From: Wyss Clemens <WYS@...>
Date: 2001-09-21T17:51:17+09:00
Subject: [ruby-talk:21482] Regexp hangs

Good morning (evening to Matz and Co ;-)),
Sorry for the long appendix.
We have a regexp(in line 159) which seems to "hang", at least it takes a
verrrry long time to match. To me it looks like this is a bug or a special
case, which might need special hadling.
The regexp tries to match function declarations. If we put the function name
of the first function declaration also to the end (i.e. line 19 -> end
H_LINT_2_UNSIGNED_INTEGER;) the regexp matches immediately. If the name is
omitted (or wrong) it takes almost 2 minutes before the script terminates.
And this time seems to increase exponentially depending on str's size. Also,
when we remove the \1 in the regexp the "problem" disappears.

A  bug, or are making wrong use of regexps (maybe the \1)?
Thanks for any feedback
Clemens
--------- appendix --------------------
#!/usr/bin/env ruby 
str = "
function H_LINT_2_UNSIGNED_INTEGER (VALUE : in LONG_INTEGER)
return UNSIGNED_INTEGER is
  CASTED_VAL : array [2] of UNSIGNED_INTEGER;
  TEMP : LONG_INTEGER;
begin
  if (VALUE < 0) then
    CASTED_VAL[1] := 0;
  else
    if (VALUE > H#FFFF#) then
      CASTED_VAL[1] := H#FFFF#;
    else
      // Value is valid
      CASTED_VAL := BI.LINT_2_UINT (VALUE);
    end if;
  end if;
  return (CASTED_VAL[1]);
end ddd ;
// ^ the function name should be here!

procedure GET_ALICE_LCL_STATUS 
  (
    P_LATCH : in INTEGER;
    P_ALICE_IS_ON : BOOLEAN;
    P_DP_READ_RESULT : BOOLEAN
  ) is
LCL_STATUS : LONG_INTEGER;
begin
  P_ALICE_IS_ON := FALSE;

  if (P_LATCH = K_PAR1_LCL_A) then
    // Get the state of LCL A
    DP.READ_SPECIFIC_BITS (PPK_NPWDA11D_ACCESS_DESC, LCL_STATUS,
                          P_DP_READ_RESULT, PPK_NPWDA11D_MASK,
                           INTEGER(PPK_NPWDA11D_SHIFT));
    if (LCL_STATUS = PPK_NPWDA11D_ON) then
      // ALICE LCL A is ON 
      P_ALICE_IS_ON := TRUE;
    end if;
  else
    // Get the value of LCL B
    DP.READ_SPECIFIC_BITS (PPK_NPWDA23D_ACCESS_DESC, LCL_STATUS,
                           P_DP_READ_RESULT, PPK_NPWDA23D_MASK,
                           INTEGER(PPK_NPWDA23D_SHIFT));
    if (LCL_STATUS = PPK_NPWDA23D_ON) then
      // ALICE LCL B is ON 
      P_ALICE_IS_ON := TRUE;
    end if;
  end if;
end GET_ALICE_LCL_STATUS;

function IS_PRECONDITION_VALID () return BOOLEAN is
RET_VALUE         : BOOLEAN;
EVENT_RESULT      : BOOLEAN;
IS_LCL_ON         : BOOLEAN;
DP_READ_RESULT    : BOOLEAN;
begin
  RET_VALUE := TRUE;

  // Check if ALICE LCL A is OFF
  GET_ALICE_LCL_STATUS (K_PAR1_LCL_A, IS_LCL_ON, DP_READ_RESULT);
  if IS_LCL_ON then
    // ALICE LCL A is ON 
    EVENT_RESULT := MSG.SEND_REPORT (K_EVENT_3 [0], K_EVENT_3 [1],
                                     K_EVENT_3 [2],
                                     K_EMPTY_EVENT_PARAM_LIST);
    RET_VALUE := FALSE;
  end if;

  if RET_VALUE then
    // Check if ALICE LCL B is OFF
    GET_ALICE_LCL_STATUS (K_PAR1_LCL_B, IS_LCL_ON, DP_READ_RESULT);
    if IS_LCL_ON then
      // ALICE LCL B is ON  
      EVENT_RESULT := MSG.SEND_REPORT (K_EVENT_4 [0], K_EVENT_4 [1],
                                       K_EVENT_4 [2],
                                       K_EMPTY_EVENT_PARAM_LIST);
      RET_VALUE := FALSE;
    end if;
  end if;

  if (RET_VALUE) then
    // The preconditions are valid
    EVENT_RESULT := MSG.SEND_REPORT (K_EVENT_50 [0], K_EVENT_50 [1], 
                                     K_EVENT_50 [2],
K_EMPTY_EVENT_PARAM_LIST);
  end if;

  return (RET_VALUE);
end IS_PRECONDITION_VALID;

function IS_SWITCH_LCL_OFF_SUCCESSFUL
  (
    P_LATCH : in INTEGER
  ) return BOOLEAN is
RET_VALUE      : BOOLEAN;
TC_RESULT      : BOOLEAN;
IS_ALICE_ON     : BOOLEAN;
IS_CHECK_VALID : BOOLEAN;
begin
  RET_VALUE := TRUE;

  // Switch ALICE LCL OFF
  if (P_LATCH = K_PAR1_LCL_A) then
    TC_RESULT := MSG.SEND_TC_AND_CONTINUE (PPK_TC_ZPWMA030SWOF_BUFFER);
    TC_RESULT := MSG.SEND_TC_AND_CONTINUE (PPK_TC_ZPWMB030SWOF_BUFFER);
  else
    TC_RESULT := MSG.SEND_TC_AND_CONTINUE (PPK_TC_ZPWMA287SWOF_BUFFER);
    TC_RESULT := MSG.SEND_TC_AND_CONTINUE (PPK_TC_ZPWMB287SWOF_BUFFER);
  end if;

  // Wait for the LCL reaction
  BI.WAIT_DELAY (K_TM_UPDATE_CYCLE_TIME);

  // Verify if ALICE LCL is off
  GET_ALICE_LCL_STATUS (P_LATCH, IS_ALICE_ON, IS_CHECK_VALID);
  if (IS_ALICE_ON) then
    RET_VALUE := FALSE;
  end if;

  return (RET_VALUE);
end IS_SWITCH_LCL_OFF_SUCCESSFUL;

function IS_SWITCH_LCL_OFF_SUCCESSFUL
  (
    P_LATCH : in INTEGER
  ) return BOOLEAN is
RET_VALUE      : BOOLEAN;
TC_RESULT      : BOOLEAN;
IS_ALICE_ON     : BOOLEAN;
IS_CHECK_VALID : BOOLEAN;
begin
  RET_VALUE := TRUE;

  // Switch ALICE LCL OFF
  if (P_LATCH = K_PAR1_LCL_A) then
    TC_RESULT := MSG.SEND_TC_AND_CONTINUE (PPK_TC_ZPWMA030SWOF_BUFFER);
    TC_RESULT := MSG.SEND_TC_AND_CONTINUE (PPK_TC_ZPWMB030SWOF_BUFFER);
  else
    TC_RESULT := MSG.SEND_TC_AND_CONTINUE (PPK_TC_ZPWMA287SWOF_BUFFER);
    TC_RESULT := MSG.SEND_TC_AND_CONTINUE (PPK_TC_ZPWMB287SWOF_BUFFER);
  end if;

  // Wait for the LCL reaction
  BI.WAIT_DELAY (K_TM_UPDATE_CYCLE_TIME);

  // Verify if ALICE LCL is off
  GET_ALICE_LCL_STATUS (P_LATCH, IS_ALICE_ON, IS_CHECK_VALID);
  if (IS_ALICE_ON) then
    RET_VALUE := FALSE;
  end if;

  return (RET_VALUE);
end IS_SWITCH_LCL_OFF_SUCCESSFUL;
  
"

# The ominous regexp ;-)
str =~
/^function\s*(.*?)\s*[(](.*?)[)]\s*return\s*(.*?)\s*is\s*(.*?)begin\s*(.*?)e
nd\s*\1\s*[;]/m