upmendex: modify heuristic detection of page number types

git-svn-id: svn://tug.org/texlive/trunk@60549 c570f23f-e606-0410-a88d-b1316a301751

13 files changed, 94 insertions, 24 deletions

diff --git a/Build/source/texk/upmendex/ChangeLog b/Build/source/texk/upmendex/ChangeLog
index 045305ef8c9..65c5ba4519a 100644
--- a/Build/source/texk/upmendex/ChangeLog
+++ b/Build/source/texk/upmendex/ChangeLog
@@ -1,3 +1,11 @@
+2021-09-19  TANAKA Takuji  <ttk@t-lab.opal.ne.jp>
+
+	* fread.c:
+	Modify heuristic detection of page number types.
+	* tests/upmendex.test, tests/range.idx, tests/range[123].ind,
+	tests/pprecB.idx, tests/pprec[34].ist, tests/pprecB-[034].ind:
+	Update tests.
+
 2021-09-18  TANAKA Takuji  <ttk@t-lab.opal.ne.jp>
 
 	* fread.c, fwrite.c, sort.c, mendex.h, {,ex}var.h:
diff --git a/Build/source/texk/upmendex/fread.c b/Build/source/texk/upmendex/fread.c
index ba1d3514856..66d3e0b1449 100644
--- a/Build/source/texk/upmendex/fread.c
+++ b/Build/source/texk/upmendex/fread.c
@@ -344,7 +344,7 @@ LOOP:
 					copy_multibyte_char(buff, table, &j, &k);
 				}
 
-				table[k]='\0';	
+				table[k]='\0';
 
 				for (k=0;k<=ind[l].num;k++) {
 					if (strcmp(ind[l].p[k].page,table)==0) {
@@ -356,9 +356,9 @@ LOOP:
 					ind[l].num++;
 					if (!((ind[l].num)%16)) ind[l].p=(struct page *)xrealloc(ind[l].p,sizeof(struct page)*((int)((ind[l].num)/16)+1)*16);
 
-					ind[l].p[ind[l].num].page=xstrdup(table);	
+					ind[l].p[ind[l].num].page=xstrdup(table);
 
-					ind[l].p[ind[l].num].enc=xstrdup(estr);	
+					ind[l].p[ind[l].num].enc=xstrdup(estr);
 					chkpageattr(&ind[l].p[ind[l].num]);
 				}
 			}
@@ -388,15 +388,15 @@ LOOP:
 						nest++;
 					if (buff[j]==arg_close) {
 						if (nest==0) {
-							table[k]='\0';	
-							ind[i].p[0].page=xstrdup(table);	
+							table[k]='\0';
+							ind[i].p[0].page=xstrdup(table);
 							break;
 						}
 						else nest--;
 					}
 					copy_multibyte_char(buff, table, &j, &k);
 				}
-				ind[l].p[0].enc=xstrdup(estr);	
+				ind[l].p[0].enc=xstrdup(estr);
 				chkpageattr(&ind[i].p[0]);
 			}
 		}
@@ -538,7 +538,7 @@ ATTRLOOP:
 			switch(page_precedence[pattr[cc]]) {
 			case 'r':
 				if (strchr("ivxlcdm",*page0)==NULL ||
-				    (strchr("lcdm",*page0) && strchr(page_precedence,'a') && strlen(buff)==1 && pcpos))  {
+				    (strchr("lcdm",*page0) && strchr(page_precedence,'a') && strlen(buff)==1 && pcpos)) {
 					/* heuristic detection as alphabet since L=50, C=100, D=100, M=1000 are quite large */
 					if (pattr[cc]<pplen-1)
 						pattr[cc]++;
@@ -549,7 +549,7 @@ ATTRLOOP:
 				break;
 			case 'R':
 				if (strchr("IVXLCDM",*page0)==NULL ||
-				    (strchr("LCDM",*page0) && strchr(page_precedence,'A') && strlen(buff)==1 && pcpos))  {
+				    (strchr("LCDM",*page0) && strchr(page_precedence,'A') && strlen(buff)==1 && pcpos)) {
 					/* heuristic detection as alphabet since L=50, C=100, D=100, M=1000 are quite large */
 					if (pattr[cc]<pplen-1)
 						pattr[cc]++;
@@ -568,7 +568,9 @@ ATTRLOOP:
 				}
 				break;
 			case 'a':
-				if (*page0<'a' || *page0>'z' || strlen(buff)>1) {
+				if (*page0<'a' || *page0>'z' || strlen(buff)>1 ||
+				    (strchr("ivx",*page0) && strchr(page_precedence,'r') && !pcpos)) {
+					/* heuristic detection as roman number since I=1, V=5, X=10 are quite small */
 					if (pattr[cc]<pplen-1)
 						pattr[cc]++;
 					else pattr[cc]=0;
@@ -577,7 +579,9 @@ ATTRLOOP:
 				}
 				break;
 			case 'A':
-				if (*page0<'A' || *page0>'Z' || strlen(buff)>1) {
+				if (*page0<'A' || *page0>'Z' || strlen(buff)>1 ||
+				    (strchr("IVX",*page0) && strchr(page_precedence,'R') && !pcpos)) {
+					/* heuristic detection as roman number since I=1, V=5, X=10 are quite small */
 					if (pattr[cc]<pplen-1)
 						pattr[cc]++;
 					else pattr[cc]=0;
diff --git a/Build/source/texk/upmendex/tests/pprec3.ist b/Build/source/texk/upmendex/tests/pprec3.ist
new file mode 100644
index 00000000000..b8f33af1dd8
--- /dev/null
+++ b/Build/source/texk/upmendex/tests/pprec3.ist
@@ -0,0 +1,11 @@
+% -*- coding: utf-8 -*-
+
+page_precedence "naA"
+
+delim_0 "[[delim0]]"
+delim_1 "[[delim1]]"
+delim_2 "[[delim2]]"
+delim_r "[[delimr]]"
+suffix_2p "[[sfx 2p]]"
+suffix_3p "[[sfx 3p]]"
+
diff --git a/Build/source/texk/upmendex/tests/pprec4.ist b/Build/source/texk/upmendex/tests/pprec4.ist
new file mode 100644
index 00000000000..768fec7642b
--- /dev/null
+++ b/Build/source/texk/upmendex/tests/pprec4.ist
@@ -0,0 +1,11 @@
+% -*- coding: utf-8 -*-
+
+page_precedence "Ana"
+
+delim_0 "[[delim0]]"
+delim_1 "[[delim1]]"
+delim_2 "[[delim2]]"
+delim_r "[[delimr]]"
+suffix_2p "[[sfx 2p]]"
+suffix_3p "[[sfx 3p]]"
+
diff --git a/Build/source/texk/upmendex/tests/pprecB-0.ind b/Build/source/texk/upmendex/tests/pprecB-0.ind
new file mode 100644
index 00000000000..1e23d9fb4a7
--- /dev/null
+++ b/Build/source/texk/upmendex/tests/pprecB-0.ind
@@ -0,0 +1,5 @@
+\begin{theindex}
+
+  \item entryA[[delim0]]1[[sfx 3p]], a[[sfx 3p]], A[[sfx 3p]]
+
+\end{theindex}
diff --git a/Build/source/texk/upmendex/tests/pprecB-3.ind b/Build/source/texk/upmendex/tests/pprecB-3.ind
new file mode 100644
index 00000000000..1e23d9fb4a7
--- /dev/null
+++ b/Build/source/texk/upmendex/tests/pprecB-3.ind
@@ -0,0 +1,5 @@
+\begin{theindex}
+
+  \item entryA[[delim0]]1[[sfx 3p]], a[[sfx 3p]], A[[sfx 3p]]
+
+\end{theindex}
diff --git a/Build/source/texk/upmendex/tests/pprecB-4.ind b/Build/source/texk/upmendex/tests/pprecB-4.ind
new file mode 100644
index 00000000000..3943ad48549
--- /dev/null
+++ b/Build/source/texk/upmendex/tests/pprecB-4.ind
@@ -0,0 +1,5 @@
+\begin{theindex}
+
+  \item entryA[[delim0]]A[[sfx 3p]], 1[[sfx 3p]], a[[sfx 3p]]
+
+\end{theindex}
diff --git a/Build/source/texk/upmendex/tests/pprecB.idx b/Build/source/texk/upmendex/tests/pprecB.idx
new file mode 100644
index 00000000000..a1d8fc31ae2
--- /dev/null
+++ b/Build/source/texk/upmendex/tests/pprecB.idx
@@ -0,0 +1,9 @@
+\indexentry{entryA}{a}
+\indexentry{entryA}{b}
+\indexentry{entryA}{c}
+\indexentry{entryA}{1}
+\indexentry{entryA}{2}
+\indexentry{entryA}{3}
+\indexentry{entryA}{A}
+\indexentry{entryA}{B}
+\indexentry{entryA}{C}
diff --git a/Build/source/texk/upmendex/tests/range.idx b/Build/source/texk/upmendex/tests/range.idx
index e77882010c6..3eccb113aa4 100644
--- a/Build/source/texk/upmendex/tests/range.idx
+++ b/Build/source/texk/upmendex/tests/range.idx
@@ -58,13 +58,13 @@
 \indexentry{entryF}{D-DIII}
 \indexentry{entryF}{D-DIV}
 
-\indexentry{entryP}{A-a-1-I-iii}
-\indexentry{entryP}{A-a-2-I-iii}
-\indexentry{entryP}{A-a-3-I-iii}
-\indexentry{entryP}{A-a-1-II-iii}
-\indexentry{entryP}{A-a-1-III-iii}
-\indexentry{entryP}{A-a-1-I-iv}
-\indexentry{entryP}{A-a-1-I-v}
+\indexentry{entryP}{A-a-1-VII-iii}
+\indexentry{entryP}{A-a-2-VII-iii}
+\indexentry{entryP}{A-a-3-VII-iii}
+\indexentry{entryP}{A-a-1-VIII-iii}
+\indexentry{entryP}{A-a-1-IX-iii}
+\indexentry{entryP}{A-a-1-VII-iv}
+\indexentry{entryP}{A-a-1-VII-v}
 
 \indexentry{entryQ}{1-2-3-4-5-6-7-8-9-10}
 \indexentry{entryQ}{1-2-3-4-5-6-7-8-9-11}
diff --git a/Build/source/texk/upmendex/tests/range1.ind b/Build/source/texk/upmendex/tests/range1.ind
index 0dc1d842a9f..c5f047a55e5 100644
--- a/Build/source/texk/upmendex/tests/range1.ind
+++ b/Build/source/texk/upmendex/tests/range1.ind
@@ -15,8 +15,8 @@
   \item entryF, C-I[[sfx mp]], D-DI[[sfx mp]], V-LIII[[sfx mp]]
   \item entryF1, C-I[[sfx mp]], \bold{D-DII}, D-DI[[sfx mp]], 
 		V-LIII[[sfx mp]]
-  \item entryP, A-a-1-I-iii[[sfx mp]], A-a-1-II-iii, A-a-1-III-iii, 
-		A-a-2-I-iii, A-a-3-I-iii
+  \item entryP, A-a-1-VII-iii[[sfx mp]], A-a-1-VIII-iii, A-a-1-IX-iii, 
+		A-a-2-VII-iii, A-a-3-VII-iii
   \item entryQ, 1-2-3-4-5-6-7-8-9-10[[sfx 2p]], 
 		1-2-3-4-5-6-7-8-9-14[[sfx mp]], 
 		1-2-3-4-5-6-7-8-9-22[[sfx mp]], 
diff --git a/Build/source/texk/upmendex/tests/range2.ind b/Build/source/texk/upmendex/tests/range2.ind
index 9e0bd8ce5e3..789eb3f75c7 100644
--- a/Build/source/texk/upmendex/tests/range2.ind
+++ b/Build/source/texk/upmendex/tests/range2.ind
@@ -16,8 +16,8 @@
 		V-LIII[[delim r]]V-LVI
   \item entryF1, C-I[[delim r]]C-IV, \bold{D-DII}, D-DI[[delim r]]D-DIV, 
 		V-LIII[[delim r]]V-LVI
-  \item entryP, A-a-1-I-iii[[sfx 3p]], A-a-1-II-iii, A-a-1-III-iii, 
-		A-a-2-I-iii, A-a-3-I-iii
+  \item entryP, A-a-1-VII-iii[[sfx 3p]], A-a-1-VIII-iii, A-a-1-IX-iii, 
+		A-a-2-VII-iii, A-a-3-VII-iii
   \item entryQ, 1-2-3-4-5-6-7-8-9-10[[sfx 2p]], 
 		1-2-3-4-5-6-7-8-9-14[[sfx 3p]], 
 		1-2-3-4-5-6-7-8-9-22[[delim r]]1-2-3-4-5-6-7-8-9-25, 
diff --git a/Build/source/texk/upmendex/tests/range3.ind b/Build/source/texk/upmendex/tests/range3.ind
index f5b5f5268ff..519730a138e 100644
--- a/Build/source/texk/upmendex/tests/range3.ind
+++ b/Build/source/texk/upmendex/tests/range3.ind
@@ -15,8 +15,8 @@
   \item entryF, C-I[[sfx mp]], D-DI[[sfx mp]], V-LIII[[sfx mp]]
   \item entryF1, C-I[[sfx mp]], \bold{D-DII}, D-DI[[sfx mp]], 
 		V-LIII[[sfx mp]]
-  \item entryP, A-a-1-I-iii[[sfx 3p]], A-a-1-II-iii, A-a-1-III-iii, 
-		A-a-2-I-iii, A-a-3-I-iii
+  \item entryP, A-a-1-VII-iii[[sfx 3p]], A-a-1-VIII-iii, A-a-1-IX-iii, 
+		A-a-2-VII-iii, A-a-3-VII-iii
   \item entryQ, 1-2-3-4-5-6-7-8-9-10[[sfx 2p]], 
 		1-2-3-4-5-6-7-8-9-14[[sfx 3p]], 
 		1-2-3-4-5-6-7-8-9-22[[sfx mp]], 
diff --git a/Build/source/texk/upmendex/tests/upmendex.test b/Build/source/texk/upmendex/tests/upmendex.test
index 54152ca7a06..6a5d935a0ae 100755
--- a/Build/source/texk/upmendex/tests/upmendex.test
+++ b/Build/source/texk/upmendex/tests/upmendex.test
@@ -7,7 +7,7 @@
 TEXMFCNF=$srcdir/../kpathsea
 export TEXMFCNF
 
-rm -f foo.* uni.* pprec*.* range*.* locale*.* attr*.* rule*.*
+rm -f foo.* uni.* range*.* pprec*.* locale*.* attr*.* rule*.*
 
 ./upmendex $srcdir/tests/foo.idx -o foo.ind1 -t foo.ilg1 \
 	&& diff $srcdir/tests/foo.ind foo.ind1 || exit 1
@@ -54,6 +54,18 @@ cat $srcdir/tests/uni.idx | \
 	-o pprecA-2.ind1 -t pprecA-2.ilg \
 	&& diff $srcdir/tests/pprecA-2.ind pprecA-2.ind1 || exit 1
 
+./upmendex -s $srcdir/tests/pprec0.ist $srcdir/tests/pprecB.idx \
+	-o pprecB-0.ind1 -t pprecB-0.ilg \
+	&& diff $srcdir/tests/pprecB-0.ind pprecB-0.ind1 || exit 1
+
+./upmendex -s $srcdir/tests/pprec3.ist $srcdir/tests/pprecB.idx \
+	-o pprecB-3.ind1 -t pprecB-3.ilg \
+	&& diff $srcdir/tests/pprecB-3.ind pprecB-3.ind1 || exit 1
+
+./upmendex -s $srcdir/tests/pprec4.ist $srcdir/tests/pprecB.idx \
+	-o pprecB-4.ind1 -t pprecB-4.ilg \
+	&& diff $srcdir/tests/pprecB-4.ind pprecB-4.ind1 || exit 1
+
 # test for icu_locale
 ./upmendex -s $srcdir/tests/uni.ist -s $srcdir/tests/locale1.ist $srcdir/tests/rule.idx \
 	-o locale1.ind1 -t locale1.ilg \