Incoming

CMakeLists.txt

@@ -9,7 +9,7 @@ option(ENABLE_SVG "Enable SVG support, for generating SVG background images and
 
 include_directories(${CMAKE_SOURCE_DIR}/src)
 
-set(PDF2HTMLEX_VERSION "0.14.6")
+set(PDF2HTMLEX_VERSION "0.15.0")
 set(ARCHIVE_NAME pdf2htmlex-${PDF2HTMLEX_VERSION})
 add_custom_target(dist
     COMMAND git archive --prefix=${ARCHIVE_NAME}/ HEAD

src/HTMLRenderer/font.cc

@@ -373,6 +373,14 @@ string HTMLRenderer::dump_type3_font (GfxFont * font, FontInfo & info)
 #endif
 }
 
+namespace {
+
+void output_map_file_header(std::ostream& out) {
+    out << "glyph_code mapped_code unicode" << std::endl;
+}
+
+} // namespace
+
 void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo & info, bool get_metric_only)
 {
     if(param.debug)
@@ -528,6 +536,7 @@ void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo
             ffw_reencode_glyph_order();
 
             GfxCIDFont * _font = dynamic_cast<GfxCIDFont*>(font);
+            assert(_font != nullptr);
 
             // To locate CID2GID for the font
             // as in CairoFontEngine.cc
@@ -574,6 +583,7 @@ void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo
             map_filename = (char*)str_fmt("%s/f%llx.map", param.tmp_dir.c_str(), info.id);
             tmp_files.add(map_filename);
             map_outf.open(map_filename);
+            output_map_file_header(map_outf);
         }
 
         unordered_set<int> codeset;
@@ -650,6 +660,7 @@ void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo
                         {
                             map_outf.close();
                             map_outf.open(map_filename);
+                            output_map_file_header(map_outf);
                         }
                         continue;
                     }

src/HTMLRenderer/general.cc

@@ -120,11 +120,13 @@ void HTMLRenderer::process(PDFDoc *doc)
     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";
+            if(param.quiet == 0)
+                cerr << "Stop processing, reach max size\n";
             break;
         }
 
-        cerr << "Working: " << (i-param.first_page) << "/" << page_count << '\r' << flush;
+        if (param.quiet == 0)
+            cerr << "Working: " << (i-param.first_page) << "/" << page_count << '\r' << flush;
 
         if(param.split_pages)
         {
@@ -153,9 +155,11 @@ void HTMLRenderer::process(PDFDoc *doc)
             f_curpage = nullptr;
         }
     }
-    if(page_count >= 0)
+    if(page_count >= 0 && param.quiet == 0)
         cerr << "Working: " << page_count << "/" << page_count;
-    cerr << endl;
+
+    if(param.quiet == 0)
+        cerr << endl;
 
     ////////////////////////
     // Process Outline
@@ -167,7 +171,8 @@ void HTMLRenderer::process(PDFDoc *doc)
     bg_renderer = nullptr;
     fallback_bg_renderer = nullptr;
 
-    cerr << endl;
+    if(param.quiet == 0)
+        cerr << endl;
 }
 
 void HTMLRenderer::setDefaultCTM(double *ctm)

src/HTMLRenderer/link.cc

@@ -56,73 +56,70 @@ static string get_linkdest_detail_str(LinkDest * dest, Catalog * catalog, int &
     // dec
     sout << "[" << pageno;
 
-    if(dest)
+    switch(dest->getKind())
     {
-        switch(dest->getKind())
-        {
-            case destXYZ:
-                {
-                    sout << ",\"XYZ\",";
-                    if(dest->getChangeLeft())
-                        sout << (dest->getLeft());
-                    else
-                        sout << "null";
-                    sout << ",";
-                    if(dest->getChangeTop())
-                        sout << (dest->getTop());
-                    else
-                        sout << "null";
-                    sout << ",";
-                    if(dest->getChangeZoom())
-                        sout << (dest->getZoom());
-                    else
-                        sout << "null";
-                }
-                break;
-            case destFit:
-                sout << ",\"Fit\"";
-                break;
-            case destFitH:
-                sout << ",\"FitH\",";
-                if(dest->getChangeTop())
-                    sout << (dest->getTop());
-                else
-                    sout << "null";
-                break;
-            case destFitV:
-                sout << ",\"FitV\",";
+        case destXYZ:
+            {
+                sout << ",\"XYZ\",";
                 if(dest->getChangeLeft())
                     sout << (dest->getLeft());
                 else
                     sout << "null";
-                break;
-            case destFitR:
-                sout << ",\"FitR\","
-                    << (dest->getLeft()) << ","
-                    << (dest->getBottom()) << ","
-                    << (dest->getRight()) << ","
-                    << (dest->getTop());
-                break;
-            case destFitB:
-                sout << ",\"FitB\"";
-                break;
-            case destFitBH:
-                sout << ",\"FitBH\",";
+                sout << ",";
                 if(dest->getChangeTop())
                     sout << (dest->getTop());
                 else
                     sout << "null";
-                break;
-            case destFitBV:
-                sout << ",\"FitBV\",";
-                if(dest->getChangeLeft())
-                    sout << (dest->getLeft());
+                sout << ",";
+                if(dest->getChangeZoom())
+                    sout << (dest->getZoom());
                 else
                     sout << "null";
-                break;
-            default:
-                break;
-        }
+            }
+            break;
+        case destFit:
+            sout << ",\"Fit\"";
+            break;
+        case destFitH:
+            sout << ",\"FitH\",";
+            if(dest->getChangeTop())
+                sout << (dest->getTop());
+            else
+                sout << "null";
+            break;
+        case destFitV:
+            sout << ",\"FitV\",";
+            if(dest->getChangeLeft())
+                sout << (dest->getLeft());
+            else
+                sout << "null";
+            break;
+        case destFitR:
+            sout << ",\"FitR\","
+                << (dest->getLeft()) << ","
+                << (dest->getBottom()) << ","
+                << (dest->getRight()) << ","
+                << (dest->getTop());
+            break;
+        case destFitB:
+            sout << ",\"FitB\"";
+            break;
+        case destFitBH:
+            sout << ",\"FitBH\",";
+            if(dest->getChangeTop())
+                sout << (dest->getTop());
+            else
+                sout << "null";
+            break;
+        case destFitBV:
+            sout << ",\"FitBV\",";
+            if(dest->getChangeLeft())
+                sout << (dest->getLeft());
+            else
+                sout << "null";
+            break;
+        default:
+            break;
     }
     sout << "]";
 
@@ -166,6 +163,7 @@ string HTMLRenderer::get_linkaction_str(LinkAction * action, string & detail)
             case actionURI:
                 {
                     auto * real_action = dynamic_cast<LinkURI*>(action);
+                    assert(real_action != nullptr);
                     dest_str = real_action->getURI()->getCString();
                 }
                 break;

src/HTMLTextLine.cc

@@ -378,13 +378,12 @@ void HTMLTextLine::optimize_normal(std::vector<HTMLTextLine*> & lines)
     new_offsets.reserve(offsets.size());
 
     auto offset_iter1 = offsets.begin();
-    for(auto state_iter2 = states.begin(), state_iter1 = state_iter2++; 
-            state_iter1 != states.end(); 
-            ++state_iter1, ++state_iter2)
+    for(auto state_iter1 = states.begin(); state_iter1 != states.end(); ++state_iter1)
     {
+        const auto state_iter2 = std::next(state_iter1);
         const size_t text_idx1 = state_iter1->start_idx;
         const size_t text_idx2 = (state_iter2 == states.end()) ? text.size() : state_iter2->start_idx;
-        size_t text_count = text_idx2 - text_idx1;
+        const size_t text_count = text_idx2 - text_idx1;
 
         // there might be some offsets before the first state
         while((offset_iter1 != offsets.end()) 

src/Param.h

@@ -79,6 +79,7 @@ struct Param
     std::string tmp_dir;
     int debug;
     int proof;
+    int quiet;
 
     std::string input_filename, output_filename;
 };

src/Preprocessor.cc

@@ -45,7 +45,8 @@ void Preprocessor::process(PDFDoc * doc)
     int page_count = (param.last_page - param.first_page + 1);
     for(int i = param.first_page; i <= param.last_page ; ++i) 
     {
-        cerr << "Preprocessing: " << (i-param.first_page) << "/" << page_count << '\r' << flush;
+        if(param.quiet == 0)
+            cerr << "Preprocessing: " << (i - param.first_page) << "/" << page_count << '\r' << flush;
 
         doc->displayPage(this, i, DEFAULT_DPI, DEFAULT_DPI,
                 0, 
@@ -54,9 +55,11 @@ void Preprocessor::process(PDFDoc * doc)
                 false, // printing
                 nullptr, nullptr, nullptr, nullptr);
     }
-    if(page_count >= 0)
+    if(page_count >= 0 && param.quiet == 0)
         cerr << "Preprocessing: " << page_count << "/" << page_count;
-    cerr << endl;
+
+    if(param.quiet == 0)
+        cerr << endl;
 }
 
 void Preprocessor::drawChar(GfxState *state, double x, double y,

src/StateManager.h

@@ -84,7 +84,7 @@ class StateManager<double, Imp>
 
 // Be careful about the mixed usage of Matrix and const double *
 // the input is usually double *, which might be changed, so we have to copy the content out
-// in the map we use Matrix instead of double * such that the array may be automatically release when deconstructing
+// in the map we use Matrix instead of double * such that the array may be automatically release when destructing
 template <class Imp>
 class StateManager<Matrix, Imp>
 {
@@ -96,7 +96,7 @@ class StateManager<Matrix, Imp>
     // return id
     long long install(const double * new_value) {
         Matrix m;
-        memcpy(m.m, new_value, sizeof(m.m));
+        memcpy(m.m, new_value, 4 * sizeof(double));
         auto iter = value_map.lower_bound(m);
         if((iter != value_map.end()) && (tm_equal(m.m, iter->first.m, 4)))
         {

src/pdf2htmlEX.cc

@@ -111,7 +111,7 @@ void prepare_directories()
 
     errno = 0;
 
-    unique_ptr<char> pBuf(new char[tmp_dir.size() + 1]);
+    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)
@@ -160,7 +160,7 @@ void parse_options (int argc, char **argv)
         .add("process-form", &param.process_form, 0, "include text fields and radio buttons")
         .add("printing", &param.printing, 1, "enable printing support")
         .add("fallback", &param.fallback, 0, "output in fallback mode")
-        .add("tmp-file-size-limit", &param.tmp_file_size_limit, -1, "Maximum size (in KB) used by temporary files, -1 for no limit.")
+        .add("tmp-file-size-limit", &param.tmp_file_size_limit, -1, "Maximum size (in KB) used by temporary files, -1 for no limit")
 
         // fonts
         .add("embed-external-font", &param.embed_external_font, 1, "embed local match for external fonts")
@@ -186,8 +186,8 @@ void parse_options (int argc, char **argv)
         // background image
         .add("bg-format", &param.bg_format, "png", "specify background image format")
         .add("svg-node-count-limit", &param.svg_node_count_limit, -1, "if node count in a svg background image exceeds this limit,"
-                " fall back this page to bitmap background; negative value means no limit.")
-        .add("svg-embed-bitmap", &param.svg_embed_bitmap, 1, "1: embed bitmaps in svg background; 0: dump bitmaps to external files if possible.")
+                " fall back this page to bitmap background; negative value means no limit")
+        .add("svg-embed-bitmap", &param.svg_embed_bitmap, 1, "1: embed bitmaps in svg background; 0: dump bitmaps to external files if possible")
 
         // encryption
         .add("owner-password,o", &param.owner_password, "", "owner password (for encrypted files)", true)
@@ -196,11 +196,12 @@ void parse_options (int argc, char **argv)
 
         // misc.
         .add("clean-tmp", &param.clean_tmp, 1, "remove temporary files after conversion")
-        .add("tmp-dir", &param.tmp_dir, param.tmp_dir, "specify the location of temporary directory.")
+        .add("tmp-dir", &param.tmp_dir, param.tmp_dir, "specify the location of temporary directory")
         .add("data-dir", &param.data_dir, param.data_dir, "specify data directory")
         .add("poppler-data-dir", &param.poppler_data_dir, param.poppler_data_dir, "specify poppler data directory")
         .add("debug", &param.debug, 0, "print debugging information")
-        .add("proof", &param.proof, 0, "texts are drawn on both text layer and background for proof.")
+        .add("proof", &param.proof, 0, "texts are drawn on both text layer and background for proof")
+        .add("quiet", &param.quiet, 0, "perform operations quietly")
 
         // meta
         .add("version,v", "print copyright and version info", &show_version_and_exit)

src/util/unicode.cc

@@ -40,7 +40,9 @@ Unicode unicode_from_font (CharCode code, GfxFont * font)
 {
     if(!font->isCIDFont())
     {
-        char * cname = dynamic_cast<Gfx8BitFont*>(font)->getCharName(code);
+        auto * font2 = dynamic_cast<Gfx8BitFont*>(font);
+        assert(font2 != nullptr);
+        char * cname = font2->getCharName(code);
         // may be untranslated ligature
         if(cname)
         {