feat: input filters

src/lib/AST/ASTVisitor.cpp

@@ -16,6 +16,7 @@
 #include "ParseJavadoc.hpp"
 #include "lib/Support/Path.hpp"
 #include "lib/Support/Debug.hpp"
+#include "lib/Support/Glob.hpp"
 #include "lib/Lib/Diagnostics.hpp"
 #include "lib/Lib/Filters.hpp"
 #include "lib/Lib/Info.hpp"
@@ -1243,7 +1244,7 @@ class ASTVisitor
         const Decl* D,
         AccessSpecifier access)
     {
-        if(config_->inaccessibleMembers !=
+        if (config_->inaccessibleMembers !=
             ConfigImpl::SettingsImpl::ExtractPolicy::Always)
         {
             // KRYSTIAN FIXME: this doesn't handle direct
@@ -1253,6 +1254,44 @@ class ASTVisitor
                 return false;
         }
 
+        if (!config_->input.include.empty())
+        {
+            // Get filename
+            FileInfo* file = getFileInfo(D->getBeginLoc());
+            if (!file)
+                return false;
+            std::string filename = file->full_path;
+            bool matchPrefix = std::ranges::any_of(
+                config_->input.include,
+                [&filename](const std::string& prefix)
+                {
+                    return filename.starts_with(prefix);
+                });
+            if (!matchPrefix)
+            {
+                return false;
+            }
+        }
+
+        if (!config_->input.filePatterns.empty())
+        {
+            // Get filename
+            FileInfo* file = getFileInfo(D->getBeginLoc());
+            if (!file)
+                return false;
+            std::string filename = file->full_path;
+            bool matchPattern = std::ranges::any_of(
+                config_->input.filePatterns,
+                [&filename](const std::string& pattern)
+                {
+                    return globMatch(pattern, filename);
+                });
+            if (!matchPattern)
+            {
+                return false;
+            }
+        }
+
     #if 0
         bool extract = inExtractedFile(D);
         // if we're extracting a declaration as a dependency,
@@ -3433,19 +3472,20 @@ class ASTVisitorConsumer
         // so errors prior to traversal are reported
         Diagnostics diags;
 
+        // loads and caches source files into memory
         SourceManager& source = Context.getSourceManager();
         // get the name of the translation unit.
         // will be std::nullopt_t if it isn't a file
         std::optional<llvm::SmallString<128>> file_name =
             source.getNonBuiltinFilenameForID(source.getMainFileID());
         // KRYSTIAN NOTE: should we report anything here?
-        if(! file_name)
+        if (!file_name)
+        {
             return;
+        }
 
         // skip the translation unit if configured to do so
-        if(! config_.shouldVisitTU(
-            convert_to_slash(*file_name)))
-            return;
+        convert_to_slash(*file_name);
 
         ASTVisitor visitor(
             config_,
@@ -3454,7 +3494,7 @@ class ASTVisitorConsumer
             Context,
             *sema_);
 
-        // traverse the translation unit
+        // Traverse the translation unit
         visitor.build();
 
         // VFALCO If we returned from the function early
@@ -3537,6 +3577,14 @@ class ASTVisitorConsumer
 //
 //------------------------------------------------
 
+/** A frontend action for visiting the AST
+
+    This is used by the tooling infrastructure to create
+    an ASTAction for each translation unit.
+
+    The ASTAction is responsible for creating the ASTConsumer
+    which will be used to traverse the AST.
+*/
 struct ASTAction
     : public clang::ASTFrontendAction
 {
@@ -3548,27 +3596,57 @@ struct ASTAction
     {
     }
 
+    /** Execute the action
+
+        This is called by the tooling infrastructure to execute
+        the action for each translation unit.
+
+        The action will parse the AST with the consumer
+        that should have been previously created with
+        CreateASTConsumer.
+
+        This consumer then creates a ASTVisitor that
+        will convert the AST into a set of MrDocs Info
+        objects.
+     */
     void
     ExecuteAction() override
     {
         CompilerInstance& CI = getCompilerInstance();
-        if(! CI.hasPreprocessor())
+        if (!CI.hasPreprocessor())
+        {
             return;
+        }
 
-        // ensure comments in system headers are retained.
-        // we may want them if e.g. a declaration was extracted
+        // Ensure comments in system headers are retained.
+        // We may want them if, e.g., a declaration was extracted
         // as a dependency
         CI.getLangOpts().RetainCommentsFromSystemHeaders = true;
 
-        if(! CI.hasSema())
+        if (!CI.hasSema())
+        {
             CI.createSema(getTranslationUnitKind(), nullptr);
+        }
 
         ParseAST(
             CI.getSema(),
             false, // ShowStats
             true); // SkipFunctionBodies
     }
 
+    /** Create the object that will traverse the AST
+
+        This is called by the tooling infrastructure to create
+        an ASTConsumer for each translation unit.
+
+        This consumer creates a ASTVisitor that will convert
+        the AST into a set of our objects.
+
+        The main function of the ASTVisitorConsumer is
+        the HandleTranslationUnit function, which is called
+        to traverse the AST.
+
+     */
     std::unique_ptr<clang::ASTConsumer>
     CreateASTConsumer(
         clang::CompilerInstance& Compiler,
@@ -3585,6 +3663,11 @@ struct ASTAction
 
 //------------------------------------------------
 
+/** A frontend action factory for ASTAction
+
+    This is used by the tooling infrastructure to create
+    an ASTAction for each translation unit.
+*/
 struct ASTActionFactory :
     tooling::FrontendActionFactory
 {

src/lib/Lib/ConfigImpl.cpp

@@ -15,6 +15,7 @@
 #include "lib/Support/Error.hpp"
 #include "lib/Support/Path.hpp"
 #include "lib/Support/Yaml.hpp"
+#include "lib/Support/Glob.hpp"
 #include <mrdocs/Support/Path.hpp>
 #include <clang/Tooling/AllTUsExecution.h>
 #include <llvm/Support/FileSystem.h>
@@ -40,6 +41,7 @@ struct llvm::yaml::MappingTraits<
         SettingsImpl::FileFilter& f)
     {
         io.mapOptional("include", f.include);
+        io.mapOptional("file-patterns", f.filePatterns);
     }
 };
 
@@ -87,20 +89,21 @@ struct llvm::yaml::MappingTraits<SettingsImpl>
     static void mapping(IO& io,
         SettingsImpl& cfg)
     {
+        io.mapOptional("include",           cfg.ignoreFailures);
         io.mapOptional("defines",           cfg.defines);
         io.mapOptional("ignore-failures",   cfg.ignoreFailures);
 
+
         // io.mapOptional("extract",           cfg.extract);
         io.mapOptional("referenced-declarations", cfg.referencedDeclarations);
-        io.mapOptional("anonymous-namespaces", cfg.anonymousNamespaces);
-        io.mapOptional("inaccessible-members", cfg.inaccessibleMembers);
-        io.mapOptional("inaccessible-bases", cfg.inaccessibleBases);
-
+        io.mapOptional("anonymous-namespaces",    cfg.anonymousNamespaces);
+        io.mapOptional("inaccessible-members",    cfg.inaccessibleMembers);
+        io.mapOptional("inaccessible-bases",      cfg.inaccessibleBases);
 
         io.mapOptional("generate",          cfg.generate);
         io.mapOptional("multipage",         cfg.multiPage);
         io.mapOptional("source-root",       cfg.sourceRoot);
-        io.mapOptional("base-url",               cfg.baseURL);
+        io.mapOptional("base-url",          cfg.baseURL);
 
         io.mapOptional("input",             cfg.input);
 
@@ -209,7 +212,7 @@ ConfigImpl(
     }
 
     // Adjust input files
-    for(auto& name : inputFileIncludes_)
+    for (auto& name : settings_.input.include)
     {
         name = files::makePosixStyle(
             files::makeAbsolute(name, settings_.workingDir));
@@ -231,14 +234,34 @@ ConfigImpl(
 
 bool
 ConfigImpl::
-shouldVisitTU(
+shouldVisitSymbol(
     llvm::StringRef filePath) const noexcept
 {
-    if(inputFileIncludes_.empty())
+    if (settings_.input.include.empty())
+    {
         return true;
-    for(auto const& s : inputFileIncludes_)
-        if(filePath == s)
+    }
+    for (auto& p: settings_.input.include)
+    {
+        // Exact match
+        if (filePath == p)
+        {
             return true;
+        }
+        // Prefix match
+        if (filePath.startswith(p))
+        {
+            bool validPattern = std::ranges::any_of(
+                    settings_.input.filePatterns,
+                    [&](auto const &pattern) {
+                        return globMatch(pattern, filePath);
+                    });
+            if (validPattern)
+            {
+                return true;
+            }
+        }
+    }
     return false;
 }
 

src/lib/Lib/ConfigImpl.hpp

@@ -96,6 +96,9 @@ class ConfigImpl
         {
             /// Directories to include
             std::vector<std::string> include;
+
+            /// File patterns
+            std::vector<std::string> filePatterns;
         };
 
         /// @copydoc FileFilter
@@ -158,7 +161,6 @@ class ConfigImpl
     SettingsImpl settings_;
     ThreadPool& threadPool_;
     llvm::SmallString<0> outputPath_;
-    std::vector<std::string> inputFileIncludes_;
     dom::Object configObj_;
 
     friend class Config;
@@ -194,7 +196,7 @@ class ConfigImpl
         to the file being processed.
     */
     bool
-    shouldVisitTU(
+    shouldVisitSymbol(
         llvm::StringRef filePath) const noexcept;
 
     /** Returns true if the file should be visited.

src/lib/Lib/CorpusImpl.cpp

@@ -127,6 +127,8 @@ build(
     #else
         InfoExecutionContext context(*config);
     #endif
+    // Create an `ASTActionFactory` to create multiple
+    // `ASTAction`s that extract the AST for each translation unit.
     std::unique_ptr<tooling::FrontendActionFactory> action =
         makeFrontendActionFactory(context, *config);
     MRDOCS_ASSERT(action);
@@ -147,11 +149,13 @@ build(
             tooling::ClangTool Tool(compilations, { path },
                 std::make_shared<PCHContainerOperations>(), FS);
 
-            // suppress error messages from the tool
+            // Suppress error messages from the tool
             Tool.setPrintErrorMessage(false);
 
-            if(Tool.run(action.get()))
+            if (Tool.run(action.get()))
+            {
                 formatError("Failed to run action on {}", path).Throw();
+            }
         };
 
     // ------------------------------------------
@@ -182,8 +186,8 @@ build(
     else
     {
         TaskGroup taskGroup(config->threadPool());
-        for(std::size_t index = 0;
-            std::string& file : files)
+        std::size_t index = 0;
+        for (std::string& file : files)
         {
             taskGroup.async(
             [&, idx = ++index, path = std::move(file)]()
@@ -205,8 +209,10 @@ build(
     if (!errors.empty())
     {
         Error err(errors);
-        if(! (*config)->ignoreFailures)
+        if (!(*config)->ignoreFailures)
+        {
             return Unexpected(err);
+        }
         report::warn(
             "Warning: mapping failed because ", err);
     }
@@ -222,11 +228,7 @@ build(
         report::format(reportLevel,
             "Reducing declarations");
 
-        auto results = context.results();
-        if(! results)
-            return Unexpected(results.error());
-        corpus->info_ = std::move(results.value());
-
+        MRDOCS_TRY(corpus->info_, context.results());
         report::format(reportLevel,
             "Reduced {} symbols in {}",
                 corpus->info_.size(),