Add multiple charsets support to img2twit, and autodetect charset when

decoding.

git-svn-id: file:///srv/caca.zoy.org/var/lib/svn/libpipi/trunk@3526 92316355-f0b4-4df1-b90c-862c8a59935f

examples/img2twit.cpp

@@ -26,40 +26,40 @@
 #include "../genethumb/mygetopt.h"
 
 /*
- * User-definable settings.
+ * Format-dependent settings. Change this and you risk making all other
+ * generated strings unusable.
  */
 
-/* The Unicode characters at disposal - XXX: must be _ordered_ */
-static const uint32_t unichars[] =
-{
-    /* Printable ASCII (except space) */
-    //0x0021, 0x007f,
-
-    /* Stupid symbols and Dingbats shit */
-    //0x25a0, 0x2600, /* Geometric Shapes */
-    //0x2600, 0x269e, 0x26a0, 0x26bd, 0x26c0, 0x26c4, /* Misc. Symbols */
-    //0x2701, 0x2705, 0x2706, 0x270a, 0x270c, 0x2728, 0x2729, 0x274c,
-    //  0x274d, 0x274e, 0x274f, 0x2753, 0x2756, 0x2757, 0x2758, 0x275f,
-    //  0x2761, 0x2795, 0x2798, 0x27b0, 0x27b1, 0x27bf, /* Dingbats */
-
-    /* Chinese-looking stuff */
-    //0x2e80, 0x2e9a, 0x2e9b, 0x2ef4, /* CJK Radicals Supplement */
-    //0x2f00, 0x2fd6, /* Kangxi Radicals */
-    //0x3400, 0x4db6, /* CJK Unified Ideographs Extension A */
-    0x4e00, 0x9fa6, /* CJK Unified Ideographs */
-
-    /* Korean - most people don't know the difference anyway */
-    //0xac00, 0xd7a4, /* Hangul Syllables */
-
-    /* More Chinese */
-    //0xf900, 0xfa2e, 0xfa30, 0xfa6b, 0xfa70, 0xfada, /* CJK Compat. Idgphs. */
-
-    /* TODO: there's also the U+20000 and U+2f800 planes, but they're
-     * not supported by the Twitter Javascript filter (yet?). */
-
-    /* End of list marker - XXX: don't remove! */
-    0x0000, 0x0000
-};
+/* Printable ASCII (except space) */
+#define RANGE_ASCII 0x0021, 0x007f
+
+/* CJK Unified Ideographs */
+#define RANGE_CJK 0x4e00, 0x9fa6
+//0x2e80, 0x2e9a, 0x2e9b, 0x2ef4, /* CJK Radicals Supplement */
+//0x2f00, 0x2fd6, /* Kangxi Radicals */
+//0x3400, 0x4db6, /* CJK Unified Ideographs Extension A */
+//0xac00, 0xd7a4, /* Hangul Syllables -- Korean, not Chinese */
+//0xf900, 0xfa2e, 0xfa30, 0xfa6b, 0xfa70, 0xfada, /* CJK Compat. Idgphs. */
+/* TODO: there's also the U+20000 and U+2f800 planes, but they're
+ * not supported by the Twitter Javascript filter (yet?). */
+
+/* Stupid symbols and Dingbats shit */
+#define RANGE_SYMBOLS 0x25a0, 0x2600, /* Geometric Shapes */ \
+  0x2600, 0x269e, 0x26a0, 0x26bd, 0x26c0, 0x26c4, /* Misc. Symbols */ \
+  0x2701, 0x2705, 0x2706, 0x270a, 0x270c, 0x2728, 0x2729, 0x274c, \
+    0x274d, 0x274e, 0x274f, 0x2753, 0x2756, 0x2757, 0x2758, 0x275f, \
+    0x2761, 0x2795, 0x2798, 0x27b0, 0x27b1, 0x27bf /* Dingbats */
+
+/* End of list marker */
+#define RANGE_END 0x0, 0x0
+
+/* Pre-defined character ranges XXX: must be _ordered_ */
+static const uint32_t unichars_ascii[] = { RANGE_ASCII, RANGE_END };
+static const uint32_t unichars_cjk[] = { RANGE_CJK, RANGE_END };
+static const uint32_t unichars_symbols[] = { RANGE_SYMBOLS, RANGE_END };
+
+/* The Unicode characters at disposal */
+static const uint32_t *unichars;
 
 /* The maximum image size we want to support */
 #define MAX_W 4000
@@ -794,11 +794,12 @@ static void analyse(pipi_image_t *src)
 
 int main(int argc, char *argv[])
 {
+    uint32_t unicode_data[4096]; /* FIXME: allocate this dynamically */
     int opstats[2 * NB_OPS];
     char const *srcname = NULL, *dstname = NULL;
     pipi_image_t *src, *tmp, *dst;
     double error = 1.0;
-    int width, height, ret = 0;
+    int width, height;
 
     /* Parse command-line options */
     for(;;)
@@ -808,12 +809,13 @@ int main(int argc, char *argv[])
         {
             { "output",      1, NULL, 'o' },
             { "length",      1, NULL, 'l' },
+            { "charset",     1, NULL, 'c' },
             { "quality",     1, NULL, 'q' },
             { "debug",       0, NULL, 'd' },
             { "help",        0, NULL, 'h' },
             { NULL,          0, NULL, 0   },
         };
-        int c = mygetopt(argc, argv, "o:l:q:dh", long_options, &option_index);
+        int c = mygetopt(argc, argv, "o:l:c:q:dh", long_options, &option_index);
 
         if(c == -1)
             break;
@@ -831,6 +833,20 @@ int main(int argc, char *argv[])
                 MAX_MSG_LEN = 16;
             }
             break;
+        case 'c':
+            if(!strcmp(myoptarg, "ascii"))
+                unichars = unichars_ascii;
+            else if(!strcmp(myoptarg, "cjk"))
+                unichars = unichars_cjk;
+            else if(!strcmp(myoptarg, "symbols"))
+                unichars = unichars_symbols;
+            else
+            {
+                fprintf(stderr, "Error: invalid char block \"%s\".", myoptarg);
+                fprintf(stderr, "Valid sets are: ascii, cjk, symbols\n");
+                return EXIT_FAILURE;
+            }
+            break;
         case 'q':
             ITERATIONS_PER_POINT = 10 * atof(myoptarg);
             if(ITERATIONS_PER_POINT < 0)
@@ -849,6 +865,7 @@ int main(int argc, char *argv[])
             printf("Mandatory arguments to long options are mandatory for short options too.\n");
             printf("  -o, --output <filename>   output resulting image to filename\n");
             printf("  -l, --length <size>       message length in characters (default 140)\n");
+            printf("  -c, --charset <block>     character set to use (ascii, [cjk], symbols)\n");
             printf("  -q, --quality <rate>      set image quality (0 - 10) (default 5)\n");
             printf("  -d, --debug               print debug information\n");
             printf("  -h, --help                display this help and exit\n");
@@ -879,6 +896,32 @@ int main(int argc, char *argv[])
     if(myoptind == argc - 1)
         srcname = argv[myoptind];
 
+    /* Decoding mode: read UTF-8 text from stdin */
+    if(dstname)
+        for(int i = 0; i < MAX_MSG_LEN; i++)
+            unicode_data[i] = fread_utf8(stdin);
+
+    /* Autodetect charset if decoding, otherwise switch to CJK. */
+    if(!unichars)
+    {
+        if(dstname)
+        {
+            if(unicode_data[0] >= 0x0021 && unicode_data[0] < 0x007f)
+                unichars = unichars_ascii;
+            else if(unicode_data[0] >= 0x4e00 && unicode_data[0] < 0x9fa6)
+                unichars = unichars_cjk;
+            else if(unicode_data[0] >= 0x25a0 && unicode_data[0] < 0x27bf)
+                unichars = unichars_symbols;
+            else
+            {
+                fprintf(stderr, "Error: unable to detect charset\n");
+                return EXIT_FAILURE;
+            }
+        }
+        else
+            unichars = unichars_cjk;
+    }
+
     pipi_set_gamma(1.0);
 
     /* Precompute bit allocation */
@@ -903,14 +946,10 @@ int main(int argc, char *argv[])
 
     if(dstname)
     {
-        /* Decoding mode: read UTF-8 text from stdin, find each
-         * character's index in our character list, and push it to our
-         * wonderful custom bitstream. */
-        uint32_t data[MAX_MSG_LEN];
-        for(int i = 0; i < MAX_MSG_LEN; i++)
-            data[i] = uni2index(fread_utf8(stdin));
+        /* Decoding mode: find each character's index in our character
+         * list, and push it to our wonderful custom bitstream. */
         for(int i = MAX_MSG_LEN; i--; )
-            b.push(data[i], NUM_CHARACTERS);
+            b.push(uni2index(unicode_data[i]), NUM_CHARACTERS);
 
         /* Read width and height from bitstream */
         src = NULL;
@@ -932,7 +971,7 @@ int main(int argc, char *argv[])
         height = pipi_get_image_height(src);
     }
 
-    /* Compute best w/h ratio */
+    /* Compute "best" w/h ratio */
     dw = 1; dh = TOTAL_CELLS;
     for(unsigned int i = 1; i <= TOTAL_CELLS; i++)
     {
@@ -1160,6 +1199,6 @@ int main(int argc, char *argv[])
         pipi_free(dst);
     }
 
-    return ret;
+    return EXIT_SUCCESS;
 }