Merge branch 'issue173' #173

source/texk/mendexk/fread.c

@@ -93,7 +93,9 @@ int idxread(char *filename, int start)
 				continue;
 			}
 
-			if (quo==0 && buff[j]==escape) {
+			if (esc==1 && buff[j]==escape) {
+				esc=0;
+			} else if (quo==0 && buff[j]==escape) {
 				esc=1;
 			}
 
@@ -150,7 +152,7 @@ int idxread(char *filename, int start)
 						if (buff[j]==encap) {
 							j++;
 							cc=getestr(&buff[j],estr);
-							if (cc<0 || strchr(estr,encap)) {
+							if (cc<0) {
 								fprintf(efp,"\nError: Bad encap string in %s, line %d.",filename,ind[i].lnum);
 								if (efp!=stderr) fprintf(stderr,"\nError: Bad encap string in %s, line %d.",filename,ind[i].lnum);
 								eflg++;
@@ -417,28 +419,37 @@ int idxread(char *filename, int start)
 /*   pic up encap string   */
 static int getestr(char *buff, char *estr)
 {
-	int i,nest=0;
-
-	for (i=0;i<strlen(buff);i++) {
-		if (buff[i]==encap) {
-			if (i>0) {
-				if ((unsigned char)buff[i-1]<0x80) {
-					estr[i]=buff[i];
-					i++;
-				}
+	int i,j,nest=0,esc=0,quo=0;
+
+	for (i=0,j=0;i<strlen(buff);i++,j++) {
+		/* If a "quote" character is found, it is removed and the
+		   following character is not treated as a special character.
+		   If a "quote" character follows an odd number of
+		   consecutive "escape" characters, it is not treated as a
+		   special character and is left as is.
+		   Note that the "escape" characters are not removed.  */
+		esc=0; quo=0;
+		if (buff[i]==escape) {
+			estr[j]=buff[i];
+			i++; j++;
+			esc=1;
+		} else if (buff[i]==quote) {
+			i++;
+			quo=1;
+		}
+		if (quo==0) {
+			if (nest==0 && esc==0 && buff[i]==arg_close) {
+				estr[j]='\0';
+				return i;
 			}
-			else {
-				estr[i]=buff[i];
-				i++;
+			if (esc==0 && buff[i]==arg_open) nest++;
+			else if (esc==0 && buff[i]==arg_close) nest--;
+			else if (buff[i]==level || buff[i]==actual || buff[i]==encap) {
+				fprintf(efp, "\nError: Extra `%c\' at position %d in encap string.",buff[i],i);
+				return -1;
 			}
 		}
-		if (nest==0 && buff[i]==arg_close) {
-			estr[i]='\0';
-			return i;
-		}
-		if (buff[i]==arg_open) nest++;
-		else if (buff[i]==arg_close) nest--;
-		copy_multibyte_char(buff, estr, &i, NULL);
+		copy_multibyte_char(buff, estr, &i, &j);
 	}
 
 	return -1;

source/texk/mendexk/styfile.c

@@ -197,6 +197,12 @@ static void convline(char *buff1, int start, char *buff2)
 				else if (buff1[i]=='t') buff2[j]='\t';
 				else if (buff1[i]=='r') buff2[j]='\r';
 				else if (buff1[i]=='\"') buff2[j]='\"';
+				else {
+					/* Otherwise, the '\' is simply ignored here and the
+					   following character is copied as is in the next loop. */
+					i--;
+					continue;
+				}
 			}
 			else buff2[j]=buff1[i];
 			j++;

source/texk/mendexk/tests/head1.ist

@@ -0,0 +1,3 @@
+headings_flag 1
+heading_prefix "aaa\tbbb\|ccc\\ddd\n\[\{\(\"\'"
+heading_suffix "\'\"\)\}\]\naaa\tbbb\|ccc\\ddd"

source/texk/mendexk/tests/mendex.test

@@ -42,5 +42,11 @@ $_mendex $srcdir/tests/tort.idx -o tort.ind1 -t tort.ilg1 \
 $_mendex $srcdir/tests/tortW.idx -o tortW.ind1 -t tortW.ilg1 \
 	&& diff $srcdir/tests/ok-tort.ind tortW.ind1 || :
 
+$_mendex $srcdir/tests/tort.idx -s $srcdir/tests/head1.ist -o tort2.ind1 -t tort2.ilg1 \
+	&& diff $srcdir/tests/ok-tort2.ind tort2.ind1 || :
+
+$_mendex $srcdir/tests/tortW.idx -s $srcdir/tests/head1.ist -o tortW2.ind1 -t tortW2.ilg1 \
+	&& diff $srcdir/tests/ok-tort2.ind tortW2.ind1 || :
+
 
 exit $rc

source/texk/mendexk/tests/ok-tort.ind

@@ -4,6 +4,11 @@
 
   \indexspace
 
+  \item encap-escape1 aaa@bbb|ccc!ddd"eee{fff}, \ppp{11}, 10--12
+  \item encap-escape2, \aaa@bbb|ccc!ddd"eee{fff}{20--22}
+
+  \indexspace
+
   \item implicit-range, 10--12
   \item incons-encaps, \ii{6}, 5--7
   \item incons-encaps2, \ii{5}, 5

source/texk/mendexk/tests/ok-tort2.ind

@@ -0,0 +1,39 @@
+\begin{theindex}
+aaa	bbb|ccc\ddd
+[{("'B'")}]
+aaa	bbb|ccc\ddd
+  \item bad-encaps, \ii{5--7}
+
+  \indexspace
+aaa	bbb|ccc\ddd
+[{("'E'")}]
+aaa	bbb|ccc\ddd
+  \item encap-escape1 aaa@bbb|ccc!ddd"eee{fff}, \ppp{11}, 10--12
+  \item encap-escape2, \aaa@bbb|ccc!ddd"eee{fff}{20--22}
+
+  \indexspace
+aaa	bbb|ccc\ddd
+[{("'I'")}]
+aaa	bbb|ccc\ddd
+  \item implicit-range, 10--12
+  \item incons-encaps, \ii{6}, 5--7
+  \item incons-encaps2, \ii{5}, 5
+  \item incons-entries, 6, \xx{6}
+
+  \indexspace
+aaa	bbb|ccc\ddd
+[{("'M'")}]
+aaa	bbb|ccc\ddd
+  \item missing(, 6
+  \item missing), 6
+  \item mixed-range, i, 6
+  \item mixed-range1, i--iv, 3--6
+  \item mixed-range2, \xx{ii}, i--ii, \yy{3--6}
+
+  \indexspace
+aaa	bbb|ccc\ddd
+[{("'O'")}]
+aaa	bbb|ccc\ddd
+  \item ok-encaps, \ii{5--7}
+
+\end{theindex}

source/texk/mendexk/tests/tort.idx

@@ -51,3 +51,11 @@
 \indexentry{mixed-range2|xx}{ii}
 \indexentry{mixed-range2|yy}{3}
 \indexentry{mixed-range2|)}{6}
+
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|(}{10}
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|ppp}{11}
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|)}{12}
+
+\indexentry{encap-escape2|(aaa"@bbb"|ccc"!ddd""eee"{fff"}}{20}
+\indexentry{encap-escape2|aaa"@bbb"|ccc"!ddd""eee"{fff"}}{21}
+\indexentry{encap-escape2|)}{22}

source/texk/mendexk/tests/tortW.idx

@@ -51,3 +51,11 @@
 \indexentry{mixed-range2|xx}{ii}
 \indexentry{mixed-range2|yy}{3}
 \indexentry{mixed-range2|)}{6}
+
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|(}{10}
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|ppp}{11}
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|)}{12}
+
+\indexentry{encap-escape2|(aaa"@bbb"|ccc"!ddd""eee"{fff"}}{20}
+\indexentry{encap-escape2|aaa"@bbb"|ccc"!ddd""eee"{fff"}}{21}
+\indexentry{encap-escape2|)}{22}

source/texk/upmendex/fread.c

@@ -93,7 +93,9 @@ int idxread(char *filename, int start)
 				continue;
 			}
 
-			if (quo==0 && buff[j]==escape) {
+			if (esc==1 && buff[j]==escape) {
+				esc=0;
+			} else if (quo==0 && buff[j]==escape) {
 				esc=1;
 			}
 
@@ -150,7 +152,7 @@ int idxread(char *filename, int start)
 						if (buff[j]==encap) {
 							j++;
 							cc=getestr(&buff[j],estr);
-							if (cc<0 || strchr(estr,encap)) {
+							if (cc<0) {
 								fprintf(efp,"\nError: Bad encap string in %s, line %d.",filename,ind[i].lnum);
 								if (efp!=stderr) fprintf(stderr,"\nError: Bad encap string in %s, line %d.",filename,ind[i].lnum);
 								eflg++;
@@ -422,28 +424,37 @@ int idxread(char *filename, int start)
 /*   pic up encap string   */
 static int getestr(char *buff, char *estr)
 {
-	int i,nest=0;
-
-	for (i=0;i<strlen(buff);i++) {
-		if (buff[i]==encap) {
-			if (i>0) {
-				if ((unsigned char)buff[i-1]<0x80) {
-					estr[i]=buff[i];
-					i++;
-				}
+	int i,j,nest=0,esc=0,quo=0;
+
+	for (i=0,j=0;i<strlen(buff);i++,j++) {
+		/* If a "quote" character is found, it is removed and the
+		   following character is not treated as a special character.
+		   If a "quote" character follows an odd number of
+		   consecutive "escape" characters, it is not treated as a
+		   special character and is left as is.
+		   Note that the "escape" characters are not removed.  */
+		esc=0; quo=0;
+		if (buff[i]==escape) {
+			estr[j]=buff[i];
+			i++; j++;
+			esc=1;
+		} else if (buff[i]==quote) {
+			i++;
+			quo=1;
+		}
+		if (quo==0) {
+			if (nest==0 && esc==0 && buff[i]==arg_close) {
+				estr[j]='\0';
+				return i;
 			}
-			else {
-				estr[i]=buff[i];
-				i++;
+			if (esc==0 && buff[i]==arg_open) nest++;
+			else if (esc==0 && buff[i]==arg_close) nest--;
+			else if (buff[i]==level || buff[i]==actual || buff[i]==encap) {
+				fprintf(efp, "\nError: Extra `%c\' at position %d in encap string.",buff[i],i);
+				return -1;
 			}
 		}
-		if (nest==0 && buff[i]==arg_close) {
-			estr[i]='\0';
-			return i;
-		}
-		if (buff[i]==arg_open) nest++;
-		else if (buff[i]==arg_close) nest--;
-		copy_multibyte_char(buff, estr, &i, NULL);
+		copy_multibyte_char(buff, estr, &i, &j);
 	}
 
 	return -1;

source/texk/upmendex/styfile.c

@@ -279,6 +279,12 @@ static void convline(char *buff1, int start, char *buff2)
 				else if (buff1[i]=='t') buff2[j]='\t';
 				else if (buff1[i]=='r') buff2[j]='\r';
 				else if (buff1[i]=='\"') buff2[j]='\"';
+				else {
+					/* Otherwise, the '\' is simply ignored here and the
+					   following character is copied as is in the next loop. */
+					i--;
+					continue;
+				}
 			}
 			else buff2[j]=buff1[i];
 			j++;

source/texk/upmendex/tests/head1.ist

@@ -0,0 +1,3 @@
+headings_flag 1
+heading_prefix "aaa\tbbb\|ccc\\ddd\n\[\{\(\"\'"
+heading_suffix "\'\"\)\}\]\naaa\tbbb\|ccc\\ddd"

source/texk/upmendex/tests/ok-tort.ind

@@ -4,6 +4,11 @@
 
   \indexspace
 
+  \item encap-escape1 aaa@bbb|ccc!ddd"eee{fff}, \ppp{11}, 10--12
+  \item encap-escape2, \aaa@bbb|ccc!ddd"eee{fff}{20--22}
+
+  \indexspace
+
   \item implicit-range, 10--12
   \item incons-encaps, \ii{6}, 5--7
   \item incons-encaps2, \ii{5}, 5

source/texk/upmendex/tests/ok-tort2.ind

@@ -0,0 +1,39 @@
+\begin{theindex}
+aaa	bbb|ccc\ddd
+[{("'B'")}]
+aaa	bbb|ccc\ddd
+  \item bad-encaps, \ii{5--7}
+
+  \indexspace
+aaa	bbb|ccc\ddd
+[{("'E'")}]
+aaa	bbb|ccc\ddd
+  \item encap-escape1 aaa@bbb|ccc!ddd"eee{fff}, \ppp{11}, 10--12
+  \item encap-escape2, \aaa@bbb|ccc!ddd"eee{fff}{20--22}
+
+  \indexspace
+aaa	bbb|ccc\ddd
+[{("'I'")}]
+aaa	bbb|ccc\ddd
+  \item implicit-range, 10--12
+  \item incons-encaps, \ii{6}, 5--7
+  \item incons-encaps2, \ii{5}, 5
+  \item incons-entries, 6, \xx{6}
+
+  \indexspace
+aaa	bbb|ccc\ddd
+[{("'M'")}]
+aaa	bbb|ccc\ddd
+  \item missing(, 6
+  \item missing), 6
+  \item mixed-range, i, 6
+  \item mixed-range1, i--iv, 3--6
+  \item mixed-range2, \xx{ii}, i--ii, \yy{3--6}
+
+  \indexspace
+aaa	bbb|ccc\ddd
+[{("'O'")}]
+aaa	bbb|ccc\ddd
+  \item ok-encaps, \ii{5--7}
+
+\end{theindex}

source/texk/upmendex/tests/tort.idx

@@ -51,3 +51,11 @@
 \indexentry{mixed-range2|xx}{ii}
 \indexentry{mixed-range2|yy}{3}
 \indexentry{mixed-range2|)}{6}
+
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|(}{10}
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|ppp}{11}
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|)}{12}
+
+\indexentry{encap-escape2|(aaa"@bbb"|ccc"!ddd""eee"{fff"}}{20}
+\indexentry{encap-escape2|aaa"@bbb"|ccc"!ddd""eee"{fff"}}{21}
+\indexentry{encap-escape2|)}{22}

source/texk/upmendex/tests/tortW.idx

@@ -51,3 +51,11 @@
 \indexentry{mixed-range2|xx}{ii}
 \indexentry{mixed-range2|yy}{3}
 \indexentry{mixed-range2|)}{6}
+
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|(}{10}
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|ppp}{11}
+\indexentry{encap-escape1 aaa"@bbb"|ccc"!ddd""eee"{fff"}|)}{12}
+
+\indexentry{encap-escape2|(aaa"@bbb"|ccc"!ddd""eee"{fff"}}{20}
+\indexentry{encap-escape2|aaa"@bbb"|ccc"!ddd""eee"{fff"}}{21}
+\indexentry{encap-escape2|)}{22}

source/texk/upmendex/tests/upmendex.test

@@ -42,5 +42,11 @@ $_upmendex $srcdir/tests/tort.idx -o tort.ind1 -t tort.ilg1 \
 $_upmendex $srcdir/tests/tortW.idx -o tortW.ind1 -t tortW.ilg1 \
 	&& diff $srcdir/tests/ok-tort.ind tortW.ind1 || :
 
+$_upmendex $srcdir/tests/tort.idx -s $srcdir/tests/head1.ist -o tort2.ind1 -t tort2.ilg1 \
+	&& diff $srcdir/tests/ok-tort2.ind tort2.ind1 || :
+
+$_upmendex $srcdir/tests/tortW.idx -s $srcdir/tests/head1.ist -o tortW2.ind1 -t tortW2.ilg1 \
+	&& diff $srcdir/tests/ok-tort2.ind tortW2.ind1 || :
+
 
 exit $rc