util/MessageId.h
| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #pragma once | ||
| 2 | |||
| 3 | #include <QRegularExpression> | ||
| 4 | #include <QString> | ||
| 5 | #include <QStringList> | ||
| 6 | |||
| 7 | // Bounded RFC 5322 Message-ID tokenization shared by the IMAP parser, cache, | ||
| 8 | // and thread builder. Message-ID local parts are case-sensitive in principle; | ||
| 9 | // only the domain is folded. The parser deliberately does not turn arbitrary | ||
| 10 | // prose containing an @ into an identity. | ||
| 11 | namespace MessageId { | ||
| 12 | |||
| 13 | inline constexpr qsizetype MaximumFieldCharacters = 64 * 1024; | ||
| 14 | inline constexpr qsizetype MaximumTokenCharacters = 998; | ||
| 15 | inline constexpr qsizetype MaximumTokens = 100; | ||
| 16 | |||
| 17 | 4065218 | inline QString canonicalize(QString token) { | |
| 18 |
1/1✓ Branch 2 → 3 taken 4065218 times.
|
4065218 | token = token.trimmed(); |
| 19 |
1/1✓ Branch 7 → 8 taken 4065218 times.
|
4065218 | if (token.startsWith(QLatin1Char('<')) && |
| 20 |
8/9✓ Branch 8 → 9 taken 35427 times.
✓ Branch 8 → 16 taken 4029791 times.
✓ Branch 11 → 12 taken 35427 times.
✓ Branch 12 → 13 taken 34485 times.
✓ Branch 12 → 16 taken 942 times.
✓ Branch 14 → 15 taken 34485 times.
✗ Branch 14 → 16 not taken.
✓ Branch 17 → 18 taken 34485 times.
✓ Branch 17 → 25 taken 4030733 times.
|
4065218 | token.endsWith(QLatin1Char('>')) && token.size() >= 2) { |
| 21 |
2/2✓ Branch 19 → 20 taken 34485 times.
✓ Branch 20 → 21 taken 34485 times.
|
34485 | token = token.mid(1, token.size() - 2).trimmed(); |
| 22 | } | ||
| 23 |
2/2✓ Branch 28 → 29 taken 3090668 times.
✓ Branch 28 → 48 taken 3 times.
|
7155889 | if (token.isEmpty() || token.size() > MaximumTokenCharacters || |
| 24 |
6/6✓ Branch 31 → 32 taken 3090668 times.
✓ Branch 32 → 33 taken 3086779 times.
✓ Branch 32 → 48 taken 3889 times.
✓ Branch 35 → 36 taken 3086779 times.
✓ Branch 36 → 37 taken 3084490 times.
✓ Branch 36 → 48 taken 2289 times.
|
3090668 | token.contains(QLatin1Char('<')) || token.contains(QLatin1Char('>')) || |
| 25 |
8/8✓ Branch 26 → 27 taken 3090671 times.
✓ Branch 26 → 48 taken 974547 times.
✓ Branch 39 → 40 taken 3084490 times.
✓ Branch 40 → 41 taken 3083803 times.
✓ Branch 40 → 48 taken 687 times.
✓ Branch 43 → 44 taken 3083803 times.
✓ Branch 44 → 45 taken 3083318 times.
✓ Branch 44 → 48 taken 485 times.
|
10239207 | token.contains(QLatin1Char('\r')) || token.contains(QLatin1Char('\n')) || |
| 26 |
5/5✓ Branch 46 → 47 taken 3083318 times.
✓ Branch 47 → 48 taken 1 time.
✓ Branch 47 → 49 taken 3083317 times.
✓ Branch 50 → 51 taken 981901 times.
✓ Branch 50 → 52 taken 3083317 times.
|
7148536 | token.contains(QChar::Null)) { |
| 27 | 981901 | return {}; | |
| 28 | } | ||
| 29 | |||
| 30 | // The last @ is the domain separator even for an obsolete quoted local | ||
| 31 | // part containing @. CFWS immediately around that separator is syntax-only. | ||
| 32 | 3083317 | const qsizetype at = token.lastIndexOf(QLatin1Char('@')); | |
| 33 |
6/6✓ Branch 55 → 56 taken 3080781 times.
✓ Branch 55 → 58 taken 2536 times.
✓ Branch 57 → 58 taken 295 times.
✓ Branch 57 → 59 taken 3080486 times.
✓ Branch 60 → 61 taken 2831 times.
✓ Branch 60 → 62 taken 3080486 times.
|
3083317 | if (at <= 0 || at >= token.size() - 1) |
| 34 | 2831 | return {}; | |
| 35 |
2/2✓ Branch 62 → 63 taken 3080486 times.
✓ Branch 63 → 64 taken 3080486 times.
|
3080486 | QString local = token.left(at).trimmed(); |
| 36 |
2/2✓ Branch 65 → 66 taken 3080486 times.
✓ Branch 66 → 67 taken 3080486 times.
|
3080486 | QString domain = token.mid(at + 1).trimmed(); |
| 37 | 3080486 | const bool quotedLocal = local.size() >= 2 && | |
| 38 |
5/5✓ Branch 69 → 70 taken 3072509 times.
✓ Branch 69 → 79 taken 7977 times.
✓ Branch 72 → 73 taken 3072509 times.
✓ Branch 73 → 74 taken 34 times.
✓ Branch 73 → 79 taken 3072475 times.
|
3080520 | local.startsWith(QLatin1Char('"')) && |
| 39 |
3/3✓ Branch 76 → 77 taken 34 times.
✓ Branch 77 → 78 taken 12 times.
✓ Branch 77 → 79 taken 22 times.
|
34 | local.endsWith(QLatin1Char('"')); |
| 40 |
4/5✓ Branch 83 → 84 taken 3080486 times.
✗ Branch 83 → 103 not taken.
✓ Branch 85 → 86 taken 3080486 times.
✓ Branch 86 → 87 taken 3080463 times.
✓ Branch 86 → 103 taken 23 times.
|
6160972 | if (local.isEmpty() || domain.isEmpty() || domain.contains(QChar::Space) || |
| 41 |
6/8✓ Branch 81 → 82 taken 3080486 times.
✗ Branch 81 → 103 not taken.
✓ Branch 89 → 90 taken 3080463 times.
✓ Branch 90 → 91 taken 3080462 times.
✓ Branch 90 → 103 taken 1 time.
✓ Branch 93 → 94 taken 3080462 times.
✓ Branch 94 → 95 taken 3080462 times.
✗ Branch 94 → 103 not taken.
|
9241434 | domain.contains(QLatin1Char('\t')) || domain.contains(QLatin1Char('@')) || |
| 42 |
2/2✓ Branch 95 → 96 taken 3080450 times.
✓ Branch 95 → 104 taken 12 times.
|
3080462 | (!quotedLocal && |
| 43 |
8/8✓ Branch 97 → 98 taken 3080450 times.
✓ Branch 98 → 99 taken 3080421 times.
✓ Branch 98 → 103 taken 29 times.
✓ Branch 101 → 102 taken 3080421 times.
✓ Branch 102 → 103 taken 1 time.
✓ Branch 102 → 104 taken 3080420 times.
✓ Branch 105 → 106 taken 54 times.
✓ Branch 105 → 107 taken 3080432 times.
|
6160936 | (local.contains(QChar::Space) || local.contains(QLatin1Char('\t'))))) |
| 44 | 54 | return {}; | |
| 45 |
4/4✓ Branch 107 → 108 taken 3080432 times.
✓ Branch 108 → 109 taken 3080432 times.
✓ Branch 115 → 110 taken 59708606 times.
✓ Branch 115 → 116 taken 3080431 times.
|
62789037 | for (const QChar ch : token) { |
| 46 |
2/2✓ Branch 111 → 112 taken 1 time.
✓ Branch 111 → 114 taken 59708605 times.
|
59708606 | if (ch.category() == QChar::Other_Control) |
| 47 | 1 | return {}; | |
| 48 | } | ||
| 49 |
3/3✓ Branch 116 → 117 taken 3080431 times.
✓ Branch 119 → 120 taken 3080431 times.
✓ Branch 120 → 121 taken 3080431 times.
|
3080431 | return local + QLatin1Char('@') + domain.toLower(); |
| 50 | 3080486 | } | |
| 51 | |||
| 52 | 2512738 | inline QStringList parseTokens(const QString &field) { | |
| 53 |
1/1✓ Branch 2 → 3 taken 2512738 times.
|
2512738 | const QString bounded = field.left(MaximumFieldCharacters); |
| 54 | 2512738 | QStringList result; | |
| 55 |
1/1✓ Branch 4 → 5 taken 2512738 times.
|
2512738 | result.reserve(qMin<qsizetype>(MaximumTokens, 16)); |
| 56 | |||
| 57 | static const QRegularExpression bracketed( | ||
| 58 |
4/7✓ Branch 5 → 6 taken 25 times.
✓ Branch 5 → 18 taken 2512713 times.
✓ Branch 7 → 8 taken 25 times.
✗ Branch 7 → 18 not taken.
✓ Branch 12 → 13 taken 25 times.
✗ Branch 86 → 87 not taken.
✗ Branch 86 → 88 not taken.
|
2513349 | QStringLiteral(R"(<\s*([^<>\r\n]{1,998}?)\s*>)")); |
| 59 |
1/1✓ Branch 19 → 20 taken 2512738 times.
|
2512738 | auto matches = bracketed.globalMatch(bounded); |
| 60 |
7/7✓ Branch 31 → 32 taken 2513543 times.
✓ Branch 32 → 33 taken 806 times.
✓ Branch 32 → 36 taken 2512737 times.
✓ Branch 34 → 35 taken 805 times.
✓ Branch 34 → 36 taken 1 time.
✓ Branch 37 → 21 taken 805 times.
✓ Branch 37 → 38 taken 2512738 times.
|
2513543 | while (matches.hasNext() && result.size() < MaximumTokens) { |
| 61 |
3/3✓ Branch 21 → 22 taken 805 times.
✓ Branch 22 → 23 taken 805 times.
✓ Branch 23 → 24 taken 805 times.
|
805 | const QString token = canonicalize(matches.next().captured(1)); |
| 62 |
2/2✓ Branch 27 → 28 taken 223 times.
✓ Branch 27 → 29 taken 582 times.
|
805 | if (!token.isEmpty()) |
| 63 |
1/1✓ Branch 28 → 29 taken 223 times.
|
223 | result.append(token); |
| 64 | 805 | } | |
| 65 |
2/2✓ Branch 39 → 40 taken 107 times.
✓ Branch 39 → 41 taken 2512631 times.
|
2512738 | if (!result.isEmpty()) |
| 66 | 107 | return result; | |
| 67 | |||
| 68 | // ENVELOPE implementations sometimes return a bare msg-id. Accept it only | ||
| 69 | // when the complete field, apart from surrounding comments/CFWS, is one ID. | ||
| 70 | 2512631 | QString bare = bounded; | |
| 71 |
4/7✓ Branch 42 → 43 taken 23 times.
✓ Branch 42 → 55 taken 2512608 times.
✓ Branch 44 → 45 taken 23 times.
✗ Branch 44 → 55 not taken.
✓ Branch 49 → 50 taken 23 times.
✗ Branch 106 → 107 not taken.
✗ Branch 106 → 108 not taken.
|
2512654 | static const QRegularExpression comments(QStringLiteral(R"(\([^()]*\))")); |
| 72 |
7/7✓ Branch 58 → 59 taken 2536355 times.
✓ Branch 58 → 64 taken 3266 times.
✓ Branch 61 → 62 taken 2536355 times.
✓ Branch 62 → 63 taken 26990 times.
✓ Branch 62 → 64 taken 2509365 times.
✓ Branch 65 → 56 taken 26990 times.
✓ Branch 65 → 66 taken 2512631 times.
|
2539621 | for (int pass = 0; pass < 8 && bare.contains(QLatin1Char('(')); ++pass) |
| 73 |
1/1✓ Branch 56 → 57 taken 26990 times.
|
26990 | bare.remove(comments); |
| 74 |
1/1✓ Branch 67 → 68 taken 2512631 times.
|
2512631 | const QString token = canonicalize(bare); |
| 75 |
2/2✓ Branch 70 → 71 taken 1724622 times.
✓ Branch 70 → 72 taken 788009 times.
|
2512631 | if (!token.isEmpty()) |
| 76 |
1/1✓ Branch 71 → 72 taken 1724622 times.
|
1724622 | result.append(token); |
| 77 | 2512631 | return result; | |
| 78 | 2512738 | } | |
| 79 | |||
| 80 | 1555666 | inline QString canonicalFirst(const QString &field) { | |
| 81 | // Fast paths for the two shapes almost every header has — a bare id or | ||
| 82 | // one bracketed id — with the result parseTokens() would give: no | ||
| 83 | // comments to strip, nothing for the bracket scan to find besides the | ||
| 84 | // whole field (Sprint 101 S101.0.6: the regex dominated the summary pass). | ||
| 85 |
1/1✓ Branch 2 → 3 taken 1555666 times.
|
1555666 | const QString bounded = field.left(MaximumFieldCharacters); |
| 86 |
4/4✓ Branch 5 → 6 taken 1555666 times.
✓ Branch 9 → 10 taken 1553603 times.
✓ Branch 10 → 11 taken 1518040 times.
✓ Branch 10 → 16 taken 35563 times.
|
3109269 | if (!bounded.contains(QLatin1Char('(')) && !bounded.contains(QLatin1Char('>')) |
| 87 |
7/7✓ Branch 6 → 7 taken 1553603 times.
✓ Branch 6 → 16 taken 2063 times.
✓ Branch 13 → 14 taken 1518040 times.
✓ Branch 14 → 15 taken 1517351 times.
✓ Branch 14 → 16 taken 689 times.
✓ Branch 17 → 18 taken 1517351 times.
✓ Branch 17 → 23 taken 38315 times.
|
3109269 | && !bounded.contains(QLatin1Char('<'))) |
| 88 |
1/1✓ Branch 19 → 20 taken 1517351 times.
|
1517351 | return canonicalize(bounded); |
| 89 |
1/1✓ Branch 23 → 24 taken 38315 times.
|
38315 | const QString trimmed = bounded.trimmed(); |
| 90 |
4/4✓ Branch 26 → 27 taken 38315 times.
✓ Branch 30 → 31 taken 34907 times.
✓ Branch 31 → 32 taken 34444 times.
✓ Branch 31 → 56 taken 463 times.
|
73222 | if (trimmed.startsWith(QLatin1Char('<')) && trimmed.endsWith(QLatin1Char('>')) |
| 91 |
4/4✓ Branch 35 → 36 taken 34425 times.
✓ Branch 35 → 56 taken 19 times.
✓ Branch 40 → 41 taken 34419 times.
✓ Branch 40 → 56 taken 6 times.
|
34444 | && trimmed.indexOf(QLatin1Char('<'), 1) < 0 && trimmed.indexOf(QLatin1Char('>')) == trimmed.size() - 1 |
| 92 |
6/6✓ Branch 43 → 44 taken 34419 times.
✓ Branch 44 → 45 taken 34412 times.
✓ Branch 44 → 56 taken 7 times.
✓ Branch 47 → 48 taken 34412 times.
✓ Branch 48 → 49 taken 34409 times.
✓ Branch 48 → 56 taken 3 times.
|
34419 | && !trimmed.contains(QLatin1Char('(')) && !trimmed.contains(QLatin1Char('\r')) |
| 93 |
9/9✓ Branch 27 → 28 taken 34907 times.
✓ Branch 27 → 56 taken 3408 times.
✓ Branch 51 → 52 taken 34409 times.
✓ Branch 52 → 53 taken 34408 times.
✓ Branch 52 → 56 taken 1 time.
✓ Branch 54 → 55 taken 34406 times.
✓ Branch 54 → 56 taken 2 times.
✓ Branch 57 → 58 taken 34406 times.
✓ Branch 57 → 63 taken 3909 times.
|
73222 | && !trimmed.contains(QLatin1Char('\n')) && trimmed.size() <= MaximumTokenCharacters + 2) |
| 94 |
1/1✓ Branch 59 → 60 taken 34406 times.
|
34406 | return canonicalize(trimmed); |
| 95 |
1/1✓ Branch 63 → 64 taken 3909 times.
|
3909 | const QStringList tokens = parseTokens(field); |
| 96 |
2/2✓ Branch 65 → 66 taken 3836 times.
✓ Branch 65 → 67 taken 73 times.
|
3909 | return tokens.isEmpty() ? QString() : tokens.first(); |
| 97 | 1555666 | } | |
| 98 | |||
| 99 | } // namespace MessageId | ||
| 100 |