added levenstein distance search

frate-dev · Oct 30, 2023 · 689d969 · 689d969
1 parent 23335e8
commit 689d969
Show file tree

Hide file tree

Showing 4 changed files with 108 additions and 176 deletions.
diff --git a/install_manifest.txt b/install_manifest.txt
@@ -1,82 +1 @@
-/usr/local/bin/cmaker
-/usr/local/lib/libz.so.1.2.13.zlib-ng
-/usr/local/lib/libz.so.1
-/usr/local/lib/libz.so
-/usr/local/include/zlib.h
-/usr/local/include/zlib_name_mangling.h
-/usr/local/include/zconf.h
-/usr/local/lib/pkgconfig/zlib.pc
-/usr/local/lib/libcurl.so.4.8.0
-/usr/local/lib/libcurl.so.4
-/usr/local/lib/libcurl.so
-/usr/local/bin/curl-config
-/usr/local/lib/pkgconfig/libcurl.pc
-/usr/local/include/curl/multi.h
-/usr/local/include/curl/options.h
-/usr/local/include/curl/urlapi.h
-/usr/local/include/curl/header.h
-/usr/local/include/curl/websockets.h
-/usr/local/include/curl/curlver.h
-/usr/local/include/curl/system.h
-/usr/local/include/curl/typecheck-gcc.h
-/usr/local/include/curl/easy.h
-/usr/local/include/curl/stdcheaders.h
-/usr/local/include/curl/curl.h
-/usr/local/include/curl/mprintf.h
-/usr/local/lib/cmake/CURL/CURLTargets.cmake
-/usr/local/lib/cmake/CURL/CURLTargets-release.cmake
-/usr/local/lib/cmake/CURL/CURLConfigVersion.cmake
-/usr/local/lib/cmake/CURL/CURLConfig.cmake
-/usr/local/lib/libcpr.so.1.11.0
-/usr/local/lib/libcpr.so.1
-/usr/local/lib/libcpr.so
-/usr/local/include/cpr/curlmultiholder.h
-/usr/local/include/cpr/file.h
-/usr/local/include/cpr/util.h
-/usr/local/include/cpr/auth.h
-/usr/local/include/cpr/range.h
-/usr/local/include/cpr/interceptor.h
-/usr/local/include/cpr/interface.h
-/usr/local/include/cpr/http_version.h
-/usr/local/include/cpr/redirect.h
-/usr/local/include/cpr/async.h
-/usr/local/include/cpr/bearer.h
-/usr/local/include/cpr/cert_info.h
-/usr/local/include/cpr/user_agent.h
-/usr/local/include/cpr/filesystem.h
-/usr/local/include/cpr/cookies.h
-/usr/local/include/cpr/ssl_ctx.h
-/usr/local/include/cpr/reserve_size.h
-/usr/local/include/cpr/low_speed.h
-/usr/local/include/cpr/unix_socket.h
-/usr/local/include/cpr/connect_timeout.h
-/usr/local/include/cpr/local_port.h
-/usr/local/include/cpr/ssl_options.h
-/usr/local/include/cpr/cpr.h
-/usr/local/include/cpr/body.h
-/usr/local/include/cpr/buffer.h
-/usr/local/include/cpr/api.h
-/usr/local/include/cpr/multiperform.h
-/usr/local/include/cpr/response.h
-/usr/local/include/cpr/threadpool.h
-/usr/local/include/cpr/error.h
-/usr/local/include/cpr/payload.h
-/usr/local/include/cpr/proxyauth.h
-/usr/local/include/cpr/accept_encoding.h
-/usr/local/include/cpr/local_port_range.h
-/usr/local/include/cpr/callback.h
-/usr/local/include/cpr/curlholder.h
-/usr/local/include/cpr/limit_rate.h
-/usr/local/include/cpr/status_codes.h
-/usr/local/include/cpr/parameters.h
-/usr/local/include/cpr/verbose.h
-/usr/local/include/cpr/resolve.h
-/usr/local/include/cpr/session.h
-/usr/local/include/cpr/cprtypes.h
-/usr/local/include/cpr/curl_container.h
-/usr/local/include/cpr/multipart.h
-/usr/local/include/cpr/proxies.h
-/usr/local/include/cpr/timeout.h
-/usr/local/include/cpr/singleton.h
-/usr/local/include/cpr/async_wrapper.h
-/usr/local/include/cpr/cprver.h
+/usr/local/bin/cmaker
diff --git a/src/Command/Command.hpp b/src/Command/Command.hpp
@@ -58,6 +58,13 @@ namespace Command {
   ParseResult removeOptions(int argc, char** argv);
   ParseResult updateOptions(int argc, char** argv);
 
+  typedef struct packageResult {
+    std::string name;
+    std::string url;
+    std::string version;
+    int score;
+  } packageResult;
+
   typedef struct dependency {
     std::string name;
     std::string url;
@@ -134,6 +141,8 @@ namespace Command {
    * What the fuck lucas
    */
 
+  std::vector<packageResult> searchPackage(std::string query);
+
   std::string downloadIndex();
   /*
    *  adds  dependency to toml and regenerates the CMakeLists.txt

diff --git a/src/Command/CommandAdd.cpp b/src/Command/CommandAdd.cpp
@@ -38,76 +38,6 @@ namespace Command {
     return true;
   }
 
-  typedef struct packageResult {
-    std::string name;
-    std::string url;
-    std::string version;
-    int score;
-  } packageResult;
-
-  std::vector<packageResult> calculatePackageScores(std::vector<std::string> queryTokens){
-    std::vector<packageResult> results;
-    json rawIndex = Utils::fetchJson("https://github.com/cmaker-dev/index/releases/latest/download/index.json"); 
-    int lengthDiffAbs = 0;
-    for(std::string query: queryTokens){
-      for(json package: rawIndex){
-        //TODO: add description that can be searched
-        results.push_back({
-            .name = package["name"],
-            .url = package["git"],
-            .score = 0
-            });
-        //if it is an exact match
-        if(results.back().name == query){
-          results.back().score += 100;
-          break;
-        }
-        //Check if the query is in the package name
-        if(results.back().name.find(query) != std::string::npos){
-          results.back().score += 50;
-        }
-
-        lengthDiffAbs = std::abs((int)results.back().name.length() - (int)query.length());
-        if(results.back().score > 10){
-          results.back().score -= lengthDiffAbs;
-        }
-      }
-    }
-    std::sort(results.begin(), results.end(), [](packageResult a, packageResult b){
-        return a.score > b.score;
-        });
-    std::vector<packageResult> resultsUnique;
-
-    for(packageResult result: results){
-      if(std::any_of(resultsUnique.begin(), resultsUnique.end(), [&result](packageResult &r){
-            r.score += result.score;
-            return r.name == result.name;
-            })){
-        continue;
-      }
-      resultsUnique.push_back(result);
-    }
-    return resultsUnique;
-  }
-
-  std::vector<packageResult> searchPackage(std::string query){
-    std::cout << "Searching for package" << std::endl;
-
-    std::vector<std::string> queryTokens = Utils::split(query, ' ');
-
-    std::vector<packageResult> results = calculatePackageScores(queryTokens);
-
-
-    std::cout << "Results:" << std::endl;
-    for(int i = results.size();i > -1; i--){
-      if(results[i].score > 10){
-        //Adjusting the indexes for the normies
-        std::cout << "[" << i << "]" << results[i].name << " - (" << results[i].url << ")" << std::endl;
-      }
-    }
-
-    return results;
-  }
 
 
   bool checkForOverlappingDependencies(std::shared_ptr<Context> ctx, std::string name){

diff --git a/src/Command/CommandSearchPackage.cpp b/src/Command/CommandSearchPackage.cpp
@@ -2,34 +2,108 @@
 #include <curl/curl.h>
 #include <curl/easy.h>
 #include <nlohmann/json.hpp>
+#include "../Utils/General.hpp"
 
 
 
 namespace Command {
-//   const std::string RepoIndex = "https://raw.githubusercontent.com/cmaker-dev/index/main/index.json";
-//   void writeCallback(char* ptr, size_t size, size_t nmemb, std::string* data) {
-//     data->append(ptr, size * nmemb);
-//   }
-//   void SearchPackage(){
-//     curl_easy_init();
-//     std::shared_ptr<CURL> curl(curl_easy_init(), curl_easy_cleanup);
-// 
-//     curl_easy_setopt(curl.get(), CURLOPT_URL, RepoIndex.c_str());
-// 
-//     std::string response;
-// 
-//     curl_easy_setopt(curl.get(), CURLOPT_WRITEFUNCTION, writeCallback);
-//     curl_easy_setopt(curl.get(), CURLOPT_WRITEDATA, &response);
-// 
-//     curl_easy_perform(curl.get());
-// 
-//     nlohmann::json j = nlohmann::json::parse(response);
-// 
-//     std::cout << j.dump(4) << std::endl;
-// 
-// 
-// 
-//   }
+  using nlohmann::json;
 
+  int levensteinDistance(std::string aStr, std::string bStr){
+    //minimize the amount of memory used
+    std::string* a = &aStr;
+    std::string* b = &bStr;
+    if(aStr.length() > bStr.length()){
+      a = &bStr;
+      b = &aStr;
+    }
+    int aLen = a->length();
+    int bLen = b->length();
+    int* prev = new int[aLen + 1];
+    int* curr = new int[aLen + 1];
+    for(int i = 0; i <= aLen; i++){
+      prev[i] = i;
+    }
+
+    for(int i = 1; i <= bLen; i++){
+      curr[0] = i;
+      for(int j = 1; j <= aLen; j++){
+        int cost = (*a)[j - 1] == (*b)[i - 1] ? 0 : 1;
+        curr[j] = std::min(std::min(curr[j - 1] + 1, prev[j] + 1), prev[j - 1] + cost);
+      }
+      int* temp = prev;
+      prev = curr;
+      curr = temp;
+    }
+    int result = prev[aLen];
+    delete[] prev;
+    delete[] curr;
+    return result;
+  }
+  std::vector<packageResult> calculatePackageScores(std::vector<std::string> queryTokens){
+    std::vector<packageResult> results;
+    json rawIndex = Utils::fetchJson("https://github.com/cmaker-dev/index/releases/latest/download/index.json"); 
+    int lengthDiffAbs = 0;
+    for(std::string query: queryTokens){
+      for(json package: rawIndex){
+        //TODO: add description that can be searched
+        results.push_back({
+            .name = package["name"],
+            .url = package["git"],
+            .score = 0
+            });
+        //if it is an exact match
+        if(results.back().name == query){
+          results.back().score += 100;
+        }
+        //Check if the query is in the package name
+        if(results.back().name.find(query) != std::string::npos){
+          results.back().score += 50;
+        }
+        if(levensteinDistance(results.back().name, query) < 3){
+          results.back().score += 10;
+        }
+
+        lengthDiffAbs = std::abs((int)results.back().name.length() - (int)query.length());
+        if(results.back().score > 20){
+          results.back().score -= lengthDiffAbs;
+        }
+      }
+    }
+    std::sort(results.begin(), results.end(), [](packageResult a, packageResult b){
+        return a.score > b.score;
+        });
+    std::vector<packageResult> resultsUnique;
+
+    for(packageResult result: results){
+      if(std::any_of(resultsUnique.begin(), resultsUnique.end(), [&result](packageResult &r){
+            r.score += result.score;
+            return r.name == result.name;
+            })){
+        continue;
+      }
+      resultsUnique.push_back(result);
+    }
+    return resultsUnique;
+  }
+
+  std::vector<packageResult> searchPackage(std::string query){
+    std::cout << "Searching for package" << std::endl;
+
+    std::vector<std::string> queryTokens = Utils::split(query, ' ');
+
+    std::vector<packageResult> results = calculatePackageScores(queryTokens);
+
+
+    std::cout << "Results:" << std::endl;
+    for(int i = results.size();i > -1; i--){
+      if(results[i].score > 10){
+        //Adjusting the indexes for the normies
+        std::cout << "[" << i << "]" << "[s:" << results[i].score << "] " << results[i].name << " - (" << results[i].url << ")" << std::endl;
+      }
+    }
+
+    return results;
+  }
 
 }