Fix #15132 build dir: cached results are discarded on the second run when includes are not in order (#8957)

Author: wjakobsson

lib/preprocessor.cpp

@@ -1114,22 +1114,29 @@ void Preprocessor::dump(std::ostream &out) const
 std::size_t Preprocessor::calculateHash(const std::string &toolinfo) const
 {
     std::string hashData = toolinfo;
-    for (const simplecpp::Token *tok = mTokens.cfront(); tok; tok = tok->next) {
-        if (!tok->comment) {
-            hashData += tok->str();
-            hashData += static_cast<char>(tok->location.line);
-            hashData += static_cast<char>(tok->location.col);
-        }
-    }
-    for (const auto &filedata : mFileCache) {
-        for (const simplecpp::Token *tok = filedata->tokens.cfront(); tok; tok = tok->next) {
+    const auto addTokens = [&hashData](const simplecpp::TokenList &tokens) {
+        for (const simplecpp::Token *tok = tokens.cfront(); tok; tok = tok->next) {
             if (!tok->comment) {
                 hashData += tok->str();
                 hashData += static_cast<char>(tok->location.line);
                 hashData += static_cast<char>(tok->location.col);
             }
         }
-    }
+    };
+    addTokens(mTokens);
+
+    // Hash the included files in a fixed order, the order they are loaded in differs between runs
+    std::vector<const simplecpp::FileData *> fileData;
+    fileData.reserve(mFileCache.size());
+    std::transform(mFileCache.cbegin(), mFileCache.cend(), std::back_inserter(fileData), [](const std::unique_ptr<simplecpp::FileData> &data) {
+        return data.get();
+    });
+    std::sort(fileData.begin(), fileData.end(), [](const simplecpp::FileData *a, const simplecpp::FileData *b) {
+        return a->filename < b->filename;
+    });
+    for (const simplecpp::FileData *data : fileData)
+        addTokens(data->tokens);
+
     return (std::hash<std::string>{})(hashData);
 }
 

test/cli/other_test.py

@@ -4575,6 +4575,47 @@ def run_and_assert_cppcheck(stdout_exp):
     # - internalError
 
 
+def test_analyzerinfo_include_order(tmp_path):  # #15132
+    with open(tmp_path / 'z.h', 'wt') as f:
+        f.write('int z;\n')
+    with open(tmp_path / 'a.h', 'wt') as f:
+        f.write('int a;\n')
+    test_file = tmp_path / 'test.c'
+    with open(test_file, 'wt') as f:
+        f.write(
+"""#include "z.h"
+#include "a.h"
+int f(void) { return a + z; }
+""")
+
+    build_dir = tmp_path / 'b1'
+    os.makedirs(build_dir)
+
+    args = [
+        '-q',
+        '--debug-analyzerinfo',
+        '--cppcheck-build-dir={}'.format(build_dir),
+        str(test_file)
+    ]
+
+    test_file_s = str(test_file).replace('\\', '/')
+    test_a1_file_s = str(build_dir / 'test.a1').replace('\\', '/')
+
+    exitcode, stdout, stderr = cppcheck(args)
+    assert exitcode == 0, stdout
+    assert stdout.splitlines() == [
+        "no cached result '{}' for '{}' found".format(test_a1_file_s, test_file_s)
+    ]
+    assert stderr.splitlines() == []
+
+    exitcode, stdout, stderr = cppcheck(args)
+    assert exitcode == 0, stdout
+    assert stdout.splitlines() == [
+        "skipping analysis - loaded 0 cached finding(s) from '{}' for '{}'".format(test_a1_file_s, test_file_s)
+    ]
+    assert stderr.splitlines() == []
+
+
 def test_ctu_function_call_path_slash(tmp_path):  # #14591
     test_file = tmp_path / 'test.cpp'
     with open(test_file, "w") as f:

test/testpreprocessor.cpp

@@ -382,6 +382,7 @@ class TestPreprocessor : public TestFixture {
         TEST_CASE(limitsDefines);
 
         TEST_CASE(hashCalculation);
+        TEST_CASE(hashCalculationIncludeOrder);
 
         TEST_CASE(standard);
 
@@ -3182,6 +3183,25 @@ class TestPreprocessor : public TestFixture {
         ASSERT(getHash(code2) != getHash(code3));
     }
 
+    void hashCalculationIncludeOrder() {
+        // #15132 - the load order of the includes must not change the hash
+        const ScopedFile headerA("hash_a.h", "int a;\n");
+        const ScopedFile headerZ("hash_z.h", "int z;\n");
+        const char code[] = "#include \"hash_z.h\"\n"
+                            "#include \"hash_a.h\"\n";
+
+        const auto getHashLoadOrder = [&](const std::vector<std::string> &loadOrder) {
+            std::vector<std::string> files;
+            simplecpp::TokenList tokens(code, files, "test.c");
+            Preprocessor preprocessor(tokens, settingsDefault, *this, Standards::Language::C);
+            for (const std::string &header : loadOrder)
+                ASSERT(preprocessor.loadFile(files, header));
+            return preprocessor.calculateHash("");
+        };
+
+        ASSERT_EQUALS(getHashLoadOrder({"hash_z.h", "hash_a.h"}), getHashLoadOrder({"hash_a.h", "hash_z.h"}));
+    }
+
     void standard() {
 
         const char code[] = "int a;\n";