Home | History | Annotate | Download | only in util
      1 /*
      2  * Copyright  2011  Google, Inc.
      3  *
      4  *  This is part of HarfBuzz, a text shaping library.
      5  *
      6  * Permission is hereby granted, without written agreement and without
      7  * license or royalty fees, to use, copy, modify, and distribute this
      8  * software and its documentation for any purpose, provided that the
      9  * above copyright notice and the following two paragraphs appear in
     10  * all copies of this software.
     11  *
     12  * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
     13  * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
     14  * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
     15  * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
     16  * DAMAGE.
     17  *
     18  * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
     19  * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
     20  * FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
     21  * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
     22  * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
     23  *
     24  * Google Author(s): Behdad Esfahbod
     25  */
     26 
     27 #ifndef OPTIONS_HH
     28 #define OPTIONS_HH
     29 
     30 #include "hb-private.hh"
     31 
     32 #include <stdlib.h>
     33 #include <stddef.h>
     34 #include <string.h>
     35 #include <stdio.h>
     36 #include <assert.h>
     37 #include <math.h>
     38 #include <locale.h>
     39 #include <errno.h>
     40 #include <fcntl.h>
     41 #ifdef HAVE_UNISTD_H
     42 #include <unistd.h> /* for isatty() */
     43 #endif
     44 #if defined(_WIN32) || defined(__CYGWIN__)
     45 #include <io.h> /* for setmode() under Windows */
     46 #endif
     47 
     48 #include <hb.h>
     49 #ifdef HAVE_OT
     50 #include <hb-ot.h>
     51 #endif
     52 #include <glib.h>
     53 #include <glib/gprintf.h>
     54 
     55 #if !GLIB_CHECK_VERSION (2, 22, 0)
     56 # define g_mapped_file_unref g_mapped_file_free
     57 #endif
     58 
     59 void fail (hb_bool_t suggest_help, const char *format, ...) G_GNUC_NORETURN G_GNUC_PRINTF (2, 3);
     60 
     61 extern hb_bool_t debug;
     62 
     63 struct option_group_t
     64 {
     65   virtual void add_options (struct option_parser_t *parser) = 0;
     66 
     67   virtual void pre_parse (GError **error G_GNUC_UNUSED) {};
     68   virtual void post_parse (GError **error G_GNUC_UNUSED) {};
     69 };
     70 
     71 
     72 struct option_parser_t
     73 {
     74   option_parser_t (const char *usage) {
     75     memset (this, 0, sizeof (*this));
     76     usage_str = usage;
     77     context = g_option_context_new (usage);
     78     to_free = g_ptr_array_new ();
     79 
     80     add_main_options ();
     81   }
     82   ~option_parser_t (void) {
     83     g_option_context_free (context);
     84     g_ptr_array_foreach (to_free, (GFunc) g_free, nullptr);
     85     g_ptr_array_free (to_free, TRUE);
     86   }
     87 
     88   void add_main_options (void);
     89 
     90   void add_group (GOptionEntry   *entries,
     91 		  const gchar    *name,
     92 		  const gchar    *description,
     93 		  const gchar    *help_description,
     94 		  option_group_t *option_group);
     95 
     96   void free_later (char *p) {
     97     g_ptr_array_add (to_free, p);
     98   }
     99 
    100   void parse (int *argc, char ***argv);
    101 
    102   G_GNUC_NORETURN void usage (void) {
    103     g_printerr ("Usage: %s [OPTION...] %s\n", g_get_prgname (), usage_str);
    104     exit (1);
    105   }
    106 
    107   private:
    108   const char *usage_str;
    109   GOptionContext *context;
    110   GPtrArray *to_free;
    111 };
    112 
    113 
    114 #define DEFAULT_MARGIN 16
    115 #define DEFAULT_FORE "#000000"
    116 #define DEFAULT_BACK "#FFFFFF"
    117 #define FONT_SIZE_UPEM 0x7FFFFFFF
    118 #define FONT_SIZE_NONE 0
    119 
    120 struct view_options_t : option_group_t
    121 {
    122   view_options_t (option_parser_t *parser) {
    123     annotate = false;
    124     fore = nullptr;
    125     back = nullptr;
    126     line_space = 0;
    127     margin.t = margin.r = margin.b = margin.l = DEFAULT_MARGIN;
    128 
    129     add_options (parser);
    130   }
    131   ~view_options_t (void)
    132   {
    133     g_free (fore);
    134     g_free (back);
    135   }
    136 
    137   void add_options (option_parser_t *parser);
    138 
    139   hb_bool_t annotate;
    140   char *fore;
    141   char *back;
    142   double line_space;
    143   struct margin_t {
    144     double t, r, b, l;
    145   } margin;
    146 };
    147 
    148 
    149 struct shape_options_t : option_group_t
    150 {
    151   shape_options_t (option_parser_t *parser)
    152   {
    153     direction = language = script = nullptr;
    154     bot = eot = preserve_default_ignorables = false;
    155     features = nullptr;
    156     num_features = 0;
    157     shapers = nullptr;
    158     utf8_clusters = false;
    159     cluster_level = HB_BUFFER_CLUSTER_LEVEL_DEFAULT;
    160     normalize_glyphs = false;
    161     verify = false;
    162     num_iterations = 1;
    163 
    164     add_options (parser);
    165   }
    166   ~shape_options_t (void)
    167   {
    168     g_free (direction);
    169     g_free (language);
    170     g_free (script);
    171     free (features);
    172     g_strfreev (shapers);
    173   }
    174 
    175   void add_options (option_parser_t *parser);
    176 
    177   void setup_buffer (hb_buffer_t *buffer)
    178   {
    179     hb_buffer_set_direction (buffer, hb_direction_from_string (direction, -1));
    180     hb_buffer_set_script (buffer, hb_script_from_string (script, -1));
    181     hb_buffer_set_language (buffer, hb_language_from_string (language, -1));
    182     hb_buffer_set_flags (buffer, (hb_buffer_flags_t) (HB_BUFFER_FLAG_DEFAULT |
    183 			 (bot ? HB_BUFFER_FLAG_BOT : 0) |
    184 			 (eot ? HB_BUFFER_FLAG_EOT : 0) |
    185 			 (preserve_default_ignorables ? HB_BUFFER_FLAG_PRESERVE_DEFAULT_IGNORABLES : 0)));
    186     hb_buffer_set_cluster_level (buffer, cluster_level);
    187     hb_buffer_guess_segment_properties (buffer);
    188   }
    189 
    190   static void copy_buffer_properties (hb_buffer_t *dst, hb_buffer_t *src)
    191   {
    192     hb_segment_properties_t props;
    193     hb_buffer_get_segment_properties (src, &props);
    194     hb_buffer_set_segment_properties (dst, &props);
    195     hb_buffer_set_flags (dst, hb_buffer_get_flags (src));
    196     hb_buffer_set_cluster_level (dst, hb_buffer_get_cluster_level (src));
    197   }
    198 
    199   void populate_buffer (hb_buffer_t *buffer, const char *text, int text_len,
    200 			const char *text_before, const char *text_after)
    201   {
    202     hb_buffer_clear_contents (buffer);
    203     if (text_before) {
    204       unsigned int len = strlen (text_before);
    205       hb_buffer_add_utf8 (buffer, text_before, len, len, 0);
    206     }
    207     hb_buffer_add_utf8 (buffer, text, text_len, 0, text_len);
    208     if (text_after) {
    209       hb_buffer_add_utf8 (buffer, text_after, -1, 0, 0);
    210     }
    211 
    212     if (!utf8_clusters) {
    213       /* Reset cluster values to refer to Unicode character index
    214        * instead of UTF-8 index. */
    215       unsigned int num_glyphs = hb_buffer_get_length (buffer);
    216       hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, nullptr);
    217       for (unsigned int i = 0; i < num_glyphs; i++)
    218       {
    219 	info->cluster = i;
    220 	info++;
    221       }
    222     }
    223 
    224     setup_buffer (buffer);
    225   }
    226 
    227   hb_bool_t shape (hb_font_t *font, hb_buffer_t *buffer, const char **error=nullptr)
    228   {
    229     hb_buffer_t *text_buffer = nullptr;
    230     if (verify)
    231     {
    232       text_buffer = hb_buffer_create ();
    233       hb_buffer_append (text_buffer, buffer, 0, -1);
    234     }
    235 
    236     if (!hb_shape_full (font, buffer, features, num_features, shapers))
    237     {
    238       if (error)
    239         *error = "all shapers failed.";
    240       return false;
    241     }
    242 
    243     if (normalize_glyphs)
    244       hb_buffer_normalize_glyphs (buffer);
    245 
    246     if (verify && !verify_buffer (buffer, text_buffer, font, error))
    247       return false;
    248 
    249     if (text_buffer)
    250       hb_buffer_destroy (text_buffer);
    251 
    252     return true;
    253   }
    254 
    255   bool verify_buffer (hb_buffer_t  *buffer,
    256 		      hb_buffer_t  *text_buffer,
    257 		      hb_font_t    *font,
    258 		      const char  **error=nullptr)
    259   {
    260     if (!verify_buffer_monotone (buffer, error))
    261       return false;
    262     if (!verify_buffer_safe_to_break (buffer, text_buffer, font, error))
    263       return false;
    264     return true;
    265   }
    266 
    267   bool verify_buffer_monotone (hb_buffer_t *buffer, const char **error=nullptr)
    268   {
    269     /* Check that clusters are monotone. */
    270     if (cluster_level == HB_BUFFER_CLUSTER_LEVEL_MONOTONE_GRAPHEMES ||
    271 	cluster_level == HB_BUFFER_CLUSTER_LEVEL_MONOTONE_CHARACTERS)
    272     {
    273       bool is_forward = HB_DIRECTION_IS_FORWARD (hb_buffer_get_direction (buffer));
    274 
    275       unsigned int num_glyphs;
    276       hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, &num_glyphs);
    277 
    278       for (unsigned int i = 1; i < num_glyphs; i++)
    279 	if (info[i-1].cluster != info[i].cluster &&
    280 	    (info[i-1].cluster < info[i].cluster) != is_forward)
    281 	{
    282 	  if (error)
    283 	    *error = "clusters are not monotone.";
    284 	  return false;
    285 	}
    286     }
    287 
    288     return true;
    289   }
    290 
    291   bool verify_buffer_safe_to_break (hb_buffer_t  *buffer,
    292 				    hb_buffer_t  *text_buffer,
    293 				    hb_font_t    *font,
    294 				    const char  **error=nullptr)
    295   {
    296     if (cluster_level != HB_BUFFER_CLUSTER_LEVEL_MONOTONE_GRAPHEMES &&
    297 	cluster_level != HB_BUFFER_CLUSTER_LEVEL_MONOTONE_CHARACTERS)
    298     {
    299       /* Cannot perform this check without monotone clusters.
    300        * Then again, unsafe-to-break flag is much harder to use without
    301        * monotone clusters. */
    302       return true;
    303     }
    304 
    305     /* Check that breaking up shaping at safe-to-break is indeed safe. */
    306 
    307     hb_buffer_t *fragment = hb_buffer_create ();
    308     hb_buffer_t *reconstruction = hb_buffer_create ();
    309     copy_buffer_properties (reconstruction, buffer);
    310 
    311     unsigned int num_glyphs;
    312     hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, &num_glyphs);
    313 
    314     unsigned int num_chars;
    315     hb_glyph_info_t *text = hb_buffer_get_glyph_infos (text_buffer, &num_chars);
    316 
    317     /* Chop text and shape fragments. */
    318     bool forward = HB_DIRECTION_IS_FORWARD (hb_buffer_get_direction (buffer));
    319     unsigned int start = 0;
    320     unsigned int text_start = forward ? 0 : num_chars;
    321     unsigned int text_end = text_start;
    322     for (unsigned int end = 1; end < num_glyphs + 1; end++)
    323     {
    324       if (end < num_glyphs &&
    325 	  (info[end].cluster == info[end-1].cluster ||
    326 	   info[end-(forward?0:1)].mask & HB_GLYPH_FLAG_UNSAFE_TO_BREAK))
    327 	  continue;
    328 
    329       /* Shape segment corresponding to glyphs start..end. */
    330       if (end == num_glyphs)
    331       {
    332         if (forward)
    333 	  text_end = num_chars;
    334 	else
    335 	  text_start = 0;
    336       }
    337       else
    338       {
    339 	if (forward)
    340 	{
    341 	  unsigned int cluster = info[end].cluster;
    342 	  while (text_end < num_chars && text[text_end].cluster < cluster)
    343 	    text_end++;
    344 	}
    345 	else
    346 	{
    347 	  unsigned int cluster = info[end - 1].cluster;
    348 	  while (text_start && text[text_start - 1].cluster >= cluster)
    349 	    text_start--;
    350 	}
    351       }
    352       assert (text_start < text_end);
    353 
    354       if (0)
    355 	printf("start %d end %d text start %d end %d\n", start, end, text_start, text_end);
    356 
    357       hb_buffer_clear_contents (fragment);
    358       copy_buffer_properties (fragment, buffer);
    359 
    360       /* TODO: Add pre/post context text. */
    361       hb_buffer_flags_t flags = hb_buffer_get_flags (fragment);
    362       if (0 < text_start)
    363         flags = (hb_buffer_flags_t) (flags & ~HB_BUFFER_FLAG_BOT);
    364       if (text_end < num_chars)
    365         flags = (hb_buffer_flags_t) (flags & ~HB_BUFFER_FLAG_EOT);
    366       hb_buffer_set_flags (fragment, flags);
    367 
    368       hb_buffer_append (fragment, text_buffer, text_start, text_end);
    369       if (!hb_shape_full (font, fragment, features, num_features, shapers))
    370       {
    371 	if (error)
    372 	  *error = "all shapers failed while shaping fragment.";
    373 	hb_buffer_destroy (reconstruction);
    374 	hb_buffer_destroy (fragment);
    375 	return false;
    376       }
    377       hb_buffer_append (reconstruction, fragment, 0, -1);
    378 
    379       start = end;
    380       if (forward)
    381 	text_start = text_end;
    382       else
    383 	text_end = text_start;
    384     }
    385 
    386     bool ret = true;
    387     hb_buffer_diff_flags_t diff = hb_buffer_diff (reconstruction, buffer, (hb_codepoint_t) -1, 0);
    388     if (diff)
    389     {
    390       if (error)
    391 	*error = "Safe-to-break test failed.";
    392       ret = false;
    393 
    394       /* Return the reconstructed result instead so it can be inspected. */
    395       hb_buffer_set_length (buffer, 0);
    396       hb_buffer_append (buffer, reconstruction, 0, -1);
    397     }
    398 
    399     hb_buffer_destroy (reconstruction);
    400     hb_buffer_destroy (fragment);
    401 
    402     return ret;
    403   }
    404 
    405   void shape_closure (const char *text, int text_len,
    406 		      hb_font_t *font, hb_buffer_t *buffer,
    407 		      hb_set_t *glyphs)
    408   {
    409     hb_buffer_reset (buffer);
    410     hb_buffer_add_utf8 (buffer, text, text_len, 0, text_len);
    411     setup_buffer (buffer);
    412     hb_ot_shape_glyphs_closure (font, buffer, features, num_features, glyphs);
    413   }
    414 
    415   /* Buffer properties */
    416   char *direction;
    417   char *language;
    418   char *script;
    419 
    420   /* Buffer flags */
    421   hb_bool_t bot;
    422   hb_bool_t eot;
    423   hb_bool_t preserve_default_ignorables;
    424 
    425   hb_feature_t *features;
    426   unsigned int num_features;
    427   char **shapers;
    428   hb_bool_t utf8_clusters;
    429   hb_buffer_cluster_level_t cluster_level;
    430   hb_bool_t normalize_glyphs;
    431   hb_bool_t verify;
    432   unsigned int num_iterations;
    433 };
    434 
    435 
    436 struct font_options_t : option_group_t
    437 {
    438   font_options_t (option_parser_t *parser,
    439 		  int default_font_size_,
    440 		  unsigned int subpixel_bits_)
    441   {
    442     variations = nullptr;
    443     num_variations = 0;
    444     default_font_size = default_font_size_;
    445     subpixel_bits = subpixel_bits_;
    446     font_file = nullptr;
    447     face_index = 0;
    448     font_size_x = font_size_y = default_font_size;
    449     font_funcs = nullptr;
    450 
    451     font = nullptr;
    452 
    453     add_options (parser);
    454   }
    455   ~font_options_t (void) {
    456     g_free (font_file);
    457     free (variations);
    458     g_free (font_funcs);
    459     hb_font_destroy (font);
    460   }
    461 
    462   void add_options (option_parser_t *parser);
    463 
    464   hb_font_t *get_font (void) const;
    465 
    466   char *font_file;
    467   int face_index;
    468   hb_variation_t *variations;
    469   unsigned int num_variations;
    470   int default_font_size;
    471   unsigned int subpixel_bits;
    472   mutable double font_size_x;
    473   mutable double font_size_y;
    474   char *font_funcs;
    475 
    476   private:
    477   mutable hb_font_t *font;
    478 };
    479 
    480 
    481 struct text_options_t : option_group_t
    482 {
    483   text_options_t (option_parser_t *parser) {
    484     text_before = nullptr;
    485     text_after = nullptr;
    486 
    487     text = nullptr;
    488     text_file = nullptr;
    489 
    490     fp = nullptr;
    491     gs = nullptr;
    492     line = nullptr;
    493     line_len = (unsigned int) -1;
    494 
    495     add_options (parser);
    496   }
    497   ~text_options_t (void) {
    498     g_free (text_before);
    499     g_free (text_after);
    500     g_free (text);
    501     g_free (text_file);
    502     if (gs)
    503       g_string_free (gs, true);
    504     if (fp)
    505       fclose (fp);
    506   }
    507 
    508   void add_options (option_parser_t *parser);
    509 
    510   void post_parse (GError **error G_GNUC_UNUSED) {
    511     if (text && text_file)
    512       g_set_error (error,
    513 		   G_OPTION_ERROR, G_OPTION_ERROR_BAD_VALUE,
    514 		   "Only one of text and text-file can be set");
    515   };
    516 
    517   const char *get_line (unsigned int *len);
    518 
    519   char *text_before;
    520   char *text_after;
    521 
    522   char *text;
    523   char *text_file;
    524 
    525   private:
    526   FILE *fp;
    527   GString *gs;
    528   char *line;
    529   unsigned int line_len;
    530 };
    531 
    532 struct output_options_t : option_group_t
    533 {
    534   output_options_t (option_parser_t *parser,
    535 		    const char **supported_formats_ = nullptr) {
    536     output_file = nullptr;
    537     output_format = nullptr;
    538     supported_formats = supported_formats_;
    539     explicit_output_format = false;
    540 
    541     fp = nullptr;
    542 
    543     add_options (parser);
    544   }
    545   ~output_options_t (void) {
    546     g_free (output_file);
    547     g_free (output_format);
    548     if (fp)
    549       fclose (fp);
    550   }
    551 
    552   void add_options (option_parser_t *parser);
    553 
    554   void post_parse (GError **error G_GNUC_UNUSED)
    555   {
    556     if (output_format)
    557       explicit_output_format = true;
    558 
    559     if (output_file && !output_format) {
    560       output_format = strrchr (output_file, '.');
    561       if (output_format)
    562       {
    563 	  output_format++; /* skip the dot */
    564 	  output_format = strdup (output_format);
    565       }
    566     }
    567 
    568     if (output_file && 0 == strcmp (output_file, "-"))
    569       output_file = nullptr; /* STDOUT */
    570   }
    571 
    572   FILE *get_file_handle (void);
    573 
    574   char *output_file;
    575   char *output_format;
    576   const char **supported_formats;
    577   bool explicit_output_format;
    578 
    579   mutable FILE *fp;
    580 };
    581 
    582 struct format_options_t : option_group_t
    583 {
    584   format_options_t (option_parser_t *parser) {
    585     show_glyph_names = true;
    586     show_positions = true;
    587     show_clusters = true;
    588     show_text = false;
    589     show_unicode = false;
    590     show_line_num = false;
    591     show_extents = false;
    592     show_flags = false;
    593     trace = false;
    594 
    595     add_options (parser);
    596   }
    597 
    598   void add_options (option_parser_t *parser);
    599 
    600   void serialize_unicode (hb_buffer_t  *buffer,
    601 			  GString      *gs);
    602   void serialize_glyphs (hb_buffer_t  *buffer,
    603 			 hb_font_t    *font,
    604 			 hb_buffer_serialize_format_t format,
    605 			 hb_buffer_serialize_flags_t flags,
    606 			 GString      *gs);
    607   void serialize_line_no (unsigned int  line_no,
    608 			  GString      *gs);
    609   void serialize_buffer_of_text (hb_buffer_t  *buffer,
    610 				 unsigned int  line_no,
    611 				 const char   *text,
    612 				 unsigned int  text_len,
    613 				 hb_font_t    *font,
    614 				 GString      *gs);
    615   void serialize_message (unsigned int  line_no,
    616 			  const char   *type,
    617 			  const char   *msg,
    618 			  GString      *gs);
    619   void serialize_buffer_of_glyphs (hb_buffer_t  *buffer,
    620 				   unsigned int  line_no,
    621 				   const char   *text,
    622 				   unsigned int  text_len,
    623 				   hb_font_t    *font,
    624 				   hb_buffer_serialize_format_t output_format,
    625 				   hb_buffer_serialize_flags_t format_flags,
    626 				   GString      *gs);
    627 
    628 
    629   hb_bool_t show_glyph_names;
    630   hb_bool_t show_positions;
    631   hb_bool_t show_clusters;
    632   hb_bool_t show_text;
    633   hb_bool_t show_unicode;
    634   hb_bool_t show_line_num;
    635   hb_bool_t show_extents;
    636   hb_bool_t show_flags;
    637   hb_bool_t trace;
    638 };
    639 
    640 /* fallback implementation for scalbn()/scalbnf() for pre-2013 MSVC */
    641 #if defined (_MSC_VER) && (_MSC_VER < 1800)
    642 
    643 #ifndef FLT_RADIX
    644 #define FLT_RADIX 2
    645 #endif
    646 
    647 __inline long double scalbn (long double x, int exp)
    648 {
    649   return x * (pow ((long double) FLT_RADIX, exp));
    650 }
    651 
    652 __inline float scalbnf (float x, int exp)
    653 {
    654   return x * (pow ((float) FLT_RADIX, exp));
    655 }
    656 #endif
    657 
    658 #endif
    659