Test and fix more comparison types in box removal

src/boxes.c

@@ -24,6 +24,7 @@
 #include <unistd.h>
 
 #include "boxes.h"
+#include "bxstring.h"
 #include "cmdline.h"
 #include "discovery.h"
 #include "generate.h"

src/remove.c

@@ -400,6 +400,9 @@ static int hmm_shiftable(shape_line_ctx_t *shapes_relevant, uint32_t *cur_pos, s
                 if (p != NULL && p < end_pos && is_blank_between(cur_pos, p)) {
                     result = hmm(shapes_relevant, p + quality, i + (shapes_relevant[i].elastic ? 0 : 1),
                             end_pos, 1, anchored_right);
+                    if (result == 0 && shapes_relevant[i].elastic) {
+                        result = hmm(shapes_relevant, p + quality, i + 1, end_pos, 1, anchored_right);
+                    }
                     break;
                 }
                 if (can_shorten_right == -1) {
@@ -521,6 +524,9 @@ static int match_horiz_line(remove_ctx_t *ctx, int hside, size_t input_line_idx,
             continue;
         }
         ctx->comp_type = comp_type;
+        #ifdef DEBUG
+            fprintf(stderr, "  Setting comparison type to: %s\n", comparison_name[comp_type]);
+        #endif
 
         shape_line_ctx_t *shapes_relevant = prepare_comp_shapes_horiz(hside, comp_type, shape_line_idx);
         debug_print_shapes_relevant(shapes_relevant);

src/tools.c

@@ -566,6 +566,14 @@ int is_csi_reset(const uint32_t *csi)
                     if (puc >= 0x40 && puc <= 0x7e) {
                         return 1;
                     }
+                    else if ((puc == '1' && *rest == '0')
+                          || (puc == '3' && *rest == '9')
+                          || (puc == '4' && *rest == '9')
+                          || (puc == '5' && *rest == '9')
+                          || (puc == '7' && *rest == '5')) {
+                        rest = u32_next(&puc, rest);
+                        break;
+                    }
                     return 0;
                 }
                 break;

src/tools.h

@@ -156,10 +156,12 @@ size_t count_invisible_chars(const uint32_t *s, size_t *num_esc, char **ascii, s
 
 
 /**
- * Determine whether the given sequence of characters is a CSI (also called "escape sequence") that resets all
- * modifications, typically `ESC[0m`.
- * @param csi a pointer into a zero-terminated UTF-32 string
- * @returns 1 if true, 0 if false
+ * Determine if `csi` points at a CSI which can be considered a "reset" sequence.
+ * https://en.wikipedia.org/wiki/ANSI_escape_code#SGR_(Select_Graphic_Rendition)_parameters
+ * We recognize `\x1b[0m`, and also 10, 39, 49, 59, and 75 instead of 0 as "resets", because they all serve to reset
+ * the terminal to some default. Especially, lolcat uses `\x1b[39m` as reset CSI.
+ * @param csi a string which might or might not be a CSI "reset" sequence (zero-terminated UTF-32 string)
+ * @return 1 if it was a CSI "reset", 0 otherwise
  */
 int is_csi_reset(const uint32_t *csi);
 

test/181_remove_box_ignore_invisible_all.txt

@@ -0,0 +1,28 @@
+:DESC
+Remove a box consisting of unicode characters and ansi escape codes, even when the colors are all wrong (input box
+has different colors than the box design).
+Boxes shall recognize a box design by the visible characters, not by the colors.
+This tests the comparison mode 'ignore_invisible_all'.
+
+:ARGS
+--config 18x_remove_box_ignore_invisible_all.cfg --remove
+:INPUT
+         .-"```"-.
+        /_______; \
+       (_________)\|
+       / / 👁 👁 \ \(_)
+      / ( \___/ ) \
+  _ooo\__\_____/__/____
+ /                     \
+|     Hello World!      |
+ \_________________ooo_/
+      /           \
+     /:.:.:.:.:.:.:\
+         |  |  |
+         \==|==/  jgs
+         /-'Y'-\
+        (__/ \__)
+:OUTPUT-FILTER
+:EXPECTED
+    Hello World!
+:EOF

test/182_remove_box_ignore_invisible_shape.txt

@@ -0,0 +1,26 @@
+:DESC
+In this test case, the box design is fully colored, but the input is monochrome.
+This tests the comparison mode 'ignore_invisible_shape'.
+
+:ARGS
+--config 18x_remove_box_ignore_invisible_all.cfg --design designA --remove
+:INPUT
+         .-"```"-.
+        /_______; \
+       (_________)\|
+       / / 👁 👁 \ \(_)
+      / ( \___/ ) \
+  _ooo\__\_____/__/____
+ /                     \
+|     Hello World!      |
+ \_________________ooo_/
+      /           \
+     /:.:.:.:.:.:.:\
+         |  |  |
+         \==|==/  jgs
+         /-'Y'-\
+        (__/ \__)
+:OUTPUT-FILTER
+:EXPECTED
+    Hello World!
+:EOF

test/183_remove_box_ignore_invisible_input.txt

@@ -0,0 +1,28 @@
+:DESC
+In this test case, the box design is monochrome, but the input is colored.
+This tests the comparison mode 'ignore_invisible_input'.
+
+:ARGS
+--design capgirl --remove
+:INPUT
+         .-"```"-.
+        /_______; \
+       (_________)\|
+       / / a a \ \(_)
+      / ( \___/ ) \
+  _ooo\__\_____/__/____
+ /                     \
+|     Hello, world!     |
+|     World, hello!     |
+ \_________________ooo_/
+      /           \
+     /:.:.:.:.:.:.:\
+         |  |  |
+         \==|==/  jgs
+         /-'Y'-\
+        (__/ \__)
+:OUTPUT-FILTER
+:EXPECTED
+    Hello, world!
+    World, hello!
+:EOF

test/18x_remove_box_ignore_invisible_all.cfg

@@ -0,0 +1,71 @@
+# A fully lolcat-colored box including unicode eyes.
+
+BOX designA
+
+sample
+             .-"```"-.
+            /_______; \
+           (_________)\|
+           / / 👁 👁 \ \(_)
+          / ( \___/ ) \
+      _ooo\__\_____/__/____
+     /                     \
+    |     Hello World!      |
+     \_________________ooo_/
+          /           \
+         /:.:.:.:.:.:.:\
+             |  |  |
+             \==|==/  jgs
+             /-'Y'-\
+            (__/ \__)
+ends
+
+# Monochrome sample:
+#             .-"```"-.
+#            /_______; \
+#           (_________)\|
+#           / / 👁 👁 \ \(_)
+#          / ( \___/ ) \
+#      _ooo\__\_____/__/____
+#     /                     \
+#    |     Hello World!      |
+#     \_________________ooo_/
+#          /           \
+#         /:.:.:.:.:.:.:\
+#             |  |  |
+#             \==|==/  jgs
+#             /-'Y'-\
+#            (__/ \__)
+
+delim ~"
+
+shapes {
+    n   ("      .-~"```~"-.    ",
+         "     /_______; \   ",
+         "    (_________)\|  ",
+         "    / / 👁 👁 \ \(_) ",
+         "   / ( \___/ ) \   ",
+         "ooo\__\_____/__/___",
+         "                   ")
+    s   ("________________ooo",
+         "   /           \   ",
+         "  /:.:.:.:.:.:.:\  ",
+         "      |  |  |      ",
+         "      \==|==/  jgs ",
+         "      /-'Y'-\      ",
+         "     (__/ \__)     ")
+    e   (" |")
+    w   ("| ")
+    sw  (" \", "  ", "  ", "  ", "  ", "  ", "  ")
+    se  ("/ ", "  ", "  ", "  ", "  ", "  ", "  ")
+    ssw ("_", " ", " ", " ", " ", " ", " ")
+    sse ("_", " ", " ", " ", " ", " ", " ")
+    nw  ("  ", "  ", "  ", "  ", "  ", "  ", " /")
+    nnw (" ", " ", " ", " ", " ", "_", " ")
+    nne (" ", " ", " ", " ", " ", "_", " ")
+    ne  ("  ", "  ", "  ", "  ", "  ", "  ", "\ ")
+}
+
+elastic (nne,nnw, sse,ssw, e, w)
+
+END designA

utest/bxstring_test.c

@@ -353,6 +353,34 @@ void test_ansi_unicode_null(void **state)
 
 
 
+void test_ansi_unicode_tc183(void **state)
+{
+    UNUSED(state);
+
+    uint32_t *ustr32 = u32_strconv_from_arg("\x1b[38;5;43m|\x1b[39m\x1b[38;5;49m \x1b[39m    X", "ASCII");
+    assert_non_null(ustr32);
+    bxstr_t *actual = bxs_from_unicode(ustr32);
+
+    assert_non_null(actual);
+    assert_non_null(actual->memory);
+    assert_string_equal("|     X", actual->ascii);
+    assert_int_equal(0, (int) actual->indent);
+    assert_int_equal(7, (int) actual->num_columns);
+    assert_int_equal(37, (int) actual->num_chars);
+    assert_int_equal(7, (int) actual->num_chars_visible);
+    assert_int_equal(30, (int) actual->num_chars_invisible);
+    assert_int_equal(0, (int) actual->trailing);
+    int expected_firstchar_idx[] = {0, 16, 32, 33, 34, 35, 36, 37};
+    assert_array_equal(expected_firstchar_idx, actual->first_char, 8);
+    int expected_vischar_idx[] = {10, 26, 32, 33, 34, 35, 36, 37};
+    assert_array_equal(expected_vischar_idx, actual->visible_char, 8);
+
+    BFREE(ustr32);
+    bxs_free(actual);
+}
+
+
+
 void test_bxs_new_empty_string(void **state)
 {
     UNUSED(state);

utest/bxstring_test.h

@@ -35,6 +35,7 @@ void test_ansi_unicode_illegalchar(void **state);
 void test_ansi_unicode_tabs(void **state);
 void test_ansi_unicode_broken_escapes(void **state);
 void test_ansi_unicode_null(void **state);
+void test_ansi_unicode_tc183(void **state);
 
 void test_bxs_new_empty_string(void **state);
 

utest/main.c

@@ -135,6 +135,7 @@ int main(void)
         cmocka_unit_test_setup(test_ansi_unicode_tabs, beforeTest),
         cmocka_unit_test_setup(test_ansi_unicode_broken_escapes, beforeTest),
         cmocka_unit_test_setup(test_ansi_unicode_null, beforeTest),
+        cmocka_unit_test_setup(test_ansi_unicode_tc183, beforeTest),
         cmocka_unit_test_setup(test_bxs_new_empty_string, beforeTest),
         cmocka_unit_test_setup(test_bxs_is_blank, beforeTest),
         cmocka_unit_test_setup(test_bxs_strdup, beforeTest),