link validation working

CMakeLists.txt

@@ -13,7 +13,7 @@ project(kc
     VERSION 
         1.0 
     DESCRIPTION 
-        "C++ scratchpad" 
+        "Knowledge crawler for analysing notes" 
     LANGUAGES 
         CXX)
     

src/CMakeLists.txt

@@ -5,6 +5,7 @@
   parse/Link.cpp
   parse/FileContext.cpp
   parse/FileContextCache.cpp
+  valid/link.cpp
   logging.cpp
   config.cpp
 )

src/config.cpp

@@ -14,9 +14,14 @@ std::shared_ptr<po::variables_map> init_config(int argc, const char *argv[])
             ("help", "produce help message")
             ("path,p", po::value<std::string>()->default_value("."), "set root path of knowledge base")
             ("config", po::value<std::string>()->default_value("kc.ini"), "config file location")
+            ("command", po::value<std::string>(), "command to execute")
+            ("subargs", po::value<std::vector<std::string> >(), "Arguments for command")
             ("index", po::value<int>()->default_value(1), "index")
         ;
 
+        po::positional_options_description pos;
+        pos.add("command", 1).add("subargs", -1);
+
         po::options_description cmdline_options;
         cmdline_options.add(desc);
 
@@ -30,6 +35,8 @@ std::shared_ptr<po::variables_map> init_config(int argc, const char *argv[])
         auto vm = std::make_shared<po::variables_map>();
         po::store(po::command_line_parser(argc, argv)
                 .options(cmdline_options)
+                .positional(pos)
+                // .allow_unregistered()
                 .run(), 
             *vm);
 

src/fs/FileEntry.cpp

@@ -11,7 +11,7 @@ FileEntry::FileEntry(fs::directory_entry entry)
     
 }
 
-bool FileEntry::content_loaded()
+bool FileEntry::content_loaded() const
 {
     return loaded;
 }
@@ -25,7 +25,7 @@ std::string FileEntry::load_content()
     return file_content;
 }
 
-std::string FileEntry::get_content()
+std::string FileEntry::get_content() const
 {
     return file_content;
 }

src/fs/FileEntry.hpp

@@ -17,9 +17,9 @@ class FileEntry {
     fs::directory_entry file_entry;
     fs::path relative_path;
 
-    bool content_loaded();
+    bool content_loaded() const;
     std::string load_content();
-    std::string get_content();
+    std::string get_content() const;
     void clear_content();
 
     private:

src/fs/fs.cpp

@@ -5,9 +5,9 @@
 
 static const std::string exclusions[] = {".git", ".obsidian"};
 
-std::vector<kc::FileEntry> kc::walk_dir(std::string dir)
+std::vector<std::shared_ptr<kc::FileEntry>> kc::walk_dir(const std::string dir)
 {
-    auto matched = std::vector<kc::FileEntry>();
+    auto matched = std::vector<std::shared_ptr<kc::FileEntry>>();
     auto base_path = fs::path(dir);
 
     for (auto const& dir_entry : fs::recursive_directory_iterator(base_path))
@@ -23,14 +23,15 @@ std::vector<kc::FileEntry> kc::walk_dir(std::string dir)
             if (dir_entry_path_string.contains(exclusion))
             {
                 excluded = true;
+                break;
             }
         }
 
         if (!excluded) 
         {
-            auto entry = kc::FileEntry(dir_entry);
+            auto entry = std::make_shared<kc::FileEntry>(dir_entry);
 
-            entry.relative_path = fs::relative(dir_entry_path, base_path);
+            entry->relative_path = fs::relative(dir_entry_path, base_path);
             
             matched.push_back(entry);
         }

src/fs/fs.hpp

@@ -3,6 +3,7 @@
 #include <string>
 #include <filesystem>
 #include <vector>
+#include <memory>
 
 #include "FileEntry.hpp"
 
@@ -10,6 +11,6 @@ namespace fs = std::filesystem;
 
 namespace kc {
 
-std::vector<kc::FileEntry> walk_dir(std::string dir);
+std::vector<std::shared_ptr<kc::FileEntry>> walk_dir(std::string dir);
 
 }

src/logging.cpp

@@ -1,5 +1,6 @@
 #include "logging.hpp"
 
+
 namespace logging = boost::log;
 namespace src = boost::log::sources;
 namespace sinks = boost::log::sinks;
@@ -13,10 +14,11 @@ void init_logging()
     (
         keywords::file_name = "kc_%N.log",
         keywords::time_based_rotation = sinks::file::rotation_at_time_point(0, 0, 0),
-        keywords::format = "[%TimeStamp%] [%ThreadID%] [%Severity%] %Message%"
+        keywords::format = "[%TimeStamp%] [%ThreadID%] [%Severity%] %Message%",
+        keywords::open_mode = std::ios::app
     );
 
-    logging::add_console_log(std::cout, boost::log::keywords::format = "[%TimeStamp%] [%Severity%] >> %Message%");
+    // logging::add_console_log(std::cout, boost::log::keywords::format = "[%TimeStamp%] [%Severity%] >> %Message%");
 
     logging::core::get()->set_filter
     (

src/logging.hpp

@@ -1,5 +1,7 @@
 #pragma once
 
+#include <iostream>
+
 #include <boost/log/core.hpp>
 #include <boost/log/trivial.hpp>
 #include <boost/log/expressions.hpp>
@@ -7,4 +9,16 @@
 #include <boost/log/utility/setup/common_attributes.hpp>
 #include <boost/log/utility/setup/console.hpp>
 
-void init_logging();
+void init_logging();
+
+inline void print_and_log(std::string log_line)
+{
+    BOOST_LOG_TRIVIAL(info) << log_line;
+    std::cout << log_line << std::endl;
+}
+
+inline void print_and_log_error(std::string log_line)
+{
+    BOOST_LOG_TRIVIAL(error) << log_line;
+    std::cout << "ERROR: " << log_line << std::endl;
+}

src/main.cpp

@@ -10,6 +10,10 @@
 #include "config.hpp"
 #include "fs/fs.hpp"
 #include "parse/FileContextCache.hpp"
+#include "valid/link.hpp"
+
+void run_validate(std::shared_ptr<boost::program_options::variables_map> config);
+
 
 int main(int argc, const char *argv[]) {
     
@@ -18,45 +22,41 @@ int main(int argc, const char *argv[]) {
     BOOST_LOG_TRIVIAL(info) << "================================";
     BOOST_LOG_TRIVIAL(info) << "              kc";
     BOOST_LOG_TRIVIAL(info) << "================================";
-    BOOST_LOG_TRIVIAL(info) << "starting up....";
+    BOOST_LOG_TRIVIAL(info) << "Starting up....";
 
     auto config = init_config(argc, argv);
 
     if(config)
     {
-        auto env_path = (*config)["path"].as<std::string>();
-        BOOST_LOG_TRIVIAL(info) << "Loading knowledge base from " << env_path;
-
-        auto file_cache = kc::FileContextCache();
-        file_cache.load(env_path);
-        file_cache.parse_all();
-
-        auto context = file_cache.get()[(*config)["index"].as<int>()];
-        
-        std::cout << context->file_entry->get_content() << std::endl << std::endl << std::endl;
-
-        std::cout << "links: " << context->links.size() << std::endl;
-        std::cout << "images: " << context->images.size() << std::endl;
-        std::cout << "tags: " << context->tags.size() << std::endl << std::endl << std::endl;;
-
-        for (auto link : context->links)
+        if (config->count("command") == 1)
         {
-            std::cout << link.original_form << "  " << link.display << " --- " << link.link << std::endl;
-        }
+            auto command = (*config)["command"].as<std::string>();
 
-        std::cout << "tag cache: " << file_cache.tag_map.size() << std::endl;
-
-        for (auto tag : file_cache.tag_map)
-        {
-            std::cout << tag.first << ": ";
-
-            for (auto tag_entry: tag.second)
+            if (command == "validate")
             {
-                std::cout << tag_entry->relative_path << ", ";
+                run_validate(config);
             }
-
-            std::cout << std::endl;
+        }
+        else
+        {
+            print_and_log_error("Command not found, exiting");
+            return 1;
         }
 
+        return 0;
     }
+
+    return 1;
+}
+
+void run_validate(std::shared_ptr<boost::program_options::variables_map> config)
+{
+    auto env_path = (*config)["path"].as<std::string>();
+    print_and_log("> Loading knowledge base from " + env_path);
+
+    auto file_cache = kc::FileContextCache();
+    file_cache.load(env_path);
+    file_cache.parse_all();
+
+    kc::validate_links(file_cache.get());
 }

src/parse/FileContext.cpp

@@ -2,8 +2,8 @@
 
 namespace kc {
 
-FileContext::FileContext(kc::FileEntry entry)
-: file_entry(std::make_shared<kc::FileEntry>(entry))
+FileContext::FileContext(std::shared_ptr<kc::FileEntry> entry)
+: file_entry(entry)
 {
     
 }
@@ -33,7 +33,7 @@ void FileContext::parse()
     std::smatch image_match;
     while(std::regex_search(file_content, image_match, image_regex)) {
         
-        images.push_back(image_match.str());
+        images.push_back(kc::Link(image_match.str()));
         file_content = image_match.suffix();
     }
 

src/parse/FileContext.hpp

@@ -13,7 +13,7 @@ namespace kc {
 class FileContext {
     public:
 
-    FileContext(kc::FileEntry entry);
+    FileContext(std::shared_ptr<kc::FileEntry> entry);
 
     std::shared_ptr<kc::FileEntry> file_entry;
     std::vector<kc::Link> links;

src/parse/FileContextCache.cpp

@@ -1,20 +1,25 @@
 #include "FileContextCache.hpp"
+
+#include <algorithm>
+#include <execution>
+
 #include "../fs/fs.hpp"
 #include "../logging.hpp"
 
 namespace kc {
 
-void FileContextCache::load(std::string root_path)
+void FileContextCache::load(const std::string root_path)
 {
     BOOST_LOG_TRIVIAL(trace) << "Beginning cache load";
 
     auto entries = kc::walk_dir(root_path);
+    this->root_path.assign(root_path);
 
     for (auto entry : entries)
     {
-        if (entry.relative_path.extension() == ".md")
+        if (entry->relative_path.extension() == ".md")
         {
-            entry.load_content();
+            entry->load_content();
         }
 
         file_contexts.push_back(std::make_shared<kc::FileContext>(entry));
@@ -26,8 +31,14 @@ void FileContextCache::load(std::string root_path)
 void FileContextCache::parse_all()
 {
     tag_map.clear();
-    for (auto context: file_contexts)
-    {
+
+#if __APPLE__
+    std::for_each(file_contexts.begin(), file_contexts.end(), [this](std::shared_ptr<kc::FileContext> &context)
+#else
+    std::for_each(std::execution::par_unseq, file_contexts.begin(), file_contexts.end(), [this](std::shared_ptr<kc::FileContext> &context)
+#endif
+
+    { 
         if (context->file_entry->relative_path.extension() == ".md")
         {
             context->parse();
@@ -40,7 +51,7 @@ void FileContextCache::parse_all()
                 }
             }
         }
-    }
+    });
 }
 
 void FileContextCache::clear()
@@ -49,14 +60,19 @@ void FileContextCache::clear()
     file_contexts.shrink_to_fit();
 }
 
-size_t FileContextCache::size()
+size_t FileContextCache::size() const
 {
     return file_contexts.size();
 }
 
-std::vector<std::shared_ptr<kc::FileContext>> FileContextCache::get()
+std::vector<std::shared_ptr<kc::FileContext>> FileContextCache::get() const
 {
     return file_contexts;
 }
 
+std::string FileContextCache::get_root_path() const
+{
+    return root_path;
+}
+
 }

src/parse/FileContextCache.hpp

@@ -12,14 +12,16 @@ class FileContextCache {
     public:
     void load(std::string root_path);
     void clear();
-    size_t size();
-    std::vector<std::shared_ptr<kc::FileContext>> get();
+    size_t size() const;
+    std::vector<std::shared_ptr<kc::FileContext>> get() const;
     void parse_all();
+    std::string get_root_path() const;
 
     std::unordered_map<std::string, std::vector<std::shared_ptr<kc::FileEntry>>> tag_map;
 
     private:
     std::vector<std::shared_ptr<kc::FileContext>> file_contexts;
+    std::string root_path;
 };
 
 }

src/parse/Link.cpp

@@ -16,12 +16,20 @@ Link::Link(std::string original)
 
     link = original_form.substr(opening_link + 1, closing_link - opening_link - 1);
 
-    auto display_pos = original_form.find('#', opening_link);
+    external = link.starts_with("http");
 
-    if(display_pos != std::string::npos)
+    auto sublink_pos = original_form.find('#', opening_link);
+
+    if(sublink_pos != std::string::npos)
     {
-        display = original_form.substr(display_pos + 1, closing_link - display_pos - 1);
+        sublink = original_form.substr(sublink_pos + 1, closing_link - sublink_pos - 1);
+        link = original_form.substr(opening_link + 1, sublink_pos - opening_link - 1);
     }
 }
 
+bool Link::is_external() const
+{
+    return external;
+}
+
 }

src/parse/Link.hpp

@@ -12,10 +12,13 @@ class Link {
     std::string display;
     std::string link;
     std::string sublink;
+    bool is_external() const;
 
     Link(std::string original);
 
     private:
+
+    bool external;
 };
 
 }

src/valid/link.cpp

@@ -0,0 +1,48 @@
+#include "link.hpp"
+
+#include <filesystem>
+#include <iostream>
+
+#include "../logging.hpp"
+
+namespace fs = std::filesystem;
+
+namespace kc {
+
+std::vector<kc::FileLinkStateResult> validate_links(const std::vector<std::shared_ptr<kc::FileContext>> &contexts)
+{
+    std::vector<kc::FileLinkStateResult> ret;
+
+    auto invalid_counter = 0;
+
+    for (auto context : contexts)
+    {
+        if (context->links.size() > 0)
+        {
+            for (auto link: context->links)
+            {
+                if(!link.is_external()) {
+                    
+                    auto composed = context->file_entry->file_entry.path().parent_path() / fs::path(link.link);
+
+                    auto entry = fs::directory_entry(composed);
+
+                    if(!entry.exists())
+                    {
+                        print_and_log("Invalid link: " + std::string(context->file_entry->file_entry.path()) + " -> " + link.original_form);
+                        invalid_counter++;
+                    }
+                }
+            }
+        }
+    }
+
+    if (invalid_counter == 0)
+    {
+        print_and_log("All links valid");
+    }
+
+    return ret;
+}
+
+}

src/valid/link.hpp

@@ -0,0 +1,25 @@
+#pragma once
+
+#include <memory>
+#include <vector>
+#include "../parse/FileContext.hpp"
+
+namespace kc {
+
+enum LinkState {
+    VALID, INVALID
+};
+
+struct LinkStateResult {
+    LinkState link_state;
+    kc::Link link;
+};
+
+struct FileLinkStateResult {
+    std::shared_ptr<FileContext> file_context;
+    std::vector<LinkStateResult> link_states;
+};
+
+std::vector<FileLinkStateResult> validate_links(const std::vector<std::shared_ptr<kc::FileContext>> &contexts);
+
+}