diff options
| author | 2024-04-14 18:10:28 +0200 | |
|---|---|---|
| committer | 2024-04-14 18:10:28 +0200 | |
| commit | 49db791a43fdf4afecc27336d84ca6ab780b7f77 (patch) | |
| tree | 32caf521bef501420e75ace3939d4aa47f827bdc /query.zig | |
| parent | Merge pull request #154 from JacobCrabill/dev/zig-0.12-modules (diff) | |
| download | zig-sqlite-49db791a43fdf4afecc27336d84ca6ab780b7f77.tar.gz zig-sqlite-49db791a43fdf4afecc27336d84ca6ab780b7f77.tar.xz zig-sqlite-49db791a43fdf4afecc27336d84ca6ab780b7f77.zip | |
fix for latest zig
Also some major refactoring around the query parsing state.
Diffstat (limited to 'query.zig')
| -rw-r--r-- | query.zig | 332 |
1 files changed, 173 insertions, 159 deletions
| @@ -20,168 +20,178 @@ fn isNamedIdentifierChar(c: u8) bool { | |||
| 20 | } | 20 | } |
| 21 | 21 | ||
| 22 | fn bindMarkerForName(comptime markers: []const BindMarker, comptime name: []const u8) ?BindMarker { | 22 | fn bindMarkerForName(comptime markers: []const BindMarker, comptime name: []const u8) ?BindMarker { |
| 23 | for (markers[0..]) |marker| { | 23 | for (markers) |marker| { |
| 24 | if (marker.name != null and std.mem.eql(u8, marker.name.?, name)) | 24 | if (marker.name != null and std.mem.eql(u8, marker.name.?, name)) |
| 25 | return marker; | 25 | return marker; |
| 26 | } | 26 | } |
| 27 | return null; | 27 | return null; |
| 28 | } | 28 | } |
| 29 | 29 | ||
| 30 | pub fn ParsedQuery(comptime query: []const u8) ParsedQueryState(query.len) { | 30 | pub fn ParsedQuery(comptime tmp_query: []const u8) type { |
| 31 | // This contains the final SQL query after parsing with our | 31 | return struct { |
| 32 | // own typed bind markers removed. | 32 | const Self = @This(); |
| 33 | comptime var buf: [query.len]u8 = undefined; | ||
| 34 | comptime var pos = 0; | ||
| 35 | comptime var state = .start; | ||
| 36 | |||
| 37 | comptime var current_bind_marker_type: [256]u8 = undefined; | ||
| 38 | comptime var current_bind_marker_type_pos = 0; | ||
| 39 | |||
| 40 | // becomes part of our result | ||
| 41 | comptime var bind_markers: [128]BindMarker = undefined; | ||
| 42 | comptime var nb_bind_markers: usize = 0; | ||
| 43 | |||
| 44 | // used for capturing slices, such as bind parameter name | ||
| 45 | comptime var hold_pos = 0; | ||
| 46 | |||
| 47 | inline for (query) |c| { | ||
| 48 | switch (state) { | ||
| 49 | .start => switch (c) { | ||
| 50 | '?', ':', '@', '$' => { | ||
| 51 | bind_markers[nb_bind_markers] = BindMarker{}; | ||
| 52 | current_bind_marker_type_pos = 0; | ||
| 53 | state = .bind_marker; | ||
| 54 | buf[pos] = c; | ||
| 55 | pos += 1; | ||
| 56 | }, | ||
| 57 | '\'', '"', '[', '`' => { | ||
| 58 | state = .inside_string; | ||
| 59 | buf[pos] = c; | ||
| 60 | pos += 1; | ||
| 61 | }, | ||
| 62 | else => { | ||
| 63 | buf[pos] = c; | ||
| 64 | pos += 1; | ||
| 65 | }, | ||
| 66 | }, | ||
| 67 | .inside_string => switch (c) { | ||
| 68 | '\'', '"', ']', '`' => { | ||
| 69 | state = .start; | ||
| 70 | buf[pos] = c; | ||
| 71 | pos += 1; | ||
| 72 | }, | ||
| 73 | else => { | ||
| 74 | buf[pos] = c; | ||
| 75 | pos += 1; | ||
| 76 | }, | ||
| 77 | }, | ||
| 78 | .bind_marker => switch (c) { | ||
| 79 | '?', ':', '@', '$' => @compileError("invalid multiple '?', ':', '$' or '@'."), | ||
| 80 | '{' => { | ||
| 81 | state = .bind_marker_type; | ||
| 82 | }, | ||
| 83 | else => { | ||
| 84 | if (isNamedIdentifierChar(c)) { | ||
| 85 | // This is the start of a named bind marker. | ||
| 86 | state = .bind_marker_identifier; | ||
| 87 | hold_pos = pos + 1; | ||
| 88 | } else { | ||
| 89 | // This is a unnamed, untyped bind marker. | ||
| 90 | state = .start; | ||
| 91 | |||
| 92 | bind_markers[nb_bind_markers].typed = null; | ||
| 93 | nb_bind_markers += 1; | ||
| 94 | } | ||
| 95 | buf[pos] = c; | ||
| 96 | pos += 1; | ||
| 97 | }, | ||
| 98 | }, | ||
| 99 | .bind_marker_identifier => switch (c) { | ||
| 100 | '?', ':', '@', '$' => @compileError("unregconised multiple '?', ':', '$' or '@'."), | ||
| 101 | '{' => { | ||
| 102 | state = .bind_marker_type; | ||
| 103 | current_bind_marker_type_pos = 0; | ||
| 104 | }, | ||
| 105 | else => { | ||
| 106 | if (!isNamedIdentifierChar(c)) { | ||
| 107 | // This marks the end of the named bind marker. | ||
| 108 | state = .start; | ||
| 109 | const name = buf[hold_pos .. pos - 1]; | ||
| 110 | if (bindMarkerForName(bind_markers[0..nb_bind_markers], name) == null) { | ||
| 111 | bind_markers[nb_bind_markers].name = name; | ||
| 112 | nb_bind_markers += 1; | ||
| 113 | } | ||
| 114 | } | ||
| 115 | buf[pos] = c; | ||
| 116 | pos += 1; | ||
| 117 | }, | ||
| 118 | }, | ||
| 119 | .bind_marker_type => switch (c) { | ||
| 120 | '}' => { | ||
| 121 | state = .start; | ||
| 122 | |||
| 123 | const type_info_string = current_bind_marker_type[0..current_bind_marker_type_pos]; | ||
| 124 | // Handles optional types | ||
| 125 | const typ = if (type_info_string[0] == '?') blk: { | ||
| 126 | const child_type = ParseType(type_info_string[1..]); | ||
| 127 | break :blk @Type(std.builtin.Type{ | ||
| 128 | .Optional = .{ | ||
| 129 | .child = child_type, | ||
| 130 | }, | ||
| 131 | }); | ||
| 132 | } else blk: { | ||
| 133 | break :blk ParseType(type_info_string); | ||
| 134 | }; | ||
| 135 | |||
| 136 | bind_markers[nb_bind_markers].typed = typ; | ||
| 137 | nb_bind_markers += 1; | ||
| 138 | }, | ||
| 139 | else => { | ||
| 140 | current_bind_marker_type[current_bind_marker_type_pos] = c; | ||
| 141 | current_bind_marker_type_pos += 1; | ||
| 142 | }, | ||
| 143 | }, | ||
| 144 | else => { | ||
| 145 | @compileError("invalid state " ++ @tagName(state)); | ||
| 146 | }, | ||
| 147 | } | ||
| 148 | } | ||
| 149 | |||
| 150 | // The last character was a bind marker prefix so this must be an untyped bind marker. | ||
| 151 | switch (state) { | ||
| 152 | .bind_marker => { | ||
| 153 | bind_markers[nb_bind_markers].typed = null; | ||
| 154 | nb_bind_markers += 1; | ||
| 155 | }, | ||
| 156 | .bind_marker_identifier => { | ||
| 157 | nb_bind_markers += 1; | ||
| 158 | }, | ||
| 159 | .start => {}, | ||
| 160 | else => @compileError("invalid final state " ++ @tagName(state) ++ ", this means you wrote an incomplete bind marker type"), | ||
| 161 | } | ||
| 162 | 33 | ||
| 163 | var parsed_state = ParsedQueryState(query.len){ | 34 | const result = parse(); |
| 164 | .bind_markers = bind_markers, | ||
| 165 | .nb_bind_markers = nb_bind_markers, | ||
| 166 | .query = undefined, | ||
| 167 | .query_len = pos, | ||
| 168 | }; | ||
| 169 | 35 | ||
| 170 | std.mem.copyForwards(u8, &parsed_state.query, &buf); | 36 | pub const bind_markers = result.bind_markers[0..result.bind_markers_len]; |
| 171 | 37 | ||
| 172 | return parsed_state; | 38 | pub fn getQuery() []const u8 { |
| 173 | } | 39 | return Self.result.query[0..Self.result.query_len]; |
| 40 | } | ||
| 174 | 41 | ||
| 175 | pub fn ParsedQueryState(comptime max_query_len: usize) type { | 42 | const ParsedQueryResult = struct { |
| 176 | return struct { | 43 | bind_markers: [128]BindMarker, |
| 177 | const Self = @This(); | 44 | bind_markers_len: usize, |
| 178 | bind_markers: [128]BindMarker, | 45 | query: [tmp_query.len]u8, |
| 179 | nb_bind_markers: usize, | 46 | query_len: usize, |
| 180 | query: [max_query_len]u8, | 47 | }; |
| 181 | query_len: usize, | ||
| 182 | 48 | ||
| 183 | pub fn getQuery(comptime self: *const Self) []const u8 { | 49 | fn parse() ParsedQueryResult { |
| 184 | return self.query[0..self.query_len]; | 50 | // This contains the final SQL query after parsing with our |
| 51 | // own typed bind markers removed. | ||
| 52 | var buf: [tmp_query.len]u8 = undefined; | ||
| 53 | var pos = 0; | ||
| 54 | var state = .start; | ||
| 55 | |||
| 56 | var current_bind_marker_type: [256]u8 = undefined; | ||
| 57 | var current_bind_marker_type_pos = 0; | ||
| 58 | |||
| 59 | // becomes part of our result | ||
| 60 | var tmp_bind_markers: [128]BindMarker = undefined; | ||
| 61 | var nb_tmp_bind_markers: usize = 0; | ||
| 62 | |||
| 63 | // used for capturing slices, such as bind parameter name | ||
| 64 | var hold_pos = 0; | ||
| 65 | |||
| 66 | for (tmp_query) |c| { | ||
| 67 | switch (state) { | ||
| 68 | .start => switch (c) { | ||
| 69 | '?', ':', '@', '$' => { | ||
| 70 | tmp_bind_markers[nb_tmp_bind_markers] = BindMarker{}; | ||
| 71 | current_bind_marker_type_pos = 0; | ||
| 72 | state = .bind_marker; | ||
| 73 | buf[pos] = c; | ||
| 74 | pos += 1; | ||
| 75 | }, | ||
| 76 | '\'', '"', '[', '`' => { | ||
| 77 | state = .inside_string; | ||
| 78 | buf[pos] = c; | ||
| 79 | pos += 1; | ||
| 80 | }, | ||
| 81 | else => { | ||
| 82 | buf[pos] = c; | ||
| 83 | pos += 1; | ||
| 84 | }, | ||
| 85 | }, | ||
| 86 | .inside_string => switch (c) { | ||
| 87 | '\'', '"', ']', '`' => { | ||
| 88 | state = .start; | ||
| 89 | buf[pos] = c; | ||
| 90 | pos += 1; | ||
| 91 | }, | ||
| 92 | else => { | ||
| 93 | buf[pos] = c; | ||
| 94 | pos += 1; | ||
| 95 | }, | ||
| 96 | }, | ||
| 97 | .bind_marker => switch (c) { | ||
| 98 | '?', ':', '@', '$' => @compileError("invalid multiple '?', ':', '$' or '@'."), | ||
| 99 | '{' => { | ||
| 100 | state = .bind_marker_type; | ||
| 101 | }, | ||
| 102 | else => { | ||
| 103 | if (isNamedIdentifierChar(c)) { | ||
| 104 | // This is the start of a named bind marker. | ||
| 105 | state = .bind_marker_identifier; | ||
| 106 | hold_pos = pos + 1; | ||
| 107 | } else { | ||
| 108 | // This is a unnamed, untyped bind marker. | ||
| 109 | state = .start; | ||
| 110 | |||
| 111 | tmp_bind_markers[nb_tmp_bind_markers].typed = null; | ||
| 112 | nb_tmp_bind_markers += 1; | ||
| 113 | } | ||
| 114 | buf[pos] = c; | ||
| 115 | pos += 1; | ||
| 116 | }, | ||
| 117 | }, | ||
| 118 | .bind_marker_identifier => switch (c) { | ||
| 119 | '?', ':', '@', '$' => @compileError("unregconised multiple '?', ':', '$' or '@'."), | ||
| 120 | '{' => { | ||
| 121 | state = .bind_marker_type; | ||
| 122 | current_bind_marker_type_pos = 0; | ||
| 123 | }, | ||
| 124 | else => { | ||
| 125 | if (!isNamedIdentifierChar(c)) { | ||
| 126 | // This marks the end of the named bind marker. | ||
| 127 | state = .start; | ||
| 128 | const name = buf[hold_pos .. pos - 1]; | ||
| 129 | // TODO(vincent): name retains a pointer to a comptime var, FIX ! | ||
| 130 | if (bindMarkerForName(tmp_bind_markers[0..nb_tmp_bind_markers], name) == null) { | ||
| 131 | const new_buf = buf; | ||
| 132 | tmp_bind_markers[nb_tmp_bind_markers].name = new_buf[hold_pos .. pos - 1]; | ||
| 133 | nb_tmp_bind_markers += 1; | ||
| 134 | } | ||
| 135 | } | ||
| 136 | buf[pos] = c; | ||
| 137 | pos += 1; | ||
| 138 | }, | ||
| 139 | }, | ||
| 140 | .bind_marker_type => switch (c) { | ||
| 141 | '}' => { | ||
| 142 | state = .start; | ||
| 143 | |||
| 144 | const type_info_string = current_bind_marker_type[0..current_bind_marker_type_pos]; | ||
| 145 | // Handles optional types | ||
| 146 | const typ = if (type_info_string[0] == '?') blk: { | ||
| 147 | const child_type = ParseType(type_info_string[1..]); | ||
| 148 | break :blk @Type(std.builtin.Type{ | ||
| 149 | .Optional = .{ | ||
| 150 | .child = child_type, | ||
| 151 | }, | ||
| 152 | }); | ||
| 153 | } else blk: { | ||
| 154 | break :blk ParseType(type_info_string); | ||
| 155 | }; | ||
| 156 | |||
| 157 | tmp_bind_markers[nb_tmp_bind_markers].typed = typ; | ||
| 158 | nb_tmp_bind_markers += 1; | ||
| 159 | }, | ||
| 160 | else => { | ||
| 161 | current_bind_marker_type[current_bind_marker_type_pos] = c; | ||
| 162 | current_bind_marker_type_pos += 1; | ||
| 163 | }, | ||
| 164 | }, | ||
| 165 | else => { | ||
| 166 | @compileError("invalid state " ++ @tagName(state)); | ||
| 167 | }, | ||
| 168 | } | ||
| 169 | } | ||
| 170 | |||
| 171 | // The last character was a bind marker prefix so this must be an untyped bind marker. | ||
| 172 | switch (state) { | ||
| 173 | .bind_marker => { | ||
| 174 | tmp_bind_markers[nb_tmp_bind_markers].typed = null; | ||
| 175 | nb_tmp_bind_markers += 1; | ||
| 176 | }, | ||
| 177 | .bind_marker_identifier => { | ||
| 178 | nb_tmp_bind_markers += 1; | ||
| 179 | }, | ||
| 180 | .start => {}, | ||
| 181 | else => @compileError("invalid final state " ++ @tagName(state) ++ ", this means you wrote an incomplete bind marker type"), | ||
| 182 | } | ||
| 183 | |||
| 184 | const final_bind_markers = tmp_bind_markers; | ||
| 185 | const final_bind_markers_len = nb_tmp_bind_markers; | ||
| 186 | const final_buf = buf; | ||
| 187 | const final_query_len = pos; | ||
| 188 | |||
| 189 | return .{ | ||
| 190 | .bind_markers = final_bind_markers, | ||
| 191 | .bind_markers_len = final_bind_markers_len, | ||
| 192 | .query = final_buf, | ||
| 193 | .query_len = final_query_len, | ||
| 194 | }; | ||
| 185 | } | 195 | } |
| 186 | }; | 196 | }; |
| 187 | } | 197 | } |
| @@ -246,7 +256,7 @@ test "parsed query: query" { | |||
| 246 | 256 | ||
| 247 | inline for (testCases) |tc| { | 257 | inline for (testCases) |tc| { |
| 248 | @setEvalBranchQuota(100000); | 258 | @setEvalBranchQuota(100000); |
| 249 | comptime var parsed_query = ParsedQuery(tc.query); | 259 | const parsed_query = ParsedQuery(tc.query); |
| 250 | try testing.expectEqualStrings(tc.expected_query, parsed_query.getQuery()); | 260 | try testing.expectEqualStrings(tc.expected_query, parsed_query.getQuery()); |
| 251 | } | 261 | } |
| 252 | } | 262 | } |
| @@ -296,7 +306,7 @@ test "parsed query: bind markers types" { | |||
| 296 | @setEvalBranchQuota(100000); | 306 | @setEvalBranchQuota(100000); |
| 297 | const parsed_query = comptime ParsedQuery(tc.query); | 307 | const parsed_query = comptime ParsedQuery(tc.query); |
| 298 | 308 | ||
| 299 | try testing.expectEqual(1, parsed_query.nb_bind_markers); | 309 | try testing.expectEqual(1, parsed_query.bind_markers.len); |
| 300 | 310 | ||
| 301 | const bind_marker = parsed_query.bind_markers[0]; | 311 | const bind_marker = parsed_query.bind_markers[0]; |
| 302 | try testing.expectEqual(tc.expected_marker.typed, bind_marker.typed); | 312 | try testing.expectEqual(tc.expected_marker.typed, bind_marker.typed); |
| @@ -344,7 +354,7 @@ test "parsed query: bind markers identifier" { | |||
| 344 | inline for (testCases) |tc| { | 354 | inline for (testCases) |tc| { |
| 345 | const parsed_query = comptime ParsedQuery(tc.query); | 355 | const parsed_query = comptime ParsedQuery(tc.query); |
| 346 | 356 | ||
| 347 | try testing.expectEqual(@as(usize, 1), parsed_query.nb_bind_markers); | 357 | try testing.expectEqual(@as(usize, 1), parsed_query.bind_markers.len); |
| 348 | 358 | ||
| 349 | const bind_marker = parsed_query.bind_markers[0]; | 359 | const bind_marker = parsed_query.bind_markers[0]; |
| 350 | try testing.expectEqual(tc.expected_marker, bind_marker); | 360 | try testing.expectEqual(tc.expected_marker, bind_marker); |
| @@ -388,9 +398,13 @@ test "parsed query: query bind identifier" { | |||
| 388 | 398 | ||
| 389 | inline for (testCases) |tc| { | 399 | inline for (testCases) |tc| { |
| 390 | @setEvalBranchQuota(100000); | 400 | @setEvalBranchQuota(100000); |
| 391 | comptime var parsed_query = ParsedQuery(tc.query); | 401 | |
| 392 | try testing.expectEqualStrings(tc.expected_query, parsed_query.getQuery()); | 402 | comptime { |
| 393 | try testing.expectEqual(tc.expected_nb_bind_markers, parsed_query.nb_bind_markers); | 403 | const parsed_query = ParsedQuery(tc.query); |
| 404 | |||
| 405 | try testing.expectEqual(tc.expected_nb_bind_markers, parsed_query.bind_markers.len); | ||
| 406 | try testing.expectEqualStrings(tc.expected_query, parsed_query.getQuery()); | ||
| 407 | } | ||
| 394 | } | 408 | } |
| 395 | } | 409 | } |
| 396 | 410 | ||
| @@ -416,9 +430,9 @@ test "parsed query: bind marker character inside string" { | |||
| 416 | 430 | ||
| 417 | inline for (testCases) |tc| { | 431 | inline for (testCases) |tc| { |
| 418 | @setEvalBranchQuota(100000); | 432 | @setEvalBranchQuota(100000); |
| 419 | comptime var parsed_query = ParsedQuery(tc.query); | 433 | const parsed_query = ParsedQuery(tc.query); |
| 420 | 434 | ||
| 421 | try testing.expectEqual(@as(usize, tc.exp_bind_markers), parsed_query.nb_bind_markers); | 435 | try testing.expectEqual(@as(usize, tc.exp_bind_markers), parsed_query.bind_markers.len); |
| 422 | try testing.expectEqualStrings(tc.exp, parsed_query.getQuery()); | 436 | try testing.expectEqualStrings(tc.exp, parsed_query.getQuery()); |
| 423 | } | 437 | } |
| 424 | } | 438 | } |