lavc/ccaption_dec: implement special and extended character sets

character sets implemented as defined in https://en.wikipedia.org/wiki/EIA-608#Characters

1 files changed, 149 insertions, 3 deletions

diff --git a/libavcodec/ccaption_dec.c b/libavcodec/ccaption_dec.c
index 5fb2ec6293..237272c5cd 100644
--- a/libavcodec/ccaption_dec.c
+++ b/libavcodec/ccaption_dec.c
@@ -63,6 +63,116 @@ enum cc_font {
     CCFONT_UNDERLINED_ITALICS,
 };
 
+enum cc_charset {
+    CCSET_BASIC_AMERICAN,
+    CCSET_SPECIAL_AMERICAN,
+    CCSET_EXTENDED_SPANISH_FRENCH_MISC,
+    CCSET_EXTENDED_PORTUGUESE_GERMAN_DANISH,
+};
+
+static const char *charset_overrides[4][128] =
+{
+    [CCSET_BASIC_AMERICAN] = {
+        [0x27] = "\u2019",
+        [0x2a] = "\u00e1",
+        [0x5c] = "\u00e9",
+        [0x5e] = "\u00ed",
+        [0x5f] = "\u00f3",
+        [0x60] = "\u00fa",
+        [0x7b] = "\u00e7",
+        [0x7c] = "\u00f7",
+        [0x7d] = "\u00d1",
+        [0x7e] = "\u00f1",
+        [0x7f] = "\u2588"
+    },
+    [CCSET_SPECIAL_AMERICAN] = {
+        [0x30] = "\u00ae",
+        [0x31] = "\u00b0",
+        [0x32] = "\u00bd",
+        [0x33] = "\u00bf",
+        [0x34] = "\u2122",
+        [0x35] = "\u00a2",
+        [0x36] = "\u00a3",
+        [0x37] = "\u266a",
+        [0x38] = "\u00e0",
+        [0x39] = "\u00A0",
+        [0x3a] = "\u00e8",
+        [0x3b] = "\u00e2",
+        [0x3c] = "\u00ea",
+        [0x3d] = "\u00ee",
+        [0x3e] = "\u00f4",
+        [0x3f] = "\u00fb",
+    },
+    [CCSET_EXTENDED_SPANISH_FRENCH_MISC] = {
+        [0x20] = "\u00c1",
+        [0x21] = "\u00c9",
+        [0x22] = "\u00d3",
+        [0x23] = "\u00da",
+        [0x24] = "\u00dc",
+        [0x25] = "\u00fc",
+        [0x26] = "\u00b4",
+        [0x27] = "\u00a1",
+        [0x28] = "*",
+        [0x29] = "\u2018",
+        [0x2a] = "-",
+        [0x2b] = "\u00a9",
+        [0x2c] = "\u2120",
+        [0x2d] = "\u00b7",
+        [0x2e] = "\u201c",
+        [0x2f] = "\u201d",
+        [0x30] = "\u00c0",
+        [0x31] = "\u00c2",
+        [0x32] = "\u00c7",
+        [0x33] = "\u00c8",
+        [0x34] = "\u00ca",
+        [0x35] = "\u00cb",
+        [0x36] = "\u00eb",
+        [0x37] = "\u00ce",
+        [0x38] = "\u00cf",
+        [0x39] = "\u00ef",
+        [0x3a] = "\u00d4",
+        [0x3b] = "\u00d9",
+        [0x3c] = "\u00f9",
+        [0x3d] = "\u00db",
+        [0x3e] = "\u00ab",
+        [0x3f] = "\u00bb",
+    },
+    [CCSET_EXTENDED_PORTUGUESE_GERMAN_DANISH] = {
+        [0x20] = "\u00c3",
+        [0x21] = "\u00e3",
+        [0x22] = "\u00cd",
+        [0x23] = "\u00cc",
+        [0x24] = "\u00ec",
+        [0x25] = "\u00d2",
+        [0x26] = "\u00f2",
+        [0x27] = "\u00d5",
+        [0x28] = "\u00f5",
+        [0x29] = "{",
+        [0x2a] = "}",
+        [0x2b] = "\\",
+        [0x2c] = "^",
+        [0x2d] = "_",
+        [0x2e] = "|",
+        [0x2f] = "~",
+        [0x30] = "\u00c4",
+        [0x31] = "\u00e4",
+        [0x32] = "\u00d6",
+        [0x33] = "\u00f6",
+        [0x34] = "\u00df",
+        [0x35] = "\u00a5",
+        [0x36] = "\u00a4",
+        [0x37] = "\u00a6",
+        [0x38] = "\u00c5",
+        [0x39] = "\u00e5",
+        [0x3a] = "\u00d8",
+        [0x3b] = "\u00f8",
+        [0x3c] = "\u250c",
+        [0x3d] = "\u2510",
+        [0x3e] = "\u2514",
+        [0x3f] = "\u2518",
+    },
+};
+
 static const unsigned char pac2_attribs[32][3] = // Color, font, ident
 {
     { CCCOL_WHITE,   CCFONT_REGULAR,            0 },  // 0x40 || 0x60
@@ -103,6 +213,7 @@ static const unsigned char pac2_attribs[32][3] = // Color, font, ident
 struct Screen {
     /* +1 is used to compensate null character of string */
     uint8_t characters[SCREEN_ROWS][SCREEN_COLUMNS+1];
+    uint8_t charsets[SCREEN_ROWS][SCREEN_COLUMNS+1];
     uint8_t colors[SCREEN_ROWS][SCREEN_COLUMNS+1];
     uint8_t fonts[SCREEN_ROWS][SCREEN_COLUMNS+1];
     /*
@@ -123,6 +234,7 @@ typedef struct CCaptionSubContext {
     uint8_t cursor_column;
     uint8_t cursor_color;
     uint8_t cursor_font;
+    uint8_t cursor_charset;
     AVBPrint buffer;
     int buffer_changed;
     int rollup;
@@ -189,6 +301,7 @@ static void flush_decoder(AVCodecContext *avctx)
     ctx->cursor_column = 0;
     ctx->cursor_font = 0;
     ctx->cursor_color = 0;
+    ctx->cursor_charset = 0;
     ctx->active_screen = 0;
     ctx->last_real_time = 0;
     ctx->screen_touched = 0;
@@ -204,10 +317,13 @@ static int write_char(CCaptionSubContext *ctx, struct Screen *screen, char ch)
     uint8_t col = ctx->cursor_column;
     char *row = screen->characters[ctx->cursor_row];
     char *font = screen->fonts[ctx->cursor_row];
+    char *charset = screen->charsets[ctx->cursor_row];
 
     if (col < SCREEN_COLUMNS) {
         row[col] = ch;
         font[col] = ctx->cursor_font;
+        charset[col] = ctx->cursor_charset;
+        ctx->cursor_charset = CCSET_BASIC_AMERICAN;
         if (ch) ctx->cursor_column++;
         return 0;
     }
@@ -306,6 +422,7 @@ static void roll_up(CCaptionSubContext *ctx)
         memcpy(screen->characters[i_row], screen->characters[i_row+1], SCREEN_COLUMNS);
         memcpy(screen->colors[i_row], screen->colors[i_row+1], SCREEN_COLUMNS);
         memcpy(screen->fonts[i_row], screen->fonts[i_row+1], SCREEN_COLUMNS);
+        memcpy(screen->charsets[i_row], screen->charsets[i_row+1], SCREEN_COLUMNS);
         if (CHECK_FLAG(screen->row_used, i_row + 1))
             SET_FLAG(screen->row_used, i_row);
     }
@@ -325,10 +442,12 @@ static int capture_screen(CCaptionSubContext *ctx)
         if (CHECK_FLAG(screen->row_used, i)) {
             const char *row = screen->characters[i];
             const char *font = screen->fonts[i];
+            const char *charset = screen->charsets[i];
+            const char *override;
             int j = 0;
 
             /* skip leading space */
-            while (row[j] == ' ')
+            while (row[j] == ' ' && charset[j] == CCSET_BASIC_AMERICAN)
                 j++;
 
             for (; j < SCREEN_COLUMNS; j++) {
@@ -362,7 +481,12 @@ static int capture_screen(CCaptionSubContext *ctx)
                     }
                 }
                 prev_font = font[j];
-                av_bprintf(&ctx->buffer, "%s%s%c", e_tag, s_tag, row[j]);
+                override = charset_overrides[(int)charset[j]][(int)row[j]];
+                if (override) {
+                    av_bprintf(&ctx->buffer, "%s%s%s", e_tag, s_tag, override);
+                } else {
+                    av_bprintf(&ctx->buffer, "%s%s%c", e_tag, s_tag, row[j]);
+                }
             }
             av_bprintf(&ctx->buffer, "\\N");
         }
@@ -419,6 +543,7 @@ static void handle_pac(CCaptionSubContext *ctx, uint8_t hi, uint8_t lo)
     ctx->cursor_row = row_map[index] - 1;
     ctx->cursor_color =  pac2_attribs[lo][0];
     ctx->cursor_font = pac2_attribs[lo][1];
+    ctx->cursor_charset = CCSET_BASIC_AMERICAN;
     ctx->cursor_column = 0;
     indent = pac2_attribs[lo][2];
     for (i = 0; i < indent; i++) {
@@ -474,7 +599,25 @@ static void handle_char(CCaptionSubContext *ctx, char hi, char lo, int64_t pts)
 
     SET_FLAG(screen->row_used, ctx->cursor_row);
 
-    write_char(ctx, screen, hi);
+    switch (hi) {
+      case 0x11:
+        ctx->cursor_charset = CCSET_SPECIAL_AMERICAN;
+        break;
+      case 0x12:
+        if (ctx->cursor_column > 0)
+            ctx->cursor_column -= 1;
+        ctx->cursor_charset = CCSET_EXTENDED_SPANISH_FRENCH_MISC;
+        break;
+      case 0x13:
+        if (ctx->cursor_column > 0)
+            ctx->cursor_column -= 1;
+        ctx->cursor_charset = CCSET_EXTENDED_PORTUGUESE_GERMAN_DANISH;
+        break;
+      default:
+        ctx->cursor_charset = CCSET_BASIC_AMERICAN;
+        write_char(ctx, screen, hi);
+        break;
+    }
 
     if (lo) {
         write_char(ctx, screen, lo);
@@ -560,6 +703,9 @@ static void process_cc608(CCaptionSubContext *ctx, int64_t pts, uint8_t hi, uint
             ff_dlog(ctx, "Unknown command 0x%hhx 0x%hhx\n", hi, lo);
             break;
         }
+    } else if (hi >= 0x11 && hi <= 0x13) {
+        /* Special characters */
+        handle_char(ctx, hi, lo, pts);
     } else if (hi >= 0x20) {
         /* Standard characters (always in pairs) */
         handle_char(ctx, hi, lo, pts);