From 4a908db655e2ebf076b2fb629bce41c3ae3dbc55 Mon Sep 17 00:00:00 2001
From: Lu Wang <coolwanglu@gmail.com>
Date: Tue, 11 Sep 2012 21:52:46 +0800
Subject: [PATCH] fixed: cannot parse arguments with spaces; add namespace
 pdf2htmlEX

---
 src/ArgParser.cc                   |  25 ++++--
 src/BackgroundRenderer.cc          |   2 +
 src/FontPreprocessor.cc            |   4 +
 src/HTMLRenderer/general.cc        |  11 ++-
 src/HTMLRenderer/install.cc        |   4 +-
 src/include/ArgParser.h            | 133 ++++++++++++++++-------------
 src/include/BackgroundRenderer.h   |   5 +-
 src/include/FontPreprocessor.h     |   3 +
 src/include/HTMLRenderer.h         |   9 +-
 src/include/Param.h                |   3 +
 src/include/ff.h                   |   3 +
 src/include/namespace.h            |   2 +
 src/include/pdf2htmlEX-config.h.in |   3 +
 src/include/util.h                 |  31 ++++---
 src/pdf2htmlEX.cc                  |   1 +
 src/util.cc                        |   3 +
 16 files changed, 154 insertions(+), 88 deletions(-)

diff --git a/src/ArgParser.cc b/src/ArgParser.cc
index a61bf2b..81ff714 100644
--- a/src/ArgParser.cc
+++ b/src/ArgParser.cc
@@ -13,6 +13,8 @@
 
 #include "ArgParser.h"
 
+namespace pdf2htmlEX {
+
 using std::ostream;
 using std::cerr;
 using std::endl;
@@ -22,6 +24,16 @@ using std::unordered_map;
 using std::make_pair;
 using std::ostringstream;
 
+void read_value(const char * arg, std::string * location)
+{
+    *location = std::string(arg);
+}
+
+void dump_value(std::ostream & out, const std::string & v)
+{
+    out << '"' << v << '"';
+}
+
 ArgParser::~ArgParser(void)
 {
     for(auto iter = arg_entries.begin(); iter != arg_entries.end(); ++iter)
@@ -111,11 +123,10 @@ void ArgParser::parse(int argc, char ** argv) const
     }
 
     {
-        int i = optind;
         auto iter = optional_arg_entries.begin();
-        while((i < argc) && (iter != optional_arg_entries.end())) 
+        while((optind < argc) && (iter != optional_arg_entries.end())) 
         {
-            (*(iter++))->parse(argv[i++]);
+            (*(iter++))->parse(argv[optind++]);
         }
     }
 }
@@ -146,10 +157,6 @@ ArgParser::ArgEntryBase::ArgEntryBase(const char * name, const char * descriptio
     }
 }
 
-
-void dump_default_value(std::ostream & out, const std::string & v)
-{
-    out << '"' << v << '"';
-}
-
 const int ArgParser::arg_col_width = 40;
+
+} // namespace pdf2htmlEX
diff --git a/src/BackgroundRenderer.cc b/src/BackgroundRenderer.cc
index 0ce5a8b..6956408 100644
--- a/src/BackgroundRenderer.cc
+++ b/src/BackgroundRenderer.cc
@@ -11,6 +11,8 @@
 #include "BackgroundRenderer.h"
 #include "util.h"
 
+using namespace pdf2htmlEX;
+
 void BackgroundRenderer::drawChar(GfxState *state, double x, double y,
   double dx, double dy,
   double originX, double originY,
diff --git a/src/FontPreprocessor.cc b/src/FontPreprocessor.cc
index 6c79f72..b5ffc1b 100644
--- a/src/FontPreprocessor.cc
+++ b/src/FontPreprocessor.cc
@@ -15,6 +15,8 @@
 #include "FontPreprocessor.h"
 #include "util.h"
 
+namespace pdf2htmlEX {
+
 FontPreprocessor::FontPreprocessor(void)
     : cur_font_id(0)
     , cur_code_map(nullptr)
@@ -59,3 +61,5 @@ const char * FontPreprocessor::get_code_map (long long font_id) const
     auto iter = code_maps.find(font_id);
     return (iter == code_maps.end()) ? nullptr : (iter->second);
 }
+
+} // namespace pdf2htmlEX
diff --git a/src/HTMLRenderer/general.cc b/src/HTMLRenderer/general.cc
index 29ab895..5715876 100644
--- a/src/HTMLRenderer/general.cc
+++ b/src/HTMLRenderer/general.cc
@@ -20,7 +20,7 @@
 using std::fixed;
 using std::flush;
 
-static void dummy(void *, ErrorCategory, int pos, char *)
+static void dummy(void *, enum ErrorCategory, int pos, char *)
 {
 }
 
@@ -144,8 +144,8 @@ void HTMLRenderer::pre_process()
         html_fout << ifstream(str_fmt("%s/%s", PDF2HTMLEX_DATA_PATH.c_str(), NECK_HTML_FILENAME.c_str()), ifstream::binary).rdbuf();
     }
 
-    html_fout << fixed << hex;
-    allcss_fout << fixed << hex;
+    fix_stream(html_fout);
+    fix_stream(allcss_fout);
 
     allcss_fout << ifstream(str_fmt("%s/%s", PDF2HTMLEX_DATA_PATH.c_str(), CSS_FILENAME.c_str()), ifstream::binary).rdbuf(); 
 }
@@ -241,6 +241,11 @@ void HTMLRenderer::process_single_html()
     out << ifstream(PDF2HTMLEX_DATA_PATH + "/" + TAIL_HTML_FILENAME, ifstream::binary).rdbuf();
 }
 
+void HTMLRenderer::fix_stream (std::ostream & out)
+{
+    out << fixed << hex;
+}
+
 void HTMLRenderer::add_tmp_file(const string & fn)
 {
     if(!param->clean_tmp)
diff --git a/src/HTMLRenderer/install.cc b/src/HTMLRenderer/install.cc
index 3301f43..c1011e8 100644
--- a/src/HTMLRenderer/install.cc
+++ b/src/HTMLRenderer/install.cc
@@ -250,8 +250,8 @@ long long HTMLRenderer::install_word_space(double word_space)
 long long HTMLRenderer::install_color(const GfxRGB * rgb)
 {
     const GfxRGB & c = *rgb;
-    auto iter = color_map.lower_bound(c);
-    if((iter != color_map.end()) && (c == (iter->first)))
+    auto iter = color_map.find(c);
+    if(iter != color_map.end())
         return iter->second;
 
     long long new_color_id = color_map.size();
diff --git a/src/include/ArgParser.h b/src/include/ArgParser.h
index 9d242fe..3bfff92 100644
--- a/src/include/ArgParser.h
+++ b/src/include/ArgParser.h
@@ -18,57 +18,81 @@
 #define nullptr (NULL)
 #endif
 
+namespace pdf2htmlEX {
+
+//helper
+template<class T>
+void read_value(const char * arg, T * location)
+{
+    std::istringstream sin(arg);
+    if((sin >> (*location)) && (sin.eof()))
+        return;
+
+    throw std::string("Invalid argument: ") + arg;
+}
+
+extern void read_value(const char * arg, std::string * location);
+
+template<class T>
+void dump_value(std::ostream & out, const T & v)
+{
+    out << v;
+}
+
+extern void dump_value(std::ostream & out, const std::string & v);
+
+
 class ArgParser
 {
-public:
-    ~ArgParser(void);
-
-    typedef void (*ArgParserCallBack) (const char * arg);
-
-    /*
-     * optname: name of the argment, should be provided as --optname
-     * description: if description is "", the argument won't be shown in show_usage()
-     */
-
-    ArgParser & add(const char * optname, const char * description, ArgParserCallBack callback = nullptr);
-
-    template <class T, class Tv>
-    ArgParser & add(const char * optname, T * location, const Tv & default_value, const char * description, ArgParserCallBack callback = nullptr);
-
-    void parse(int argc, char ** argv) const;
-    void show_usage(std::ostream & out) const;
-
-private:
-    class ArgEntryBase
-    {
     public:
-        ArgEntryBase(const char * name, const char * description, bool need_arg);
-        virtual ~ArgEntryBase() { }
-        char shortname;
-        std::string name;
-        std::string description;
-        bool need_arg;
-        virtual void parse (const char * arg) const = 0;
-        virtual void show_usage (std::ostream & out) const = 0;
-    };
+        ~ArgParser(void);
 
-    template <class T, class Tv>
-    class ArgEntry : public ArgEntryBase
-    {
-    public:
-        ArgEntry(const char * name, T * location, const Tv & deafult_value, ArgParserCallBack callback, const char * description);
+        typedef void (*ArgParserCallBack) (const char * arg);
 
-        virtual void parse (const char * arg) const;
-        virtual void show_usage (std::ostream & out) const;
+        /*
+         * optname: name of the argment, should be provided as --optname
+         * description: if description is "", the argument won't be shown in show_usage()
+         */
+
+        ArgParser & add(const char * optname, const char * description, ArgParserCallBack callback = nullptr);
+
+        template <class T, class Tv>
+            ArgParser & add(const char * optname, T * location, const Tv & default_value, const char * description, ArgParserCallBack callback = nullptr);
+
+        void parse(int argc, char ** argv) const;
+        void show_usage(std::ostream & out) const;
 
     private:
-        T * location;
-        T default_value;
-        ArgParserCallBack callback;
-    };
+        class ArgEntryBase
+        {
+            public:
+                ArgEntryBase(const char * name, const char * description, bool need_arg);
+                virtual ~ArgEntryBase() { }
+                char shortname;
+                std::string name;
+                std::string description;
+                bool need_arg;
+                virtual void parse (const char * arg) const = 0;
+                virtual void show_usage (std::ostream & out) const = 0;
+        };
 
-    std::vector<ArgEntryBase *> arg_entries, optional_arg_entries;
-    static const int arg_col_width;
+        template <class T, class Tv>
+        class ArgEntry : public ArgEntryBase
+        {
+            public:
+                ArgEntry(const char * name, T * location, const Tv & deafult_value, ArgParserCallBack callback, const char * description);
+
+                virtual void parse (const char * arg) const;
+                virtual void show_usage (std::ostream & out) const;
+
+            private:
+                T * location;
+                T default_value;
+                ArgParserCallBack callback;
+        };
+
+        std::vector<ArgEntryBase *> arg_entries, optional_arg_entries;
+        static const int arg_col_width;
 };
 
 template<class T, class Tv>
@@ -86,9 +110,9 @@ ArgParser & ArgParser::add(const char * optname, T * location, const Tv & defaul
 template<class T, class Tv>
 ArgParser::ArgEntry<T, Tv>::ArgEntry(const char * name, T * location, const Tv & default_value,  ArgParserCallBack callback, const char * description)
     : ArgEntryBase(name, description, (location != nullptr))
-    , location(location)
-    , default_value(default_value)
-    , callback(callback)
+      , location(location)
+      , default_value(default_value)
+      , callback(callback)
 { 
     if(need_arg)
         *location = T(default_value);
@@ -102,24 +126,13 @@ void ArgParser::ArgEntry<T, Tv>::parse(const char * arg) const
         if(!arg)
             throw std::string("Missing argument of option: --") + name;
 
-        std::istringstream sin(arg);
-        if(!(sin >> (*location)))
-            throw std::string("Cannot parse argment of option: --") + name;
+        read_value(arg, location);
     }
 
     if(callback)
         (*callback)(arg); 
 }
 
-// helper
-template<class T>
-void dump_default_value(std::ostream & out, const T & v)
-{
-    out << v;
-}
-
-extern void dump_default_value(std::ostream & out, const std::string & v);
-
 template<class T, class Tv>
 void ArgParser::ArgEntry<T, Tv>::show_usage(std::ostream & out) const
 { 
@@ -144,7 +157,7 @@ void ArgParser::ArgEntry<T, Tv>::show_usage(std::ostream & out) const
     if(need_arg)
     {
         sout << " <arg> (=";
-        dump_default_value(sout, default_value);
+        dump_value(sout, default_value);
         sout << ")";
     }
 
@@ -157,4 +170,6 @@ void ArgParser::ArgEntry<T, Tv>::show_usage(std::ostream & out) const
     out << " " << description << std::endl;
 }
 
+} // namespace ArgParser
+
 #endif //ARGPARSER_H__
diff --git a/src/include/BackgroundRenderer.h b/src/include/BackgroundRenderer.h
index 335184f..39cbd07 100644
--- a/src/include/BackgroundRenderer.h
+++ b/src/include/BackgroundRenderer.h
@@ -12,6 +12,8 @@
 
 #include <SplashOutputDev.h>
 
+namespace pdf2htmlEX {
+
 // Based on BackgroundRenderer from poppler
 class BackgroundRenderer : public SplashOutputDev {
 public:
@@ -31,7 +33,6 @@ public:
       CharCode code, int nBytes, Unicode *u, int uLen);
 };
 
-
-
+}
 
 #endif //BACKGROUND_RENDERER_H__
diff --git a/src/include/FontPreprocessor.h b/src/include/FontPreprocessor.h
index 96bc8e5..bfdf4fd 100644
--- a/src/include/FontPreprocessor.h
+++ b/src/include/FontPreprocessor.h
@@ -15,6 +15,8 @@
 
 #include <OutputDev.h>
 
+namespace pdf2htmlEX {
+
 class FontPreprocessor : public OutputDev {
 public:
     FontPreprocessor(void);
@@ -38,5 +40,6 @@ protected:
     std::unordered_map<long long, char*> code_maps;
 };
 
+} // namespace pdf2htmlEX
 
 #endif //FONTPREPROCESSOR_H__
diff --git a/src/include/HTMLRenderer.h b/src/include/HTMLRenderer.h
index a451db9..59f66c6 100644
--- a/src/include/HTMLRenderer.h
+++ b/src/include/HTMLRenderer.h
@@ -53,6 +53,8 @@
  *
  */
 
+namespace pdf2htmlEX {
+
 class HTMLRenderer : public OutputDev
 {
     public:
@@ -123,6 +125,9 @@ class HTMLRenderer : public OutputDev
         // misc
         ////////////////////////////////////////////////////
         
+        // set flags 
+        void fix_stream (std::ostream & out);
+
         void add_tmp_file (const std::string & fn);
         void clean_tmp_files ();
         std::string dump_embedded_font (GfxFont * font, long long fn_id);
@@ -346,7 +351,7 @@ class HTMLRenderer : public OutputDev
         std::map<TM, long long> transform_matrix_map;
         std::map<double, long long> letter_space_map;
         std::map<double, long long> word_space_map;
-        std::map<GfxRGB, long long> color_map; 
+        std::unordered_map<GfxRGB, long long, GfxRGB_hash, GfxRGB_equal> color_map; 
         std::map<double, long long> whitespace_map;
         std::map<double, long long> rise_map;
 
@@ -363,4 +368,6 @@ class HTMLRenderer : public OutputDev
         static const std::string CSS_FILENAME;
 };
 
+} //namespace pdf2htmlEX
+
 #endif /* HTMLRENDERER_H_ */
diff --git a/src/include/Param.h b/src/include/Param.h
index 0359846..ab3bbb8 100644
--- a/src/include/Param.h
+++ b/src/include/Param.h
@@ -11,6 +11,8 @@
 
 #include <string>
 
+namespace pdf2htmlEX {
+
 struct Param
 {
     // PDF stuff
@@ -45,5 +47,6 @@ struct Param
     int clean_tmp;
 };
 
+} // namespace pdf2htmlEX
 
 #endif //PARAM_h__
diff --git a/src/include/ff.h b/src/include/ff.h
index df8eb33..f368ea6 100644
--- a/src/include/ff.h
+++ b/src/include/ff.h
@@ -13,11 +13,13 @@
 
 #ifdef __cplusplus
 #include <cstdint>
+namespace pdf2htmlEX {
 extern "C" {
 #else
 #include <stdint.h>
 #endif
 
+
 void ff_init(void);
 void ff_fin(void);
 void ff_load_font(const char * filename);
@@ -35,4 +37,5 @@ void ff_set_descent(int d);
 
 #ifdef __cplusplus
 }
+}
 #endif
diff --git a/src/include/namespace.h b/src/include/namespace.h
index a74f936..57b64da 100644
--- a/src/include/namespace.h
+++ b/src/include/namespace.h
@@ -19,5 +19,7 @@ using std::make_pair;
 using std::ifstream;
 using std::ofstream;
 
+using namespace pdf2htmlEX;
+
 #endif // NAMESPACE_H__
 
diff --git a/src/include/pdf2htmlEX-config.h.in b/src/include/pdf2htmlEX-config.h.in
index 836d34c..7a7ef62 100644
--- a/src/include/pdf2htmlEX-config.h.in
+++ b/src/include/pdf2htmlEX-config.h.in
@@ -11,9 +11,12 @@
 
 #include <string>
 
+namespace pdf2htmlEX {
+
 static const std::string PDF2HTMLEX_VERSION = "@PDF2HTMLEX_VERSION@";
 static const std::string PDF2HTMLEX_PREFIX = "@CMAKE_INSTALL_PREFIX@";
 static const std::string PDF2HTMLEX_DATA_PATH = "@CMAKE_INSTALL_PREFIX@""/share/pdf2htmlEX";
 
+} // namespace pdf2htmlEX
 
 #endif //PDF2HTMLEX_CONFIG_H__
diff --git a/src/include/util.h b/src/include/util.h
index 2cfae36..53342f2 100644
--- a/src/include/util.h
+++ b/src/include/util.h
@@ -24,6 +24,8 @@
 #define nullptr (NULL)
 #endif
 
+namespace pdf2htmlEX {
+
 static const double EPS = 1e-6;
 extern const double id_matrix[6];
 
@@ -36,7 +38,7 @@ extern const std::map<std::string, std::string> GB_ENCODED_FONT_NAME_MAP;
 namespace
 {
     template <class T>
-    void dummy(){ 
+    void _(){ 
         auto _1 = &mapUCS2; 
         auto _2 = &mapUTF8;
     }
@@ -78,19 +80,23 @@ Unicode check_unicode(Unicode * u, int len, CharCode code, GfxFont * font);
 
 void outputUnicodes(std::ostream & out, const Unicode * u, int uLen);
 
-static inline bool operator < (const GfxRGB & rgb1, const GfxRGB & rgb2)
+class GfxRGB_hash 
 {
-    if(rgb1.r < rgb2.r) return true;
-    if(rgb1.r > rgb2.r) return false;
-    if(rgb1.g < rgb2.g) return true;
-    if(rgb1.g > rgb2.g) return false;
-    return (rgb1.b < rgb2.b);
-}
+public:
+    size_t operator () (const GfxRGB & rgb) const
+    {
+        return (colToByte(rgb.r) << 16) | (colToByte(rgb.g) << 8) | (colToByte(rgb.b));
+    }
+};
 
-static inline bool operator == (const GfxRGB & rgb1, const GfxRGB & rgb2)
-{
-    return ((rgb1.r == rgb2.r) && (rgb1.g == rgb2.g) && (rgb1.b == rgb1.b));
-}
+class GfxRGB_equal
+{ 
+public:
+    bool operator ()(const GfxRGB & rgb1, const GfxRGB & rgb2) const
+    {
+        return ((rgb1.r == rgb2.r) && (rgb1.g == rgb2.g) && (rgb1.b == rgb1.b));
+    }
+};
 
 // we may need more info of a font in the future
 class FontInfo
@@ -222,4 +228,5 @@ bool is_truetype_suffix(const std::string & suffix);
 std::string get_filename(const std::string & path);
 std::string get_suffix(const std::string & path);
 
+} // namespace util
 #endif //UTIL_H__
diff --git a/src/pdf2htmlEX.cc b/src/pdf2htmlEX.cc
index 3289bb7..4406893 100644
--- a/src/pdf2htmlEX.cc
+++ b/src/pdf2htmlEX.cc
@@ -25,6 +25,7 @@
 #include "ArgParser.h"
 
 using namespace std;
+using namespace pdf2htmlEX;
 
 Param param;
 ArgParser argparser;
diff --git a/src/util.cc b/src/util.cc
index 0b07c5b..5495557 100644
--- a/src/util.cc
+++ b/src/util.cc
@@ -26,6 +26,8 @@ using std::string;
 using std::map;
 using std::ostream;
 
+namespace pdf2htmlEX {
+
 const double id_matrix[6] = {1.0, 0.0, 0.0, 1.0, 0.0, 0.0};
 
 const map<string, string> BASE_14_FONT_CSS_FONT_MAP({
@@ -195,3 +197,4 @@ string get_suffix(const string & path)
         return fn.substr(idx);
 }
 
+} // namespace pdf2htmlEX