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        // A literal string carries bytes, not text. `/U`, `/O`, `/Perms` and the
361        // strings of an encrypted document are binary, and decoding them as text
362        // to re-encode the result as UTF-8 changes their length and their content
363        // — that is what rejected a correct empty password in issue #459. The
364        // bytes therefore reach the object model unchanged; the strings that ARE
365        // text are decoded where they are read as text, by
366        // [`PdfString::to_text`](super::objects::PdfString::to_text).
367        Ok(Token::String(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        // If a peeked byte is buffered, the lexer's logical position is one
780        // byte BEFORE `current_pos`. That byte is at the start of the scan
781        // window and must be checked as a candidate keyword char, otherwise
782        // the caller (which consumes peek_buffer via `read_bytes` and so
783        // operates in logical space) ends up off-by-one when it reads
784        // the returned offset. (Issue #260 root cause.)
785        if let Some(buffered) = start_buffer_state {
786            bytes_read = 1;
787            match_buffer.push(buffered);
788            if match_buffer.len() == keyword_bytes.len() && match_buffer == keyword_bytes {
789                self.reader.seek(SeekFrom::Start(current_pos))?;
790                self.peek_buffer = start_buffer_state;
791                return Ok(Some(bytes_read - keyword_bytes.len()));
792            }
793        }
794
795        // Search for the keyword
796        while bytes_read < max_bytes {
797            let mut byte = [0u8; 1];
798            match self.reader.read_exact(&mut byte) {
799                Ok(_) => {
800                    bytes_read += 1;
801                    match_buffer.push(byte[0]);
802
803                    // Keep only the last keyword.len() bytes in match_buffer
804                    if match_buffer.len() > keyword_bytes.len() {
805                        match_buffer.remove(0);
806                    }
807
808                    // Check if we found the keyword
809                    if match_buffer.len() == keyword_bytes.len() && match_buffer == keyword_bytes {
810                        // Restore position
811                        self.reader.seek(SeekFrom::Start(current_pos))?;
812                        self.peek_buffer = start_buffer_state;
813                        return Ok(Some(bytes_read - keyword_bytes.len()));
814                    }
815                }
816                Err(_) => break, // EOF or error
817            }
818        }
819
820        // Restore position
821        self.reader.seek(SeekFrom::Start(current_pos))?;
822        self.peek_buffer = start_buffer_state;
823        Ok(None)
824    }
825
826    /// Peek ahead n bytes without consuming them
827    pub fn peek_ahead(&mut self, n: usize) -> ParseResult<Vec<u8>>
828    where
829        R: Seek,
830    {
831        use std::io::{Read, Seek, SeekFrom};
832
833        // Save current position
834        let current_pos = self.reader.stream_position()?;
835        let start_buffer_state = self.peek_buffer;
836
837        // Read n bytes
838        let mut bytes = vec![0u8; n];
839        let bytes_read = self.reader.read(&mut bytes)?;
840        bytes.truncate(bytes_read);
841
842        // Restore position
843        self.reader.seek(SeekFrom::Start(current_pos))?;
844        self.peek_buffer = start_buffer_state;
845
846        Ok(bytes)
847    }
848
849    /// Save the current position for later restoration
850    pub fn save_position(&mut self) -> ParseResult<(u64, Option<u8>)>
851    where
852        R: Seek,
853    {
854        use std::io::Seek;
855        let pos = self.reader.stream_position()?;
856        Ok((pos, self.peek_buffer))
857    }
858
859    /// Restore a previously saved position
860    pub fn restore_position(&mut self, saved: (u64, Option<u8>)) -> ParseResult<()>
861    where
862        R: Seek,
863    {
864        use std::io::{Seek, SeekFrom};
865        self.reader.seek(SeekFrom::Start(saved.0))?;
866        self.peek_buffer = saved.1;
867        self.position = saved.0 as usize;
868        Ok(())
869    }
870
871    /// Peek the next token without consuming it.
872    ///
873    /// Restores the lexer position whether `next_token` succeeds or fails.
874    /// Prior to issue #260 the error path propagated via `?` before the
875    /// position was restored, leaving the cursor mid-word — recovery code
876    /// using `find_keyword_ahead` from the post-error position then could
877    /// not find the keyword it expected to discover ahead.
878    pub fn peek_token(&mut self) -> ParseResult<Token>
879    where
880        R: Seek,
881    {
882        let saved_pos = self.save_position()?;
883        let result = self.next_token();
884        self.restore_position(saved_pos)?;
885        result
886    }
887
888    /// Check if a character is likely a problematic encoding character
889    fn is_problematic_encoding_char(&self, ch: u8) -> bool {
890        // Control characters and Latin-1 supplement range that often indicate encoding issues
891        (0x80..=0x9F).contains(&ch) ||
892        ch == 0x07 || // Bell character
893        (ch <= 0x1F && ch != 0x09 && ch != 0x0A && ch != 0x0D) || // Control chars except tab, LF, CR
894        // In lenient mode, also handle extended Latin-1 characters that may appear in corrupted streams
895        (self.options.lenient_syntax && ch >= 0xA0) // Extended Latin-1 range (u8 max is 0xFF)
896    }
897
898    /// Handle problematic encoding characters in the main token stream
899    fn handle_encoding_char_in_token_stream(&mut self, ch: u8) -> ParseResult<Token> {
900        if self.options.lenient_encoding {
901            // Consume the problematic character and continue
902            self.consume_char()?;
903
904            // Log warning about the character recovery
905            if self.options.collect_warnings {
906                let replacement_char = match ch {
907                    0x07 => "bell",
908                    0x00..=0x1F => "control",
909                    0x80..=0x9F => "latin1-supplement",
910                    _ => "unknown",
911                };
912
913                self.warnings.push(ParseWarning::InvalidEncoding {
914                    position: self.position,
915                    recovered_text: format!(
916                        "Skipped problematic {replacement_char} character (0x{ch:02X})"
917                    ),
918                    encoding_used: None,
919                    replacement_count: 1,
920                });
921            }
922
923            // Skip this character and try to get the next token
924            self.skip_whitespace()?;
925            if let Ok(Some(_)) = self.peek_char() {
926                self.next_token() // Recursively try next token
927            } else {
928                Err(ParseError::SyntaxError {
929                    position: self.position,
930                    message: "Unexpected end of file after problematic character".to_string(),
931                })
932            }
933        } else {
934            // In strict mode, generate a more descriptive error
935            let char_description = match ch {
936                0x07 => "Bell character (\\u{07})".to_string(),
937                0x00..=0x1F => format!("Control character (\\u{{{ch:02X}}})"),
938                0x80..=0x9F => format!("Latin-1 supplement character (\\u{{{ch:02X}}})"),
939                _ => format!("Problematic character (\\u{{{ch:02X}}})"),
940            };
941
942            Err(ParseError::CharacterEncodingError {
943                position: self.position,
944                message: format!(
945                    "Unexpected character: {char_description} - Consider using lenient parsing mode"
946                ),
947            })
948        }
949    }
950}
951
952#[cfg(test)]
953mod tests {
954    use super::*;
955    use std::io::Cursor;
956
957    #[test]
958    fn test_lexer_basic_tokens() {
959        // Test positive and negative numbers
960        let input = b"123 -456 3.14 true false null /Name";
961        let mut lexer = Lexer::new(Cursor::new(input));
962
963        assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
964        assert_eq!(lexer.next_token().unwrap(), Token::Integer(-456));
965        assert_eq!(lexer.next_token().unwrap(), Token::Real(3.14));
966        assert_eq!(lexer.next_token().unwrap(), Token::Boolean(true));
967        assert_eq!(lexer.next_token().unwrap(), Token::Boolean(false));
968        assert_eq!(lexer.next_token().unwrap(), Token::Null);
969        assert_eq!(lexer.next_token().unwrap(), Token::Name("Name".to_string()));
970        assert_eq!(lexer.next_token().unwrap(), Token::Eof);
971    }
972
973    #[test]
974    fn test_lexer_negative_numbers() {
975        // Test negative numbers without space
976        let input = b"-123 -45.67";
977        let mut lexer = Lexer::new(Cursor::new(input));
978
979        assert_eq!(lexer.next_token().unwrap(), Token::Integer(-123));
980        assert_eq!(lexer.next_token().unwrap(), Token::Real(-45.67));
981    }
982
983    #[test]
984    fn test_lexer_strings() {
985        let input = b"(Hello World) <48656C6C6F>";
986        let mut lexer = Lexer::new(Cursor::new(input));
987
988        assert_eq!(
989            lexer.next_token().unwrap(),
990            Token::String(b"Hello World".to_vec())
991        );
992        assert_eq!(
993            lexer.next_token().unwrap(),
994            Token::String(b"Hello".to_vec())
995        );
996    }
997
998    #[test]
999    fn test_lexer_dictionaries() {
1000        let input = b"<< /Type /Page >>";
1001        let mut lexer = Lexer::new(Cursor::new(input));
1002
1003        assert_eq!(lexer.next_token().unwrap(), Token::DictStart);
1004        assert_eq!(lexer.next_token().unwrap(), Token::Name("Type".to_string()));
1005        assert_eq!(lexer.next_token().unwrap(), Token::Name("Page".to_string()));
1006        assert_eq!(lexer.next_token().unwrap(), Token::DictEnd);
1007    }
1008
1009    #[test]
1010    fn test_lexer_arrays() {
1011        let input = b"[1 2 3]";
1012        let mut lexer = Lexer::new(Cursor::new(input));
1013
1014        assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1015        assert_eq!(lexer.next_token().unwrap(), Token::Integer(1));
1016        assert_eq!(lexer.next_token().unwrap(), Token::Integer(2));
1017        assert_eq!(lexer.next_token().unwrap(), Token::Integer(3));
1018        assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1019    }
1020
1021    #[test]
1022    fn test_lexer_references() {
1023        let input = b"1 0 R 25 1 R";
1024        let mut lexer = Lexer::new(Cursor::new(input));
1025
1026        // Now references are parsed as separate tokens
1027        assert_eq!(lexer.next_token().unwrap(), Token::Integer(1));
1028        assert_eq!(lexer.next_token().unwrap(), Token::Integer(0));
1029        // 'R' should be parsed as a keyword or name
1030        match lexer.next_token().unwrap() {
1031            Token::Name(s) if s == "R" => {} // Could be a name
1032            other => panic!("Expected R token, got {other:?}"),
1033        }
1034
1035        assert_eq!(lexer.next_token().unwrap(), Token::Integer(25));
1036        assert_eq!(lexer.next_token().unwrap(), Token::Integer(1));
1037        match lexer.next_token().unwrap() {
1038            Token::Name(s) if s == "R" => {} // Could be a name
1039            other => panic!("Expected R token, got {other:?}"),
1040        }
1041    }
1042
1043    #[test]
1044    fn test_lexer_comments() {
1045        let input = b"%PDF-1.7\n123";
1046        let mut lexer = Lexer::new(Cursor::new(input));
1047
1048        assert_eq!(
1049            lexer.next_token().unwrap(),
1050            Token::Comment("PDF-1.7".to_string())
1051        );
1052        assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1053    }
1054
1055    // Comprehensive tests for Lexer
1056    mod comprehensive_tests {
1057        use super::*;
1058        use std::io::Cursor;
1059
1060        #[test]
1061        fn test_token_debug_trait() {
1062            let token = Token::Integer(42);
1063            let debug_str = format!("{token:?}");
1064            assert!(debug_str.contains("Integer"));
1065            assert!(debug_str.contains("42"));
1066        }
1067
1068        #[test]
1069        fn test_token_clone() {
1070            let token = Token::String(b"hello".to_vec());
1071            let cloned = token.clone();
1072            assert_eq!(token, cloned);
1073        }
1074
1075        #[test]
1076        fn test_token_equality() {
1077            assert_eq!(Token::Integer(42), Token::Integer(42));
1078            assert_ne!(Token::Integer(42), Token::Integer(43));
1079            assert_eq!(Token::Boolean(true), Token::Boolean(true));
1080            assert_ne!(Token::Boolean(true), Token::Boolean(false));
1081            assert_eq!(Token::Null, Token::Null);
1082            assert_ne!(Token::Null, Token::Integer(0));
1083        }
1084
1085        #[test]
1086        fn test_lexer_empty_input() {
1087            let input = b"";
1088            let mut lexer = Lexer::new(Cursor::new(input));
1089            assert_eq!(lexer.next_token().unwrap(), Token::Eof);
1090        }
1091
1092        #[test]
1093        fn test_lexer_whitespace_only() {
1094            let input = b"   \t\n\r  ";
1095            let mut lexer = Lexer::new(Cursor::new(input));
1096            assert_eq!(lexer.next_token().unwrap(), Token::Eof);
1097        }
1098
1099        #[test]
1100        fn test_lexer_integer_edge_cases() {
1101            let input = b"0 +123 -0 9876543210";
1102            let mut lexer = Lexer::new(Cursor::new(input));
1103
1104            assert_eq!(lexer.next_token().unwrap(), Token::Integer(0));
1105            assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1106            assert_eq!(lexer.next_token().unwrap(), Token::Integer(0));
1107            assert_eq!(lexer.next_token().unwrap(), Token::Integer(9876543210));
1108        }
1109
1110        #[test]
1111        fn test_lexer_real_edge_cases() {
1112            let input = b"0.0 +3.14 -2.71828 .5 5. 123.456789";
1113            let mut lexer = Lexer::new(Cursor::new(input));
1114
1115            assert_eq!(lexer.next_token().unwrap(), Token::Real(0.0));
1116            assert_eq!(lexer.next_token().unwrap(), Token::Real(3.14));
1117            assert_eq!(lexer.next_token().unwrap(), Token::Real(-2.71828));
1118            assert_eq!(lexer.next_token().unwrap(), Token::Real(0.5));
1119            assert_eq!(lexer.next_token().unwrap(), Token::Real(5.0));
1120            assert_eq!(lexer.next_token().unwrap(), Token::Real(123.456789));
1121        }
1122
1123        #[test]
1124        fn test_lexer_scientific_notation() {
1125            let input = b"1.23e10 -4.56E-5 1e0 2E+3";
1126            let mut lexer = Lexer::new(Cursor::new(input));
1127
1128            assert_eq!(lexer.next_token().unwrap(), Token::Real(1.23e10));
1129            assert_eq!(lexer.next_token().unwrap(), Token::Real(-4.56e-5));
1130            assert_eq!(lexer.next_token().unwrap(), Token::Real(1e0));
1131            assert_eq!(lexer.next_token().unwrap(), Token::Real(2e3));
1132        }
1133
1134        #[test]
1135        fn test_lexer_string_literal_escapes() {
1136            let input = b"(Hello\\nWorld) (Tab\\tChar) (Quote\\\"Mark) (Backslash\\\\)";
1137            let mut lexer = Lexer::new(Cursor::new(input));
1138
1139            assert_eq!(
1140                lexer.next_token().unwrap(),
1141                Token::String(b"Hello\nWorld".to_vec())
1142            );
1143            assert_eq!(
1144                lexer.next_token().unwrap(),
1145                Token::String(b"Tab\tChar".to_vec())
1146            );
1147            assert_eq!(
1148                lexer.next_token().unwrap(),
1149                Token::String(b"Quote\"Mark".to_vec())
1150            );
1151            assert_eq!(
1152                lexer.next_token().unwrap(),
1153                Token::String(b"Backslash\\".to_vec())
1154            );
1155        }
1156
1157        #[test]
1158        fn test_lexer_string_literal_nested_parens() {
1159            let input = b"(Nested (parentheses) work)";
1160            let mut lexer = Lexer::new(Cursor::new(input));
1161
1162            assert_eq!(
1163                lexer.next_token().unwrap(),
1164                Token::String(b"Nested (parentheses) work".to_vec())
1165            );
1166        }
1167
1168        #[test]
1169        fn test_lexer_string_literal_empty() {
1170            let input = b"()";
1171            let mut lexer = Lexer::new(Cursor::new(input));
1172
1173            assert_eq!(lexer.next_token().unwrap(), Token::String(b"".to_vec()));
1174        }
1175
1176        #[test]
1177        fn test_lexer_hexadecimal_strings() {
1178            let input = b"<48656C6C6F> <20576F726C64> <>";
1179            let mut lexer = Lexer::new(Cursor::new(input));
1180
1181            assert_eq!(
1182                lexer.next_token().unwrap(),
1183                Token::String(b"Hello".to_vec())
1184            );
1185            assert_eq!(
1186                lexer.next_token().unwrap(),
1187                Token::String(b" World".to_vec())
1188            );
1189            assert_eq!(lexer.next_token().unwrap(), Token::String(b"".to_vec()));
1190        }
1191
1192        #[test]
1193        fn test_lexer_hexadecimal_strings_odd_length() {
1194            let input = b"<48656C6C6F2> <1> <ABC>";
1195            let mut lexer = Lexer::new(Cursor::new(input));
1196
1197            // Odd length hex strings should pad with 0
1198            assert_eq!(
1199                lexer.next_token().unwrap(),
1200                Token::String(b"Hello ".to_vec())
1201            );
1202            assert_eq!(lexer.next_token().unwrap(), Token::String(b"\x10".to_vec()));
1203            assert_eq!(
1204                lexer.next_token().unwrap(),
1205                Token::String(b"\xAB\xC0".to_vec())
1206            );
1207        }
1208
1209        #[test]
1210        fn test_lexer_hexadecimal_strings_whitespace() {
1211            let input = b"<48 65 6C 6C 6F>";
1212            let mut lexer = Lexer::new(Cursor::new(input));
1213
1214            assert_eq!(
1215                lexer.next_token().unwrap(),
1216                Token::String(b"Hello".to_vec())
1217            );
1218        }
1219
1220        #[test]
1221        fn test_lexer_names() {
1222            let input = b"/Type /Page /Root /Kids /Count /MediaBox";
1223            let mut lexer = Lexer::new(Cursor::new(input));
1224
1225            assert_eq!(lexer.next_token().unwrap(), Token::Name("Type".to_string()));
1226            assert_eq!(lexer.next_token().unwrap(), Token::Name("Page".to_string()));
1227            assert_eq!(lexer.next_token().unwrap(), Token::Name("Root".to_string()));
1228            assert_eq!(lexer.next_token().unwrap(), Token::Name("Kids".to_string()));
1229            assert_eq!(
1230                lexer.next_token().unwrap(),
1231                Token::Name("Count".to_string())
1232            );
1233            assert_eq!(
1234                lexer.next_token().unwrap(),
1235                Token::Name("MediaBox".to_string())
1236            );
1237        }
1238
1239        #[test]
1240        fn test_lexer_names_with_special_chars() {
1241            let input = b"/Name#20with#20spaces /Name#2Fwith#2Fslashes";
1242            let mut lexer = Lexer::new(Cursor::new(input));
1243
1244            assert_eq!(
1245                lexer.next_token().unwrap(),
1246                Token::Name("Name with spaces".to_string())
1247            );
1248            assert_eq!(
1249                lexer.next_token().unwrap(),
1250                Token::Name("Name/with/slashes".to_string())
1251            );
1252        }
1253
1254        #[test]
1255        fn test_lexer_names_edge_cases() {
1256            let input = b"/ /A /123 /true /false /null";
1257            let mut lexer = Lexer::new(Cursor::new(input));
1258
1259            assert_eq!(lexer.next_token().unwrap(), Token::Name("".to_string()));
1260            assert_eq!(lexer.next_token().unwrap(), Token::Name("A".to_string()));
1261            assert_eq!(lexer.next_token().unwrap(), Token::Name("123".to_string()));
1262            assert_eq!(lexer.next_token().unwrap(), Token::Name("true".to_string()));
1263            assert_eq!(
1264                lexer.next_token().unwrap(),
1265                Token::Name("false".to_string())
1266            );
1267            assert_eq!(lexer.next_token().unwrap(), Token::Name("null".to_string()));
1268        }
1269
1270        #[test]
1271        fn test_lexer_nested_dictionaries() {
1272            let input = b"<< /Type /Page /Resources << /Font << /F1 123 0 R >> >> >>";
1273            let mut lexer = Lexer::new(Cursor::new(input));
1274
1275            assert_eq!(lexer.next_token().unwrap(), Token::DictStart);
1276            assert_eq!(lexer.next_token().unwrap(), Token::Name("Type".to_string()));
1277            assert_eq!(lexer.next_token().unwrap(), Token::Name("Page".to_string()));
1278            assert_eq!(
1279                lexer.next_token().unwrap(),
1280                Token::Name("Resources".to_string())
1281            );
1282            assert_eq!(lexer.next_token().unwrap(), Token::DictStart);
1283            assert_eq!(lexer.next_token().unwrap(), Token::Name("Font".to_string()));
1284            assert_eq!(lexer.next_token().unwrap(), Token::DictStart);
1285            assert_eq!(lexer.next_token().unwrap(), Token::Name("F1".to_string()));
1286            assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1287            assert_eq!(lexer.next_token().unwrap(), Token::Integer(0));
1288            assert_eq!(lexer.next_token().unwrap(), Token::Name("R".to_string()));
1289            assert_eq!(lexer.next_token().unwrap(), Token::DictEnd);
1290            assert_eq!(lexer.next_token().unwrap(), Token::DictEnd);
1291            assert_eq!(lexer.next_token().unwrap(), Token::DictEnd);
1292        }
1293
1294        #[test]
1295        fn test_lexer_nested_arrays() {
1296            let input = b"[[1 2] [3 4] [5 [6 7]]]";
1297            let mut lexer = Lexer::new(Cursor::new(input));
1298
1299            assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1300            assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1301            assert_eq!(lexer.next_token().unwrap(), Token::Integer(1));
1302            assert_eq!(lexer.next_token().unwrap(), Token::Integer(2));
1303            assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1304            assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1305            assert_eq!(lexer.next_token().unwrap(), Token::Integer(3));
1306            assert_eq!(lexer.next_token().unwrap(), Token::Integer(4));
1307            assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1308            assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1309            assert_eq!(lexer.next_token().unwrap(), Token::Integer(5));
1310            assert_eq!(lexer.next_token().unwrap(), Token::ArrayStart);
1311            assert_eq!(lexer.next_token().unwrap(), Token::Integer(6));
1312            assert_eq!(lexer.next_token().unwrap(), Token::Integer(7));
1313            assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1314            assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1315            assert_eq!(lexer.next_token().unwrap(), Token::ArrayEnd);
1316        }
1317
1318        #[test]
1319        fn test_lexer_mixed_content() {
1320            let input = b"<< /Type /Page /MediaBox [0 0 612 792] /Resources << /Font << /F1 << /Type /Font /Subtype /Type1 >> >> >> >>";
1321            let mut lexer = Lexer::new(Cursor::new(input));
1322
1323            // Just test that we can parse this without errors
1324            let mut tokens = Vec::new();
1325            loop {
1326                match lexer.next_token().unwrap() {
1327                    Token::Eof => break,
1328                    token => tokens.push(token),
1329                }
1330            }
1331            assert!(tokens.len() > 10);
1332        }
1333
1334        #[test]
1335        fn test_lexer_keywords() {
1336            let input = b"obj endobj stream endstream startxref";
1337            let mut lexer = Lexer::new(Cursor::new(input));
1338
1339            assert_eq!(lexer.next_token().unwrap(), Token::Obj);
1340            assert_eq!(lexer.next_token().unwrap(), Token::EndObj);
1341            assert_eq!(lexer.next_token().unwrap(), Token::Stream);
1342            assert_eq!(lexer.next_token().unwrap(), Token::EndStream);
1343            assert_eq!(lexer.next_token().unwrap(), Token::StartXRef);
1344        }
1345
1346        #[test]
1347        fn test_lexer_multiple_comments() {
1348            let input = b"%First comment\n%Second comment\n123";
1349            let mut lexer = Lexer::new(Cursor::new(input));
1350
1351            assert_eq!(
1352                lexer.next_token().unwrap(),
1353                Token::Comment("First comment".to_string())
1354            );
1355            assert_eq!(
1356                lexer.next_token().unwrap(),
1357                Token::Comment("Second comment".to_string())
1358            );
1359            assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1360        }
1361
1362        #[test]
1363        fn test_lexer_comment_without_newline() {
1364            let input = b"%Comment at end";
1365            let mut lexer = Lexer::new(Cursor::new(input));
1366
1367            assert_eq!(
1368                lexer.next_token().unwrap(),
1369                Token::Comment("Comment at end".to_string())
1370            );
1371            assert_eq!(lexer.next_token().unwrap(), Token::Eof);
1372        }
1373
1374        #[test]
1375        fn test_lexer_special_characters_in_streams() {
1376            let input = b"<< /Length 5 >> stream\nHello endstream";
1377            let mut lexer = Lexer::new(Cursor::new(input));
1378
1379            assert_eq!(lexer.next_token().unwrap(), Token::DictStart);
1380            assert_eq!(
1381                lexer.next_token().unwrap(),
1382                Token::Name("Length".to_string())
1383            );
1384            assert_eq!(lexer.next_token().unwrap(), Token::Integer(5));
1385            assert_eq!(lexer.next_token().unwrap(), Token::DictEnd);
1386            assert_eq!(lexer.next_token().unwrap(), Token::Stream);
1387            // The actual stream content would be handled by a higher-level parser
1388        }
1389
1390        #[test]
1391        fn test_lexer_push_token() {
1392            let input = b"123 456";
1393            let mut lexer = Lexer::new(Cursor::new(input));
1394
1395            let token1 = lexer.next_token().unwrap();
1396            assert_eq!(token1, Token::Integer(123));
1397
1398            let token2 = lexer.next_token().unwrap();
1399            assert_eq!(token2, Token::Integer(456));
1400
1401            // Push token2 back
1402            lexer.push_token(token2.clone());
1403
1404            // Should get token2 again
1405            let token3 = lexer.next_token().unwrap();
1406            assert_eq!(token3, token2);
1407
1408            // Should get EOF
1409            let token4 = lexer.next_token().unwrap();
1410            assert_eq!(token4, Token::Eof);
1411        }
1412
1413        #[test]
1414        fn test_lexer_push_multiple_tokens() {
1415            let input = b"123";
1416            let mut lexer = Lexer::new(Cursor::new(input));
1417
1418            let original_token = lexer.next_token().unwrap();
1419            assert_eq!(original_token, Token::Integer(123));
1420
1421            // Push multiple tokens
1422            lexer.push_token(Token::Boolean(true));
1423            lexer.push_token(Token::Boolean(false));
1424            lexer.push_token(Token::Null);
1425
1426            // Should get them back in reverse order (stack behavior)
1427            assert_eq!(lexer.next_token().unwrap(), Token::Null);
1428            assert_eq!(lexer.next_token().unwrap(), Token::Boolean(false));
1429            assert_eq!(lexer.next_token().unwrap(), Token::Boolean(true));
1430            assert_eq!(lexer.next_token().unwrap(), Token::Eof);
1431        }
1432
1433        #[test]
1434        fn test_lexer_read_newline() {
1435            let input = b"123\n456\r\n789";
1436            let mut lexer = Lexer::new(Cursor::new(input));
1437
1438            // Read first digits
1439            let digits1 = lexer.read_digits().unwrap();
1440            assert_eq!(digits1, "123");
1441            assert!(lexer.read_newline().is_ok());
1442
1443            // Read second digits
1444            let digits2 = lexer.read_digits().unwrap();
1445            assert_eq!(digits2, "456");
1446            assert!(lexer.read_newline().is_ok());
1447
1448            // Read final digits
1449            let digits3 = lexer.read_digits().unwrap();
1450            assert_eq!(digits3, "789");
1451        }
1452
1453        #[test]
1454        fn test_lexer_read_bytes() {
1455            let input = b"Hello World";
1456            let mut lexer = Lexer::new(Cursor::new(input));
1457
1458            let bytes = lexer.read_bytes(5).unwrap();
1459            assert_eq!(bytes, b"Hello");
1460
1461            let bytes = lexer.read_bytes(6).unwrap();
1462            assert_eq!(bytes, b" World");
1463        }
1464
1465        #[test]
1466        fn test_lexer_read_until_sequence() {
1467            let input = b"Hello endstream World";
1468            let mut lexer = Lexer::new(Cursor::new(input));
1469
1470            let result = lexer.read_until_sequence(b"endstream").unwrap();
1471            assert_eq!(result, b"Hello ");
1472
1473            // Continue reading after the sequence
1474            let rest = lexer.read_digits().unwrap();
1475            assert_eq!(rest, ""); // read_digits only reads digits, " World" has no digits
1476        }
1477
1478        #[test]
1479        fn test_lexer_read_until_sequence_not_found() {
1480            let input = b"Hello World";
1481            let mut lexer = Lexer::new(Cursor::new(input));
1482
1483            let result = lexer.read_until_sequence(b"notfound");
1484            assert!(result.is_err());
1485        }
1486
1487        #[test]
1488        fn test_lexer_position_tracking() {
1489            let input = b"123 456";
1490            let mut lexer = Lexer::new(Cursor::new(input));
1491
1492            let initial_pos = lexer.position();
1493            assert_eq!(initial_pos, 0);
1494
1495            lexer.next_token().unwrap(); // "123"
1496            let pos_after_first = lexer.position();
1497            assert!(pos_after_first > initial_pos);
1498
1499            lexer.next_token().unwrap(); // "456"
1500            let pos_after_second = lexer.position();
1501            assert!(pos_after_second > pos_after_first);
1502        }
1503
1504        #[test]
1505        fn test_lexer_large_numbers() {
1506            let input = b"2147483647 -2147483648 9223372036854775807 -9223372036854775808";
1507            let mut lexer = Lexer::new(Cursor::new(input));
1508
1509            assert_eq!(lexer.next_token().unwrap(), Token::Integer(2147483647));
1510            assert_eq!(lexer.next_token().unwrap(), Token::Integer(-2147483648));
1511            assert_eq!(
1512                lexer.next_token().unwrap(),
1513                Token::Integer(9223372036854775807)
1514            );
1515            assert_eq!(
1516                lexer.next_token().unwrap(),
1517                Token::Integer(-9223372036854775808)
1518            );
1519        }
1520
1521        #[test]
1522        fn test_lexer_very_long_string() {
1523            let long_str = "A".repeat(1000);
1524            let input = format!("({long_str})");
1525            let mut lexer = Lexer::new(Cursor::new(input.as_bytes()));
1526
1527            if let Token::String(s) = lexer.next_token().unwrap() {
1528                assert_eq!(s.len(), 1000);
1529                assert_eq!(s, long_str.as_bytes());
1530            } else {
1531                panic!("Expected string token");
1532            }
1533        }
1534
1535        #[test]
1536        fn test_lexer_very_long_name() {
1537            let long_name = "A".repeat(500);
1538            let input = format!("/{long_name}");
1539            let mut lexer = Lexer::new(Cursor::new(input.as_bytes()));
1540
1541            if let Token::Name(name) = lexer.next_token().unwrap() {
1542                assert_eq!(name.len(), 500);
1543                assert_eq!(name, long_name);
1544            } else {
1545                panic!("Expected name token");
1546            }
1547        }
1548
1549        #[test]
1550        fn test_lexer_error_handling_invalid_hex() {
1551            let input = b"<48656C6C6FG>";
1552            let mut lexer = Lexer::new(Cursor::new(input));
1553
1554            // Should handle invalid hex gracefully
1555            let result = lexer.next_token();
1556            assert!(result.is_ok() || result.is_err()); // Either works or fails gracefully
1557        }
1558
1559        #[test]
1560        fn test_lexer_all_token_types() {
1561            let input = b"true false null 123 -456 3.14 (string) <48656C6C6F> /Name [ ] << >> obj endobj stream endstream startxref % comment\n";
1562            let mut lexer = Lexer::new(Cursor::new(input));
1563
1564            let mut token_types = Vec::new();
1565            loop {
1566                match lexer.next_token().unwrap() {
1567                    Token::Eof => break,
1568                    token => token_types.push(std::mem::discriminant(&token)),
1569                }
1570            }
1571
1572            // Should have multiple different token types
1573            assert!(token_types.len() > 10);
1574        }
1575
1576        #[test]
1577        fn test_lexer_performance() {
1578            let input = "123 456 789 ".repeat(1000);
1579            let mut lexer = Lexer::new(Cursor::new(input.as_bytes()));
1580
1581            let start_time = std::time::Instant::now();
1582            let mut count = 0;
1583            loop {
1584                match lexer.next_token().unwrap() {
1585                    Token::Eof => break,
1586                    _ => count += 1,
1587                }
1588            }
1589            let elapsed = start_time.elapsed();
1590
1591            assert_eq!(count, 3000); // 1000 repetitions * 3 tokens each
1592            assert!(elapsed.as_millis() < 1000); // Should complete within 1 second
1593        }
1594    }
1595
1596    #[test]
1597    fn test_lexer_find_keyword_ahead() {
1598        let input = b"some data here endstream more data";
1599        let mut lexer = Lexer::new(Cursor::new(input));
1600
1601        // Find endstream keyword
1602        let result = lexer.find_keyword_ahead("endstream", 100);
1603        assert!(result.is_ok());
1604        assert_eq!(result.unwrap(), Some(15)); // Position of endstream
1605
1606        // Try to find non-existent keyword
1607        let result2 = lexer.find_keyword_ahead("notfound", 100);
1608        assert!(result2.is_ok());
1609        assert_eq!(result2.unwrap(), None);
1610
1611        // Test with limited search distance
1612        let result3 = lexer.find_keyword_ahead("endstream", 10);
1613        assert!(result3.is_ok());
1614        assert_eq!(result3.unwrap(), None); // Not found within 10 bytes
1615    }
1616
1617    #[test]
1618    fn test_lexer_peek_token() {
1619        let input = b"123 456 /Name";
1620        let mut lexer = Lexer::new(Cursor::new(input));
1621
1622        // Peek first token
1623        let peeked = lexer.peek_token();
1624        assert!(peeked.is_ok());
1625        assert_eq!(peeked.unwrap(), Token::Integer(123));
1626
1627        // Verify peek doesn't consume
1628        let next = lexer.next_token();
1629        assert!(next.is_ok());
1630        assert_eq!(next.unwrap(), Token::Integer(123));
1631
1632        // Peek and consume next tokens
1633        assert_eq!(lexer.peek_token().unwrap(), Token::Integer(456));
1634        assert_eq!(lexer.next_token().unwrap(), Token::Integer(456));
1635
1636        assert_eq!(lexer.peek_token().unwrap(), Token::Name("Name".to_string()));
1637        assert_eq!(lexer.next_token().unwrap(), Token::Name("Name".to_string()));
1638    }
1639
1640    #[test]
1641    fn test_lexer_expect_keyword() {
1642        let input = b"endstream obj endobj";
1643        let mut lexer = Lexer::new(Cursor::new(input));
1644
1645        // Expect correct keyword
1646        assert!(lexer.expect_keyword("endstream").is_ok());
1647
1648        // Expect another correct keyword
1649        assert!(lexer.expect_keyword("obj").is_ok());
1650
1651        // Expect wrong keyword (should fail)
1652        let result = lexer.expect_keyword("stream");
1653        assert!(result.is_err());
1654        match result {
1655            Err(ParseError::UnexpectedToken { expected, found }) => {
1656                assert!(expected.contains("stream"));
1657                assert!(found.contains("EndObj"));
1658            }
1659            _ => panic!("Expected UnexpectedToken error"),
1660        }
1661    }
1662
1663    #[test]
1664    fn test_lexer_save_restore_position() {
1665        let input = b"123 456 789";
1666        let mut lexer = Lexer::new(Cursor::new(input));
1667
1668        // Read first token
1669        assert_eq!(lexer.next_token().unwrap(), Token::Integer(123));
1670
1671        // Save position
1672        let saved = lexer.save_position();
1673        assert!(saved.is_ok());
1674        let saved_pos = saved.unwrap();
1675
1676        // Read more tokens
1677        assert_eq!(lexer.next_token().unwrap(), Token::Integer(456));
1678        assert_eq!(lexer.next_token().unwrap(), Token::Integer(789));
1679
1680        // Restore position
1681        assert!(lexer.restore_position(saved_pos).is_ok());
1682
1683        // Should be back at second token
1684        assert_eq!(lexer.next_token().unwrap(), Token::Integer(456));
1685    }
1686
1687    #[test]
1688    fn test_lexer_character_encoding_recovery() {
1689        // Test string with encoding issues (Windows-1252 bytes)
1690        let input = b"(Caf\x80 \x91Hello\x92)"; // "Café 'Hello'"
1691        let options = ParseOptions::lenient();
1692        let mut lexer = Lexer::new_with_options(Cursor::new(input), options);
1693
1694        match lexer.next_token().unwrap() {
1695            Token::String(bytes) => {
1696                // Should contain the text, potentially with encoding recovery
1697                let text = String::from_utf8_lossy(&bytes);
1698                tracing::debug!("Recovered text: {text}");
1699                assert!(!text.is_empty()); // Should not be empty
1700            }
1701            other => panic!("Expected String token, got {other:?}"),
1702        }
1703
1704        // Check that warnings were collected
1705        let warnings = lexer.warnings();
1706        if !warnings.is_empty() {
1707            tracing::debug!("Encoding warnings: {warnings:?}");
1708        }
1709    }
1710
1711    /// Helper to create a lexer without encoding recovery for raw byte tests
1712    fn lexer_no_encoding(data: &[u8]) -> Lexer<Cursor<&[u8]>> {
1713        let mut opts = ParseOptions::default();
1714        opts.lenient_encoding = false;
1715        Lexer::new_with_options(Cursor::new(data), opts)
1716    }
1717
1718    #[test]
1719    fn test_lexer_octal_escape_overflow_777_raw() {
1720        // \777 = octal 777 = 511 decimal, overflows u8.
1721        // Per ISO 32000-1:2008 §7.3.4.2: "high-order overflow shall be ignored"
1722        // 511 as u8 = 255 (0x1FF truncated to 0xFF)
1723        let mut lexer = lexer_no_encoding(b"(\\777)");
1724        match lexer.next_token().unwrap() {
1725            Token::String(bytes) => assert_eq!(bytes, vec![0xFF]),
1726            other => panic!("Expected String token, got {other:?}"),
1727        }
1728    }
1729
1730    #[test]
1731    fn test_lexer_octal_escape_overflow_400_raw() {
1732        // \400 = 256 decimal, just overflows u8 → 0
1733        let mut lexer = lexer_no_encoding(b"(\\400)");
1734        match lexer.next_token().unwrap() {
1735            Token::String(bytes) => assert_eq!(bytes, vec![0x00]),
1736            other => panic!("Expected String token, got {other:?}"),
1737        }
1738    }
1739
1740    #[test]
1741    fn test_lexer_octal_escape_max_valid_377_raw() {
1742        // \377 = 255, max valid octal for u8
1743        let mut lexer = lexer_no_encoding(b"(\\377)");
1744        match lexer.next_token().unwrap() {
1745            Token::String(bytes) => assert_eq!(bytes, vec![0xFF]),
1746            other => panic!("Expected String token, got {other:?}"),
1747        }
1748    }
1749
1750    #[test]
1751    fn test_lexer_octal_escape_overflow_mixed_raw() {
1752        // Mix of overflow octal and normal text
1753        let mut lexer = lexer_no_encoding(b"(A\\777B\\101C)");
1754        match lexer.next_token().unwrap() {
1755            Token::String(bytes) => {
1756                assert_eq!(bytes, vec![b'A', 0xFF, b'B', b'A', b'C']);
1757            }
1758            other => panic!("Expected String token, got {other:?}"),
1759        }
1760    }
1761
1762    #[test]
1763    fn test_lexer_octal_escape_overflow_no_panic_with_encoding() {
1764        // With default encoding recovery, overflow octals must not panic
1765        let mut lexer = Lexer::new(Cursor::new(b"(\\777\\400\\577)" as &[u8]));
1766        match lexer.next_token().unwrap() {
1767            Token::String(bytes) => {
1768                // Encoding recovery may transform high bytes to UTF-8,
1769                // but the key assertion is: no panic on overflow
1770                assert!(!bytes.is_empty());
1771            }
1772            other => panic!("Expected String token, got {other:?}"),
1773        }
1774    }
1775}
oxidize_pdf/parser/lexer.rs

oxidize_pdf/parser/
lexer.rs
