new option --embed, --embed-*; better ArgEntry

pdf2htmlEX.1.in

@@ -35,6 +35,7 @@ Specify the last page to process
 
 .SS Dimensions
 
+.TP
 .B --zoom <ratio>, --fit-width <width>, --fit-height <height>
 --zoom specifies the zoom factor directly; --fit-width/height specifies the maximum width/height of a page, the values are in pixels.
 
@@ -53,16 +54,26 @@ Specify the horizontal and vertical DPI for images
 
 .SS Output 
 
-.TP
+.B --embed <string>
-.B --single-html <0|1> (Default: 1)
+.br
-Whether to embed everything into one HTML file.
+.B --embed-css <0|1> (Default: 1)
+.br
+.B --embed-font <0|1> (Default: 1)
+.br
+.B --embed-image <0|1> (Default: 1)
+.br
+.B --embed-javascript <0|1> (Default: 1)
+.br
+.B --embed-outline <0|1> (Default: 1)
+.RS
+Specify which elements should be embedded into the output HTML file.
 
-If switched off, there will be several files generated along with the HTML file including files for fonts, css, images.
+If switched off, separated files will be generated along with the HTML file for the corresponding elements.
-
-Note that the outline will always be embedded into the main HTML file no matter if this switch is on or not. 
-And only when this switch is off will there be a separate .outline file contains the outline. 
-You need to modify the manifest if you do not want outline embedded.
 
+--embed accepts a string as argument. Each letter of the string must be one of `cCfFiIjJoO`, which corresponds
+to one of the --embed-*** switches. Lower case letters for 0 and upper case letters for 1. For example,
+`--embed cFIJo` means to embed everything but CSS files and outlines.
+.RE
 .TP
 .B --split-pages <0|1> (Default: 0)
 If turned on, the content of each page is stored in a separated file.

src/ArgParser.cc

@@ -41,9 +41,20 @@ void dump_value(std::ostream & out, const std::string & v)
     out << '"' << v << '"';
 }
 
-ArgParser & ArgParser::add(const char * optname, const char * description, ArgParserCallBack callback)
+ArgParser & ArgParser::add(const char * optname, const char * description, ArgParserCallBack callback, bool need_arg)
 {
-    return add<char>(optname, nullptr, 0, description, callback, true);
+    // ArgEntry does not accept nullptr as optname nor description
+    if((!optname) || (!optname[0]))
+    {
+        // when optname is nullptr or "", it's optional, and description is dropped
+        optional_arg_entries.emplace_back(new ArgEntry<string, string>("", "", callback, need_arg));
+    }
+    else
+    {
+        arg_entries.emplace_back(new ArgEntry<string, string>(optname, (description ? description : ""), callback, need_arg));
+    }
+
+    return *this;
 }
 
 void ArgParser::parse(int argc, char ** argv) const

src/ArgParser.h

@@ -42,79 +42,86 @@ extern void dump_value(std::ostream & out, const std::string & v);
 
 class ArgParser
 {
-    public:
+public:
-        typedef void (*ArgParserCallBack) (const char * arg);
+    typedef void (*ArgParserCallBack) (const char * arg);
 
-        /*
+    /*
-         * The 1st is for arg without arguments (i.e. flags), and the 2nd is for general args.
+     * The 1st is for arguments with callbacks(i.e. flags)
-         * optname:
+     * The 2nd is for arguments linked to variables
-         *  - if not nullptr, it should be the name of the arg, should be in the format of "<long name>[,<short char>]", e.g. "help,h"
+     *
-         *  - if nullptr, it denotes an optional arg, and description will be ignored
+     * optname:
-         * description:
+     *  - if not nullptr, it should be the name of the arg, should be in the format of "<long name>[,<short char>]", e.g. "help,h"
-         *  - if description is nullptr or "", the argument won't be shown in show_usage()
+     *  - if nullptr, it denotes an optional arg, and description will be ignored
-         */
+     * description:
-        ArgParser & add(const char * optname, const char * description, ArgParserCallBack callback = nullptr);
+     *  - if description is nullptr or "", the argument won't be shown in show_usage()
-        template <class T, class Tv>
+     *
-        ArgParser & add(const char * optname, T * location, const Tv & default_value, const char * description, ArgParserCallBack callback = nullptr, bool dont_show_default = false);
+     * location:
+     *  - if not nullptr, the argument for this arg is stored there
+     *  - if nullptr, this arg does not need arguments
+     */
+    ArgParser & add(const char * optname, const char * description, ArgParserCallBack callback, bool need_arg = false);
+    template <class T, class Tv>
+    ArgParser & add(const char * optname, T * location, const Tv & default_value, const char * description, bool dont_show_default = false);
 
-        void parse(int argc, char ** argv) const;
+    void parse(int argc, char ** argv) const;
-        void show_usage(std::ostream & out) const;
+    void show_usage(std::ostream & out) const;
+
+private:
+    // type names helper
+    template<class> 
+    static const char * get_type_name(void) { return "unknown"; }
+
+    struct ArgEntryBase
+    {
+        /* name or description cannot be nullptr */
+        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;
+    };
+
+    template <class T, class Tv>
+    struct ArgEntry : public ArgEntryBase
+    {
+        ArgEntry(const char * name, 
+                const char * description,
+                ArgParserCallBack callback,
+                bool need_arg);
+
+        ArgEntry(const char * name, 
+                T * location, const Tv & default_value, 
+                const char * description, bool dont_show_default);
+
+        virtual void parse (const char * arg) const;
+        virtual void show_usage (std::ostream & out) const;
 
     private:
-        // type names helper
+        T * location;
-        template<class> 
+        T default_value;
-        static const char * get_type_name(void) { return "unknown"; }
+        ArgParserCallBack callback;
+        bool dont_show_default;
+    };
 
-        class ArgEntryBase
+    std::vector<std::unique_ptr<ArgEntryBase>> arg_entries, optional_arg_entries;
-        {
+    static const int arg_col_width;
-            public:
-                /* name or description cannot be nullptr */
-                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;
-        };
-
-        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, bool dont_show_default);
-                
-
-                virtual void parse (const char * arg) const;
-                virtual void show_usage (std::ostream & out) const;
-
-            private:
-                T * location;
-                T default_value;
-                ArgParserCallBack callback;
-                bool dont_show_default;
-        };
-
-        std::vector<std::unique_ptr<ArgEntryBase>> arg_entries, optional_arg_entries;
-        static const int arg_col_width;
 };
 
 template<class T, class Tv>
-ArgParser & ArgParser::add(const char * optname, T * location, const Tv & default_value, const char * description, ArgParserCallBack callback, bool dont_show_default)
+ArgParser & ArgParser::add(const char * optname, T * location, const Tv & default_value, const char * description, bool dont_show_default)
 {
     // ArgEntry does not accept nullptr as optname nor description
     if((!optname) || (!optname[0]))
     {
         // when optname is nullptr or "", it's optional, and description is dropped
-        optional_arg_entries.emplace_back(new ArgEntry<T, Tv>("", location, default_value, callback, "", dont_show_default));
+        optional_arg_entries.emplace_back(new ArgEntry<T, Tv>("", location, default_value, "", dont_show_default));
     }
     else
     {
-        arg_entries.emplace_back(new ArgEntry<T, Tv>(optname, location, default_value, callback, (description ? description : ""), dont_show_default));
+        arg_entries.emplace_back(new ArgEntry<T, Tv>(optname, location, default_value, (description ? description : ""), dont_show_default));
     }
 
     return *this;
@@ -126,12 +133,22 @@ template<> const char * ArgParser::get_type_name<double>      (void);
 template<> const char * ArgParser::get_type_name<std::string> (void);
 
 template<class T, class Tv>
-ArgParser::ArgEntry<T, Tv>::ArgEntry(const char * name, T * location, const Tv & default_value,  ArgParserCallBack callback, const char * description, bool dont_show_default)
+ArgParser::ArgEntry<T, Tv>::ArgEntry(const char * name, const char * description, ArgParserCallBack callback, bool need_arg)
+    : ArgEntryBase(name, description, need_arg)
+    , location(nullptr)
+    , default_value()
+    , callback(callback)
+    , dont_show_default(true)
+{
+}
+    
+template<class T, class Tv>
+ArgParser::ArgEntry<T, Tv>::ArgEntry(const char * name, T * location, const Tv & default_value, const char * description, bool dont_show_default)
     : ArgEntryBase(name, description, (location != nullptr))
-      , location(location)
+    , location(location)
-      , default_value(default_value)
+    , default_value(default_value)
-      , callback(callback)
+    , callback(nullptr)
-      , dont_show_default(dont_show_default)
+    , dont_show_default(dont_show_default)
 { 
     if(need_arg)
         *location = T(default_value);
@@ -145,7 +162,7 @@ void ArgParser::ArgEntry<T, Tv>::parse(const char * arg) const
         if(!arg)
             throw std::string("Missing argument of option: --") + name;
 
-        if(!read_value(arg, location))
+        if((location != nullptr) && (!read_value(arg, location)))
             throw std::string("Invalid argument: ") + arg;
     }
 

src/HTMLRenderer/font.cc

@@ -562,10 +562,10 @@ void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo
      * We need to reload in order to retrieve/fix accurate ascent/descent, some info won't be written to the font by fontforge until saved.
      */
     string fn = (char*)str_fmt("%s/f%llx%s", 
-        (param.single_html ? param.tmp_dir : param.dest_dir).c_str(),
+        (param.embed_font ? param.tmp_dir : param.dest_dir).c_str(),
         info.id, param.font_suffix.c_str());
 
-    if(param.single_html)
+    if(param.embed_font)
         tmp_files.add(fn);
 
     ffw_load_font(cur_tmp_fn.c_str());
@@ -763,7 +763,7 @@ void HTMLRenderer::export_remote_font(const FontInfo & info, const string & suff
 
     {
         auto fn = str_fmt("f%llx%s", info.id, suffix.c_str());
-        if(param.single_html)
+        if(param.embed_font)
         {
             auto path = param.tmp_dir + "/" + (char*)fn;
             ifstream fin(path, ifstream::binary);

src/HTMLRenderer/general.cc

@@ -122,8 +122,8 @@ void HTMLRenderer::process(PDFDoc *doc)
 
         if(param.process_nontext)
         {
-            auto fn = str_fmt("%s/bg%x.png", (param.single_html ? param.tmp_dir : param.dest_dir).c_str(), i);
+            auto fn = str_fmt("%s/bg%x.png", (param.embed_image ? param.tmp_dir : param.dest_dir).c_str(), i);
-            if(param.single_html)
+            if(param.embed_image)
                 tmp_files.add((char*)fn);
 
             bg_renderer->render_page(doc, i, (char*)fn);
@@ -196,7 +196,7 @@ void HTMLRenderer::startPage(int pageNum, GfxState *state, XRef * xref)
     {
         f_pages.fs << "<img class=\"" << CSS::BACKGROUND_IMAGE_CN 
             << "\" alt=\"\" src=\"";
-        if(param.single_html)
+        if(param.embed_image)
         {
             auto path = str_fmt("%s/bg%x.png", param.tmp_dir.c_str(), pageNum);
             ifstream fin((char*)path, ifstream::binary);
@@ -280,23 +280,19 @@ void HTMLRenderer::pre_process(PDFDoc * doc)
     // we may output utf8 characters, so always use binary
     {
         /*
-         * If single-html
+         * If embed-css
          * we have to keep the generated css file into a temporary place
          * and embed it into the main html later
          *
-         *
+         * otherwise
-         * If single-html
-         * as there's no place to embed the css file, just leave it alone (into param.dest_dir)
-         *
-         * If !single-html
          * leave it in param.dest_dir
          */
 
-        auto fn = (param.single_html)
+        auto fn = (param.embed_css)
             ? str_fmt("%s/__css", param.tmp_dir.c_str())
             : str_fmt("%s/%s", param.dest_dir.c_str(), param.css_filename.c_str());
 
-        if(param.single_html)
+        if(param.embed_css)
             tmp_files.add((char*)fn);
 
         f_css.path = (char*)fn;
@@ -312,11 +308,11 @@ void HTMLRenderer::pre_process(PDFDoc * doc)
          * The logic for outline is similar to css
          */
 
-        auto fn = (param.single_html)
+        auto fn = (param.embed_outline)
             ? str_fmt("%s/__outline", param.tmp_dir.c_str())
             : str_fmt("%s/%s", param.dest_dir.c_str(), param.outline_filename.c_str());
 
-        if(param.single_html)
+        if(param.embed_outline)
             tmp_files.add((char*)fn);
 
         f_outline.path = (char*)fn;
@@ -333,7 +329,6 @@ void HTMLRenderer::pre_process(PDFDoc * doc)
     if(!param.split_pages)
     {
         /*
-         * If single-html
          * we have to keep the html file for pages into a temporary place
          * because we'll have to embed css before it
          *
@@ -430,7 +425,7 @@ void HTMLRenderer::post_process(void)
             }
             else if (line == "$outline")
             {
-                if (param.process_outline)
+                if (param.process_outline && param.embed_outline)
                 {
                     ifstream fin(f_outline.path, ifstream::binary);
                     if(!fin)
@@ -522,28 +517,31 @@ void HTMLRenderer::embed_file(ostream & out, const string & path, const string &
     string fn = get_filename(path);
     string suffix = (type == "") ? get_suffix(fn) : type; 
     
-    auto iter = EMBED_STRING_MAP.find(make_pair(suffix, (bool)param.single_html));
+    // TODO
+    auto iter = EMBED_STRING_MAP.find(suffix);
     if(iter == EMBED_STRING_MAP.end())
     {
         cerr << "Warning: unknown suffix: " << suffix << endl;
         return;
     }
 
-    if(param.single_html)
+    const auto & entry = iter->second;
+    
+    if(param.*(entry.embed_flag))
     {
         ifstream fin(path, ifstream::binary);
         if(!fin)
             throw string("Cannot open file ") + path + " for embedding";
-        out << iter->second.first << endl
+        out << entry.prefix_embed << endl
             << fin.rdbuf();
         out.clear(); // out will set fail big if fin is empty
-        out << iter->second.second << endl;
+        out << entry.suffix_embed << endl;
     }
     else
     {
-        out << iter->second.first;
+        out << entry.prefix_external;
         outputURL(out, fn);
-        out << iter->second.second << endl;
+        out << entry.suffix_external << endl;
 
         if(copy)
         {

src/Param.h

@@ -25,7 +25,11 @@ struct Param
     double h_dpi, v_dpi;
     
     // output 
-    int single_html;
+    int embed_css;
+    int embed_font;
+    int embed_image;
+    int embed_javascript;
+    int embed_outline;
     int split_pages;
     std::string dest_dir;
     std::string css_filename;

src/pdf2htmlEX.cc

@@ -36,9 +36,9 @@ using namespace pdf2htmlEX;
 Param param;
 ArgParser argparser;
 
-void deprecated_embed_base_font(const char * dummy = nullptr)
+void deprecated_single_html(const char * dummy = nullptr)
 {
-    cerr << "--embed-base-font is deprecated. Use --embed-external-font instead." << endl;
+    cerr << "--single_html is deprecated. Use `--embed CFIJO` instead." << endl;
     exit(EXIT_FAILURE);
 }
 
@@ -60,6 +60,31 @@ void show_version_and_exit(const char * dummy = nullptr)
     exit(EXIT_SUCCESS);
 }
 
+void embed_parser (const char * str)
+{
+    while(true)
+    {
+        switch(*str)
+        {
+            case '\0': return; break;
+            case 'c': param.embed_css = 0; break;
+            case 'C': param.embed_css = 1; break;
+            case 'f': param.embed_font = 0; break;
+            case 'F': param.embed_font = 1; break;
+            case 'i': param.embed_image = 0; break;
+            case 'I': param.embed_image = 1; break;
+            case 'j': param.embed_javascript = 0; break;
+            case 'J': param.embed_javascript = 1; break;
+            case 'o': param.embed_outline = 0; break;
+            case 'O': param.embed_outline = 1; break;
+            default:
+                cerr << "Unknown character `" << (*str) << "` for --embed" << endl;
+                break;
+        }
+        ++ str;
+    }
+}
+
 void parse_options (int argc, char **argv)
 {
     argparser
@@ -68,15 +93,20 @@ void parse_options (int argc, char **argv)
         .add("last-page,l", &param.last_page, numeric_limits<int>::max(), "last page to convert")
         
         // dimensions
-        .add("zoom", &param.zoom, 0, "zoom ratio", nullptr, true)
+        .add("zoom", &param.zoom, 0, "zoom ratio", true)
-        .add("fit-width", &param.fit_width, 0, "fit width to <fp> pixels", nullptr, true) 
+        .add("fit-width", &param.fit_width, 0, "fit width to <fp> pixels", true) 
-        .add("fit-height", &param.fit_height, 0, "fit height to <fp> pixels", nullptr, true)
+        .add("fit-height", &param.fit_height, 0, "fit height to <fp> pixels", true)
         .add("use-cropbox", &param.use_cropbox, 1, "use CropBox instead of MediaBox")
         .add("hdpi", &param.h_dpi, 144.0, "horizontal resolution for graphics in DPI")
         .add("vdpi", &param.v_dpi, 144.0, "vertical resolution for graphics in DPI")
         
         // output files
-        .add("single-html", &param.single_html, 1, "generate a single HTML file")
+        .add("embed", "specify which elements should be embedded into output", embed_parser, true)
+        .add("embed-css", &param.embed_css, 1, "embed CSS files into output")
+        .add("embed-font", &param.embed_font, 1, "embed font files into output")
+        .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("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")
@@ -107,8 +137,8 @@ void parse_options (int argc, char **argv)
         .add("optimize-text", &param.optimize_text, 0, "try to reduce the number of HTML elements used for text")
         
         // encryption
-        .add("owner-password,o", &param.owner_password, "", "owner password (for encrypted files)", nullptr, true)
+        .add("owner-password,o", &param.owner_password, "", "owner password (for encrypted files)", true)
-        .add("user-password,u", &param.user_password, "", "user password (for encrypted files)", nullptr, true)
+        .add("user-password,u", &param.user_password, "", "user password (for encrypted files)", true)
         .add("no-drm", &param.no_drm, 0, "override document DRM settings")
         
         // misc.
@@ -123,7 +153,7 @@ void parse_options (int argc, char **argv)
         .add("help,h", "print usage information", &show_usage_and_exit)
 
         // deprecated
-        .add("embed-base-font", "", &deprecated_embed_base_font)
+        .add("single-html", "", &deprecated_single_html)
         
         .add("", &param.input_filename, "", "")
         .add("", &param.output_filename, "", "")

src/util/const.cc

@@ -22,10 +22,16 @@ const map<string, string> GB_ENCODED_FONT_NAME_MAP({
     {"\xC1\xA5\xCA\xE9", "SimLi"},
 });
 
-const std::map<std::pair<std::string, bool>, std::pair<std::string, std::string> > EMBED_STRING_MAP({
+const std::map<std::string, EmbedStringEntry> EMBED_STRING_MAP({
-        {{".css", 0}, {"<link rel=\"stylesheet\" type=\"text/css\" href=\"", "\"/>"}},
+        {".css", {&Param::embed_css, 
-        {{".css", 1}, {"<style type=\"text/css\">", "</style>"}},
+                  "<style type=\"text/css\">", 
-        {{".js", 0}, {"<script type=\"text/javascript\" src=\"", "\"></script>"}},
+                  "</style>",
-        {{".js", 1}, {"<script type=\"text/javascript\">", "</script>"}}
+                  "<link rel=\"stylesheet\" type=\"text/css\" href=\"", 
+                  "\"/>" }},
+        {".js", {&Param::embed_javascript,
+                 "<script type=\"text/javascript\">", 
+                 "</script>",
+                 "<script type=\"text/javascript\" src=\"",
+                 "\"></script>" }}
 });
 } //namespace pdf2htmlEX

src/util/const.h

@@ -11,6 +11,8 @@
 #include <map>
 #include <string>
 
+#include "Param.h"
+
 namespace pdf2htmlEX {
 
 #ifndef nullptr
@@ -24,9 +26,17 @@ extern const double ID_MATRIX[6];
 // For GB encoded font names
 extern const std::map<std::string, std::string> GB_ENCODED_FONT_NAME_MAP;
 // map to embed files into html
-// key: (suffix, if_embed_content)
+struct EmbedStringEntry
-// value: (prefix string, suffix string)
+{
-extern const std::map<std::pair<std::string, bool>, std::pair<std::string, std::string> > EMBED_STRING_MAP;
+    int Param::*embed_flag; 
+    // used when *embed_flag == true
+    std::string prefix_embed;
+    std::string suffix_embed;
+    // used when *embed_flag == false
+    std::string prefix_external;
+    std::string suffix_external;
+};
+extern const std::map<std::string, EmbedStringEntry> EMBED_STRING_MAP;
 
 } // namespace pdf2htmlEX