Merge pull request #254 from marcsanfacon/master

MinGW Port & 2 new options

AUTHORS

@@ -11,6 +11,7 @@ hasufell <julian.ospald@googlemail.com>
 Herbert Jones <herbert@mediafire.com>
 Hongliang Tian <tatetian@gmail.com>
 John Hewson <john@jahewson.com>
+Marc Sanfacon <marc.sanfacon@gmail.com>
 Michele Redolfi <michele@tecnicaict.com>
 Mick Giles <mick@mickgiles.com>
 Ryan Morlok <ryan.morlok@morlok.com>

CMakeLists.txt

@@ -130,7 +130,6 @@ else()
     set(PDF2HTMLEX_LIBS ${PDF2HTMLEX_LIBS} ${PYTHON_LIBRARIES})
 endif()
 
-
 # debug build flags (overwrite default cmake debug flags)
 set(CMAKE_C_FLAGS_DEBUG "-ggdb -pg")
 set(CMAKE_CXX_FLAGS_DEBUG "-ggdb -pg")
@@ -159,7 +158,7 @@ endif()
 include(CheckCXXCompilerFlag)
 check_cxx_compiler_flag("${CMAKE_CXX_FLAGS}" CXX0X_SUPPORT)
 if(NOT CXX0X_SUPPORT)
-    message(FATAL_ERROR "Error: you compiler does not support C++0x, please update it.")
+    message(FATAL_ERROR "Error: your compiler does not support C++0x, please update it.")
 endif()
 
 

pdf2htmlEX.1.in

@@ -82,6 +82,13 @@ This switch is useful if you want pages to be loaded separately & dynamically --
 
 Also see --page-filename.
 
+.TP
+.B --tmp-file-size-limit <limit> (Default: -1)
+This limits the total size of the temporary files which will also limit the total size of the output file.
+This is an estimate and it will stop after a page, once the total temporary files size is greater than this number.
+
+-1 means no limit and is the default.
+
 .TP
 .B --dest-dir <dir> (Default: .)
 Specify destination folder.
@@ -262,6 +269,10 @@ If switched off, intermediate files won't be cleaned in the end.
 .B --data-dir <dir> (Default: @CMAKE_INSTALL_PREFIX@/share/pdf2htmlEX)
 Specify the folder holding the manifest and other files (see below for the manifest file)`
 
+.TP
+.B --tmp-dir <dir> (Default: /tmp)
+Specify the temporary folder to use for temporary files
+
 .TP
 .B --css-draw <0|1> (Default: 0)
 Experimental and unsupported CSS drawing

src/HTMLRenderer/general.cc

@@ -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.tmp_file_size_limit != -1 && tmp_files.get_total_size() > param.tmp_file_size_limit * 1024) {
+            cerr << "Stop processing, reach max size\n";
+            break;
+        }
+
         cerr << "Working: " << (i-param.first_page) << "/" << page_count << '\r' << flush;
 
         if(param.split_pages)
@@ -394,20 +399,6 @@ void HTMLRenderer::post_process(void)
     long line_no = 0;
     while(getline(manifest_fin, line))
     {
-        ++line_no;
-
-        if(line == "\"\"\"")
-        {
-            embed_string = !embed_string;
-            continue;
-        }
-
-        if(embed_string)
-        {
-            output << line << endl;
-            continue;
-        }
-
         // trim space at both sides
         {
             static const char * whitespaces = " \t\n\v\f\r";
@@ -424,6 +415,20 @@ void HTMLRenderer::post_process(void)
             }
         }
 
+        ++line_no;
+
+        if(line == "\"\"\"")
+        {
+            embed_string = !embed_string;
+            continue;
+        }
+
+        if(embed_string)
+        {
+            output << line << endl;
+            continue;
+        }
+
         if(line.empty() || line[0] == '#')
             continue;
 

src/Param.h

@@ -31,6 +31,7 @@ struct Param
     int embed_javascript;
     int embed_outline;
     int split_pages;
+    int tmp_file_size_limit;
     std::string dest_dir;
     std::string css_filename;
     std::string page_filename;
@@ -69,13 +70,11 @@ struct Param
     // misc.
     int clean_tmp;
     std::string data_dir;
+    std::string tmp_dir;
     int css_draw;
     int debug;
 
     std::string input_filename, output_filename;
-
-    // not a paramater
-    std::string tmp_dir;
 };
 
 } // namespace pdf2htmlEX

src/TmpFiles.cc

@@ -9,12 +9,22 @@
 
 #include <iostream>
 #include <cstdio>
+#include <sys/stat.h>
 
 #include "TmpFiles.h"
 #include "Param.h"
 
 using namespace std;
 
+#ifndef _WIN32
+#   define STAT stat
+#   define RMDIR rmdir
+#else
+#   include <direct.h>
+#   define STAT _stat
+#   define RMDIR _rmdir
+#endif
+
 namespace pdf2htmlEX {
 
 
@@ -49,10 +59,23 @@ void TmpFiles::clean()
             cerr << "Remove temporary file: " << fn << endl;
     }
 
-    remove(param.tmp_dir.c_str());
+    RMDIR(param.tmp_dir.c_str());
     if(param.debug)
         cerr << "Remove temporary directory: " << param.tmp_dir << endl;
 }
 
+// Return the total size of the temporary files in bytes
+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

@@ -14,6 +14,7 @@ public:
     ~TmpFiles();
 
     void add( const std::string& fn);
+    double get_total_size() const;
 
 private:
     void clean();

src/pdf2htmlEX.cc

@@ -42,6 +42,12 @@ using namespace pdf2htmlEX;
 Param param;
 ArgParser argparser;
 
+#ifdef _WIN32
+#   include <iomanip>
+#   include <libgen.h>
+#   include <direct.h>
+#endif
+
 void deprecated_font_suffix(const char * dummy = nullptr)
 {
     cerr << "--font-suffix is deprecated. Use `--font-format` instead." << endl;
@@ -65,7 +71,7 @@ void show_version_and_exit(const char * dummy = nullptr)
 #if ENABLE_SVG
     cerr << "  cairo " << cairo_version_string() << endl;
 #endif
-    cerr << "Default data-dir: " << PDF2HTMLEX_DATA_PATH << endl;
+    cerr << "Default data-dir: " << param.data_dir << endl;
     cerr << "Supported image format:";
 #ifdef ENABLE_LIBPNG
     cerr << " png";
@@ -106,6 +112,41 @@ void embed_parser (const char * str)
     }
 }
 
+void prepare_directories()
+{
+    std::string tmp_dir = param.tmp_dir + "/pdf2htmlEX-XXXXXX";
+#ifndef _WIN32
+    errno = 0;
+
+    unique_ptr<char> pBuf(new char[tmp_dir.size() + 1]);
+    strcpy(pBuf.get(), tmp_dir.c_str());
+    auto p = mkdtemp(pBuf.get());
+    if(p == nullptr)
+    {
+        const char * errmsg = strerror(errno);
+        if(!errmsg)
+        {
+            errmsg = "unknown error";
+        }
+        cerr << "Cannot create temp directory: " << errmsg << endl;
+        exit(EXIT_FAILURE);
+    }
+    param.tmp_dir = pBuf.get();
+#else
+    srand((unsigned)time(0));
+    int rand_value = (int)((rand() / ((double)RAND_MAX+1.0)) * 1e6);
+    stringstream ss;
+    ss << setw(6) << rand_value;
+
+    tmp_dir.erase(tmp_dir.size() - 6);
+    param.tmp_dir = tmp_dir + ss.str();
+    if (mkdir(param.tmp_dir.c_str())) {
+        cerr << "Cannot create temp directory (" << param.tmp_dir << "): " << strerror(errno) << endl;
+        exit(EXIT_FAILURE);
+    }
+#endif
+}
+
 void parse_options (int argc, char **argv)
 {
     argparser
@@ -128,6 +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("tmp-file-size-limit", &param.tmp_file_size_limit, -1, "Limit the temporary file output size, in KB (-1 for no limit). This is only an estimate, the output may be bigger")
         .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")
@@ -168,7 +210,8 @@ void parse_options (int argc, char **argv)
 
         // misc.
         .add("clean-tmp", &param.clean_tmp, 1, "remove temporary files after conversion")
-        .add("data-dir", &param.data_dir, PDF2HTMLEX_DATA_PATH, "specify data directory")
+        .add("tmp-dir", &param.tmp_dir, param.tmp_dir, "specify the location of tempory directory.")
+        .add("data-dir", &param.data_dir, param.data_dir, "specify data directory")
         // TODO: css drawings are hidden on print, for annot links, need to fix it for other drawings
 //        .add("css-draw", &param.css_draw, 0, "[experimental and unsupported] CSS drawing")
         .add("debug", &param.debug, 0, "print debugging information")
@@ -317,29 +360,33 @@ void check_param()
 
 int main(int argc, char **argv)
 {
+    // We need to adjust these directories before parsing the options.
+#ifndef _WIN32
+    param.tmp_dir = "/tmp";
+    param.data_dir = PDF2HTMLEX_DATA_PATH;
+#else
+    {
+        // Under Windows, the default data_dir is under /data in the pdf2htmlEX directory
+        stringstream ss;
+        ss << dirname(argv[0]) << "/data";
+        param.data_dir = ss.str();
+
+        // Under Windows, the temp path is not under /tmp, find it.
+        char temppath[MAX_PATH];
+        ::GetTempPath(MAX_PATH, temppath);
+        param.tmp_dir = temppath;
+    }
+#endif
+
     parse_options(argc, argv);
     check_param();
 
     //prepare the directories
-    {
-        char buf[] = "/tmp/pdf2htmlEX-XXXXXX";
-        errno = 0;
-        auto p = mkdtemp(buf);
-        if(p == nullptr)
-        {
-            const char * errmsg = strerror(errno);
-            if(!errmsg)
-            {
-                errmsg = "unknown error";
-            }
-            cerr << "Cannot create temp directory: " << errmsg << endl;
-            exit(EXIT_FAILURE);
-        }
-        param.tmp_dir = buf;
-    }
+    prepare_directories();
 
-    if(param.debug)
+    if(param.debug) {
         cerr << "temporary dir: " << (param.tmp_dir) << endl;
+    }
 
     try
     {

src/util/path.cc

@@ -12,6 +12,15 @@
 
 #include "path.h"
 
+#ifdef _WIN32
+#   include <direct.h>
+#   define STAT _stat
+#   define MKDIR(A, B) _mkdir(A)
+#else
+#   define STAT stat
+#   define MKDIR(A, B) mkdir(A, B)
+#endif
+
 using std::string;
 
 namespace pdf2htmlEX {
@@ -26,13 +35,13 @@ void create_directories(const string & path)
         create_directories(path.substr(0, idx));
     }
 
-    int r = mkdir(path.c_str(), S_IRWXU);
+    int r = MKDIR(path.c_str(), S_IRWXU);
     if(r != 0)
     {
         if(errno == EEXIST)
         {
-            struct stat stat_buf;
-            if((stat(path.c_str(), &stat_buf) == 0) && S_ISDIR(stat_buf.st_mode))
+            struct STAT stat_buf;
+            if((STAT(path.c_str(), &stat_buf) == 0) && S_ISDIR(stat_buf.st_mode))
                 return;
         }