Ignore more useless (to us, for now) CSI

.github/workflows/test-build.yml

@@ -46,7 +46,7 @@ jobs:
       - name: run tests
         timeout-minutes: 2
         run: |
-          make prove
+          make prove-j2
   build-static:
     runs-on: ubuntu-latest
     container: alpine:latest
@@ -89,7 +89,7 @@ jobs:
       - name: run tests [with iTerm2 palettes]
         timeout-minutes: 1
         run: |
-          prove -v tests/*.sh
+          prove -v -j 2 tests/*.sh
       - name: static compilation [no iTerm2 palettes] with mingw32-w64
         timeout-minutes: 2
         run: |
@@ -154,7 +154,7 @@ jobs:
       - name: run tests
         timeout-minutes: 2
         run: |
-          make prove
+          make prove-j2
       - name: "Compress for ${{matrix.arch}} [with iTerm2 palettes]"
         timeout-minutes: 1
         run: |

Makefile

@@ -93,3 +93,7 @@ run-tests:
 .PHONY: prove
 prove:
 	@prove tests/*.sh
+
+.PHONY: prove-j2
+prove-j2:
+	@prove -j 2 tests/*.sh

src/main.c

@@ -138,6 +138,8 @@ just_strip_it(bool ignore_sgr_errors)
         STATE_TEXT,
         STATE_GOT_ESC,
         STATE_GOT_ESC_BRACKET,
+        STATE_GOT_ESC_BRACKET_EQUALS,
+        STATE_GOT_ESC_BRACKET_QUESTION,
     } state = STATE_TEXT;
 
     char buffer[INPUT_BUFFER_SIZE];
@@ -269,6 +271,19 @@ just_strip_it(bool ignore_sgr_errors)
                     // Ignore "Erase in Line" sequence.
                     state = STATE_TEXT;
                 }
+                else if (c == 'J')
+                {
+                    // Ignore "Erase in Display" sequence.
+                    state = STATE_TEXT;
+                }
+                else if (c == '=')
+                {
+                    state = STATE_GOT_ESC_BRACKET_EQUALS;
+                }
+                else if (c == '?')
+                {
+                    state = STATE_GOT_ESC_BRACKET_QUESTION;
+                }
                 else
                 {
                     if (ignore_sgr_errors)
@@ -282,6 +297,42 @@ just_strip_it(bool ignore_sgr_errors)
                         );
                 }
             }
+            else if (state == STATE_GOT_ESC_BRACKET_EQUALS)
+            {
+                // Read numbers until "h" or "l", ignore everything in between.
+                if (c == 'h' || c == 'l')
+                    state = STATE_TEXT;
+                else if (c < '0' || c > '9')
+                {
+                    if (ignore_sgr_errors)
+                        state = STATE_TEXT;
+                    else
+                        ERROR(
+                            "SGR ]= sequence contains invalid character "
+                            "'%c' at %zu characters read / %zu in SGR sequence "
+                            "which begun at %zu characters read.\n",
+                            c, read, sgr_chars_len, begun_at
+                        );
+                }
+            }
+            else if (state == STATE_GOT_ESC_BRACKET_QUESTION)
+            {
+                // Read numbers until "h" or "l", ignore everything in between.
+                if (c == 'h' || c == 'l')
+                    state = STATE_TEXT;
+                else if (c < '0' || c > '9')
+                {
+                    if (ignore_sgr_errors)
+                        state = STATE_TEXT;
+                    else
+                        ERROR(
+                            "SGR ]? sequence contains invalid character "
+                            "'%c' at %zu characters read / %zu in SGR sequence "
+                            "which begun at %zu characters read.\n",
+                            c, read, sgr_chars_len, begun_at
+                        );
+                }
+            }
             else
             {
                 ERROR("Invalid state %d.\n", (int)state);
@@ -328,6 +379,8 @@ static inline __attribute__((always_inline)) void ansi2html(
         STATE_TEXT,
         STATE_GOT_ESC,
         STATE_GOT_ESC_BRACKET,
+        STATE_GOT_ESC_BRACKET_EQUALS,
+        STATE_GOT_ESC_BRACKET_QUESTION,
     } state = STATE_TEXT;
 
     char buffer[INPUT_BUFFER_SIZE];
@@ -534,6 +587,24 @@ static inline __attribute__((always_inline)) void ansi2html(
                         span_outputted = true;
                     }
                 }
+                else if (c == 'J')
+                {
+                    // Ignore "Erase in Display" sequence.
+                    state = STATE_TEXT;
+                    if (span && !span_outputted)
+                    {
+                        append_to_buffer(span);
+                        span_outputted = true;
+                    }
+                }
+                else if (c == '=')
+                {
+                    state = STATE_GOT_ESC_BRACKET_EQUALS;
+                }
+                else if (c == '?')
+                {
+                    state = STATE_GOT_ESC_BRACKET_QUESTION;
+                }
                 else
                 {
                     // The character should be a digit, or semicolon.
@@ -588,6 +659,30 @@ static inline __attribute__((always_inline)) void ansi2html(
                         );
                 }
                 break;
+            case STATE_GOT_ESC_BRACKET_EQUALS:
+                // Read numbers until "h" or "l", ignore everything in between.
+                if (c == 'h' || c == 'l')
+                    state = STATE_TEXT;
+                else if (c < '0' || c > '9')
+                    ERROR(
+                        "SGR ]= sequence contains invalid character "
+                        "'%c' at %zu characters read / %zu in SGR sequence "
+                        "which begun at %zu characters read.\n",
+                        c, read, sgr_chars_len, begun_at
+                    );
+                break;
+            case STATE_GOT_ESC_BRACKET_QUESTION:
+                // Read numbers until "h" or "l", ignore everything in between.
+                if (c == 'h' || c == 'l')
+                    state = STATE_TEXT;
+                else if (c < '0' || c > '9')
+                    ERROR(
+                        "SGR ]? sequence contains invalid character "
+                        "'%c' at %zu characters read / %zu in SGR sequence "
+                        "which begun at %zu characters read.\n",
+                        c, read, sgr_chars_len, begun_at
+                    );
+                break;
             }
         }
     } while (1);

tests/csi-equals.sh

@@ -0,0 +1,21 @@
+#!/bin/bash
+
+set -e
+
+. "$(dirname "$0")/".functions.sh
+
+# CSI = 0 ... 19 h and CSI = ... l are no-op
+for l in h l; do
+    for k in $(seq 0 19); do
+        str=$(printf 'Hello, \e[0;1mBold\e[=%s%s\e[0m world!' "$k" "$l")
+        want=$'Hello, <span style="font-weight:bold;">Bold</span> world!'
+        got=$(printf '%s' "$str" | ./ansi2html -p vga)
+        str_eq_html "$str" "$want" "$got"
+
+        want=$'Hello, Bold world!'
+        got=$(printf '%s' "$str" | ./ansi2html -S)
+        str_eq_html "$str" "$want" "$got"
+    done
+done
+
+done_testing

tests/csi-j.sh

@@ -0,0 +1,29 @@
+#!/bin/bash
+
+set -e
+
+. "$(dirname "$0")/".functions.sh
+
+# CSI J is a no-op when stripping OR converting to HTML
+str=$'Hello, \e[0;1mBold\e[J\e[0m world!'
+want=$'Hello, <span style="font-weight:bold;">Bold</span> world!'
+got=$(printf '%s' "$str" | ./ansi2html -p vga)
+str_eq_html "$str" "$want" "$got"
+
+want=$'Hello, Bold world!'
+got=$(printf '%s' "$str" | ./ansi2html -S)
+str_eq_html "$str" "$want" "$got"
+
+# CSI 0 J is also a no-op, same for CSI 1 J, CSI 2 J and CSI 3 J
+for k in 0 1 2 3; do
+    str=$(printf 'Hello, \e[0;1mBold\e[%sJ\e[0m world!' "$k")
+    want=$'Hello, <span style="font-weight:bold;">Bold</span> world!'
+    got=$(printf '%s' "$str" | ./ansi2html -p vga)
+    str_eq_html "$str" "$want" "$got"
+
+    want=$'Hello, Bold world!'
+    got=$(printf '%s' "$str" | ./ansi2html -S)
+    str_eq_html "$str" "$want" "$got"
+done
+
+done_testing

tests/csi-k.sh

@@ -14,4 +14,16 @@ want=$'Hello, Bold world!'
 got=$(printf '%s' "$str" | ./ansi2html -S)
 str_eq_html "$str" "$want" "$got"
 
+# CSI 0 K is also a no-op, same for CSI 1 K and CSI 2 K
+for k in 0 1 2; do
+    str=$(printf 'Hello, \e[0;1mBold\e[%sK\e[0m world!' "$k")
+    want=$'Hello, <span style="font-weight:bold;">Bold</span> world!'
+    got=$(printf '%s' "$str" | ./ansi2html -p vga)
+    str_eq_html "$str" "$want" "$got"
+
+    want=$'Hello, Bold world!'
+    got=$(printf '%s' "$str" | ./ansi2html -S)
+    str_eq_html "$str" "$want" "$got"
+done
+
 done_testing

tests/csi-question.sh

@@ -0,0 +1,21 @@
+#!/bin/bash
+
+set -e
+
+. "$(dirname "$0")/".functions.sh
+
+# CSI ? ... h and CSI ? ... l are no-op
+for l in h l; do
+    for k in 25 47 1049; do
+        str=$(printf 'Hello, \e[0;1mBold\e[?%s%s\e[0m world!' "$k" "$l")
+        want=$'Hello, <span style="font-weight:bold;">Bold</span> world!'
+        got=$(printf '%s' "$str" | ./ansi2html -p vga)
+        str_eq_html "$str" "$want" "$got"
+
+        want=$'Hello, Bold world!'
+        got=$(printf '%s' "$str" | ./ansi2html -S)
+        str_eq_html "$str" "$want" "$got"
+    done
+done
+
+done_testing