Added a parameter to limit the output size. This is an estimate, but should be good enough.

src/HTMLRenderer/general.cc

@@ -45,7 +45,7 @@ HTMLRenderer::HTMLRenderer(const Param & param)
     ,param(param)
     ,html_text_page(param, all_manager)
     ,preprocessor(param)
-	,tmp_files(param)
+    ,tmp_files(param)
 {
     if(!(param.debug))
     {
@@ -109,6 +109,11 @@ void HTMLRenderer::process(PDFDoc *doc)
     int page_count = (param.last_page - param.first_page + 1);
     for(int i = param.first_page; i <= param.last_page ; ++i)
     {
+        if (param.max_size != -1 && tmp_files.get_total_size() > param.max_size * 1024) {
+            cerr << "Stop processing, reach max size\n";
+            break;
+        }
+
         cerr << "Working: " << (i-param.first_page) << "/" << page_count << '\r' << flush;
 
         if(param.split_pages)

src/Param.h

@@ -31,6 +31,7 @@ struct Param
     int embed_javascript;
     int embed_outline;
     int split_pages;
+    int max_size;
     std::string dest_dir;
     std::string css_filename;
     std::string page_filename;
@@ -69,6 +70,7 @@ struct Param
     // misc.
     int clean_tmp;
     std::string data_dir;
+    std::string basetmp_dir;
     int css_draw;
     int debug;
 

src/TmpFiles.cc

@@ -9,6 +9,7 @@
 
 #include <iostream>
 #include <cstdio>
+#include <sys/stat.h>
 
 #include "TmpFiles.h"
 #include "Param.h"
@@ -19,7 +20,7 @@ namespace pdf2htmlEX {
 
 
 TmpFiles::TmpFiles( const Param& param )
-	: param( param )
+    : param( param )
 { }
 
 TmpFiles::~TmpFiles()
@@ -54,5 +55,16 @@ void TmpFiles::clean()
         cerr << "Remove temporary directory: " << param.tmp_dir << endl;
 }
 
+double TmpFiles::get_total_size() const
+{
+    double total_size = 0;
+    struct _stat st;
+    for(auto iter = tmp_files.begin(); iter != tmp_files.end(); ++iter) {
+        _stat(iter->c_str(), &st);
+        total_size += st.st_size;
+    }
+
+    return total_size;
+}
 
 } // namespace pdf2htmlEX

src/TmpFiles.h

@@ -13,13 +13,14 @@ public:
     explicit TmpFiles( const Param& param );
     ~TmpFiles();
 
-	void add( const std::string& fn);
+    void add( const std::string& fn);
+    double get_total_size() const;
 
 private:
-	void clean();
+    void clean();
 
     const Param& param;
-	std::set<std::string> tmp_files;
+    std::set<std::string> tmp_files;
 };
 
 } // namespace pdf2htmlEX

src/pdf2htmlEX.cc

@@ -141,7 +141,6 @@ void prepare_directories()
     stringstream ss;
     ss << setw(6) << rand_value;
 
-    std::cout << "1- " << tmp_dir << endl;
     tmp_dir.erase(tmp_dir.size() - 6);
     param.tmp_dir = tmp_dir + ss.str();
     ::CreateDirectory(param.tmp_dir.c_str(), NULL);
@@ -170,6 +169,7 @@ void parse_options (int argc, char **argv)
         .add("embed-image", &param.embed_image, 1, "embed image files into output")
         .add("embed-javascript", &param.embed_javascript, 1, "embed JavaScript files into output")
         .add("embed-outline", &param.embed_outline, 1, "embed outlines into output")
+        .add("max-output-size", &param.max_size, -1, "maximum output size, in KB (-1 for no max)")
         .add("split-pages", &param.split_pages, 0, "split pages into separate files")
         .add("dest-dir", &param.dest_dir, ".", "specify destination directory")
         .add("css-filename", &param.css_filename, "", "filename of the generated css file")
@@ -390,7 +390,6 @@ int main(int argc, char **argv)
         cerr << "temporary dir: " << (param.tmp_dir) << endl;
     }
 
-    exit(0);
     try
     {
         create_directories(param.dest_dir);