Fixed: fully consume unknown CSI sequences containing non-numeric parameter byte

To-do: this is untested, would appreciate help.

Standard ECMA-48: Control Functions for Coded Character Sets specifies
the format of CSI commands.  Wikipedia has a concise description:
https://en.wikipedia.org/wiki/ANSI_escape_code#Control_Sequence_Introducer_commands

Do this for at least for sequences like "CSI = 5 u" that
contain non-numeric parameter bytes.

This fixes a problem with fish shell 4.0.0 which (for better or worse)
uses that sequence.

This patch introduces a slight inconsistency: for the above example,
"unknownSequence('u')" will be called.  The resulting log message may
be confusing, because we *do* support "CSI u". We should fix this to
log the entire unknown sequence. I can try doing that.

In future we should also fully consume all other unknown CSI commands.
A contrived example would be "CSI ! ! a".  If desired, I'm happy to
fix those as well, as I have some context already (but I don't think
that would block this patch).

Closes termux#4338

Author: krobelus

terminal-emulator/src/main/java/com/termux/terminal/TerminalEmulator.java

@@ -83,6 +83,10 @@ public final class TerminalEmulator {
     private static final int ESC_APC = 20;
     /** Escape processing: "ESC _" or Application Program Command (APC), followed by Escape. */
     private static final int ESC_APC_ESCAPE = 21;
+    /** Escape processing: ESC [ <parameter bytes> */
+    private static final int ESC_CSI_UNKNOWN_PARAM = 22;
+    /** Escape processing: ESC [ <parameter bytes> <intermediate bytes> */
+    private static final int ESC_CSI_UNKNOWN_INTERMEDIATE = 23;
 
     /** The number of parameter arguments including colon separated sub-parameters. */
     private static final int MAX_ESCAPE_PARAMETERS = 32;
@@ -656,6 +660,8 @@ public void processCodePoint(int b) {
                         mUseLineDrawingG1 = (b == '0');
                         break;
                     case ESC_CSI:
+                    case ESC_CSI_UNKNOWN_PARAM:
+                    case ESC_CSI_UNKNOWN_INTERMEDIATE:
                         doCsi(b);
                         break;
                     case ESC_CSI_EXCLAMATION:
@@ -1488,6 +1494,18 @@ private void restoreCursor() {
 
     /** Following a CSI - Control Sequence Introducer, "\033[". {@link #ESC_CSI}. */
     private void doCsi(int b) {
+        if (mEscapeState == ESC_CSI_UNKNOWN_PARAM || mEscapeState == ESC_CSI_UNKNOWN_INTERMEDIATE) {
+            if (mEscapeState == ESC_CSI_UNKNOWN_PARAM && b >= 0x30 && b <= 0x3F) {
+                continueSequence(mEscapeState); // parameter byte
+            } else if (b >= 0x40 && b <= 0x7E) {
+                unknownSequence(b); // final byte
+            } else if (b >= 0x20 && b <= 0x2F) {
+                continueSequence(ESC_CSI_UNKNOWN_INTERMEDIATE); // intermediate byte
+            } else {
+                unknownSequence(b);
+            }
+            return;
+        }
         switch (b) {
             case '!':
                 continueSequence(ESC_CSI_EXCLAMATION);
@@ -2218,6 +2236,8 @@ private void parseArg(int b) {
                 logError("Too many parameters when in state: " + mEscapeState);
             }
             continueSequence(mEscapeState);
+        } else if (b == '<' || b == '=' || b == '>' || b == '?') {
+            continueSequence(ESC_CSI_UNKNOWN_PARAM);
         } else {
             unknownSequence(b);
         }