Skip to main content

form_net/urlencoded/
raw.rs

1//! The allocation-free urlencoded scanner.
2
3use core::fmt;
4
5use crate::error::{FormError, Limit, Location};
6use crate::limits::{Limits, PercentPolicy};
7use crate::urlencoded::decode::decode_component;
8
9const HAS_EQUALS: u8 = 1;
10const NAME_DECODE: u8 = 2;
11const VALUE_DECODE: u8 = 4;
12
13/// Iterates the pairs of a urlencoded body without allocating or copying.
14///
15/// Budgets are checked per sequence before the sequence is handed out: body
16/// size (every byte, empty sequences included), field count, raw name and
17/// value lengths, and the running total of raw value lengths. After the
18/// first error the iterator returns `None`.
19///
20/// # Examples
21///
22/// ```
23/// use form_net::Limits;
24///
25/// let pairs: Vec<_> = form_net::raw_pairs(b"a=1&&b", &Limits::DEFAULT)
26///     .map(|p| p.map(|p| (p.name_raw(), p.value_raw(), p.has_equals())))
27///     .collect::<Result<_, _>>()?;
28/// assert_eq!(pairs, [(&b"a"[..], &b"1"[..], true), (&b"b"[..], &b""[..], false)]);
29/// # Ok::<(), form_net::FormError>(())
30/// ```
31#[must_use]
32pub fn raw_pairs<'a>(input: &'a [u8], limits: &Limits) -> RawPairs<'a> {
33    RawPairs {
34        input,
35        pos: 0,
36        budget: Budget::new(limits),
37        done: false,
38        checked_body: false,
39    }
40}
41
42/// The per-sequence budgets, shared by [`raw_pairs`] and the streaming
43/// parser so both refuse the same sequence with the same error.
44#[derive(Clone, Copy, Debug)]
45pub(crate) struct Budget {
46    // The effective limits this budget checks, not the whole `Limits`: the
47    // iterators that carry a budget are moved by value through adapters.
48    max_fields: u32,
49    max_name: u32,
50    max_value: u32,
51    max_text: u64,
52    pub(crate) body: u64,
53    fields: u32,
54    text: u64,
55}
56
57impl Budget {
58    pub(crate) const fn new(limits: &Limits) -> Budget {
59        Budget {
60            max_fields: limits.max_fields,
61            max_name: limits.name(),
62            max_value: limits.value(),
63            max_text: limits.text(),
64            body: limits.body(),
65            fields: 0,
66            text: 0,
67        }
68    }
69
70    /// Counts and checks one complete, non-empty sequence: field count,
71    /// name length, value length, then the text total.
72    pub(crate) fn admit(&mut self, pair: &RawPair<'_>) -> Result<(), FormError> {
73        let at = Location::at(pair.offset);
74        self.fields = self.fields.saturating_add(1);
75        if self.fields > self.max_fields {
76            return Err(FormError::limit(Limit::Fields, at));
77        }
78        if pair.name.len() as u64 > u64::from(self.max_name) {
79            return Err(FormError::limit(Limit::NameLength, at));
80        }
81        let value_at = pair.value_offset();
82        if pair.value.len() as u64 > u64::from(self.max_value) {
83            return Err(FormError::limit(Limit::ValueLength, value_at));
84        }
85        self.text = self.text.saturating_add(pair.value.len() as u64);
86        if self.text > self.max_text {
87            return Err(FormError::limit(Limit::TextTotal, value_at));
88        }
89        Ok(())
90    }
91
92    /// Checks the start of a sequence that is not complete yet: when its
93    /// complete form is certain to be refused, refuses it now with the error
94    /// [`admit`](Self::admit) would give, so a held sequence never outgrows
95    /// `urlencoded_buffer_len`.
96    pub(crate) fn admit_partial(
97        &self,
98        offset: u64,
99        name_len: usize,
100        has_equals: bool,
101        value_len: usize,
102    ) -> Result<(), FormError> {
103        let name_over = name_len as u64 > u64::from(self.max_name);
104        let value_over = has_equals && value_len as u64 > u64::from(self.max_value);
105        if !name_over && !value_over {
106            return Ok(());
107        }
108        if self.fields.saturating_add(1) > self.max_fields {
109            return Err(FormError::limit(Limit::Fields, Location::at(offset)));
110        }
111        if name_over {
112            return Err(FormError::limit(Limit::NameLength, Location::at(offset)));
113        }
114        let value_at = offset.saturating_add(name_len as u64).saturating_add(1);
115        Err(FormError::limit(Limit::ValueLength, Location::at(value_at)))
116    }
117
118    pub(crate) fn reset(&mut self) {
119        self.fields = 0;
120        self.text = 0;
121    }
122}
123
124/// Scans one sequence from `start`: returns where it ends (the next `&`, or
125/// the end), where its first `=` is, and its flags. The byte set narrows as
126/// facts are learned: once a side is known to need decoding, its `%` and
127/// `+` stop mattering, so a value full of escapes is crossed in eight-byte
128/// steps instead of stopping at every `%`.
129#[inline]
130fn scan(input: &[u8], start: usize) -> (usize, Option<usize>, u8) {
131    let mut flags = 0u8;
132    let mut i = start;
133    // The name: up to the first `=` or `&`.
134    let eq = loop {
135        i = if flags & NAME_DECODE == 0 {
136            crate::scan::find_any(input, i, *b"&=%+")
137        } else {
138            crate::scan::find_any(input, i, *b"&=")
139        };
140        match input.get(i) {
141            None | Some(b'&') => return (i, None, flags),
142            Some(b'=') => break i,
143            Some(_) => flags |= NAME_DECODE,
144        }
145        i = i.wrapping_add(1);
146    };
147    flags |= HAS_EQUALS;
148    i = eq.wrapping_add(1);
149    // The value: `=` is data now; after the first `%` or `+` only `&` ends it.
150    loop {
151        i = if flags & VALUE_DECODE == 0 {
152            crate::scan::find_any(input, i, *b"&%+")
153        } else {
154            crate::scan::find_byte(input, i, b'&')
155        };
156        match input.get(i) {
157            None | Some(b'&') => return (i, Some(eq), flags),
158            Some(_) => flags |= VALUE_DECODE,
159        }
160        i = i.wrapping_add(1);
161    }
162}
163
164/// The iterator returned by [`raw_pairs`].
165#[derive(Clone)]
166pub struct RawPairs<'a> {
167    input: &'a [u8],
168    pos: usize,
169    budget: Budget,
170    done: bool,
171    checked_body: bool,
172}
173
174impl fmt::Debug for RawPairs<'_> {
175    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
176        f.debug_struct("RawPairs")
177            .field("position", &self.pos)
178            .field("fields", &self.budget.fields)
179            .finish_non_exhaustive()
180    }
181}
182
183impl<'a> RawPairs<'a> {
184    fn fail(&mut self, err: FormError) -> Option<Result<RawPair<'a>, FormError>> {
185        self.done = true;
186        Some(Err(err))
187    }
188}
189
190impl<'a> Iterator for RawPairs<'a> {
191    type Item = Result<RawPair<'a>, FormError>;
192
193    fn next(&mut self) -> Option<Self::Item> {
194        if self.done {
195            return None;
196        }
197        if !self.checked_body {
198            self.checked_body = true;
199            if self.input.len() as u64 > self.budget.body {
200                return self.fail(FormError::limit(
201                    Limit::Body,
202                    Location::at(self.budget.body),
203                ));
204            }
205        }
206        loop {
207            if self.pos >= self.input.len() {
208                self.done = true;
209                return None;
210            }
211            let start = self.pos;
212            let (end, eq, flags) = scan(self.input, start);
213            self.pos = end.saturating_add(1).min(self.input.len());
214            if end == start {
215                continue;
216            }
217            let seq = self.input.get(start..end).unwrap_or(&[]);
218            let (name, value) = match eq {
219                Some(e) => {
220                    let at = e.wrapping_sub(start);
221                    (
222                        seq.get(..at).unwrap_or(&[]),
223                        seq.get(at.wrapping_add(1)..).unwrap_or(&[]),
224                    )
225                }
226                None => (seq, &[][..]),
227            };
228            let pair = RawPair {
229                name,
230                value,
231                flags,
232                offset: start as u64,
233            };
234            if let Err(e) = self.budget.admit(&pair) {
235                return self.fail(e);
236            }
237            return Some(Ok(pair));
238        }
239    }
240}
241
242impl core::iter::FusedIterator for RawPairs<'_> {}
243
244/// One `name=value` sequence, undecoded.
245///
246/// `Debug` prints the offset and the lengths only: values can be secrets.
247#[derive(Clone, Copy)]
248pub struct RawPair<'a> {
249    name: &'a [u8],
250    value: &'a [u8],
251    flags: u8,
252    offset: u64,
253}
254
255impl fmt::Debug for RawPair<'_> {
256    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
257        f.debug_struct("RawPair")
258            .field("offset", &self.offset)
259            .field("name_len", &self.name.len())
260            .field("value_len", &self.value.len())
261            .finish()
262    }
263}
264
265impl<'a> RawPair<'a> {
266    /// Splits one sequence (no `&` in it) found at `offset` in the body.
267    pub(crate) fn split(seq: &'a [u8], offset: u64) -> RawPair<'a> {
268        let (_, eq, flags) = scan(seq, 0);
269        let (name, value) = match eq {
270            Some(e) => (
271                seq.get(..e).unwrap_or(&[]),
272                seq.get(e.wrapping_add(1)..).unwrap_or(&[]),
273            ),
274            None => (seq, &[][..]),
275        };
276        RawPair {
277            name,
278            value,
279            flags,
280            offset,
281        }
282    }
283
284    /// The name as text when it needs no decoding. With `whole` (the body,
285    /// already validated as UTF-8 in one pass) it is cut from it: the cut
286    /// falls at ASCII `&` or `=`, so it is an O(1) boundary check instead of
287    /// a UTF-8 pass per name (D-104).
288    #[cfg(feature = "alloc")]
289    #[inline]
290    pub(crate) fn name_text(&self, whole: Option<&'a str>) -> Option<&'a str> {
291        if self.name_needs_decoding() {
292            return None;
293        }
294        let start = usize::try_from(self.offset).ok()?;
295        match whole {
296            Some(w) => w.get(start..start.checked_add(self.name.len())?),
297            None => core::str::from_utf8(self.name).ok(),
298        }
299    }
300
301    /// As [`name_text`](Self::name_text), for the value.
302    #[cfg(feature = "alloc")]
303    #[inline]
304    pub(crate) fn value_text(&self, whole: Option<&'a str>) -> Option<&'a str> {
305        if self.value_needs_decoding() {
306            return None;
307        }
308        if self.value.is_empty() {
309            return Some("");
310        }
311        let start = usize::try_from(self.offset)
312            .ok()?
313            .checked_add(self.name.len())?
314            .checked_add(1)?;
315        match whole {
316            Some(w) => w.get(start..start.checked_add(self.value.len())?),
317            None => core::str::from_utf8(self.value).ok(),
318        }
319    }
320
321    fn value_offset(&self) -> Location {
322        Location::at(
323            self.offset
324                .saturating_add(self.name.len() as u64)
325                .saturating_add(1),
326        )
327    }
328
329    /// The name as sent, before decoding.
330    ///
331    /// # Examples
332    ///
333    /// ```
334    /// let p = form_net::raw_pairs(b"a+b=1", &form_net::Limits::DEFAULT).next().unwrap()?;
335    /// assert_eq!(p.name_raw(), b"a+b");
336    /// # Ok::<(), form_net::FormError>(())
337    /// ```
338    #[must_use]
339    pub const fn name_raw(&self) -> &'a [u8] {
340        self.name
341    }
342
343    /// The value as sent, before decoding.
344    ///
345    /// # Examples
346    ///
347    /// ```
348    /// let p = form_net::raw_pairs(b"a=%41", &form_net::Limits::DEFAULT).next().unwrap()?;
349    /// assert_eq!(p.value_raw(), b"%41");
350    /// # Ok::<(), form_net::FormError>(())
351    /// ```
352    #[must_use]
353    pub const fn value_raw(&self) -> &'a [u8] {
354        self.value
355    }
356
357    /// Whether the sequence contained `=`: `a` and `a=` both have an empty
358    /// value, but only `a=` has an equals sign.
359    ///
360    /// # Examples
361    ///
362    /// ```
363    /// use form_net::Limits;
364    ///
365    /// let mut it = form_net::raw_pairs(b"a&b=", &Limits::DEFAULT);
366    /// assert!(!it.next().unwrap()?.has_equals());
367    /// assert!(it.next().unwrap()?.has_equals());
368    /// # Ok::<(), form_net::FormError>(())
369    /// ```
370    #[must_use]
371    pub const fn has_equals(&self) -> bool {
372        self.flags & HAS_EQUALS != 0
373    }
374
375    /// Whether the name contains `%` or `+`.
376    ///
377    /// # Examples
378    ///
379    /// ```
380    /// let p = form_net::raw_pairs(b"a+b=1", &form_net::Limits::DEFAULT).next().unwrap()?;
381    /// assert!(p.name_needs_decoding() && !p.value_needs_decoding());
382    /// # Ok::<(), form_net::FormError>(())
383    /// ```
384    #[must_use]
385    pub const fn name_needs_decoding(&self) -> bool {
386        self.flags & NAME_DECODE != 0
387    }
388
389    /// Whether the value contains `%` or `+`.
390    ///
391    /// # Examples
392    ///
393    /// ```
394    /// let p = form_net::raw_pairs(b"a=1%20", &form_net::Limits::DEFAULT).next().unwrap()?;
395    /// assert!(p.value_needs_decoding());
396    /// # Ok::<(), form_net::FormError>(())
397    /// ```
398    #[must_use]
399    pub const fn value_needs_decoding(&self) -> bool {
400        self.flags & VALUE_DECODE != 0
401    }
402
403    /// The name with no copy, when it needs no decoding and is valid UTF-8.
404    ///
405    /// # Examples
406    ///
407    /// ```
408    /// let p = form_net::raw_pairs(b"email=x", &form_net::Limits::DEFAULT).next().unwrap()?;
409    /// assert_eq!(p.name_str(), Some("email"));
410    /// # Ok::<(), form_net::FormError>(())
411    /// ```
412    #[must_use]
413    pub fn name_str(&self) -> Option<&'a str> {
414        if self.name_needs_decoding() {
415            None
416        } else {
417            core::str::from_utf8(self.name).ok()
418        }
419    }
420
421    /// The value with no copy, when it needs no decoding and is valid UTF-8.
422    ///
423    /// # Examples
424    ///
425    /// ```
426    /// let p = form_net::raw_pairs(b"q=a+b", &form_net::Limits::DEFAULT).next().unwrap()?;
427    /// assert_eq!(p.value_str(), None);
428    /// # Ok::<(), form_net::FormError>(())
429    /// ```
430    #[must_use]
431    pub fn value_str(&self) -> Option<&'a str> {
432        if self.value_needs_decoding() {
433            None
434        } else {
435            core::str::from_utf8(self.value).ok()
436        }
437    }
438
439    /// Decodes the name into `out` (see [`decode_into`](crate::decode_into)).
440    ///
441    /// # Errors
442    ///
443    /// `Encoding { InvalidPercent | InvalidUtf8 }` located in the body;
444    /// `BufferTooSmall` when `out` is shorter than the raw name.
445    ///
446    /// # Examples
447    ///
448    /// ```
449    /// use form_net::PercentPolicy;
450    ///
451    /// let p = form_net::raw_pairs(b"a%20b=1", &form_net::Limits::DEFAULT).next().unwrap()?;
452    /// let mut out = [0u8; 16];
453    /// assert_eq!(p.decode_name(&mut out, PercentPolicy::Refuse)?, "a b");
454    /// # Ok::<(), form_net::FormError>(())
455    /// ```
456    pub fn decode_name<'b>(
457        &self,
458        out: &'b mut [u8],
459        policy: PercentPolicy,
460    ) -> Result<&'b str, FormError> {
461        decode_component(self.name, out, policy).map_err(|e| e.shifted(self.offset))
462    }
463
464    /// Decodes the value into `out` (see [`decode_into`](crate::decode_into)).
465    ///
466    /// # Errors
467    ///
468    /// As [`decode_name`](Self::decode_name).
469    ///
470    /// # Examples
471    ///
472    /// ```
473    /// use form_net::PercentPolicy;
474    ///
475    /// let p = form_net::raw_pairs(b"q=rust+forms", &form_net::Limits::DEFAULT).next().unwrap()?;
476    /// let mut out = [0u8; 16];
477    /// assert_eq!(p.decode_value(&mut out, PercentPolicy::Refuse)?, "rust forms");
478    /// # Ok::<(), form_net::FormError>(())
479    /// ```
480    pub fn decode_value<'b>(
481        &self,
482        out: &'b mut [u8],
483        policy: PercentPolicy,
484    ) -> Result<&'b str, FormError> {
485        let at = self
486            .offset
487            .saturating_add(self.name.len() as u64)
488            .saturating_add(1);
489        decode_component(self.value, out, policy).map_err(|e| e.shifted(at))
490    }
491
492    /// The byte offset of the sequence in the body.
493    ///
494    /// # Examples
495    ///
496    /// ```
497    /// let mut it = form_net::raw_pairs(b"a=1&bb=2", &form_net::Limits::DEFAULT);
498    /// let _ = it.next();
499    /// assert_eq!(it.next().unwrap()?.offset(), 4);
500    /// # Ok::<(), form_net::FormError>(())
501    /// ```
502    #[must_use]
503    pub const fn offset(&self) -> u64 {
504        self.offset
505    }
506}
507
508#[cfg(test)]
509mod tests {
510    use super::*;
511    use alloc::vec::Vec;
512
513    type Pairs<'a> = Vec<(&'a [u8], &'a [u8])>;
514
515    fn pairs(input: &[u8]) -> Result<Pairs<'_>, FormError> {
516        raw_pairs(input, &Limits::DEFAULT)
517            .map(|p| p.map(|p| (p.name_raw(), p.value_raw())))
518            .collect()
519    }
520
521    #[test]
522    fn test_raw_pairs_whatwg_splitting() {
523        assert_eq!(pairs(b"").unwrap(), []);
524        assert_eq!(pairs(b"&&&").unwrap(), []);
525        assert_eq!(pairs(b"=").unwrap(), [(&b""[..], &b""[..])]);
526        assert_eq!(pairs(b"a==b").unwrap(), [(&b"a"[..], &b"=b"[..])]);
527        assert_eq!(pairs(b"=a").unwrap(), [(&b""[..], &b"a"[..])]);
528        assert_eq!(pairs(b"a;b=c").unwrap(), [(&b"a;b"[..], &b"c"[..])]);
529        assert_eq!(
530            pairs(b"a=1&b=2&").unwrap(),
531            [(&b"a"[..], &b"1"[..]), (&b"b"[..], &b"2"[..])]
532        );
533    }
534
535    #[test]
536    fn test_raw_pairs_flags_per_side() {
537        let p = raw_pairs(b"a%=b", &Limits::DEFAULT)
538            .next()
539            .unwrap()
540            .unwrap();
541        assert!(p.name_needs_decoding() && !p.value_needs_decoding());
542        let p = raw_pairs(b"a=b+%", &Limits::DEFAULT)
543            .next()
544            .unwrap()
545            .unwrap();
546        assert!(!p.name_needs_decoding() && p.value_needs_decoding());
547        let p = raw_pairs(b"a=b=+", &Limits::DEFAULT)
548            .next()
549            .unwrap()
550            .unwrap();
551        assert!(p.value_needs_decoding() && p.has_equals());
552    }
553
554    #[test]
555    fn test_raw_pairs_field_limit_n_and_n_plus_one() {
556        let l = Limits::DEFAULT.with_max_fields(2);
557        assert_eq!(raw_pairs(b"a&b", &l).count(), 2);
558        let r: Result<Vec<_>, _> = raw_pairs(b"a&b&c", &l).collect();
559        assert!(matches!(
560            r,
561            Err(FormError::Limit {
562                limit: Limit::Fields,
563                at
564            }) if at.offset == 4
565        ));
566    }
567
568    #[test]
569    fn test_raw_pairs_length_limits_at_the_edge() {
570        let l = Limits::DEFAULT
571            .with_max_name_bytes(3)
572            .with_max_value_bytes(3);
573        assert!(raw_pairs(b"abc=def", &l).all(|p| p.is_ok()));
574        let e = raw_pairs(b"abcd=1", &l).next().unwrap().unwrap_err();
575        assert!(matches!(
576            e,
577            FormError::Limit {
578                limit: Limit::NameLength,
579                ..
580            }
581        ));
582        let e = raw_pairs(b"a=defg", &l).next().unwrap().unwrap_err();
583        assert!(matches!(e, FormError::Limit { limit: Limit::ValueLength, at } if at.offset == 2));
584        let l = Limits::DEFAULT.with_max_text_bytes(4);
585        let r: Result<Vec<_>, _> = raw_pairs(b"a=12&b=345", &l).collect();
586        assert!(matches!(
587            r,
588            Err(FormError::Limit {
589                limit: Limit::TextTotal,
590                ..
591            })
592        ));
593    }
594
595    #[test]
596    fn test_raw_pairs_body_limit_counts_empty_sequences() {
597        let l = Limits::DEFAULT.with_max_body_bytes(4);
598        assert_eq!(raw_pairs(b"&&&&", &l).count(), 0);
599        let e = raw_pairs(b"&&&&&", &l).next().unwrap().unwrap_err();
600        assert!(matches!(
601            e,
602            FormError::Limit {
603                limit: Limit::Body,
604                ..
605            }
606        ));
607    }
608
609    #[test]
610    fn test_raw_pairs_stops_after_error() {
611        let l = Limits::DEFAULT.with_max_fields(0);
612        let mut it = raw_pairs(b"a&b", &l);
613        assert!(it.next().unwrap().is_err());
614        assert!(it.next().is_none());
615    }
616
617    #[test]
618    fn test_raw_pair_debug_hides_values() {
619        let p = raw_pairs(b"password=hunter2", &Limits::DEFAULT)
620            .next()
621            .unwrap()
622            .unwrap();
623        let s = alloc::format!("{p:?}");
624        assert!(!s.contains("hunter2") && !s.contains("password"));
625    }
626}