On Mon, Nov 24, 2014 at 4:26 PM, Norihiro Tanaka <nori...@kcn.ne.jp> wrote: > On Mon, 13 Oct 2014 19:39:21 +0900 > Norihiro Tanaka <nori...@kcn.ne.jp> wrote: >> If compile grep with -DDEBUG option, it outputs tokens etc. In addition >> to them, DFA states and transitions are output by this patch. Debugging >> them, the patch will be very useful. >> >> BTW, the patch doesn't make any changes without -DDEBUG option. > > I fixed some warnings output in compilation, and rebased it.
Thank you for your patience. I applied this, changed many %zd to %zu (to ensure compilation succeeded with -DDEBUG also when configured with --enable-gcc-warnings) and adjusted the commit log entry. Will push once you've ACK'd the result.
From 8f675e7c026ce56a8ef0cc33dc5fcd37f49f38a2 Mon Sep 17 00:00:00 2001 From: Norihiro Tanaka <nori...@kcn.ne.jp> Date: Sat, 22 Mar 2014 17:00:36 +0900 Subject: [PATCH] dfa: DEBUG: print detail of DFA states When compiled with -DDEBUG, grep outputs tokens etc. With this change, also print DFA states and transitions. This change is very useful when debugging those. * src/dfa.c (prtok) [DEBUG]: Change `%c' to `%02x' in printf format. (state_index) [DEBUG]: Print detail of new state. (dfastate) [DEBUG]: Print detail of DFA states. Reported as http://debbugs.gnu.org/18707 --- src/dfa.c | 86 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 78 insertions(+), 8 deletions(-) diff --git a/src/dfa.c b/src/dfa.c index f9b3da9..b5ca825 100644 --- a/src/dfa.c +++ b/src/dfa.c @@ -509,8 +509,8 @@ prtok (token t) fprintf (stderr, "END"); else if (t < NOTCHAR) { - int ch = t; - fprintf (stderr, "%c", ch); + unsigned int ch = t; + fprintf (stderr, "0x%02x", ch); } else { @@ -2139,6 +2139,28 @@ state_index (struct dfa *d, position_set const *s, int context) return i; } +#ifdef DEBUG + fprintf (stderr, "new state %zd\n nextpos:", i); + for (j = 0; j < s->nelem; ++j) + { + fprintf (stderr, " %zu:", s->elems[j].index); + prtok (d->tokens[s->elems[j].index]); + } + fprintf (stderr, "\n context:"); + if (context ^ CTX_ANY) + { + if (context & CTX_NONE) + fprintf (stderr, " CTX_NONE"); + if (context & CTX_LETTER) + fprintf (stderr, " CTX_LETTER"); + if (context & CTX_NEWLINE) + fprintf (stderr, " CTX_NEWLINE"); + } + else + fprintf (stderr, " CTX_ANY"); + fprintf (stderr, "\n"); +#endif + /* We'll have to create a new state. */ d->states = maybe_realloc (d->states, d->sindex, &d->salloc, sizeof *d->states); @@ -2369,7 +2391,7 @@ dfaanalyze (struct dfa *d, int searchflag) fprintf (stderr, "dfaanalyze:\n"); for (i = 0; i < d->tindex; ++i) { - fprintf (stderr, " %zd:", i); + fprintf (stderr, " %zu:", i); prtok (d->tokens[i]); } putc ('\n', stderr); @@ -2483,7 +2505,7 @@ dfaanalyze (struct dfa *d, int searchflag) } #ifdef DEBUG /* ... balance the above nonsyntactic #ifdef goo... */ - fprintf (stderr, "node %zd:", i); + fprintf (stderr, "node %zu:", i); prtok (d->tokens[i]); putc ('\n', stderr); fprintf (stderr, @@ -2491,13 +2513,13 @@ dfaanalyze (struct dfa *d, int searchflag) fprintf (stderr, " firstpos:"); for (j = stk[-1].nfirstpos; j-- > 0;) { - fprintf (stderr, " %zd:", firstpos[j].index); + fprintf (stderr, " %zu:", firstpos[j].index); prtok (d->tokens[firstpos[j].index]); } fprintf (stderr, "\n lastpos:"); for (j = stk[-1].nlastpos; j-- > 0;) { - fprintf (stderr, " %zd:", lastpos[j].index); + fprintf (stderr, " %zu:", lastpos[j].index); prtok (d->tokens[lastpos[j].index]); } putc ('\n', stderr); @@ -2512,12 +2534,12 @@ dfaanalyze (struct dfa *d, int searchflag) || d->tokens[i] >= CSET) { #ifdef DEBUG - fprintf (stderr, "follows(%zd:", i); + fprintf (stderr, "follows(%zu:", i); prtok (d->tokens[i]); fprintf (stderr, "):"); for (j = d->follows[i].nelem; j-- > 0;) { - fprintf (stderr, " %zd:", d->follows[i].elems[j].index); + fprintf (stderr, " %zu:", d->follows[i].elems[j].index); prtok (d->tokens[d->follows[i].elems[j].index]); } putc ('\n', stderr); @@ -2607,6 +2629,10 @@ dfastate (state_num s, struct dfa *d, state_num trans[]) bool next_isnt_1st_byte = false; /* We can't add state0. */ size_t i, j, k; +#ifdef DEBUG + fprintf (stderr, "build state %td\n", s); +#endif + zeroset (matches); for (i = 0; i < d->states[s].elems.nelem; ++i) @@ -2658,6 +2684,16 @@ dfastate (state_num s, struct dfa *d, state_num trans[]) continue; } +#ifdef DEBUG + fprintf (stderr, " nextpos %zu:", pos.index); + prtok (d->tokens[pos.index]); + fprintf (stderr, " of"); + for (j = 0; j < NOTCHAR; j++) + if (tstbit (j, matches)) + fprintf (stderr, " 0x%02zx", j); + fprintf (stderr, "\n"); +#endif + for (j = 0; j < ngrps; ++j) { /* If matches contains a single character only, and the current @@ -2818,6 +2854,29 @@ dfastate (state_num s, struct dfa *d, state_num trans[]) else state_letter = state; +#ifdef DEBUG + fprintf (stderr, "group %zu\n nextpos:", i); + for (j = 0; j < grps[i].nelem; ++j) + { + fprintf (stderr, " %zu:", grps[i].elems[j]); + prtok (d->tokens[grps[i].elems[j]]); + } + fprintf (stderr, "\n follows:"); + for (j = 0; j < follows.nelem; ++j) + { + fprintf (stderr, " %zu:", follows.elems[j].index); + prtok (d->tokens[follows.elems[j].index]); + } + fprintf (stderr, "\n states:"); + if (possible_contexts & CTX_NEWLINE) + fprintf (stderr, " CTX_NEWLINE:%td", state_newline); + if (possible_contexts & CTX_LETTER) + fprintf (stderr, " CTX_LETTER:%td", state_letter); + if (possible_contexts & CTX_NONE) + fprintf (stderr, " CTX_NONE:%td", state); + fprintf (stderr, "\n"); +#endif + /* Set the transitions for each character in the current label. */ for (j = 0; j < CHARCLASS_WORDS; ++j) for (k = 0; k < CHARCLASS_WORD_BITS; ++k) @@ -2834,6 +2893,17 @@ dfastate (state_num s, struct dfa *d, state_num trans[]) } } +#ifdef DEBUG + fprintf (stderr, "trans table %td", s); + for (i = 0; i < NOTCHAR; ++i) + { + if (!(i & 0xf)) + fprintf (stderr, "\n"); + fprintf (stderr, " %2td", trans[i]); + } + fprintf (stderr, "\n"); +#endif + for (i = 0; i < ngrps; ++i) free (grps[i].elems); free (follows.elems); -- 2.3.7
