added command-line option --cpp-header-probe to probe headers and extension-less files for Emacs C++ marker [skip ci]

Author: firewave

cli/cmdlineparser.cpp

@@ -544,6 +544,10 @@ CmdLineParser::Result CmdLineParser::parseFromArgs(int argc, const char* const a
                 }
             }
 
+            else if (std::strcmp(argv[i], "--cpp-header-probe") == 0) {
+                mSettings.cppHeaderProbe = true;
+            }
+
             // Show --debug output after the first simplifications
             else if (std::strcmp(argv[i], "--debug") == 0 ||
                      std::strcmp(argv[i], "--debug-normal") == 0)
@@ -887,6 +891,10 @@ CmdLineParser::Result CmdLineParser::parseFromArgs(int argc, const char* const a
                     return Result::Fail;
             }
 
+            else if (std::strcmp(argv[i], "--no-cpp-header-probe") == 0) {
+                mSettings.cppHeaderProbe = false;
+            }
+
             // Write results in file
             else if (std::strncmp(argv[i], "--output-file=", 14) == 0)
                 mSettings.outputFile = Path::simplifyPath(Path::fromNativeSeparators(argv[i] + 14));

lib/cppcheck.cpp

@@ -182,7 +182,7 @@ static void createDumpFile(const Settings& settings,
     case Standards::Language::None:
     {
         // TODO: error out on unknown language?
-        const Standards::Language lang = Path::identify(filename);
+        const Standards::Language lang = Path::identify(filename, settings.cppHeaderProbe);
         if (lang == Standards::Language::CPP)
             language = " language=\"cpp\"";
         else if (lang == Standards::Language::C)
@@ -420,7 +420,7 @@ unsigned int CppCheck::checkClang(const std::string &path)
         mErrorLogger.reportOut(std::string("Checking ") + path + " ...", Color::FgGreen);
 
     // TODO: this ignores the configured language
-    const bool isCpp = Path::identify(path) == Standards::Language::CPP;
+    const bool isCpp = Path::identify(path, mSettings.cppHeaderProbe) == Standards::Language::CPP;
     const std::string langOpt = isCpp ? "-x c++" : "-x c";
     const std::string analyzerInfo = mSettings.buildDir.empty() ? std::string() : AnalyzerInformation::getAnalyzerInfoFile(mSettings.buildDir, path, emptyString);
     const std::string clangcmd = analyzerInfo + ".clang-cmd";
@@ -783,7 +783,7 @@ unsigned int CppCheck::checkFile(const std::string& filename, const std::string
             TokenList tokenlist(&mSettings);
             std::istringstream istr2(code);
             // TODO: asserts when file has unknown extension
-            tokenlist.createTokens(istr2, Path::identify(*files.begin())); // TODO: check result?
+            tokenlist.createTokens(istr2, Path::identify(*files.begin(), false)); // TODO: check result?
             executeRules("define", tokenlist);
         }
 #endif

lib/path.cpp

@@ -20,11 +20,17 @@
 #undef __STRICT_ANSI__
 #endif
 
+#define LOG_EMACS_MARKER
+
 #include "path.h"
 #include "utils.h"
 
 #include <algorithm>
+#include <cstdio>
 #include <cstdlib>
+#ifdef LOG_EMACS_MARKER
+#include <iostream>
+#endif
 #include <sys/stat.h>
 #include <unordered_set>
 #include <utility>
@@ -235,7 +241,7 @@ bool Path::isCPP(const std::string &path)
 bool Path::acceptFile(const std::string &path, const std::set<std::string> &extra)
 {
     bool header = false;
-    return (identify(path, &header) != Standards::Language::None && !header) || extra.find(getFilenameExtension(path)) != extra.end();
+    return (identify(path, false, &header) != Standards::Language::None && !header) || extra.find(getFilenameExtension(path)) != extra.end();
 }
 
 // cppcheck-suppress unusedFunction
@@ -245,13 +251,98 @@ bool Path::isHeader(const std::string &path)
     return startsWith(extension, ".h");
 }
 
-Standards::Language Path::identify(const std::string &path, bool *header)
+static bool hasEmacsCppMarker(const char* path)
+{
+    // TODO: identify is called three times for each file
+    // Preprocessor::loadFiles() -> createDUI()
+    // Preprocessor::preprocess() -> createDUI()
+    // TokenList::createTokens() -> TokenList::determineCppC()
+#ifdef LOG_EMACS_MARKER
+    std::cout << path << '\n';
+#endif
+
+    FILE *fp = fopen(path, "rt");
+    if (!fp)
+        return false;
+    std::string buf(128, '\0');
+    {
+        // TODO: read the whole first line only
+        const char * const res = fgets(const_cast<char*>(buf.data()), buf.size(), fp);
+        fclose(fp);
+        fp = nullptr;
+        if (!res)
+            return false; // failed to read file
+    }
+    // TODO: replace with regular expression
+    const auto pos1 = buf.find("-*-");
+    if (pos1 == std::string::npos)
+        return false; // no start marker
+    const auto pos_nl = buf.find_first_of("\r\n");
+    if (pos_nl != std::string::npos && (pos_nl < pos1)) {
+#ifdef LOG_EMACS_MARKER
+        std::cout << path << " - Emacs marker not on the first line" << '\n';
+#endif
+        return false; // not on first line
+    }
+    const auto pos2 = buf.find("-*-", pos1 + 3);
+    // TODO: make sure we have read the whole line before bailing out
+    if (pos2 == std::string::npos) {
+#ifdef LOG_EMACS_MARKER
+        std::cout << path << " - Emacs marker not terminated" << '\n';
+#endif
+        return false; // no end marker
+    }
+#ifdef LOG_EMACS_MARKER
+    std::cout << "Emacs marker: '"  << buf.substr(pos1, (pos2 + 3) - pos1) << "'" << '\n';
+#endif
+    const std::string buf_trim = trim(buf); // trim whitespaces
+    if (buf_trim[0] != '/' || buf_trim[1] != '/') {
+#ifdef LOG_EMACS_MARKER
+        std::cout << path << " - Emacs marker not in a comment: '"  << buf.substr(pos1, (pos2 + 3) - pos1) << "'" << '\n';
+#endif
+        return false; // not a comment
+    }
+
+    // there are more variations with lowercase and no whitespaces
+    // -*- C++ -*-
+    // -*- Mode: C++; -*-
+    // -*- Mode: C++; c-basic-offset: 8 -*-
+    std::string marker = trim(buf.substr(pos1 + 3, pos2 - pos1 - 3), " ;");
+    // cut off additional attributes
+    const auto pos_semi = marker.find(';');
+    if (pos_semi != std::string::npos)
+        marker.resize(pos_semi);
+    findAndReplace(marker, "mode:", "");
+    findAndReplace(marker, "Mode:", "");
+    marker = trim(marker);
+    if (marker == "C++" || marker == "c++")
+        return true; // C++ marker found
+
+    //if (marker == "C" || marker == "c")
+    //    return false;
+#ifdef LOG_EMACS_MARKER
+    std::cout << path << " - unmatched Emacs marker: '"  << marker << "'" << '\n';
+#endif
+
+    return false; // marker is not a C++ one
+}
+
+Standards::Language Path::identify(const std::string &path, bool cppHeaderProbe, bool *header)
 {
     // cppcheck-suppress uninitvar - TODO: FP
     if (header)
         *header = false;
 
     std::string ext = getFilenameExtension(path);
+    // standard library headers have no extension
+    if (cppHeaderProbe && ext.empty()) {
+        if (hasEmacsCppMarker(path.c_str())) {
+            if (header)
+                *header = true;
+            return Standards::Language::CPP;
+        }
+        return Standards::Language::None;
+    }
     if (ext == ".C")
         return Standards::Language::CPP;
     if (c_src_exts.find(ext) != c_src_exts.end())
@@ -262,7 +353,9 @@ Standards::Language Path::identify(const std::string &path, bool *header)
     if (ext == ".h") {
         if (header)
             *header = true;
-        return Standards::Language::C; // treat as C for now
+        if (cppHeaderProbe && hasEmacsCppMarker(path.c_str()))
+            return Standards::Language::CPP;
+        return Standards::Language::C;
     }
     if (cpp_src_exts.find(ext) != cpp_src_exts.end())
         return Standards::Language::CPP;
@@ -277,7 +370,7 @@ Standards::Language Path::identify(const std::string &path, bool *header)
 bool Path::isHeader2(const std::string &path)
 {
     bool header;
-    (void)Path::identify(path, &header);
+    (void)identify(path, false, &header);
     return header;
 }
 

lib/path.h

@@ -187,10 +187,11 @@ class CPPCHECKLIB Path {
     /**
      * @brief Identify the language based on the file extension
      * @param path filename to check. path info is optional
+     * @param cppHeaderProbe check optional Emacs marker to identify extension-less and *.h files as C++
      * @param header if provided indicates if the file is a header
      * @return the language type
      */
-    static Standards::Language identify(const std::string &path, bool *header = nullptr);
+    static Standards::Language identify(const std::string &path, bool cppHeaderProbe, bool *header = nullptr);
 
     /**
      * @brief Get filename without a directory path part.

lib/preprocessor.cpp

@@ -684,7 +684,7 @@ static simplecpp::DUI createDUI(const Settings &mSettings, const std::string &cf
     dui.includes = mSettings.userIncludes;  // --include
     // TODO: use mSettings.standards.stdValue instead
     // TODO: error out on unknown language?
-    const Standards::Language lang = Path::identify(filename);
+    const Standards::Language lang = Path::identify(filename, mSettings.cppHeaderProbe);
     if (lang == Standards::Language::CPP) {
         dui.std = mSettings.standards.getCPP();
         splitcfg(mSettings.platform.getLimitsDefines(Standards::getCPP(dui.std)), dui.defines, "");

lib/settings.h

@@ -167,6 +167,9 @@ class CPPCHECKLIB WARN_UNUSED Settings {
     /** cppcheck.cfg: About text */
     std::string cppcheckCfgAbout;
 
+    /** @brief check Emacs marker to detect extension-less and *.h files as C++ */
+    bool cppHeaderProbe{};
+
     /** @brief Are we running from DACA script? */
     bool daca{};
 

lib/tokenlist.cpp

@@ -96,7 +96,7 @@ void TokenList::determineCppC()
     // only try to determine if it wasn't enforced
     if (mLang == Standards::Language::None) {
         ASSERT_LANG(!getSourceFilePath().empty());
-        mLang = Path::identify(getSourceFilePath());
+        mLang = Path::identify(getSourceFilePath(), mSettings ? mSettings->cppHeaderProbe : false);
         // TODO: cannot enable assert as this might occur for unknown extensions
         //ASSERT_LANG(mLang != Standards::Language::None);
         if (mLang == Standards::Language::None) {

releasenotes.txt

@@ -16,4 +16,5 @@ Deprecations:
 -
 
 Other:
+- added command-line option `--cpp-header-probe` (and `--no-cpp-header-probe`) to probe headers and extension-less files for Emacs marker (see https://trac.cppcheck.net/ticket/10692 for more details)
 -

test/cli/other_test.py

@@ -1428,3 +1428,31 @@ def test_filelist(tmpdir):
     for i in range(1, len(expected)+1):
         lines.remove('{}/11 files checked 0% done'.format(i, len(expected)))
     assert lines == expected
+
+
+def test_cpp_probe(tmpdir):
+    test_file = os.path.join(tmpdir, 'test.h')
+    with open(test_file, 'wt') as f:
+        f.writelines([
+            'class A {};'
+        ])
+
+    args = ['-q', '--template=simple', '--cpp-header-probe', test_file]
+    err_lines = [
+        "{}:1:1: error: Code 'classA{{' is invalid C code. Use --std or --language to configure the language. [syntaxError]".format(test_file)
+    ]
+
+    assert_cppcheck(args, ec_exp=0, err_exp=err_lines, out_exp=[])
+
+
+def test_cpp_probe_2(tmpdir):
+    test_file = os.path.join(tmpdir, 'test.h')
+    with open(test_file, 'wt') as f:
+        f.writelines([
+            '// -*- C++ -*-',
+            'class A {};'
+        ])
+
+    args = ['-q', '--template=simple', '--cpp-header-probe', test_file]
+
+    assert_cppcheck(args, ec_exp=0, err_exp=[], out_exp=[])

test/cli/testutils.py

@@ -163,7 +163,7 @@ def assert_cppcheck(args, ec_exp=None, out_exp=None, err_exp=None, env=None):
         assert exitcode == ec_exp, stdout
     if out_exp is not None:
         out_lines = stdout.splitlines()
-        assert out_lines == out_exp, stdout
+        assert out_lines == out_exp, out_lines
     if err_exp is not None:
         err_lines = stderr.splitlines()
-        assert err_lines == err_exp, stderr
+        assert err_lines == err_exp, err_lines