*** empty log message ***

[idzebra-moved-to-github.git] / recctrl / regxread.c

diff --git a/recctrl/regxread.c b/recctrl/regxread.c
index 0662a1a..6fdbd89 100644 (file)
--- a/recctrl/regxread.c
+++ b/recctrl/regxread.c
@@ -1,10 +1,32 @@
 /*
- * Copyright (C) 1994-1999, Index Data
+ * Copyright (C) 1994-2001, Index Data
  * All rights reserved.
- * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: regxread.c,v $
- * Revision 1.30  1999-07-14 10:55:28  adam
+ * Revision 1.37  2001-05-29 08:51:59  adam
+ * More fixes for character encodings.
+ *
+ * Revision 1.36  2001/05/22 21:02:26  adam
+ * Fixes for Tcl UTF8 character handling.
+ *
+ * Revision 1.35  2001/03/29 21:31:31  adam
+ * Fixed "record begin" for Tcl filter.
+ *
+ * Revision 1.34  2000/11/29 14:24:01  adam
+ * Script configure uses yaz pthreads options. Added locking for
+ * zebra_register_{lock,unlock}.
+ *
+ * Revision 1.33  1999/11/30 13:48:04  adam
+ * Improved installation. Updated for inclusion of YAZ header files.
+ *
+ * Revision 1.32  1999/09/07 07:19:21  adam
+ * Work on character mapping. Implemented replace rules.
+ *
+ * Revision 1.31  1999/07/14 13:05:29  adam
+ * Tcl filter works with objects when TCL is version 8 or later; filter
+ * works with strings otherwise (slow).
+ *
+ * Revision 1.30  1999/07/14 10:55:28  adam
  * Fixed memory leak.
  *
  * Revision 1.29  1999/07/12 07:27:54  adam
@@ -185,13 +207,17 @@
 #include <string.h>
 #include <ctype.h>
 
-#include <tpath.h>
+#include <yaz/tpath.h>
 #include <zebrautl.h>
 #include <dfa.h>
 #include "grsread.h"
 
 #if HAVE_TCL_H
 #include <tcl.h>
+
+#if MAJOR_VERSION >= 8
+#define HAVE_TCL_OBJECTS
+#endif
 #endif
 
 #define REGX_DEBUG 0
@@ -210,7 +236,7 @@
 
 struct regxCode {
     char *str;
-#if HAVE_TCL_H
+#if HAVE_TCL_OBJECTS
     Tcl_Obj *tcl_obj;
 #endif
 };
@@ -358,7 +384,7 @@ static void regxCodeDel (struct regxCode **pp)
     struct regxCode *p = *pp;
     if (p)
     {
-#if HAVE_TCL_H
+#if HAVE_TCL_OBJECTS
        if (p->tcl_obj)
            Tcl_DecrRefCount (p->tcl_obj);
 #endif
@@ -376,7 +402,7 @@ static void regxCodeMk (struct regxCode **pp, const char *buf, int len)
     p->str = (char *) xmalloc (len+1);
     memcpy (p->str, buf, len);
     p->str[len] = '\0';
-#if HAVE_TCL_H
+#if HAVE_TCL_OBJECTS
     p->tcl_obj = Tcl_NewStringObj ((char *) buf, len);
     if (p->tcl_obj)
        Tcl_IncrRefCount (p->tcl_obj);
@@ -506,7 +532,7 @@ static void lexSpecDestroy (struct lexSpec **pp)
        lexContextDestroy (lt);
        lt = lt_next;
     }
-#if HAVE_TCL_H
+#if HAVE_TCL_OBJECTS
     if (p->tcl_interp)
        Tcl_DeleteInterp (p->tcl_interp);
 #endif
@@ -692,7 +718,7 @@ int readOneSpec (struct lexSpec *spec, const char *s)
        break;
     case REGX_PATTERN:
 #if REGX_DEBUG
-       logf (LOG_DEBUG, "rule %d %s", spec->context->ruleNo, s);
+       logf (LOG_LOG, "rule %d %s", spec->context->ruleNo, s);
 #endif
         r = dfa_parse (spec->context->dfa, &s);
         if (r)
@@ -824,12 +850,12 @@ static void execData (struct lexSpec *spec,
        return ;
 #if REGX_DEBUG
     if (elen > 40)
-        logf (LOG_DEBUG, "data (%d bytes) %.15s ... %.*s", elen,
+        logf (LOG_LOG, "data (%d bytes) %.15s ... %.*s", elen,
              ebuf, 15, ebuf + elen-15);
     else if (elen > 0)
-        logf (LOG_DEBUG, "data (%d bytes) %.*s", elen, elen, ebuf);
+        logf (LOG_LOG, "data (%d bytes) %.*s", elen, elen, ebuf);
     else 
-        logf (LOG_DEBUG, "data (%d bytes)", elen);
+        logf (LOG_LOG, "data (%d bytes)", elen);
 #endif
         
     if (spec->d1_level <= 1)
@@ -937,7 +963,7 @@ static void variantBegin (struct lexSpec *spec,
     ttype[type_len] = '\0';
 
 #if REGX_DEBUG 
-    logf (LOG_DEBUG, "variant begin %s %s (%d)", tclass, ttype,
+    logf (LOG_LOG, "variant begin %s %s (%d)", tclass, ttype,
          spec->d1_level);
 #endif
 
@@ -974,7 +1000,7 @@ static void variantBegin (struct lexSpec *spec,
        }
 
 #if REGX_DEBUG 
-    logf (LOG_DEBUG, "variant node (%d)", spec->d1_level);
+    logf (LOG_LOG, "variant node (%d)", spec->d1_level);
 #endif
     parent = spec->d1_stack[spec->d1_level-1];
     res = data1_mk_node (spec->dh, spec->m);
@@ -1018,9 +1044,9 @@ static void tagStrip (const char **tag, int *len)
 static void tagBegin (struct lexSpec *spec, 
                       const char *tag, int len)
 {
-    struct data1_node *parent = spec->d1_stack[spec->d1_level -1];
+    struct data1_node *parent;
     data1_element *elem = NULL;
-    data1_node *partag = get_parent_tag(spec->dh, parent);
+    data1_node *partag;
     data1_node *res;
     data1_element *e = NULL;
     int localtag = 0;
@@ -1031,11 +1057,12 @@ static void tagBegin (struct lexSpec *spec,
         return ;
     }
     tagStrip (&tag, &len);
+
+    parent = spec->d1_stack[spec->d1_level -1];
+    partag = get_parent_tag(spec->dh, parent);
    
-    res = data1_mk_node (spec->dh, spec->m);
+    res = data1_mk_node_type (spec->dh, spec->m, DATA1N_tag);
     res->parent = parent;
-    res->which = DATA1N_tag;
-    res->u.tag.get_bytes = -1;
 
     if (len >= DATA1_LOCALDATA)
        res->u.tag.tag = (char *) nmem_malloc (spec->m, len+1);
@@ -1046,7 +1073,7 @@ static void tagBegin (struct lexSpec *spec,
     res->u.tag.tag[len] = '\0';
    
 #if REGX_DEBUG 
-    logf (LOG_DEBUG, "begin tag %s (%d)", res->u.tag.tag, spec->d1_level);
+    logf (LOG_LOG, "begin tag %s (%d)", res->u.tag.tag, spec->d1_level);
 #endif
     if (parent->which == DATA1N_variant)
         return ;
@@ -1058,9 +1085,6 @@ static void tagBegin (struct lexSpec *spec,
                                      spec->d1_stack[0]->u.root.absyn,
                                      e, res->u.tag.tag);
     res->u.tag.element = elem;
-    res->u.tag.node_selected = 0;
-    res->u.tag.make_variantlist = 0;
-    res->u.tag.no_data_requested = 0;
     res->root = parent->root;
 
     parent->last_child = res;
@@ -1093,7 +1117,7 @@ static void tagEnd (struct lexSpec *spec, int min_level,
             break;
     }
 #if REGX_DEBUG
-    logf (LOG_DEBUG, "end tag (%d)", spec->d1_level);
+    logf (LOG_LOG, "end tag (%d)", spec->d1_level);
 #endif
 }
 
@@ -1251,7 +1275,7 @@ static int cmd_tcl_begin (ClientData clientData, Tcl_Interp *interp,
        data1_absyn *absyn;
 
 #if REGX_DEBUG
-       logf (LOG_DEBUG, "begin record %s", absynName);
+       logf (LOG_LOG, "begin record %s", absynName);
 #endif
        if (!(absyn = data1_get_absyn (spec->dh, absynName)))
            logf (LOG_WARN, "Unknown tagset: %s", absynName);
@@ -1261,7 +1285,8 @@ static int cmd_tcl_begin (ClientData clientData, Tcl_Interp *interp,
            
            res = data1_mk_node (spec->dh, spec->m);
            res->which = DATA1N_root;
-           res->u.root.type = absynName;
+           res->u.root.type =
+                data1_insert_string(spec->dh, res, spec->m, absynName);
            res->u.root.absyn = absyn;
            res->root = res;
            
@@ -1283,7 +1308,7 @@ static int cmd_tcl_begin (ClientData clientData, Tcl_Interp *interp,
     {
        struct lexContext *lc = spec->context;
 #if REGX_DEBUG
-       logf (LOG_DEBUG, "begin context %s",argv[2]);
+       logf (LOG_LOG, "begin context %s",argv[2]);
 #endif
        while (lc && strcmp (argv[2], lc->name))
            lc = lc->next;
@@ -1314,7 +1339,7 @@ static int cmd_tcl_end (ClientData clientData, Tcl_Interp *interp,
            (spec->d1_level)--;
        }
 #if REGX_DEBUG
-       logf (LOG_DEBUG, "end record");
+       logf (LOG_LOG, "end record");
 #endif
        spec->stop_flag = 1;
     }
@@ -1335,7 +1360,7 @@ static int cmd_tcl_end (ClientData clientData, Tcl_Interp *interp,
        if (spec->d1_level == 0)
        {
 #if REGX_DEBUG
-           logf (LOG_DEBUG, "end element end records");
+           logf (LOG_LOG, "end element end records");
 #endif
            spec->stop_flag = 1;
        }
@@ -1343,7 +1368,7 @@ static int cmd_tcl_end (ClientData clientData, Tcl_Interp *interp,
     else if (!strcmp (argv[1], "context"))
     {
 #if REGX_DEBUG
-       logf (LOG_DEBUG, "end context");
+       logf (LOG_LOG, "end context");
 #endif
        if (spec->context_stack_top)
            (spec->context_stack_top)--;
@@ -1382,7 +1407,14 @@ static int cmd_tcl_data (ClientData clientData, Tcl_Interp *interp,
 
     while (argi < argc)
     {
+#if TCL_MAJOR_VERSION > 8 || (TCL_MAJOR_VERSION == 8 && TCL_MINOR_VERSION > 0)
+       Tcl_DString ds;
+       char *native = Tcl_UtfToExternalDString(0, argv[argi], -1, &ds);
+       execData (spec, native, strlen(native), textFlag);
+       Tcl_DStringFree (&ds);
+#else
        execData (spec, argv[argi], strlen(argv[argi]), textFlag);
+#endif
        argi++;
     }
     if (element)
@@ -1441,10 +1473,11 @@ static void execTcl (struct lexSpec *spec, struct regxCode *code)
            var_buf[var_len] = ch;
        }
     }
-    if (code->tcl_obj)
-       ret = Tcl_GlobalEvalObj(spec->tcl_interp, code->tcl_obj);
-    else
-       ret = Tcl_GlobalEval (spec->tcl_interp, code->str);
+#if HAVE_TCL_OBJECTS
+    ret = Tcl_GlobalEvalObj(spec->tcl_interp, code->tcl_obj);
+#else
+    ret = Tcl_GlobalEval (spec->tcl_interp, code->str);
+#endif
     if (ret != TCL_OK)
     {
        const char *err = Tcl_GetVar(spec->tcl_interp, "errorInfo", 0);
@@ -1499,7 +1532,7 @@ static void execCode (struct lexSpec *spec, struct regxCode *code)
                     absynName[cmd_len] = '\0';
 
 #if REGX_DEBUG
-                    logf (LOG_DEBUG, "begin record %s", absynName);
+                    logf (LOG_LOG, "begin record %s", absynName);
 #endif
                     if (!(absyn = data1_get_absyn (spec->dh, absynName)))
                         logf (LOG_WARN, "Unknown tagset: %s", absynName);
@@ -1566,7 +1599,7 @@ static void execCode (struct lexSpec *spec, struct regxCode *code)
                    r = execTok (spec, &s, &cmd_str, &cmd_len);
                    p = regxStrz (cmd_str, cmd_len, ptmp);
 #if REGX_DEBUG
-                   logf (LOG_DEBUG, "begin context %s", p);
+                   logf (LOG_LOG, "begin context %s", p);
 #endif
                    while (lc && strcmp (p, lc->name))
                        lc = lc->next;
@@ -1601,7 +1634,7 @@ static void execCode (struct lexSpec *spec, struct regxCode *code)
                }
                r = execTok (spec, &s, &cmd_str, &cmd_len);
 #if REGX_DEBUG
-               logf (LOG_DEBUG, "end record");
+               logf (LOG_LOG, "end record");
 #endif
                spec->stop_flag = 1;
            }
@@ -1623,7 +1656,7 @@ static void execCode (struct lexSpec *spec, struct regxCode *code)
                 if (spec->d1_level == 0)
                 {
 #if REGX_DEBUG
-                   logf (LOG_DEBUG, "end element end records");
+                   logf (LOG_LOG, "end element end records");
 #endif
                    spec->stop_flag = 1;
                 }
@@ -1632,7 +1665,7 @@ static void execCode (struct lexSpec *spec, struct regxCode *code)
            else if (!strcmp (p, "context"))
            {
 #if REGX_DEBUG
-               logf (LOG_DEBUG, "end context");
+               logf (LOG_LOG, "end context");
 #endif
                if (spec->context_stack_top)
                    (spec->context_stack_top)--;
@@ -1828,7 +1861,7 @@ static int execRule (struct lexSpec *spec, struct lexContext *context,
                      int ruleNo, int start_ptr, int *pptr)
 {
 #if REGX_DEBUG
-    logf (LOG_DEBUG, "exec rule %d", ruleNo);
+    logf (LOG_LOG, "exec rule %d", ruleNo);
 #endif
     return execAction (spec, context->fastRule[ruleNo]->actionList,
                        start_ptr, pptr);
@@ -1906,7 +1939,7 @@ data1_node *lexNode (struct lexSpec *spec, int *ptr)
                         if (spec->f_win_ef && *ptr != F_WIN_EOF)
                        {
 #if REGX_DEBUG
-                           logf (LOG_DEBUG, "regx: endf ptr=%d", *ptr);
+                           logf (LOG_LOG, "regx: endf ptr=%d", *ptr);
 #endif
                             (*spec->f_win_ef)(spec->f_win_fh, *ptr);
                        }
@@ -2018,7 +2051,7 @@ data1_node *grs_read_regx (struct grs_read_info *p)
     struct lexSpec **curLexSpec = &specs->spec;
 
 #if REGX_DEBUG
-    logf (LOG_DEBUG, "grs_read_regx");
+    logf (LOG_LOG, "grs_read_regx");
 #endif
     if (!*curLexSpec || strcmp ((*curLexSpec)->name, p->type))
     {
@@ -2064,7 +2097,7 @@ data1_node *grs_read_tcl (struct grs_read_info *p)
     struct lexSpec **curLexSpec = &specs->spec;
 
 #if REGX_DEBUG
-    logf (LOG_DEBUG, "grs_read_tcl");
+    logf (LOG_LOG, "grs_read_tcl");
 #endif
     if (!*curLexSpec || strcmp ((*curLexSpec)->name, p->type))
     {
@@ -2072,7 +2105,9 @@ data1_node *grs_read_tcl (struct grs_read_info *p)
         if (*curLexSpec)
             lexSpecDestroy (curLexSpec);
         *curLexSpec = lexSpecCreate (p->type, p->dh);
+       Tcl_FindExecutable("");
        tcl_interp = (*curLexSpec)->tcl_interp = Tcl_CreateInterp();
+       Tcl_Init(tcl_interp);
        Tcl_CreateCommand (tcl_interp, "begin", cmd_tcl_begin, *curLexSpec, 0);
        Tcl_CreateCommand (tcl_interp, "end", cmd_tcl_end, *curLexSpec, 0);
        Tcl_CreateCommand (tcl_interp, "data", cmd_tcl_data, *curLexSpec, 0);