link decomp, tag cache, image capture

src/const.hpp

@@ -3,6 +3,6 @@
 #include <string>
 
 static const std::string MD_LINK_REGEX = R"(\[.*?\]\(.*?\))";
-static const std::string MD_MD_LINK_REGEX = R"(\[.*?\]\(.*?\.md\))";
+static const std::string MD_MD_LINK_REGEX = R"(\[.*?\]\(.*?\.md(#.*?)*\))";
 static const std::string MD_IMAGE_LINK_REGEX = R"(!\[.*?\]\(.*?\.png\))";
-static const std::string MD_TAG_REGEX = R"(#{1}[^\s#.]+)";
+static const std::string MD_TAG_REGEX = R"((^|[[:blank:]])#{1}[^\s#.]+)";

src/main.cpp

@@ -33,14 +33,29 @@ int main(int argc, const char *argv[]) {
 
         auto context = file_cache.get()[(*config)["index"].as<int>()];
         
-        std::cout << context->file_entry.get_content() << std::endl << std::endl << std::endl;
+        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)
         {
-            std::cout << link.original_form << std::endl;
+            std::cout << link.original_form << "  " << link.display << " --- " << link.link << std::endl;
+        }
+
+        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)
+            {
+                std::cout << tag_entry->relative_path << ", ";
+            }
+
+            std::cout << std::endl;
         }
 
     }

src/parse/FileContext.cpp

@@ -3,23 +3,24 @@
 namespace kc {
 
 FileContext::FileContext(kc::FileEntry entry)
-: file_entry(entry)
+: file_entry(std::make_shared<kc::FileEntry>(entry))
 {
     
 }
 
 void FileContext::parse()
 {
-    if (!file_entry.content_loaded())
+    if (!file_entry->content_loaded())
     {
         throw std::logic_error("cannot parse from file entry as it has not been loaded");
     }
 
     links.clear();
+    images.clear();
     tags.clear();
 
     std::regex link_regex(MD_MD_LINK_REGEX);
-    std::string file_content = file_entry.get_content();
+    std::string file_content = file_entry->get_content();
     std::smatch link_match;
     while(std::regex_search(file_content, link_match, link_regex)) {
         
@@ -27,12 +28,21 @@ void FileContext::parse()
         file_content = link_match.suffix();
     }
 
-    std::regex tag_regex(MD_TAG_REGEX);
-    file_content = file_entry.get_content();
+    std::regex image_regex(MD_IMAGE_LINK_REGEX);
+    file_content = file_entry->get_content();
+    std::smatch image_match;
+    while(std::regex_search(file_content, image_match, image_regex)) {
+        
+        images.push_back(image_match.str());
+        file_content = image_match.suffix();
+    }
+
+    std::regex tag_regex(MD_TAG_REGEX, std::regex::multiline);
+    file_content = file_entry->get_content();
     std::smatch tag_match;
     while(std::regex_search(file_content, tag_match, tag_regex)) {
         
-        tags.push_back(tag_match.str());
+        tags.push_back(tag_match.str().substr(1));
         file_content = tag_match.suffix();
     }
 

src/parse/FileContext.hpp

@@ -1,6 +1,7 @@
 #pragma once
 
 #include <vector>
+#include <memory>
 #include <regex>
 
 #include "../fs/FileEntry.hpp"
@@ -14,8 +15,9 @@ class FileContext {
 
     FileContext(kc::FileEntry entry);
 
-    kc::FileEntry file_entry;
+    std::shared_ptr<kc::FileEntry> file_entry;
     std::vector<kc::Link> links;
+    std::vector<kc::Link> images;
     std::vector<std::string> tags;
 
     void parse();

src/parse/FileContextCache.cpp

@@ -25,11 +25,20 @@ void FileContextCache::load(std::string root_path)
 
 void FileContextCache::parse_all()
 {
+    tag_map.clear();
     for (auto context: file_contexts)
     {
-        if (context->file_entry.relative_path.extension() == ".md")
+        if (context->file_entry->relative_path.extension() == ".md")
         {
             context->parse();
+
+            if (context->tags.size() != 0)
+            {
+                for (auto tag : context->tags)
+                {
+                    tag_map[tag].push_back(context->file_entry);
+                }
+            }
         }
     }
 }

src/parse/FileContextCache.hpp

@@ -2,6 +2,8 @@
 
 #include <vector>
 #include <memory>
+#include <string>
+#include <unordered_map>
 #include "FileContext.hpp"
 
 namespace kc {
@@ -14,6 +16,8 @@ class FileContextCache {
     std::vector<std::shared_ptr<kc::FileContext>> get();
     void parse_all();
 
+    std::unordered_map<std::string, std::vector<std::shared_ptr<kc::FileEntry>>> tag_map;
+
     private:
     std::vector<std::shared_ptr<kc::FileContext>> file_contexts;
 };

src/parse/Link.cpp

@@ -3,9 +3,25 @@
 namespace kc {
 
 Link::Link(std::string original)
-: original_form(original)
 {
+    original_form = std::regex_replace(original, std::regex("%20"), " ");
 
+    auto opening_display = original_form.find('[');
+    auto closing_display = original_form.find(']', opening_display);
+
+    display = original_form.substr(opening_display + 1, closing_display - opening_display - 1);
+
+    auto opening_link = closing_display + 1;
+    auto closing_link = original_form.rfind(')');
+
+    link = original_form.substr(opening_link + 1, closing_link - opening_link - 1);
+
+    auto display_pos = original_form.find('#', opening_link);
+
+    if(display_pos != std::string::npos)
+    {
+        display = original_form.substr(display_pos + 1, closing_link - display_pos - 1);
+    }
 }
 
 }

src/parse/Link.hpp

@@ -1,6 +1,7 @@
 #pragma once
 
 #include <string>
+#include <regex>
 
 namespace kc {