add option 'decompose-ligature'

TODO

@@ -1,13 +1,9 @@
-cmake - pkgconfig - fontforge - python
-
 fix glyph width as provided in pdf
 
 option to break ligatures
 
 detect duplicate base fonts when embedding
 
-consider left-shift in optimization
-
 multiple charcode mapped to a same glyph
 re-encoded only used glyphs
 

pdf2htmlEX.1

@@ -69,7 +69,10 @@ There are several base font defined in PDF standards, which are supposed to be p
 If this switch is on, local matched font will be used and embedded; otherwise only font names are exported such that web browsers may try to find proper fonts themselves.
 .TP
 .B --embed-external-font <0|1> (Default: 0)
-Similar as above but for non-base fonts
+Similar as above but for non-base fonts.
+.TP
+.B --decompose-ligature <0|1> (Default: 0)
+Decompose ligatures. For example 'fi' -> 'f''i'.
 .TP
 .B --heps <len>, --veps <len> (Default: 1)
 Specify the maximum tolerable horizontal/vertical offset (in pixels).
@@ -77,7 +80,7 @@ Specify the maximum tolerable horizontal/vertical offset (in pixels).
 pdf2htmlEX would try to optimize the generated HTML file moving Text within this distance.
 .TP
 .B --space-threshold <ratio> (Default: 1.0/6)
-pdf2htmlEX would insert a whitespace character ' ' if the distance between two consecutive letters in the same line is wider than ratio * font_size
+pdf2htmlEX would insert a whitespace character ' ' if the distance between two consecutive letters in the same line is wider than ratio * font_size.
 .TP
 .B --font-size-multiplier <ratio> (Default: 10)
 Many web browsers limit the minimum font size, and many would round the given font size, which results in incorrect rendering.
@@ -104,10 +107,10 @@ If switched off, intermediate files won't be cleaned in the end.
 Convert file.pdf into file.html
 .TP
 .B pdf2htmlEX --tmp-dir tmp --clean-tmp 0 --debug 1 /path/to/file.pdf
-Convert file.pdf and leave all intermediate files
+Convert file.pdf and leave all intermediate files.
 .TP
 .B pdf2htmlEX --dest-dir out --single-html 0 --debug 1 /path/to/file.pdf
-Convert file.pdf into out/file.html and leave font/image files separated
+Convert file.pdf into out/file.html and leave font/image files separated.
 
 .SH COPYRIGHT
 .PP

src/HTMLRenderer/text.cc

@@ -26,6 +26,7 @@
 using boost::algorithm::to_lower;
 using std::unordered_set;
 using std::min;
+using std::all_of;
 
 path HTMLRenderer::dump_embedded_font (GfxFont * font, long long fn_id)
 {
@@ -422,8 +423,15 @@ void HTMLRenderer::drawString(GfxState * state, GooString * s)
             ++nSpaces;
         }
         
+        if((param->decompose_ligature) && all_of(u, u+uLen, isLegalUnicode))
+        {
+            line_buf.append_unicodes(u, uLen);
+        }
+        else
+        {
             Unicode uu = (cur_font_info->use_tounicode ? check_unicode(u, uLen, code, font) : unicode_from_font(code, font));
             line_buf.append_unicodes(&uu, 1);
+        }
 
         dx += dx1;
         dy += dy1;

src/Param.h

@@ -30,6 +30,7 @@ struct Param
     int single_html;
     int embed_base_font;
     int embed_external_font;
+    int decompose_ligature;
 
     // Advanced tweak
     double h_eps, v_eps;

src/pdf2htmlEX.cc

@@ -78,6 +78,7 @@ po::variables_map parse_options (int argc, char **argv)
         ("single-html", po::value<int>(&param.single_html)->default_value(1), "combine everything into one single HTML file")
         ("embed-base-font", po::value<int>(&param.embed_base_font)->default_value(0), "embed local matched font for base 14 fonts in the PDF file")
         ("embed-external-font", po::value<int>(&param.embed_external_font)->default_value(0), "embed local matched font for external fonts in the PDF file")
+        ("decompose-ligature", po::value<int>(&param.decompose_ligature)->default_value(0), "decompose ligatures, for example 'fi' -> 'f''i'")
 
         ("heps", po::value<double>(&param.h_eps)->default_value(1.0), "max tolerated horizontal offset (in pixels)")
         ("veps", po::value<double>(&param.v_eps)->default_value(1.0), "max tolerated vertical offset (in pixels)")