xref: /llvm-project/llvm/unittests/Support/DataExtractorTest.cpp (revision b0abd4893fa1bfae7f71b6b6e98770c9b1c07620)
1 //===- llvm/unittest/Support/DataExtractorTest.cpp - DataExtractor tests --===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 
9 #include "llvm/Support/DataExtractor.h"
10 
11 #include "llvm/ADT/StringExtras.h"
12 #include "llvm/Testing/Support/Error.h"
13 #include "gtest/gtest.h"
14 using namespace llvm;
15 
16 namespace {
17 
18 const char numberData[] = "\x80\x90\xFF\xFF\x80\x00\x00\x00";
19 const char leb128data[] = "\xA6\x49";
20 const char bigleb128data[] = "\xAA\xA9\xFF\xAA\xFF\xAA\xFF\x4A";
21 
TEST(DataExtractorTest,OffsetOverflow)22 TEST(DataExtractorTest, OffsetOverflow) {
23   DataExtractor DE(StringRef(numberData, sizeof(numberData)-1), false, 8);
24   EXPECT_FALSE(DE.isValidOffsetForDataOfSize(-2U, 5));
25 }
26 
TEST(DataExtractorTest,UnsignedNumbers)27 TEST(DataExtractorTest, UnsignedNumbers) {
28   DataExtractor DE(StringRef(numberData, sizeof(numberData)-1), false, 8);
29   uint64_t offset = 0;
30 
31   EXPECT_EQ(0x80U, DE.getU8(&offset));
32   EXPECT_EQ(1U, offset);
33   offset = 0;
34   EXPECT_EQ(0x8090U, DE.getU16(&offset));
35   EXPECT_EQ(2U, offset);
36   offset = 0;
37   EXPECT_EQ(0x8090FFFFU, DE.getU32(&offset));
38   EXPECT_EQ(4U, offset);
39   offset = 0;
40   EXPECT_EQ(0x8090FFFF80000000ULL, DE.getU64(&offset));
41   EXPECT_EQ(8U, offset);
42   offset = 0;
43   EXPECT_EQ(0x8090FFFF80000000ULL, DE.getAddress(&offset));
44   EXPECT_EQ(8U, offset);
45   offset = 0;
46 
47   uint32_t data[2];
48   EXPECT_EQ(data, DE.getU32(&offset, data, 2));
49   EXPECT_EQ(0x8090FFFFU, data[0]);
50   EXPECT_EQ(0x80000000U, data[1]);
51   EXPECT_EQ(8U, offset);
52   offset = 0;
53 
54   // Now for little endian.
55   DE = DataExtractor(StringRef(numberData, sizeof(numberData)-1), true, 4);
56   EXPECT_EQ(0x9080U, DE.getU16(&offset));
57   EXPECT_EQ(2U, offset);
58   offset = 0;
59   EXPECT_EQ(0xFFFF9080U, DE.getU32(&offset));
60   EXPECT_EQ(4U, offset);
61   offset = 0;
62   EXPECT_EQ(0x80FFFF9080ULL, DE.getU64(&offset));
63   EXPECT_EQ(8U, offset);
64   offset = 0;
65   EXPECT_EQ(0xFFFF9080U, DE.getAddress(&offset));
66   EXPECT_EQ(4U, offset);
67   offset = 0;
68 
69   EXPECT_EQ(data, DE.getU32(&offset, data, 2));
70   EXPECT_EQ(0xFFFF9080U, data[0]);
71   EXPECT_EQ(0x80U, data[1]);
72   EXPECT_EQ(8U, offset);
73 }
74 
TEST(DataExtractorTest,SignedNumbers)75 TEST(DataExtractorTest, SignedNumbers) {
76   DataExtractor DE(StringRef(numberData, sizeof(numberData)-1), false, 8);
77   uint64_t offset = 0;
78 
79   EXPECT_EQ(-128, DE.getSigned(&offset, 1));
80   EXPECT_EQ(1U, offset);
81   offset = 0;
82   EXPECT_EQ(-32624, DE.getSigned(&offset, 2));
83   EXPECT_EQ(2U, offset);
84   offset = 0;
85   EXPECT_EQ(-2137980929, DE.getSigned(&offset, 4));
86   EXPECT_EQ(4U, offset);
87   offset = 0;
88   EXPECT_EQ(-9182558167379214336LL, DE.getSigned(&offset, 8));
89   EXPECT_EQ(8U, offset);
90 }
91 
TEST(DataExtractorTest,Strings)92 TEST(DataExtractorTest, Strings) {
93   const char stringData[] = "hellohello\0hello";
94   DataExtractor DE(StringRef(stringData, sizeof(stringData)-1), false, 8);
95   uint64_t offset = 0;
96 
97   EXPECT_EQ(stringData, DE.getCStr(&offset));
98   EXPECT_EQ(11U, offset);
99   EXPECT_EQ(nullptr, DE.getCStr(&offset));
100   EXPECT_EQ(11U, offset);
101 
102   DataExtractor::Cursor C(0);
103   EXPECT_EQ(stringData, DE.getCStr(C));
104   EXPECT_EQ(11U, C.tell());
105   EXPECT_EQ(nullptr, DE.getCStr(C));
106   EXPECT_EQ(11U, C.tell());
107   EXPECT_THAT_ERROR(
108       C.takeError(),
109       FailedWithMessage("no null terminated string at offset 0xb"));
110 }
111 
TEST(DataExtractorTest,LEB128)112 TEST(DataExtractorTest, LEB128) {
113   DataExtractor DE(StringRef(leb128data, sizeof(leb128data)-1), false, 8);
114   uint64_t offset = 0;
115 
116   EXPECT_EQ(9382ULL, DE.getULEB128(&offset));
117   EXPECT_EQ(2U, offset);
118   offset = 0;
119   EXPECT_EQ(-7002LL, DE.getSLEB128(&offset));
120   EXPECT_EQ(2U, offset);
121 
122   DataExtractor BDE(StringRef(bigleb128data, sizeof(bigleb128data)-1), false,8);
123   offset = 0;
124   EXPECT_EQ(42218325750568106ULL, BDE.getULEB128(&offset));
125   EXPECT_EQ(8U, offset);
126   offset = 0;
127   EXPECT_EQ(-29839268287359830LL, BDE.getSLEB128(&offset));
128   EXPECT_EQ(8U, offset);
129 }
130 
TEST(DataExtractorTest,LEB128_error)131 TEST(DataExtractorTest, LEB128_error) {
132   DataExtractor DE(StringRef("\x81"), false, 8);
133   uint64_t Offset = 0;
134   EXPECT_EQ(0U, DE.getULEB128(&Offset));
135   EXPECT_EQ(0U, Offset);
136 
137   Offset = 0;
138   EXPECT_EQ(0U, DE.getSLEB128(&Offset));
139   EXPECT_EQ(0U, Offset);
140 
141   DataExtractor::Cursor C(0);
142   EXPECT_EQ(0U, DE.getULEB128(C));
143   EXPECT_THAT_ERROR(
144       C.takeError(),
145       FailedWithMessage("unable to decode LEB128 at offset 0x00000000: "
146                         "malformed uleb128, extends past end"));
147 
148   C = DataExtractor::Cursor(0);
149   EXPECT_EQ(0U, DE.getSLEB128(C));
150   EXPECT_THAT_ERROR(
151       C.takeError(),
152       FailedWithMessage("unable to decode LEB128 at offset 0x00000000: "
153                         "malformed sleb128, extends past end"));
154 
155   // Show non-zero offsets are reported appropriately.
156   C = DataExtractor::Cursor(1);
157   EXPECT_EQ(0U, DE.getULEB128(C));
158   EXPECT_THAT_ERROR(
159       C.takeError(),
160       FailedWithMessage("unable to decode LEB128 at offset 0x00000001: "
161                         "malformed uleb128, extends past end"));
162 }
163 
TEST(DataExtractorTest,Cursor_tell)164 TEST(DataExtractorTest, Cursor_tell) {
165   DataExtractor DE(StringRef("AB"), false, 8);
166   DataExtractor::Cursor C(0);
167   // A successful read operation advances the cursor
168   EXPECT_EQ('A', DE.getU8(C));
169   EXPECT_EQ(1u, C.tell());
170 
171   // An unsuccessful one doesn't.
172   EXPECT_EQ(0u, DE.getU16(C));
173   EXPECT_EQ(1u, C.tell());
174 
175   // And neither do any subsequent operations.
176   EXPECT_EQ(0, DE.getU8(C));
177   EXPECT_EQ(1u, C.tell());
178 
179   consumeError(C.takeError());
180 }
181 
TEST(DataExtractorTest,Cursor_seek)182 TEST(DataExtractorTest, Cursor_seek) {
183   DataExtractor::Cursor C(5);
184 
185   C.seek(3);
186   EXPECT_EQ(3u, C.tell());
187 
188   C.seek(8);
189   EXPECT_EQ(8u, C.tell());
190 
191   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
192 }
193 
TEST(DataExtractorTest,Cursor_takeError)194 TEST(DataExtractorTest, Cursor_takeError) {
195   DataExtractor DE(StringRef("AB"), false, 8);
196   DataExtractor::Cursor C(0);
197   // Initially, the cursor is in the "success" state.
198   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
199 
200   // It remains "success" after a successful read.
201   EXPECT_EQ('A', DE.getU8(C));
202   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
203 
204   // An unsuccessful read sets the error state.
205   EXPECT_EQ(0u, DE.getU32(C));
206   EXPECT_THAT_ERROR(C.takeError(), Failed());
207 
208   // Once set the error sticks until explicitly cleared.
209   EXPECT_EQ(0u, DE.getU32(C));
210   EXPECT_EQ(0, DE.getU8(C));
211   EXPECT_THAT_ERROR(C.takeError(), Failed());
212 
213   // At which point reads can be succeed again.
214   EXPECT_EQ('B', DE.getU8(C));
215   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
216 }
217 
TEST(DataExtractorTest,Cursor_chaining)218 TEST(DataExtractorTest, Cursor_chaining) {
219   DataExtractor DE(StringRef("ABCD"), false, 8);
220   DataExtractor::Cursor C(0);
221 
222   // Multiple reads can be chained without trigerring any assertions.
223   EXPECT_EQ('A', DE.getU8(C));
224   EXPECT_EQ('B', DE.getU8(C));
225   EXPECT_EQ('C', DE.getU8(C));
226   EXPECT_EQ('D', DE.getU8(C));
227   // And the error checked at the end.
228   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
229 }
230 
231 #if defined(GTEST_HAS_DEATH_TEST) && defined(_DEBUG) &&                        \
232     LLVM_ENABLE_ABI_BREAKING_CHECKS
TEST(DataExtractorDeathTest,Cursor)233 TEST(DataExtractorDeathTest, Cursor) {
234   DataExtractor DE(StringRef("AB"), false, 8);
235 
236   // Even an unused cursor must be checked for errors:
237   EXPECT_DEATH(DataExtractor::Cursor(0),
238                "Success values must still be checked prior to being destroyed");
239 
240   {
241     auto C = std::make_unique<DataExtractor::Cursor>(0);
242     EXPECT_EQ(0u, DE.getU32(*C));
243     // It must also be checked after an unsuccessful operation.
244     // destruction.
245     EXPECT_DEATH(C.reset(), "unexpected end of data");
246     EXPECT_THAT_ERROR(C->takeError(), Failed());
247   }
248   {
249     auto C = std::make_unique<DataExtractor::Cursor>(0);
250     EXPECT_EQ('A', DE.getU8(*C));
251     // Same goes for a successful one.
252     EXPECT_DEATH(
253         C.reset(),
254         "Success values must still be checked prior to being destroyed");
255     EXPECT_THAT_ERROR(C->takeError(), Succeeded());
256   }
257   {
258     auto C = std::make_unique<DataExtractor::Cursor>(0);
259     EXPECT_EQ('A', DE.getU8(*C));
260     EXPECT_EQ(0u, DE.getU32(*C));
261     // Even if a successful operation is followed by an unsuccessful one.
262     EXPECT_DEATH(C.reset(), "unexpected end of data");
263     EXPECT_THAT_ERROR(C->takeError(), Failed());
264   }
265   {
266     auto C = std::make_unique<DataExtractor::Cursor>(0);
267     EXPECT_EQ(0u, DE.getU32(*C));
268     EXPECT_EQ(0, DE.getU8(*C));
269     // Even if an unsuccessful operation is followed by one that would normally
270     // succeed.
271     EXPECT_DEATH(C.reset(), "unexpected end of data");
272     EXPECT_THAT_ERROR(C->takeError(), Failed());
273   }
274 }
275 #endif
276 
TEST(DataExtractorTest,getU8_vector)277 TEST(DataExtractorTest, getU8_vector) {
278   DataExtractor DE(StringRef("AB"), false, 8);
279   DataExtractor::Cursor C(0);
280   SmallVector<uint8_t, 2> S;
281 
282   DE.getU8(C, S, 4);
283   EXPECT_THAT_ERROR(C.takeError(), Failed());
284   EXPECT_EQ("", toStringRef(S));
285 
286   DE.getU8(C, S, 2);
287   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
288   EXPECT_EQ("AB", toStringRef(S));
289 
290   C = DataExtractor::Cursor(0x47);
291   DE.getU8(C, S, 2);
292   EXPECT_THAT_ERROR(
293       C.takeError(),
294       FailedWithMessage("offset 0x47 is beyond the end of data at 0x2"));
295 }
296 
TEST(DataExtractorTest,getU24)297 TEST(DataExtractorTest, getU24) {
298   DataExtractor DE(StringRef("ABCD"), false, 8);
299   DataExtractor::Cursor C(0);
300 
301   EXPECT_EQ(0x414243u, DE.getU24(C));
302   EXPECT_EQ(0u, DE.getU24(C));
303   EXPECT_EQ(3u, C.tell());
304   EXPECT_THAT_ERROR(C.takeError(), Failed());
305 }
306 
TEST(DataExtractorTest,skip)307 TEST(DataExtractorTest, skip) {
308   DataExtractor DE(StringRef("AB"), false, 8);
309   DataExtractor::Cursor C(0);
310 
311   DE.skip(C, 4);
312   EXPECT_THAT_ERROR(C.takeError(), Failed());
313   EXPECT_EQ(0u, C.tell());
314 
315   DE.skip(C, 2);
316   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
317   EXPECT_EQ(2u, C.tell());
318 }
319 
TEST(DataExtractorTest,eof)320 TEST(DataExtractorTest, eof) {
321   DataExtractor DE(StringRef("A"), false, 8);
322   DataExtractor::Cursor C(0);
323 
324   EXPECT_FALSE(DE.eof(C));
325 
326   EXPECT_EQ(0, DE.getU16(C));
327   EXPECT_FALSE(DE.eof(C));
328   EXPECT_THAT_ERROR(C.takeError(), Failed());
329 
330   EXPECT_EQ('A', DE.getU8(C));
331   EXPECT_TRUE(DE.eof(C));
332   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
333 }
334 
TEST(DataExtractorTest,size)335 TEST(DataExtractorTest, size) {
336   uint8_t Data[] = {'A', 'B', 'C', 'D'};
337   DataExtractor DE1(StringRef(reinterpret_cast<char *>(Data), sizeof(Data)),
338                     false, 8);
339   EXPECT_EQ(DE1.size(), sizeof(Data));
340   DataExtractor DE2(ArrayRef<uint8_t>(Data), false, 8);
341   EXPECT_EQ(DE2.size(), sizeof(Data));
342 }
343 
TEST(DataExtractorTest,FixedLengthString)344 TEST(DataExtractorTest, FixedLengthString) {
345   const char Data[] = "hello\x00\x00\x00world  \thola\x00";
346   DataExtractor DE(StringRef(Data, sizeof(Data)-1), false, 8);
347   uint64_t Offset = 0;
348   StringRef Str;
349   // Test extracting too many bytes doesn't modify Offset and returns
350   // std::nullopt.
351   Str = DE.getFixedLengthString(&Offset, sizeof(Data));
352   EXPECT_TRUE(Str.empty());
353   EXPECT_EQ(Offset, 0u);
354 
355   // Test extracting a fixed width C string with trailing NULL characters.
356   Str = DE.getFixedLengthString(&Offset, 8);
357   EXPECT_EQ(Offset, 8u);
358   EXPECT_EQ(Str.size(), 5u);
359   EXPECT_EQ(Str, "hello");
360   // Test extracting a fixed width C string with trailing space and tab
361   // characters.
362   Str = DE.getFixedLengthString(&Offset, 8, " \t");
363   EXPECT_EQ(Offset, 16u);
364   EXPECT_EQ(Str.size(), 5u);
365   EXPECT_EQ(Str, "world");
366   // Now extract a normal C string.
367   Str = DE.getCStrRef(&Offset);
368   EXPECT_EQ(Str.size(), 4u);
369   EXPECT_EQ(Str, "hola");
370 }
371 
372 
TEST(DataExtractorTest,GetBytes)373 TEST(DataExtractorTest, GetBytes) {
374   // Use data with an embedded NULL character for good measure.
375   const char Data[] = "\x01\x02\x00\x04";
376   StringRef Bytes(Data, sizeof(Data)-1);
377   DataExtractor DE(Bytes, false, 8);
378   uint64_t Offset = 0;
379   StringRef Str;
380   // Test extracting too many bytes doesn't modify Offset and returns
381   // std::nullopt.
382   Str = DE.getBytes(&Offset, sizeof(Data));
383   EXPECT_TRUE(Str.empty());
384   EXPECT_EQ(Offset, 0u);
385   // Test extracting 4 bytes from the stream.
386   Str = DE.getBytes(&Offset, 4);
387   EXPECT_EQ(Offset, 4u);
388   EXPECT_EQ(Str.size(), 4u);
389   EXPECT_EQ(Str, Bytes);
390 
391   DataExtractor::Cursor C(0);
392   EXPECT_EQ(StringRef("\x01\x02"), DE.getBytes(C, 2));
393   EXPECT_EQ(StringRef("\x00\x04", 2), DE.getBytes(C, 2));
394   EXPECT_EQ(StringRef(), DE.getBytes(C, 2));
395   EXPECT_EQ(StringRef(), DE.getBytes(C, 2));
396   EXPECT_EQ(4u, C.tell());
397   EXPECT_THAT_ERROR(C.takeError(), Failed());
398 }
399 
400 }
401