From 06e4672540094a851542b47abaf022f934b63b09 Mon Sep 17 00:00:00 2001
From: Star Rauchenberger <fefferburbia@gmail.com>
Date: Fri, 9 Dec 2022 16:53:33 -0500
Subject: Generator now splits ImageNet list into per-notion files

---
 generator/generator.cpp | 23 +++++++++++++++--------
 generator/generator.h   |  5 ++++-
 generator/main.cpp      |  7 ++++---
 3 files changed, 23 insertions(+), 12 deletions(-)

diff --git a/generator/generator.cpp b/generator/generator.cpp
index fdea10f..969b629 100644
--- a/generator/generator.cpp
+++ b/generator/generator.cpp
@@ -23,13 +23,15 @@ namespace verbly {
       std::string wordNetPath,
       std::string cmudictPath,
       std::string imageNetPath,
-      std::string outputPath) :
+      std::string outputPath,
+      std::string imageNetOutput) :
         verbNetPath_(verbNetPath),
         agidPath_(agidPath),
         wordNetPath_(wordNetPath),
         cmudictPath_(cmudictPath),
         imageNetPath_(imageNetPath),
-        db_(outputPath, hatkirby::dbmode::create)
+        db_(outputPath, hatkirby::dbmode::create),
+        imageNetOutput_(imageNetOutput)
     {
       // Ensure VerbNet directory exists
       DIR* dir;
@@ -75,6 +77,9 @@ namespace verbly {
       {
         throw std::invalid_argument("ImageNet urls.txt file not found");
       }
+
+      // Create the ImageNet output directory
+      std::filesystem::create_directory(imageNetOutput_);
     }
 
     void generator::run()
@@ -86,7 +91,7 @@ namespace verbly {
       readAdjectivePositioning();
 
       // Counts the number of URLs ImageNet has per notion
-      //readImageNetUrls();
+      readImageNetUrls();
 
       // Creates a word by WordNet sense key lookup table
       readWordNetSenseKeys();
@@ -118,17 +123,15 @@ namespace verbly {
       // Writes the database version
       writeVersion();
 
+      // Dumps data to the database
+      dumpObjects();
+
       // Calculates and writes form merography
       writeMerography();
 
       // Calculates and writes pronunciation merophony
       writeMerophony();
 
-      // Dumps data to the database
-      dumpObjects();
-
-
-
       // Populates the antonymy relationship from WordNet
       readWordNetAntonymy();
 
@@ -286,6 +289,10 @@ namespace verbly {
         {
           // We know that this notion has a wnid and is a noun.
           notionByWnid_.at(wnid)->incrementNumOfImages();
+
+          std::filesystem::path imagefile = imageNetOutput_ / std::to_string(notionByWnid_[wnid]->getId());
+          std::ofstream imageoutput(imagefile, std::ios::app);
+          imageoutput << line.substr(line.find("\t") + 1) << std::endl;
         }
       }
     }
diff --git a/generator/generator.h b/generator/generator.h
index 70f0657..3d51c35 100644
--- a/generator/generator.h
+++ b/generator/generator.h
@@ -9,6 +9,7 @@
 #include <set>
 #include <libxml/parser.h>
 #include <hkutil/database.h>
+#include <filesystem>
 #include "notion.h"
 #include "word.h"
 #include "lemma.h"
@@ -34,7 +35,8 @@ namespace verbly {
         std::string wordNetPath,
         std::string cmudictPath,
         std::string imageNetPath,
-        std::string outputPath);
+        std::string outputPath,
+        std::string imageNetOutput);
 
       // Action
 
@@ -127,6 +129,7 @@ namespace verbly {
       // Output
 
       hatkirby::database db_;
+      std::filesystem::path imageNetOutput_;
 
       // Data
 
diff --git a/generator/main.cpp b/generator/main.cpp
index 1b07706..7db7203 100644
--- a/generator/main.cpp
+++ b/generator/main.cpp
@@ -4,22 +4,23 @@
 
 void printUsage()
 {
-  std::cout << "usage: generator verbnet agid wordnet cmudict imagenet output" << std::endl;
+  std::cout << "usage: generator verbnet agid wordnet cmudict imagenet output ino" << std::endl;
   std::cout << "verbnet  :: path to a VerbNet data directory" << std::endl;
   std::cout << "agid     :: path to an AGID infl.txt file" << std::endl;
   std::cout << "wordnet  :: path to a WordNet prolog data directory" << std::endl;
   std::cout << "cmudict  :: path to a CMUDICT pronunciation file" << std::endl;
   std::cout << "imagenet :: path to an ImageNet urls.txt file" << std::endl;
   std::cout << "output   :: datafile output path" << std::endl;
+  std::cout << "ino      :: imagenet directory output path" << std::endl;
 }
 
 int main(int argc, char** argv)
 {
-  if (argc == 7)
+  if (argc == 8)
   {
     try
     {
-      verbly::generator::generator app(argv[1], argv[2], argv[3], argv[4], argv[5], argv[6]);
+      verbly::generator::generator app(argv[1], argv[2], argv[3], argv[4], argv[5], argv[6], argv[7]);
 
       try
       {
-- 
cgit 1.4.1