From 342b3c9db6d7c357bcc30093dad1b66392ecbf4f Mon Sep 17 00:00:00 2001 From: hyung-hwan Date: Sun, 22 Mar 2009 03:19:28 +0000 Subject: [PATCH] working on sed --- qse/include/qse/utl/sed.h | 11 +++- qse/lib/utl/sed.c | 129 +++++++++++++++++++++++++++++--------- qse/lib/utl/sed.h | 8 ++- 3 files changed, 118 insertions(+), 30 deletions(-) diff --git a/qse/include/qse/utl/sed.h b/qse/include/qse/utl/sed.h index 8ed5377c..a6156ba6 100644 --- a/qse/include/qse/utl/sed.h +++ b/qse/include/qse/utl/sed.h @@ -37,11 +37,15 @@ enum qse_sed_errnum_t QSE_SED_EA2PHB, /* address 2 prohibited */ QSE_SED_ENEWLN, /* a new line is expected */ QSE_SED_EBSEXP, /* \ is expected */ + QSE_SED_EGBABS, /* garbage after \ */ QSE_SED_ESCEXP, /* ; is expected */ QSE_SED_ELABTL, /* label too long */ QSE_SED_ELABEM, /* label name is empty */ QSE_SED_ELABDU, /* duplicate label name */ - QSE_SED_ETSINV /* invalid translation set */ + QSE_SED_ETSNTR, /* translation set not terminated */ + QSE_SED_ETSNSL, /* translation set not the same length*/ + QSE_SED_EGRNBA, /* group brackets not balanced */ + QSE_SED_EGRNTD /* group nested too deeply */ }; enum qse_sed_option_t @@ -82,6 +86,11 @@ struct qse_sed_t } cmd; qse_map_t labs; /* label map */ + + /* current level of command group nesting */ + int grplvl; + /* temporary storage to keep track of the begining of a command group */ + qse_sed_cmd_t* grpcmd[128]; }; diff --git a/qse/lib/utl/sed.c b/qse/lib/utl/sed.c index ad0ad94a..1cb29ecf 100644 --- a/qse/lib/utl/sed.c +++ b/qse/lib/utl/sed.c @@ -108,6 +108,11 @@ for (c = sed->cmd.buf; c != sed->cmd.cur; c++) if (c->u.branch.text != QSE_NULL) qse_str_close (c->u.branch.text); } + else if (c->type == QSE_SED_CMD_Y) + { + if (c->u.transet.ptr != QSE_NULL) + QSE_MMGR_FREE (sed->mmgr, c->u.transet.ptr); + } } QSE_MMGR_FREE (sed->mmgr, sed->cmd.buf); @@ -129,11 +134,15 @@ const qse_char_t* qse_sed_geterrmsg (qse_sed_t* sed) QSE_T("address 2 prohibited"), QSE_T("a new line expected"), QSE_T("a backslash expected"), + QSE_T("garbage after a backslash"), QSE_T("a semicolon expected"), QSE_T("label name too long"), QSE_T("empty label name"), QSE_T("duplicate label name"), - QSE_T("invalid translation set") + QSE_T("translation set not terminated"), + QSE_T("strings in translation set not the same length"), + QSE_T("group brackets not balanced"), + QSE_T("group nesting too deep") }; return (sed->errnum > 0 && sed->errnum < QSE_COUNTOF(errmsg))? @@ -498,16 +507,17 @@ oops: return -1; } -static int get_trans_set (qse_sed_t* sed, qse_sed_cmd_t* cmd) +static int get_transet (qse_sed_t* sed, qse_sed_cmd_t* cmd) { qse_cint_t c, delim; qse_str_t* t = QSE_NULL; + qse_size_t pos; c = CURSC (sed); if (c == QSE_CHAR_EOF || IS_LINTERM(c)) { - /* invalid translation set */ - sed->errnum = QSE_SED_ETSINV; + /* translation set terminated prematurely*/ + sed->errnum = QSE_SED_ETSNTR; goto oops; } @@ -527,8 +537,7 @@ static int get_trans_set (qse_sed_t* sed, qse_sed_cmd_t* cmd) if (c == QSE_CHAR_EOF || IS_LINTERM(c)) { - /* invalid translation set */ - sed->errnum = QSE_SED_ETSINV; + sed->errnum = QSE_SED_ETSNTR; goto oops; } @@ -537,10 +546,11 @@ static int get_trans_set (qse_sed_t* sed, qse_sed_cmd_t* cmd) c = NXTSC (sed); if (c == QSE_CHAR_EOF || IS_LINTERM(c)) { - /* invalid translation set */ - sed->errnum = QSE_SED_ETSINV; + sed->errnum = QSE_SED_ETSNTR; goto oops; } + + if (c == QSE_T('n')) c = QSE_T('\n'); } b[0] = c; @@ -549,23 +559,53 @@ static int get_trans_set (qse_sed_t* sed, qse_sed_cmd_t* cmd) sed->errnum = QSE_SED_ENOMEM; goto oops; } + + c = NXTSC (sed); } c = NXTSC (sed); - while (c != delim) + for (pos = 1; c != delim; pos += 2) { if (c == QSE_CHAR_EOF || IS_LINTERM(c)) { - /* invalid translation set */ - sed->errnum = QSE_SED_ETSINV; + sed->errnum = QSE_SED_ETSNTR; goto oops; } - if (c == QSE_T('\n')) + if (c == QSE_T('\\')) { + c = NXTSC (sed); + if (c == QSE_CHAR_EOF || IS_LINTERM(c)) + { + sed->errnum = QSE_SED_ETSNTR; + goto oops; + } + + if (c == QSE_T('n')) c = QSE_T('\n'); } + + if (pos >= QSE_STR_LEN(t)) + { + /* source and target not the same length */ + sed->errnum = QSE_SED_ETSNSL; + goto oops; + } + + QSE_STR_CHAR(t,pos) = c; + c = NXTSC (sed); } + if (pos < QSE_STR_LEN(t)) + { + /* source and target not the same length */ + sed->errnum = QSE_SED_ETSNSL; + goto oops; + } + + ADVSCP (sed); + if (terminate_command (sed) == -1) goto oops; + + qse_str_yield (t, &cmd->u.transet, 0); qse_str_close (t); return 0; @@ -609,10 +649,29 @@ qse_printf (QSE_T("command not recognized [%c]\n"), c); * corresponding } is met. */ cmd->type = QSE_SED_CMD_B; cmd->negfl = !cmd->negfl; + + if (sed->grplvl >= QSE_COUNTOF(sed->grpcmd)) + { + /* group nesting too deep */ + sed->errnum = QSE_SED_EGRNTD; + return -1; + } + + sed->grpcmd[sed->grplvl++] = cmd; + ADVSCP (sed); break; case QSE_T('}'): - break; + if (sed->grplvl <= 0) + { + /* group not balanced */ + sed->errnum = QSE_SED_EGRNBA; + return -1; + } + + sed->grpcmd[--sed->grplvl]->u.branch.target = cmd; + ADVSCP (sed); + return 0; case QSE_T('a'): case QSE_T('i'): @@ -640,11 +699,9 @@ qse_printf (QSE_T("command not recognized [%c]\n"), c); c = NXTSC (sed); while (IS_SPACE(c)) c = NXTSC (sed); - if (c != QSE_T('\n')) + if (c != QSE_CHAR_EOF && c != QSE_T('\n')) { - /* TODO: change error code. garbage after - * backslash... */ - sed->errnum = QSE_SED_EBSEXP; + sed->errnum = QSE_SED_EGBABS; return -1; } @@ -699,7 +756,6 @@ printf ("command %c\n", cmd->type); printf ("command %c\n", cmd->type); break; - case QSE_T('b'): case QSE_T('t'): cmd->type = c; @@ -716,13 +772,17 @@ qse_printf (QSE_T("cmd->u.branch.target = [%p]\n"), cmd->u.branch.target); break; case QSE_T('r'): + /* TODO */ break; case QSE_T('R'): + /* TODO */ break; case QSE_T('w'): + /* TODO */ break; case QSE_T('W'): + /* TODO */ break; case QSE_T('q'): @@ -740,11 +800,11 @@ qse_printf (QSE_T("cmd->u.branch.target = [%p]\n"), cmd->u.branch.target); case QSE_T('y'): cmd->type = c; ADVSCP (sed); - if (get_trans_set (sed, cmd) == -1) return -1; + if (get_transet (sed, cmd) == -1) return -1; break; } - return 0; + return 1; } static int compile_source ( @@ -767,6 +827,8 @@ static int compile_source ( */ while (1) { + int n; + c = CURSC (sed); /* skip white spaces and comments*/ @@ -824,16 +886,27 @@ static int compile_source ( cmd->negfl = 1; } - if (command (sed) == -1) return -1; - - if (sed->cmd.cur >= sed->cmd.end) + n = command (sed); + if (n == -1) return -1; + if (n > 0) { - /* TODO: too many commands */ - sed->errnum = QSE_SED_ENOMEM; /* TODO change it. */ - return -1; - } + QSE_ASSERT (n == 1); - cmd = ++sed->cmd.cur; + if (sed->cmd.cur >= sed->cmd.end) + { + /* TODO: too many commands. change errnum */ + sed->errnum = QSE_SED_ENOMEM; + return -1; + } + + cmd = ++sed->cmd.cur; + } + } + + if (sed->grplvl != 0) + { + sed->errnum = QSE_SED_EGRNBA; + return -1; } return 0; diff --git a/qse/lib/utl/sed.h b/qse/lib/utl/sed.h index 07bccc45..88b04a69 100644 --- a/qse/lib/utl/sed.h +++ b/qse/lib/utl/sed.h @@ -51,13 +51,19 @@ struct qse_sed_cmd_t union { + /* text for the a, i, c commands */ qse_str_t* text; - void* rex; + + /* translation set for the y command */ + qse_xstr_t transet; + struct { qse_str_t* text; qse_sed_cmd_t* target; } branch; + + void* rex; } u; qse_char_t* rhs; /* right-hand side of sustitution */