let digits2 = lexer.read_digits().unwrap();
1596            assert_eq!(digits2, "456");
1597            assert!(lexer.read_newline().is_ok());
1598
1599            // Read final digits
1600            let digits3 = lexer.read_digits().unwrap();
1601            assert_eq!(digits3, "789");
1602        }
1603
1604        #[test]
1605        fn test_lexer_read_bytes() {
1606            let input = b"Hello World";
1607            let mut lexer = Lexer::new(Cursor::new(input));
1608
1609            let bytes = lexer.read_bytes(5).unwrap();
1610            assert_eq!(bytes, b"Hello");
1611
1612            let bytes = lexer.read_bytes(6).unwrap();
1613            assert_eq!(bytes, b" World");
1614        }
1615
1616        #[test]
1617        fn test_lexer_read_until_sequence() {
1618            let input = b"Hello endstream World";
1619            let mut lexer = Lexer::new(Cursor::new(input));
1620
1621            let result = lexer.read_until_sequence(b"endstream").unwrap();
1622            assert_eq!(result, b"Hello ");
1623
1624            // Continue reading after the sequence
1625            let rest = lexer.read_digits().unwrap();
1626            assert_eq!(rest, ""); // read_digits only reads digits, " World" has no digits
1627        }
1628
1629        #[test]
1630        fn test_lexer_read_until_sequence_not_found() {
1631            let input = b"Hello World";
1632            let mut lexer = Lexer::new(Cursor::new(input));
1633
1634            let result = lexer.read_until_sequence(b"notfound");
1635            assert!(result.is_err());
1636        }
1637
1638        #[test]
1639        fn test_lexer_position_tracking() {
1640            let input = b"123 456";
1641            let mut lexer = Lexer::new(Cursor::new(input));
1642
1643            let initial_pos = lexer.position();
1644            assert_eq!(initial_pos, 0);
1645
1646            lexer.next_token().unwrap(); // "123"
1647            let pos_after_first = lexer.position();
1648            assert!(pos_after_first > initial_pos);
1649
1650            lexer.next_token().unwrap(); // "456"
1651            let pos_after_second = lexer.position();
1652            assert!(pos_after_second > pos_after_first);
1653        }
1654
1655        #[test]
1656        fn test_lexer_large_numbers() {
1657            let input = b"2147483647 -2147483648 9223372036854775807 -9223372036854775808";
1658            let mut lexer = Lexer::new(Cursor::new(input));
1659
1660            assert_eq!(lexer.next_token().unwrap(), Token::Integer(2147483647));
1661            assert_eq!(lexer.next_token().unwrap(), Token::Integer(-2147483648));
1662            assert_eq!(
1663                lexer.next_token().unwrap(),
1664                Token::Integer(9223372036854775807)
1665            );
1666            assert_eq!(
1667                lexer.next_token().unwrap(),
1668                Token::Integer(-9223372036854775808)
1669            );
1670        }
1671
1672        #[test]
1673        fn test_lexer_very_long_string() {
1674            let long_str = "A".repeat(1000);
1675            let input = format!("({long_str})");
1676            let mut lexer = Lexer::new(Cursor::new(input.as_bytes()));
1677
1678            if let Token::String(s) = lexer.next_token().unwrap() {
1679                assert_eq!(s.len(), 1000);
1680                assert_eq!(s, long_str.as_bytes());
1681            } else {
1682                panic!("Expected string token");
1683            }
1684        }
1685
1686        #[test]
1687        fn test_lexer_very_long_name() {
1688            let long_name = "A".repeat(500);
1689            let input = format!("/{long_name}");
1690            let mut lexer = Lexer::new(Cursor::new(input.as_bytes()));
1691
1692            if let Token::Name(name) = lexer.next_token().unwrap() {
1693                assert_eq!(name.len(), 500);
1694                assert_eq!(name, long_name);
1695            } else {
1696                panic!("Expected name token");
1697            }
1698        }
1699
1700        #[test]
1701        fn test_lexer_error_handling_invalid_hex() {
1702            let input = b"<48656C6C6FG>";
1703            let mut lexer = Lexer::new(Cursor::new(input));
1704
1705            // Should handle invalid hex gracefully
1706            let result = lexer.next_token();
1707            assert!(result.is_ok() || result.is_err()); // Either works or fails gracefully
1708        }
1709
1710        #[test]
1711        fn test_lexer_all_token_types() {
1712            let input = b"true false null 123 -456 3.14 (string) <48656C6C6F> /Name [ ] << >> obj endobj stream endstream startxref % comment\n";
1713            let mut lexer = Lexer::new(Cursor::new(input));
1714
1715            let mut token_types = Vec::new();
1716            loop {
1717                match lexer.next_token().unwrap() {
1718                    Token::Eof => break,
1719                    token => token_types.push(std::mem::discriminant(&token)),
1720                }
1721            }
1722
1723            // Should have multiple different token types
1724            assert!(token_types.len() > 10);
1725        }
1726
1727        #[test]
1728        fn test_lexer_performance() {
1729            let input = "123 456 789 ".repeat(1000);
1730            let mut lexer = Lexer::new(Cursor::new(input.as_bytes()));
1731
1732            let start_time = std::time::Instant::now();
1733            let mut count = 0;
1734            loop {
1735                match lexer.next_token().unwrap() {
1736                    Token::Eof => break,
1737                    _ => count += 1,
1738                }
1739            }
1740            let elapsed = start_time.elapsed();
1741
1742            assert_eq!(count, 3000); // 1000 repetitions * 3 tokens each
1743            assert!(elapsed.as_millis() < 1000); // Should complete within 1 second
1744        }
1745    }
1746
1747    #[test]
1748    fn test_lexer_find_keyword_ahead() {
1749        let input = b"some data here endstream more data";
1750        let mut lexer = Lexer::new(Cursor::new(input));
1751
1752        // Find endstream keyword
1753        let result = lexer.find_keyword_ahead("endstream", 100);
1754        assert!(result.is_ok());
1755        assert_eq!(result.unwrap(), Some(15)); // Position of endstream
1756
1757        // Try to find non-existent keyword
1758        let result2 = lexer.find_keyword_ahead("notfound", 100);
1759        assert!(result2.is_ok());
1760        assert_eq!(result2.unwrap(), None);
1761
1762        // Test with limited search distance
1763        let result3 = lexer.find_keyword_ahead("endstream", 10);
1764        assert!(result3.is_ok());
1765        assert_eq!(result3.unwrap(), None); // Not found within 10 bytes
1766    }
1767
1768    #[test]
1769    fn test_lexer_peek_token() {
1770        let input = b"123 456 /Name";
1771        let mut lexer = Lexer::new(Cursor::new(input));
1772
1773        // Peek first token
1774        let peeked = lexer.peek_token();
1775        assert!(peeked.is_ok());
1776        assert_eq!(peeked.unwrap(), Token::Integer(123));
1777
1778        // Verify peek doesn't consume
1779        let next = lexer.next_token();
1780        assert!(next.is_ok());
1781        assert_eq!(next.unwrap(), Token::Integer(123));
1782
1783        // Peek and consume next tokens
1784        assert_eq!(lexer.peek_token().unwrap(), Token::Integer(456));
1785        assert_eq!(lexer.next_token().unwrap(), Token::Integer(456));
1786
1787        assert_eq!(lexer.peek_token().unwrap(), Token::Name("Name".to_string()));
1788        assert_eq!(lexer.next_token().unwrap(), Token::Name("Name".to_string()));
1789    }
1790
1791    #[test]
1792    fn test_lexer_expect_keyword() {
1793        let input = b"endstream obj endobj";
1794        let mut lexer = Lexer::new(Cursor::new(input));
1795
1796        // Expect correct keyword
1797        assert!(lexer.expect_keyword("endstream").is_ok());
1798
1799        // Expect another correct keyword
1800        assert!(lexer.expect_keyword("obj").is_ok());
1801
1802        // Expect wrong keyword (should fail)
1803        let result = lexer.expect_keyword("stream");
1804        assert!(result.is_err());
1805        match result {
1806            Err(ParseError::UnexpectedToken { expected, found }) => {
1807                assert!(expected.contains("stream"));
1808                assert!(found.contains("EndObj"));
1809            }
1810            _ => panic!("Expected UnexpectedToken error"),
1811        }
1812    }
1813
1814    #[test]
1815    fn test_lexer_save_restore_position() {
1816        let input = b"123 456 789";
1817        let mut lexer = Lexer::new(Cursor::new(input));
1818
1819        // Read first token
1820        assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1821
1822        // Save position
1823        let saved = lexer.save_position();
1824        assert!(saved.is_ok());
1825        let saved_pos = saved.unwrap();
1826
1827        // Read more tokens
1828        assert_eq!(lexer.next_token().unwrap(), Token::Integer(456));
1829        assert_eq!(lexer.next_token().unwrap(), Token::Integer(789));
1830
1831        // Restore position
1832        assert!(lexer.restore_position(saved_pos).is_ok());
1833
1834        // Should be back at second token
1835        assert_eq!(lexer.next_token().unwrap(), Token::Integer(456));
1836    }
1837
1838    #[test]
1839    fn test_lexer_character_encoding_recovery() {
1840        // Test string with encoding issues (Windows-1252 bytes)
1841        let input = b"(Caf\x80 \x91Hello\x92)"; // "Café 'Hello'"
1842        let options = ParseOptions::lenient();
1843        let mut lexer = Lexer::new_with_options(Cursor::new(input), options);
1844
1845        match lexer.next_token().unwrap() {
1846            Token::String(bytes) => {
1847                // Should contain the text, potentially with encoding recovery
1848                let text = String::from_utf8_lossy(&bytes);
1849                tracing::debug!("Recovered text: {text}");
1850                assert!(!text.is_empty()); // Should not be empty
1851            }
1852            other => panic!("Expected String token, got {other:?}"),
1853        }
1854
1855        // Check that warnings were collected
1856        let warnings = lexer.warnings();
1857        if !warnings.is_empty() {
1858            tracing::debug!("Encoding warnings: {warnings:?}");
1859        }
1860    }
1861
1862    /// Helper to create a lexer without encoding recovery for raw byte tests
1863    fn lexer_no_encoding(data: &[u8]) -> Lexer<Cursor<&[u8]>> {
1864        let mut opts = ParseOptions::default();
1865        opts.lenient_encoding = false;
1866        Lexer::new_with_options(Cursor::new(data), opts)
1867    }
1868
1869    #[test]
1870    fn test_lexer_octal_escape_overflow_777_raw() {
1871        // \777 = octal 777 = 511 decimal, overflows u8.
1872        // Per ISO 32000-1:2008 §7.3.4.2: "high-order overflow shall be ignored"
1873        // 511 as u8 = 255 (0x1FF truncated to 0xFF)
1874        let mut lexer = lexer_no_encoding(b"(\\777)");
1875        match lexer.next_token().unwrap() {
1876            Token::String(bytes) => assert_eq!(bytes, vec![0xFF]),
1877            other => panic!("Expected String token, got {other:?}"),
1878        }
1879    }
1880
1881    #[test]
1882    fn test_lexer_octal_escape_overflow_400_raw() {
1883        // \400 = 256 decimal, just overflows u8 → 0
1884        let mut lexer = lexer_no_encoding(b"(\\400)");
1885        match lexer.next_token().unwrap() {
1886            Token::String(bytes) => assert_eq!(bytes, vec![0x00]),
1887            other => panic!("Expected String token, got {other:?}"),
1888        }
1889    }
1890
1891    #[test]
1892    fn test_lexer_octal_escape_max_valid_377_raw() {
1893        // \377 = 255, max valid octal for u8
1894        let mut lexer = lexer_no_encoding(b"(\\377)");
1895        match lexer.next_token().unwrap() {
1896            Token::String(bytes) => assert_eq!(bytes, vec![0xFF]),
1897            other => panic!("Expected String token, got {other:?}"),
1898        }
1899    }
1900
1901    #[test]
1902    fn test_lexer_octal_escape_overflow_mixed_raw() {
1903        // Mix of overflow octal and normal text
1904        let mut lexer = lexer_no_encoding(b"(A\\777B\\101C)");
1905        match lexer.next_token().unwrap() {
1906            Token::String(bytes) => {
1907                assert_eq!(bytes, vec![b'A', 0xFF, b'B', b'A', b'C']);
1908            }
1909            other => panic!("Expected String token, got {other:?}"),
1910        }
1911    }
1912
1913    #[test]
1914    fn test_lexer_octal_escape_overflow_no_panic_with_encoding() {
1915        // With default encoding recovery, overflow octals must not panic
1916        let mut lexer = Lexer::new(Cursor::new(b"(\\777\\400\\577)" as &[u8]));
1917        match lexer.next_token().unwrap() {
1918            Token::String(bytes) => {
1919                // Encoding recovery may transform high bytes to UTF-8,
1920                // but the key assertion is: no panic on overflow
1921                assert!(!bytes.is_empty());
1922            }
1923            other => panic!("Expected String token, got {other:?}"),
1924        }
1925    }
1926}
oxidize_pdf/parser/lexer.rs

oxidize_pdf/parser/
lexer.rs
