Add more box removal test cases

src/boxes.in.h

@@ -20,7 +20,7 @@
 #ifndef BOXES_H
 #define BOXES_H
 
-/* #define DEBUG 1 */
+#define DEBUG 1
 /* #define REGEXP_DEBUG 1 */
 /* #define PARSER_DEBUG 1 */
 /* #define LEXER_DEBUG 1 */

src/remove.c

@@ -18,6 +18,7 @@
  */
 
 #include "config.h"
+
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
@@ -155,8 +156,10 @@ static int best_match(const line_t *line,
                 cs = opt.design->shape + east_side[++w];
             }
             #ifdef DEBUG
-                fprintf(stderr, "\nj %d, k %d, w %d, cs->chars[k] = \"%s\"\n",
-                        (int) j, (int) k, w, cs->chars[k] ? cs->chars[k] : "(null)");
+                char *mbcs_temp = bxs_to_output(cs->mbcs[k]);
+                fprintf(stderr, "\nj %d, k %d, w %d, cs->chars[k] = \"%s\", cs->mbcs[k] = \"%s\"\n",
+                        (int) j, (int) k, w, cs->chars[k] ? cs->chars[k] : "(null)", mbcs_temp);
+                BFREE(mbcs_temp);
             #endif
 
             chkline.text = cs->chars[k];
@@ -528,8 +531,7 @@ static design_t *detect_design()
 
     for (dcnt = 0; dcnt < num_designs; ++dcnt, ++d) {
         #ifdef DEBUG
-        fprintf (stderr, "CONSIDERING DESIGN ---- \"%s\" ---------------\n",
-                d->name);
+            fprintf(stderr, "CONSIDERING DESIGN ---- \"%s\" ---------------\n", d->name);
         #endif
         hits = 0;
 
@@ -537,7 +539,7 @@ static design_t *detect_design()
             empty[j] = empty_side(d->shape, j);
         }
         #ifdef DEBUG
-        fprintf (stderr, "Empty sides: TOP %d, LEFT %d, BOTTOM %d, RIGHT %d\n",
+            fprintf (stderr, "Empty sides: TOP %d, LEFT %d, BOTTOM %d, RIGHT %d\n",
                 empty[BTOP], empty[BLEF], empty[BBOT], empty[BRIG]);
         #endif
 
@@ -556,7 +558,7 @@ static design_t *detect_design()
                         break;
                     }
                     for (j = 0; j < d->shape[scnt].height; ++j) {
-                        shpln.text = d->shape[scnt].chars[j];
+                        shpln.text = d->shape[scnt].chars[j];   // TODO HERE
                         shpln.len = d->shape[scnt].width;
                         if (empty_line(&shpln)) {
                             continue;
@@ -578,9 +580,9 @@ static design_t *detect_design()
                             }
                         }
                     }
-#ifdef DEBUG
-                    fprintf (stderr, "After %s corner check:\t%ld hits.\n", shape_name[scnt], hits);
-#endif
+                    #ifdef DEBUG
+                        fprintf(stderr, "After %s corner check:\t%ld hits.\n", shape_name[scnt], hits);
+                    #endif
                     break;
 
                 case NE:
@@ -1089,7 +1091,9 @@ void output_input(const int trim_only)
             indentspc[ntabs + nspcs] = '\0';
         }
         else if (opt.tabexp == 'k') {
-            indentspc = tabbify_indent(j, NULL, input.indent);
+            uint32_t *indent32 = tabbify_indent(j, NULL, input.indent);
+            indentspc = u32_strconv_to_output(indent32);
+            BFREE(indent32);
             indent = input.indent;
         }
         else {
@@ -1104,5 +1108,4 @@ void output_input(const int trim_only)
 }
 
 
-
-/*EOF*/                                                 /* vim: set sw=4: */
+/* vim: set sw=4: */

test/178_remove_broken_box.txt

@@ -0,0 +1,35 @@
+:DESC
+Remove a broken box which is still recognized by the design autodetector because of its intact west side.
+Boxes usually get damaged on their east side because their contents got edited.
+
+:ARGS
+-r
+:INPUT
+._____. ._____. .________________________. ._____. ._____.
+| ._. | | ._. | | .____________________. | | ._. | | ._. |
+| !_| |_|_|_! | | !____________________! | | !_| |_|_|_! |
+!___| |_______! !________________________! !___| |_______!
+.___|_|_| |____________________________________|_|_| |___.
+| ._____| |________________________________________| |_. |
+| !_! | | |                                    | | ! !_! |
+!_____! | |                                
+._____. | |                                
+| ._. | | |    Lorem ipsum                        | | | ._. |
+| | | | | |    dolor sit amet         | | | | | |
+| !_! | | |    consectetur adipiscing elit,     | | ! !_! |
+!_____! | |    sed do eiusmod tempor.          | | !_____!
+._____. | |                                       | | ._____.
+| ._. | | |                                    
+| !_| |_|_|____________________________________| |_|_|_! |
+!___| |________________________________________| |_______!
+.___|_|_| |___. .________________________. .___|_|_| |___.
+| ._____| |_. | | .____________________. | | ._____| |_. |
+| !_! | | !_! | | !____________________! | | !_! | | !_! |
+!_____! !_____! !________________________! !_____! !_____!
+:OUTPUT-FILTER
+:EXPECTED
+   Lorem ipsum
+   dolor sit amet
+   consectetur adipiscing elit,
+   sed do eiusmod tempor.
+:EOF

test/179_remove_box_ansi_unicode.cfg

@@ -0,0 +1,51 @@
+# A partially colored box of unicode characters, including double-wide characters.
+
+BOX designA
+
+sample
+    ┏━┳━━━━━━━━━━ 沐宸 若汐 ━━━━━━━━━┳━┓
+    ┃ ┃                              ┃ ┃
+    ほ┃ Lorem ipsum                  ┃ふ
+    げ┃ dolor sit amet               ┃が
+    ほ┃ consectetur adipiscing elit, ┃ふ
+    げ┃ sed do eiusmod tempor.       ┃が
+    ┣━╋━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╋━┫
+    ┗━┻━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┻━┛
+ends
+
+# Monochrome sample:
+# ┏━┳━━━━━━━━━━ 沐宸 若汐 ━━━━━━━━━┳━┓
+# ┃ ┃                              ┃ ┃
+# ほ┃ Lorem ipsum                  ┃ふ
+# げ┃ dolor sit amet               ┃が
+# ほ┃ consectetur adipiscing elit, ┃ふ
+# げ┃ sed do eiusmod tempor.       ┃が
+# ┣━╋━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╋━┫
+# ┗━┻━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┻━┛
+
+shapes {
+     n (" 沐宸 若汐 ",
+        "           ")
+     nw ("┏━┳", "┃ ┃")
+     nnw ("━", " ")
+     nne ("━", " ")
+     ne ("┳━┓", "┃ ┃")
+     e ("┃ふ",
+        "┃が")
+     w ("ほ┃",
+        "げ┃")
+     se ("╋━┫",
+         "┻━┛")
+     s ("━", "━")
+     sw ("┣━╋", "┗━┻")
+}
+
+elastic (
+    nnw, nne, e, w, s
+)
+
+padding {
+    horiz 1
+}
+
+END designA

test/179_remove_box_ansi_unicode.txt

@@ -0,0 +1,21 @@
+:DESC
+Remove a box consisting of unicode characters, double-wide characters, and ansi escape codes.
+
+:ARGS
+-f 179_remove_box_ansi_unicode.cfg -r
+:INPUT
+┏━┳━━━━━━━━━━ 沐宸 若汐 ━━━━━━━━━┳━┓
+┃ ┃                              ┃ ┃
+ほ┃ Lorem ipsum                  ┃ふ
+げ┃ dolor sit amet               ┃が
+ほ┃ consectetur adipiscing elit, ┃ふ
+げ┃ sed do eiusmod tempor.       ┃が
+┣━╋━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╋━┫
+┗━┻━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┻━┛
+:OUTPUT-FILTER
+:EXPECTED
+Lorem ipsum
+dolor sit amet
+consectetur adipiscing elit,
+sed do eiusmod tempor.
+:EOF

test/180_remove_box_ansi_unicode_broken_indented.cfg

@@ -0,0 +1,51 @@
+# A partially colored box of unicode characters, include double-wide characters.
+
+BOX designA
+
+sample
+    ┏━┳━━━━━━━━━━ 沐宸 若汐 ━━━━━━━━━┳━┓
+    ┃ ┃                              ┃ ┃
+    ほ┃ Lorem ipsum                  ┃ふ
+    げ┃ dolor sit amet               ┃が
+    ほ┃ consectetur adipiscing elit, ┃ふ
+    げ┃ sed do eiusmod tempor.       ┃が
+    ┣━╋━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╋━┫
+    ┗━┻━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┻━┛
+ends
+
+# Monochrome sample:
+# ┏━┳━━━━━━━━━━ 沐宸 若汐 ━━━━━━━━━┳━┓
+# ┃ ┃                              ┃ ┃
+# ほ┃ Lorem ipsum                  ┃ふ
+# げ┃ dolor sit amet               ┃が
+# ほ┃ consectetur adipiscing elit, ┃ふ
+# げ┃ sed do eiusmod tempor.       ┃が
+# ┣━╋━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╋━┫
+# ┗━┻━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┻━┛
+
+shapes {
+     n (" 沐宸 若汐 ",
+        "           ")
+     nw ("┏━┳", "┃ ┃")
+     nnw ("━", " ")
+     nne ("━", " ")
+     ne ("┳━┓", "┃ ┃")
+     e ("┃ふ",
+        "┃が")
+     w ("ほ┃",
+        "げ┃")
+     se ("╋━┫",
+         "┻━┛")
+     s ("━", "━")
+     sw ("┣━╋", "┗━┻")
+}
+
+elastic (
+    nnw, nne, e, w, s
+)
+
+padding {
+    horiz 1
+}
+
+END designA

test/180_remove_box_ansi_unicode_broken_indented.txt

@@ -0,0 +1,23 @@
+:DESC
+Remove a box consisting of unicode characters, double-wide characters, and ansi escape codes,
+while properly restoring tabbed indentation and handling the fact that the box is broken due to editing of the text
+inside.
+
+:ARGS
+--config 180_remove_box_ansi_unicode_broken_indented.cfg --remove --tabs 4u
+:INPUT
+    ┏━┳━━━━━━━━━━ 沐宸 若汐 ━━━━━━━━━┳━┓
+    ┃ ┃                              ┃ ┃
+    ほ┃ Lorem ipsum            ┃ふ
+    げ┃ dolor sit amet       
+    ほ┃ consectetur adipiscing elit,   ┃ふ
+    げ┃ sed do eiusmod tempor. ┃が
+    ┣━╋━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╋━┫
+    ┗━┻━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┻━┛
+:OUTPUT-FILTER
+:EXPECTED
+	Lorem ipsum
+	dolor sit amet
+	consectetur adipiscing elit,
+	sed do eiusmod tempor.
+:EOF