rpz

tests.rs (19161B)

---

 use super::{
     Adblock, DomainOnly, FirefoxDomainErr, Hosts, ParsedDomain as _, RpzDomain, Value, Wildcard,
 };
 use ascii_domain::dom::DomainErr;
 use num_bigint::BigUint;
 use superset_map::SupersetSet;
 #[test]
 fn adblock_parse() {
     // Test subdomains.
     assert!(
         Adblock::parse_value("||www.example.com").is_ok_and(|val| match val {
             Value::Domain(ref dom) => dom.subdomains && dom.domain.as_bytes() == b"www.example.com",
             Value::Comment(_) | Value::Blank => false,
         })
     );
     // Test whitespace and '^' removal.
     assert!(
         Adblock::parse_value("   \t\t  ||\t\t  \twww.example.com   \t\t  ^  \t\t  ").is_ok_and(
             |val| match val {
                 Value::Domain(ref dom) =>
                     dom.subdomains && dom.domain.as_bytes() == b"www.example.com",
                 Value::Comment(_) | Value::Blank => false,
             }
         )
     );
     assert!(
         Adblock::parse_value("\t\t  \twww.example.com   \t\t   \t\t  ").is_ok_and(|val| {
             match val {
                 Value::Domain(ref dom) => {
                     !dom.subdomains && dom.domain.as_bytes() == b"www.example.com"
                 }
                 Value::Comment(_) | Value::Blank => false,
             }
         })
     );
     assert!(Adblock::parse_value("www .example.com").is_err_and(
             |err| err == FirefoxDomainErr::InvalidDomain(DomainErr::InvalidByte(b' ')),
         ));
     assert!(Adblock::parse_value("||www.ExAMPle.COm").is_ok_and(|val| {
         match val {
             Value::Domain(ref dom) => {
                 Adblock::parse_value("||www.example.com").is_ok_and(|val_2| match val_2 {
                     Value::Domain(ref dom2) => {
                         dom == dom2 && dom.subdomains && dom2.subdomains && dom.cmp(dom2).is_eq()
                     }
                     Value::Comment(_) | Value::Blank => false,
                 })
             }
             Value::Comment(_) | Value::Blank => false,
         }
     }));
     // Test comment
     assert!(
         Adblock::parse_value("  \t\t   #hi").is_ok_and(|val| match val {
             Value::Comment(com) => com == "hi",
             Value::Domain(_) | Value::Blank => false,
         })
     );
     assert!(
         Adblock::parse_value("  \t\t   !!  foo").is_ok_and(|val| match val {
             Value::Comment(com) => com == "!  foo",
             Value::Domain(_) | Value::Blank => false,
         })
     );
     // Test blank
     assert!(Adblock::parse_value("  \t\t   ").is_ok_and(|val| matches!(val, Value::Blank)));
 }
 #[test]
 fn domain_only_parse_value() {
     // Test whitespace and comment.
     assert!(
         DomainOnly::parse_value("   \t\t  \t\t  \twww.example.com#asdflkj asdf alskdfj  ")
             .is_ok_and(|val| match val {
                 Value::Domain(ref dom) => dom.domain.as_bytes() == b"www.example.com",
                 Value::Comment(_) | Value::Blank => false,
             })
     );
     assert!(
         DomainOnly::parse_value("   \t\t  \t\t  \twww.example.com   \t\t  ^  \t\t  ")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidDomain(DomainErr::InvalidByte(b' ')))
     );
     // Test case-insensitivity.
     assert!(
         DomainOnly::parse_value("www.ExAMPle.CoM").is_ok_and(|val| match val {
             Value::Domain(ref dom) =>
                 DomainOnly::parse_value("www.example.com").is_ok_and(|val2| match val2 {
                     Value::Domain(ref dom2) => dom.cmp(dom2).is_eq(),
                     Value::Comment(_) | Value::Blank => false,
                 }),
             Value::Comment(_) | Value::Blank => false,
         })
     );
     // Test comment.
     assert!(
         DomainOnly::parse_value("   \t\t  \t\t  \t # hi").is_ok_and(|val| match val {
             Value::Comment(com) => com == " hi",
             Value::Domain(_) | Value::Blank => false,
         })
     );
     // Test blank.
     assert!(
         DomainOnly::parse_value("   \t\t  \t\t  \t ").is_ok_and(|val| matches!(val, Value::Blank))
     );
     // Test blank.
     assert!(
         DomainOnly::parse_value("example.xn--abc").is_ok_and(|val| matches!(val, Value::Domain(_)))
     );
     // Test invalid TLD.
     assert!(
         DomainOnly::parse_value("www.c1m").is_err_and(|err| err == FirefoxDomainErr::InvalidTld)
     );
 }
 #[test]
 fn hosts_parse_value() {
     // Test whitespace and comment.
     assert!(
         Hosts::parse_value("   \t\t  127.0.0.1\t\t  \twww.example.com#asdflkj asdf alskdfj  ")
             .is_ok_and(|val| match val {
                 Value::Domain(ref dom) => dom.domain.as_bytes() == b"www.example.com",
                 Value::Comment(_) | Value::Blank => false,
             })
     );
     assert!(
         Hosts::parse_value("   \t\t  0.0.0.0\t\t  \twww.example.com   \t\t  ^  \t\t  ")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidDomain(DomainErr::InvalidByte(b' ')))
     );
     assert!(
         Hosts::parse_value("::1\twww .example.com")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidDomain(DomainErr::InvalidByte(b' ')))
     );
     // Test invalid IP
     assert!(
         Hosts::parse_value("::2 www.example.com")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidHostsIP)
     );
     assert!(
         Hosts::parse_value(":2 www.example.com")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidHostsIP)
     );
     assert!(
         Hosts::parse_value("www.example.com").is_err_and(|e| e == FirefoxDomainErr::InvalidHostsIP)
     );
     assert!(
         Hosts::parse_value("10.4.2.256 www.example.com")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidHostsIP)
     );
     // Test case-insensitivity.
     assert!(
         Hosts::parse_value(":: www.ExAMPle.Com").is_ok_and(|val| match val {
             Value::Domain(ref dom) =>
                 Hosts::parse_value("127.0.0.1 www.example.com").is_ok_and(|val2| match val2 {
                     Value::Domain(ref dom2) => dom.cmp(dom2).is_eq(),
                     Value::Comment(_) | Value::Blank => false,
                 }),
             Value::Comment(_) | Value::Blank => false,
         })
     );
     // Test comment.
     assert!(
         Hosts::parse_value("   \t\t  \t\t  \t # hi").is_ok_and(|val| match val {
             Value::Comment(com) => com == " hi",
             Value::Domain(_) | Value::Blank => false,
         })
     );
     // Test blank.
     assert!(Hosts::parse_value("   \t\t  \t\t  \t ").is_ok_and(|val| matches!(val, Value::Blank)));
 }
 #[test]
 fn wildcard_parse_value() {
     // Test bad asterisk.
     assert!(
         Wildcard::parse_value("*")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidDomain(DomainErr::InvalidByte(b'*')))
     );
     assert!(
         Wildcard::parse_value("www*.example.com")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidDomain(DomainErr::InvalidByte(b'*')))
     );
     assert!(
         Wildcard::parse_value("www.*.com")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidDomain(DomainErr::InvalidByte(b'*')))
     );
     assert!(
         Wildcard::parse_value("*..com")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidDomain(DomainErr::EmptyLabel))
     );
     assert!(
         Wildcard::parse_value("www.com*")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidDomain(DomainErr::InvalidByte(b'*')))
     );
     assert!(
         Wildcard::parse_value("ww*w.com")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidDomain(DomainErr::InvalidByte(b'*')))
     );
     // Test case-insensitivity.
     assert!(
         Wildcard::parse_value("*.wWw.ExamPLE.com").is_ok_and(|val| match val {
             Value::Domain(ref dom) =>
                 Wildcard::parse_value("*.www.example.com").is_ok_and(|val2| match val2 {
                     Value::Domain(ref dom2) =>
                         dom.cmp(dom2).is_eq()
                             && dom == dom2
                             && dom.proper_subdomains
                             && dom2.proper_subdomains,
                     Value::Comment(_) | Value::Blank => false,
                 }),
             Value::Comment(_) | Value::Blank => false,
         })
     );
     // Test proper subdomains.
     assert!(
         Wildcard::parse_value("*.www.example.com").is_ok_and(|val| match val {
             Value::Domain(ref dom) =>
                 dom.domain.as_bytes() == b"www.example.com" && dom.proper_subdomains,
             Value::Comment(_) | Value::Blank => false,
         })
     );
     // Test comment.
     assert!(
         Wildcard::parse_value("   \t\t  \t\t  \t*.www.example.com#asdflkj asdf alskdfj  ")
             .is_ok_and(|val| match val {
                 Value::Domain(ref dom) =>
                     dom.domain.as_bytes() == b"www.example.com" && dom.proper_subdomains,
                 Value::Comment(_) | Value::Blank => false,
             })
     );
     assert!(
         Wildcard::parse_value("   \t\t  \t\t  \twww.example.com #asdflkj asdf alskdfj  ")
             .is_ok_and(|val| match val {
                 Value::Domain(ref dom) =>
                     dom.domain.as_bytes() == b"www.example.com" && !dom.proper_subdomains,
                 Value::Comment(_) | Value::Blank => false,
             })
     );
     // Test whitespace removal.
     assert!(
         Wildcard::parse_value("   \t\t  *.www.example.com   \t\t \t  ").is_ok_and(|val| {
             match val {
                 Value::Domain(ref dom) => {
                     dom.domain.as_bytes() == b"www.example.com" && dom.proper_subdomains
                 }
                 Value::Comment(_) | Value::Blank => false,
             }
         })
     );
     assert!(
         Wildcard::parse_value("\t\t  \twww.example.com   \t\t   \t\t  ").is_ok_and(|val| {
             match val {
                 Value::Domain(ref dom) => {
                     dom.domain.as_bytes() == b"www.example.com" && !dom.proper_subdomains
                 }
                 Value::Comment(_) | Value::Blank => false,
             }
         })
     );
     assert!(
         Wildcard::parse_value("www .example.com")
             .is_err_and(|e| e == FirefoxDomainErr::InvalidDomain(DomainErr::InvalidByte(b' ')))
     );
     // Test 127 labels after wildcard error.
     assert!(Wildcard::parse_value("*.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a").is_err_and(|e| e == FirefoxDomainErr::InvalidWildcardDomain));
     // Test 126 labels after wildcard is ok.
     assert!(Wildcard::parse_value("*.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a").is_ok_and(|val| match val {
             Value::Domain(ref dom) => dom.domain.iter().count() == 126 && dom.proper_subdomains,
             Value::Comment(_) | Value::Blank => false,
         }));
     // Test comment.
     assert!(
         Wildcard::parse_value("   \t\t  \t\t  \t # hi").is_ok_and(|val| match val {
             Value::Comment(com) => com == " hi",
             Value::Domain(_) | Value::Blank => false,
         })
     );
     // Test blank.
     assert!(
         Wildcard::parse_value("   \t\t  \t\t  \t ").is_ok_and(|val| matches!(val, Value::Blank))
     );
 }
 #[test]
 fn rpz_parse_value() {
     assert!(
         RpzDomain::parse_value("*.www.example.com").is_ok_and(|val| {
             let dom = val.unwrap_domain();
             dom.is_proper_subdomains() && dom.domain().as_bytes() == b"www.example.com"
         })
     );
     assert!(
         RpzDomain::parse_value("||www.example.com").is_ok_and(|val| {
             let dom = val.unwrap_domain();
             dom.is_subdomains() && dom.domain().as_bytes() == b"www.example.com"
         })
     );
     assert!(
         RpzDomain::parse_value("0.0.0.0 www.example.com").is_ok_and(|val| {
             let dom = val.unwrap_domain();
             !(dom.is_subdomains() || dom.is_proper_subdomains())
                 && dom.domain().as_bytes() == b"www.example.com"
         })
     );
     assert!(RpzDomain::parse_value("www.example.com").is_ok_and(|val| {
         let dom = val.unwrap_domain();
         !(dom.is_subdomains() || dom.is_proper_subdomains())
             && dom.domain().as_bytes() == b"www.example.com"
     }));
     // Test case-insensitivity.
     assert!(
         RpzDomain::parse_value("*.Www.ExaMPle.COm").is_ok_and(|val| {
             let dom = val.unwrap_domain();
             RpzDomain::parse_value("*.www.example.com").is_ok_and(|val2| {
                 let dom2 = val2.unwrap_domain();
                 dom.is_proper_subdomains()
                     && dom2.is_proper_subdomains()
                     && dom == dom2
                     && dom.cmp(&dom2).is_eq()
             })
         })
     );
     // Test comment.
     assert!(
         RpzDomain::parse_value("   \t\t  \t\t  \t # hi").is_ok_and(|val| match val {
             Value::Comment(com) => com == " hi",
             Value::Domain(_) | Value::Blank => false,
         })
     );
     assert!(
         RpzDomain::parse_value("   \t\t  \t\t  \t ! hi").is_ok_and(|val| match val {
             Value::Comment(com) => com == " hi",
             Value::Domain(_) | Value::Blank => false,
         })
     );
     // Test blank.
     assert!(
         RpzDomain::parse_value("   \t\t  \t\t  \t ").is_ok_and(|val| matches!(val, Value::Blank))
     );
 }
 #[expect(clippy::expect_used, clippy::unwrap_in_result, reason = "OK in tests")]
 #[expect(clippy::indexing_slicing, reason = "comments justify correctness")]
 #[expect(clippy::nonminimal_bool, reason = "want to test all ord ops")]
 #[test]
 fn rpz_ord_and_eq() -> Result<(), &'static str> {
     "www.bar.com,*.www.bar.com,||www.bar.com,*.bar.com,||bar.com,Example.com,WwW.exaMple.com,*.www.example.com,||www.example.com,*.example.com,||example.com,FOo.coM,Www.foo.com,*.foo.com,*.coM,example.net,*.net".split(',').try_fold(RpzDomain::DomainOnly(DomainOnly::parse_value("bar.com").expect("bug in DomainOnly::parse_value").unwrap_domain()), |prev, slice| {
             let cur = if slice.as_bytes()[0] == b'|' {
                 RpzDomain::Adblock(Adblock::parse_value(slice).expect("Bug in Adblock::parse_value").unwrap_domain())
             } else {
                 RpzDomain::Wildcard(Wildcard::parse_value(slice).expect("Bug in Wildcard::parse_value").unwrap_domain())
             };
             if prev < cur && cur > prev && prev == prev && cur == cur {
                 Ok(cur)
             } else {
                 Err("PartialEq or Ord are not correctly implemented for RpzDomain.")
             }
         }).map(|_| ())
 }
 #[expect(clippy::expect_used, reason = "OK in tests")]
 #[expect(clippy::indexing_slicing, reason = "comments justify correctness")]
 #[test]
 fn superset_set() {
     let mut iter = "*.NeT,*.net,www.bar.com,*.net,*.www.bar.com,||www.bar.com,*.bar.com,||bar.com,example.com,www.example.com,*.www.example.com,||www.example.com,*.example.com,||example.com,foo.com,www.foo.com,*.foo.com,*.com,example.net,*.abc.abc,||aawww.abc,abc.abc".split(',').fold(SupersetSet::new(), |mut doms, slice| {
             _ = doms.insert(if slice.as_bytes()[0] == b'|' {
                 RpzDomain::Adblock(Adblock::parse_value(slice).expect("Bug in Adblock::parse_value").unwrap_domain())
             } else {
                 RpzDomain::Wildcard(Wildcard::parse_value(slice).expect("Bug in Wildcard::parse_value").unwrap_domain())
             });
             doms
         }).into_iter();
     assert!(
         iter.next()
             .is_some_and(|d| { d.domain().as_bytes() == b"aawww.abc" && d.is_subdomains() })
     );
     assert!(
         iter.next()
             .is_some_and(|d| { d.domain().as_bytes() == b"abc.abc" && d.is_domain() })
     );
     assert!(
         iter.next()
             .is_some_and(|d| { d.domain().as_bytes() == b"abc.abc" && d.is_proper_subdomains() })
     );
     assert!(
         iter.next()
             .is_some_and(|d| { d.domain().as_bytes() == b"com" && d.is_proper_subdomains() })
     );
     assert!(
         iter.next()
             .is_some_and(|d| { d.domain().as_bytes() == b"NeT" && d.is_proper_subdomains() })
     );
     assert!(iter.next().is_none());
 }
 #[test]
 fn card() {
     // Geometric series.
     // We can have two labels each with one character,
     // one label with one to three characters, or 0 labels.
     // This is 1 + 52 + 52^2 + 52^3 + 52^2 = (1-52^4)/(1-52) + 52^2 = (52^4 - 1)/51 + 52^2 = 146069.
     assert!(Adblock::parse_value("||a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a").is_ok_and(|val| { let dom = val.unwrap_domain(); dom.domain.len().get() == 249 && dom.domain.iter().count() == 125 && dom.domain_count() == BigUint::new(vec![146_069]) }));
     // A subdomain of length 252 or 253 gets converted to a domain.
     assert!(Adblock::parse_value("||a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a.a").is_ok_and(|val| { let dom = val.unwrap_domain(); dom.domain.iter().count() == 127 && !dom.subdomains && dom.domain_count() == BigUint::new(vec![1]) }));
     // Pre-calculated manually.
     // This is the number of domains possible between 2 and 252 characters.
     assert!(Wildcard::parse_value("*.a").is_ok_and(|val| {
         val.unwrap_domain().domain_count()
             == BigUint::new(vec![
                 375_288_404,
                 2_460_223_985,
                 1_334_358_771,
                 2_543_621_408,
                 2_519_466_280,
                 1_133_682_239,
                 3_589_178_618,
                 348_125_705,
                 1_709_233_643,
                 958_334_503,
                 3_780_539_710,
                 2_181_893_897,
                 2_457_156_833,
                 3_204_765_645,
                 2_728_103_430,
                 1_817_547_150,
                 3_102_358_416,
                 444_185_044,
                 3_659_003_776,
                 10_341_713,
                 306_326_206,
                 1_336_386_425,
                 3_942_332_649,
                 2_036_577_878,
                 2_460_939_277,
                 3_976_861_337,
                 2_101_094_571,
                 2_241_770_079,
                 2_667_853_164,
                 3_687_350_273,
                 109_356_153,
                 3_455_569_358,
                 2_333_076_459,
                 2_433_207_896,
                 1_553_903_141,
                 2_621_943_843,
                 4_223_295_645,
                 1_753_858_368,
                 130_924_388,
                 965_594_304,
                 3_942_586_845,
                 1_573_844_087,
                 4_237_886_128,
                 481_383_133,
                 56_931_017,
             ])
     }));
 }