From 315c2c3793987752fd0713d17764ca49aa3275c5 Mon Sep 17 00:00:00 2001
From: Leonardo Domingues <leonardo.araujo@lavid.ufpb.br>
Date: Thu, 12 May 2016 15:52:57 -0300
Subject: [PATCH] Fix the problem to read empty lines in the field text from SRT subtitle

---
 extrator/src/extratorSRT.cpp       | 42 +++++++++++++++++++++++++++++++++---------
 extrator/src/include/extratorSRT.h |  1 +
 recognize/src/recognizer.cpp       |  4 ++--
 3 files changed, 36 insertions(+), 11 deletions(-)

diff --git a/extrator/src/extratorSRT.cpp b/extrator/src/extratorSRT.cpp
index bcde839..4fa45a4 100644
--- a/extrator/src/extratorSRT.cpp
+++ b/extrator/src/extratorSRT.cpp
@@ -4,11 +4,15 @@ ExtratorSRT::ExtratorSRT(){
     listeners = new list<ListenerSub*>();
     finish = false;
     seek_pos = 0;
+
+    // Used to watch a sequence of the subtitles during the extraction
+    index_counter = 1; 
+    
     hasNextSub = true;
     PRINTL(util::_DEBUG, "ExtratorSTR Done!\n");
 }
 
-ExtratorSRT::~ExtratorSRT(){
+ExtratorSRT::~ExtratorSRT() {
     listeners->clear();
     delete listeners;
     if (file_io) delete file_io;
@@ -110,14 +114,26 @@ Subtitle* ExtratorSRT::next() {
         string text_sub = "";
 
         try {
-            /* ID */
-            int id = 0;
-            line = bff_reader->readLine();
-            seek_pos += (int64_t) line.size() + SIZE_CSCAPE;
-            id = atoi(line.c_str());
-            sub->setID(id);
+            // ID
+            int id_sub = -1;
+
+            // Case exists some blank lines before index, read until find
+            do {
+                line = bff_reader->readLine();
+                printf("%s\n", line.c_str());
+                seek_pos++;
+            } while (line.size() < 1);
+            
+            //seek_pos += (int64_t) line.size() + SIZE_CSCAPE;
+            id_sub = atoi(line.c_str());
+            if (id_sub != index_counter)
+            {
+                PRINTL(util::_DEBUG, "[Error] The SRT file is bad formmated: indexes is not continuous.\n");
+                exit (1);
+            }
+            sub->setID(id_sub);
 
-            /* TimeIn and TimeOut */
+            // TimeIn and TimeOut
             int64_t t_in = 0, t_out = 0;
             line = bff_reader->readLine();
             seek_pos += (int64_t) line.size() + SIZE_CSCAPE;
@@ -128,24 +144,32 @@ Subtitle* ExtratorSRT::next() {
             t_out = str_to_time(line.substr(target_pos + strlen(TARGET_TIME)+1, line.size()));
             sub->setTimeOut(t_out);
 
-            /* Text: read until line be empty */
+            // Read until find an empty line
             while ((line = bff_reader->readLine()).size() > 0) {
                 text_sub += line;
                 text_sub.append(" ");
             }
+            printf("size of text: %d\n", text_sub.size());
             seek_pos += (int64_t) text_sub.size() + SIZE_CSCAPE;
 
         } catch (lavidlib::EOFException &ex) {
+
             if(text_sub == "")
                 sub->setTimeIn(0); //seta o valor 0 para nao gerar um valor aleatório
 
             sub->setSubtitleText(formatText(text_sub));
             seek_pos += (int64_t) text_sub.size() + SIZE_CSCAPE;
+
             hasNextSub = false;
             delete(bff_reader);
             return sub;
         }
+        index_counter++;
+
         sub->setSubtitleText(formatText(text_sub));
+        
+        printf("%s\n\n", sub->toString().c_str());
+
         delete(bff_reader);
         return sub;
     }
diff --git a/extrator/src/include/extratorSRT.h b/extrator/src/include/extratorSRT.h
index ecd8c65..742e5e7 100644
--- a/extrator/src/include/extratorSRT.h
+++ b/extrator/src/include/extratorSRT.h
@@ -104,6 +104,7 @@ private:
     Subtitle *subtitle;
     int64_t seek_pos;
     bool hasNextSub;
+    int index_counter;
 
     void encodingfiletoUTF8();
     string formatText(string line);
diff --git a/recognize/src/recognizer.cpp b/recognize/src/recognizer.cpp
index 74e58af..79d2879 100644
--- a/recognize/src/recognizer.cpp
+++ b/recognize/src/recognizer.cpp
@@ -2,10 +2,10 @@
 
 
 #include "recognizer.h"
-#include <json/json.h>
+#include <jsoncpp/json/json.h>
 
 
-static boolean reconhecendo ;
+static boolean reconhecendo;
 static string sentence;
 static float confidence;
 string lenguage = "pt-BR";
--
libgit2 0.21.2