From: "Igor B. Poretsky" <poretsky@mlbox.ru>
Date: Wed, 15 Feb 2023 13:44:43 +0300
Subject: C library regexp support

---
 src/Makefile      |  8 ++++----
 src/phon_rules.c  | 56 +++++++++++++++++++++++++++++++++++++++++++++++--------
 src/rule_engine.c | 32 +++++++++++++++----------------
 src/rule_engine.h |  4 ++--
 src/space.c       |  8 +-------
 src/t2s.h         |  8 ++++----
 6 files changed, 75 insertions(+), 41 deletions(-)

diff --git a/src/Makefile b/src/Makefile
index 704b098..393de2e 100644
--- a/src/Makefile
+++ b/src/Makefile
@@ -45,9 +45,9 @@ OBJS = go.o t2s.o space.o tags.o grammar.o \
 	phon_rules.o edin2sampa.o
 #coeffs.o excitation.o durations.o pitch.o audio.o ulaw.o \
 
-RULE_SRC = rule_engine.c regerror.c regexp.c regsub.c read_rules.c
+RULE_SRC = rule_engine.c read_rules.c
 
-RULE_OBJS = rule_engine.o regerror.o regexp.o regsub.o read_rules.o
+RULE_OBJS = rule_engine.o read_rules.o
 
 
 NRL_SRC = interface.c english.c nrl_edin.c phoneme.c saynum.c spellwor.c
@@ -63,7 +63,7 @@ MAIN_SRC = main.c
 
 MAIN = main.o
 
-HEADERS = regexp.h regmagic.h rule_engine.h t2s.h prototypes.h
+HEADERS = rule_engine.h t2s.h prototypes.h
 
 all: freephone
 
@@ -96,5 +96,5 @@ distribution:
 $(OBJS) : t2s.h
 $(TEST_OBJS) : t2s.h
 $(MAIN) : t2s.h
-$(RULE_OBJS) : regexp.h regmagic.h
+$(RULE_OBJS) :
 rule-engine.c : rule_engine.h
diff --git a/src/phon_rules.c b/src/phon_rules.c
index 9f88d95..dab7f67 100644
--- a/src/phon_rules.c
+++ b/src/phon_rules.c
@@ -62,24 +62,64 @@ export void load_context_rules(char * filename)
     */
 }
 
-/* replace string in situ  */
-export void phon_rules_init()
+export void phon_rules_free()
 {
   int i;
 
   for(i=0;i<nrules;i++) {
-    alloetc[i].lc = regcomp(alloetc[i].left_context);
-    alloetc[i].rc = regcomp(alloetc[i].right_context);
+    if (alloetc[i].lc) {
+      regfree(alloetc[i].lc);
+      free (alloetc[i].lc);
+      alloetc[i].lc = NULL;
+    }
+    if (alloetc[i].rc) {
+      regfree(alloetc[i].rc);
+      free(alloetc[i].rc);
+      alloetc[i].rc = NULL;
+    }
   }
 }
 
-export void phon_rules_free()
+/* replace string in situ  */
+export void phon_rules_init()
 {
-  int i;
+  int i, err, msgsize;
+  char *msg;
 
   for(i=0;i<nrules;i++) {
-    free(alloetc[i].lc);
-    free(alloetc[i].rc);
+    alloetc[i].lc = malloc(sizeof(regex_t));
+    if (!alloetc[i].lc) {
+      fprintf(stderr, "Memory allocation error\n");
+      phon_rules_free();
+      exit(EXIT_FAILURE);
+    }
+    err = regcomp(alloetc[i].lc, alloetc[i].left_context,
+		  REG_EXTENDED | REG_NEWLINE);
+    if (err) {
+      msgsize = regerror(err, alloetc[i].lc, NULL, 0);
+      msg = malloc(msgsize);
+      (void)regerror(err, alloetc[i].lc, msg, msgsize);
+      (void)fprintf(stderr, "regcomp: %s\n", msg);
+      phon_rules_free();
+      free(msg);
+      exit(EXIT_FAILURE);
+    }
+    alloetc[i].rc = malloc(sizeof(regex_t));
+    if (!alloetc[i].rc) {
+      fprintf(stderr, "Memory allocation error\n");
+      exit(EXIT_FAILURE);
+    }
+    err = regcomp(alloetc[i].rc, alloetc[i].right_context,
+		  REG_EXTENDED | REG_NEWLINE);
+    if (err) {
+      msgsize = regerror(err, alloetc[i].rc, NULL, 0);
+      msg = malloc(msgsize);
+      (void)regerror(err, alloetc[i].rc, msg, msgsize);
+      (void)fprintf(stderr, "regcomp: %s\n", msg);
+      phon_rules_free();
+      free(msg);
+      exit(EXIT_FAILURE);
+    }
   }
 }
 
diff --git a/src/rule_engine.c b/src/rule_engine.c
index 6bdc45b..939685e 100644
--- a/src/rule_engine.c
+++ b/src/rule_engine.c
@@ -22,7 +22,7 @@
 export void rule_exec(int type,int nrules, RULE *rule, SENT *sent)
 {
   int i;
-  int lmatch = 0, rmatch = 0;
+  int lmatch = REG_NOMATCH, rmatch = REG_NOMATCH;
   char *input = buffer_text(&(sent->list));
   BUFFER ib;
   BUFFER prev;
@@ -31,13 +31,13 @@ export void rule_exec(int type,int nrules, RULE *rule, SENT *sent)
   buffer_init(&prev);
   while (*input) {
     for(i=0;i<nrules;i++) {
-      lmatch = 0;
-      rmatch = 0;
+      lmatch = REG_NOMATCH;
+      rmatch = REG_NOMATCH;
       if(!strncmp(input,rule[i].target,strlen(rule[i].target))) {
 	/* candidate for rule match  */
-	lmatch = regexec(rule[i].lc,buffer_text(&prev));
-	rmatch = regexec(rule[i].rc,input+strlen(rule[i].target));
-	if(lmatch && rmatch) {
+	lmatch = regexec(rule[i].lc,buffer_text(&prev),0,NULL,0);
+	rmatch = regexec(rule[i].rc,input+strlen(rule[i].target),0,NULL,0);
+	if(!(lmatch || rmatch)) {
 	  buffer_add_str(&ib, rule[i].output);
 	  input += strlen(rule[i].target);
 	  buffer_add_str(&prev, rule[i].target);
@@ -45,16 +45,16 @@ export void rule_exec(int type,int nrules, RULE *rule, SENT *sent)
 	}
       }
     }
-    if(lmatch && rmatch) {
-      ;
-    } else if (type|SAME) {
-      buffer_add_char(&ib, *input);
-      buffer_add_char(&prev, *input);
-      input++;
-    } else {
-      buffer_free(&prev);
-      buffer_free(&ib);
-      return;
+    if(lmatch || rmatch) {
+      if (type|SAME) {
+	buffer_add_char(&ib, *input);
+	buffer_add_char(&prev, *input);
+	input++;
+      } else {
+	buffer_free(&prev);
+	buffer_free(&ib);
+	return;
+      }
     }
   }
   buffer_clear(&(sent->list));
diff --git a/src/rule_engine.h b/src/rule_engine.h
index 805f610..54cce8f 100644
--- a/src/rule_engine.h
+++ b/src/rule_engine.h
@@ -1,10 +1,10 @@
 
 typedef struct {
 	char *left_context;
-	regexp *lc;
+	regex_t *lc;
 	char *target;
 	char *right_context;
-	regexp *rc;
+	regex_t *rc;
 	char *output;
 } RULE;
 
diff --git a/src/space.c b/src/space.c
index 70bfbfb..848d0a8 100644
--- a/src/space.c
+++ b/src/space.c
@@ -112,13 +112,7 @@ export void init(CONFIG *config, BUFFER *buffer, LING_LIST *ling_list, SENT *sen
 
   /*	what follows is an example for use as a template  */
   load_context_rules("context_rules");
-  /* this goes with the rule engine code...
-     for(i=0;i<nrules;i++) {
-     rule[i].lc = regcomp(rule[i].left_context);
-     rule[i].rc = regcomp(rule[i].right_context);
-     }
-     */
- 
+
   phon_rules_init();
 
 
diff --git a/src/t2s.h b/src/t2s.h
index ffe4c5b..70f570a 100644
--- a/src/t2s.h
+++ b/src/t2s.h
@@ -3,8 +3,8 @@
 #include <string.h>
 #include <stdlib.h>
 #include <ctype.h>
-#include "regexp.h"
-#include "regmagic.h"
+#include <sys/types.h>
+#include <regex.h>
 
 #define export
 
@@ -172,10 +172,10 @@ typedef struct {
 
 typedef struct {
   char *left_context;
-  regexp *lc;
+  regex_t *lc;
   char *target;
   char *right_context;
-  regexp *rc;
+  regex_t *rc;
   char *output;
 } RULE;