Added empmatchbefore and empmatchafter prefix opcodes and updated UEB…

… tables to use them.

Allow capitals in checkEmphasisChange.

liblouis/compileTranslationTable.c

@@ -279,6 +279,8 @@ static const char *opcodeNames[CTO_None] = {
   "before",
   "noback",
   "nofor",
+  "empmatchbefore",
+  "empmatchafter",
   "swapcc",
   "swapcd",
   "swapdd",
@@ -3836,7 +3838,7 @@ compileRule (FileInfo * nested)
 			getCharacters(nested, &ptn_after);
 			getRuleDotsPattern(nested, &ruleDots);
 
-			if(!addRule(nested, opcode, &ruleChars, &ruleDots, 0, 0))
+			if(!addRule(nested, opcode, &ruleChars, &ruleDots, after, before))
 				ok = 0;
 
 			if(ptn_before.chars[0] == '-' && ptn_before.length == 1)
@@ -4689,6 +4691,18 @@ compileRule (FileInfo * nested)
       goto doOpcode;
     case CTO_NoFor:
       nofor = 1;
+
+		case CTO_EmpMatchBefore:
+
+			before |= CTC_EmpMatch;
+			goto doOpcode;
+
+		case CTO_EmpMatchAfter:
+
+			after |= CTC_EmpMatch;
+			goto doOpcode;
+
+
       goto doOpcode;
     case CTO_SwapCc:
     case CTO_SwapCd:

liblouis/lou_translateString.c

@@ -2070,7 +2070,7 @@ checkEmphasisChange(const int skip)
 {
 	int i;	
 	for(i = src + (skip + 1); i < src + transRule->charslen; i++)
-	if((emphasisBuffer[i] & ~CAPS_EMPHASIS) || transNoteBuffer[i])
+	if(emphasisBuffer[i] || transNoteBuffer[i])
 		return 1;
 	return 0;
 }
@@ -2219,11 +2219,27 @@ for_selectRule ()
 	  transCharslen = transRule->charslen;
 	  if (tryThis == 1 || ((transCharslen <= length) && validMatch ()))
 	    {
+			/*   check before emphasis match   */
+			if(transRule->before & CTC_EmpMatch)
+			{
+				if(   emphasisBuffer[src]
+				   || transNoteBuffer[src])
+					break;
+			}
+
+			/*   check before emphasis match   */
+			if(transRule->after & CTC_EmpMatch)
+			{
+				if(   emphasisBuffer[src + transCharslen]
+				   || transNoteBuffer[src + transCharslen])
+					break;
+			}
+
 	      /* check this rule */
 	      setAfter (transCharslen);
-	      if ((!transRule->after || (beforeAttributes
+	      if ((!(transRule->after & ~CTC_EmpMatch) || (beforeAttributes
 					 & transRule->after)) &&
-		  (!transRule->before || (afterAttributes
+		  (!(transRule->before & ~CTC_EmpMatch) || (afterAttributes
 					  & transRule->before)))
 		switch (transOpcode)
 		  {		/*check validity of this Translation */

liblouis/louis.h

@@ -110,7 +110,8 @@ typedef enum {
   CTC_CapsMode = 0x1000000,
   CTC_NumericMode = 0x2000000,
   CTC_NumericNoContract = 0x4000000,
-  CTC_EndOfInput = 0x8000000  //   used by pattern matcher
+  CTC_EndOfInput = 0x8000000,   //   only used by pattern matcher
+  CTC_EmpMatch = 0x10000000,   //   only used in TranslationTableRule->before and TranslationTableRule->after
 } TranslationTableCharacterAttribute;
 
 typedef enum {
@@ -261,6 +262,8 @@ typedef enum { /*Op codes */
                CTO_Before, /*only match if before character in class 30      */
                CTO_NoBack,
                CTO_NoFor,
+               CTO_EmpMatchBefore,
+               CTO_EmpMatchAfter,
                CTO_SwapCc,
                CTO_SwapCd,
                CTO_SwapDd,

liblouis/pattern.c

@@ -1380,7 +1380,7 @@ static int pattern_check_attrs(const widechar input_char, const widechar *expr_d
 {
 	int attrs;
 
-	attrs = ((expr_data[0] << 16) | expr_data[1]) & ~CTC_EndOfInput;
+	attrs = ((expr_data[0] << 16) | expr_data[1]) & ~(CTC_EndOfInput | CTC_EmpMatch);
 	if(!checkAttr(input_char, attrs, 0))
 		return 0;
 	return 1;

tables/en-ueb-g2.ctb

@@ -535,50 +535,38 @@ sufword subbasement 234-136-12-12-1-234-15-56-2345
 sufword subbing 234-136-12-12-346
 
 # be   10.6.1   10.6.2   10.6.3   10.6.4
-match %[^_~]%<* be ati!n 23              beati*
-match %[^_~]%<* be atr 23                bea*
-match %[^_~]%<* be c![hkq] 23            bec*
-match %[^_~]%<* be dra 23                bedr*
-match %[^_~]%<* be de!(%>*%[_~^]) 23     bede*
-match %[^_~]%<* be da 23                 bed*
-match %[^_~]%<* be elz 23                bee*
-match %[^_~]%<* be g![gs] 23             beg*
-match %[^_~]%<* be in 23                 bei*
-match %[^_~]%<* be l[aeiouwy] 23         bel*
-match %[^_~]%<* be neficen 23            beneficen*
-match %[^_~]%<* be net(['’]s)?%>*%[_~^] 23  benet
-match %[^_~]%<* be ne![dft] 23           bene*
-match %[^_~]%<* be n[aiu] 23             ben*
-match %[^_~]%<* be r[a] 23               bere*
-match %[^_~]%<* be re!n 23               ber*
-match %[^_~]%<* be stir 23               besti*
-match %[^_~]%<* be st[or] 23             best*
-match %[^_~]%<* be s![st] 23             bes*
-match %[^_~]%<* be t![chst'’] 23          bet*
-match %[^_~]%<* be [bfhjmopqwxz] 23
-
-#nofor begword beatif 23-1-2345-24-124
-#nofor begword beatit 23-1-2345-24-2345
-#nofor begword beb 23-12
-#nofor begword becal 23-14-1-123
-#nofor begword beco 23-14-135
-#nofor begword beda 23-145-1
-#nofor begword bede 23-145-15
-#nofor begword bedr 23-145-1235
-#nofor begword bef 23-124
-#nofor begword beg 23-1245
-#nofor begword beh 23-125
+empmatchafter match %[^_~]%<* be [Aa][Tt][Ii]![Nn] 23               beati*
+empmatchafter match %[^_~]%<* be [Aa][Tt][Rr] 23                    bea*
+empmatchafter match %[^_~]%<* be [Cc]![HKQhkq] 23                   bec*
+empmatchafter match %[^_~]%<* be [Dd][Rr][Aa] 23                    bedr*
+empmatchafter match %[^_~]%<* be [Dd][Ee]!(%>*%[_~^]) 23            bede*
+empmatchafter match %[^_~]%<* be [Dd][Aa] 23                        bed*
+empmatchafter match %[^_~]%<* be [Ee][Ll][Zz] 23                    bee*
+empmatchafter match %[^_~]%<* be [Gg]![GSgs] 23                     beg*
+empmatchafter match %[^_~]%<* be [Ii][Nn] 23                        bei*
+empmatchafter match %[^_~]%<* be [Ll][AEIOUWYaeiouwy] 23            bel*
+empmatchafter match %[^_~]%<* be [Nn][Ee][Ff][Ii][Cc][Ee][Nn] 23    beneficen*
+empmatchafter match %[^_~]%<* be [Nn][Ee][Tt](['’]s)?%>*%[_~^] 23   benet
+empmatchafter match %[^_~]%<* be [Nn][Ee]![DFTdft] 23               bene*
+empmatchafter match %[^_~]%<* be [Nn][AIUaiu] 23                    ben*
+empmatchafter match %[^_~]%<* be [Rr][Aa] 23                        bere*
+empmatchafter match %[^_~]%<* be [Rr][Ee]![Nn] 23                   ber*
+empmatchafter match %[^_~]%<* be [Ss][Tt][Ii][Rr] 23                besti*
+empmatchafter match %[^_~]%<* be [Ss][Tt][ORor] 23                  best*
+empmatchafter match %[^_~]%<* be [Ss]![STst] 23                     bes*
+empmatchafter match %[^_~]%<* be [Tt]![CHSTchst'’] 23               bet*
+empmatchafter match %[^_~]%<* be [BFHJMOPQWXZbfhjmopqwxz] 23
 
 nofor sufword be 23
 
 #   easier if after "be" cause both are dots 23
 # bb   10.6.5   10.6.6
-match %a bb %a 23
+empmatchbefore empmatchafter match %a bb %a 23
 
 nofor always bb 23
 
 # cc   10.6.5   10.6.6
-match %a cc %a 25
+empmatchbefore empmatchafter match %a cc %a 25
 
 nofor always cc 25
 
@@ -591,30 +579,28 @@ sufword saccharin 234-1-14-16-345-35
 sufword zucchini 1356-136-14-16-35-24
 
 # con   10.6.1   10.6.2   10.6.3   10.6.4
-match %[^_~]%<* con c!h 25
-match %[^_~]%<* con est 25
-match %[^_~]%<* con s%a 25  cons  "mod cons" 10.6.4
-sufword conned 14-135-1345-1345-1246
-#match %[^_~]%<* con n(!e|%>*%[^_~]) 25   conned
-#match %[^_~]%<* con ne!d 25              "
-match %[^_~]%<* con [abdfghijlmnopqrtuvwxyz] 25
+empmatchafter match %[^_~]%<* con [Cc]![Hh] 25
+empmatchafter match %[^_~]%<* con [Ee][Ss][Tt] 25
+empmatchafter match %[^_~]%<* con s%a 25  [Cc][Oo][Nn][Ss]  "mod cons" 10.6.4
+empmatchafter match %[^_~]%<* con [ABDFGHIJLMNOPQRTUVWXYZabdfghijlmnopqrtuvwxyz] 25
 sufword conakry 14-135-1345-1-13-1235-13456
 sufword conan  14-135-1345-1-1345
+sufword conned 14-135-1345-1345-1246
 
 nofor begword con 25
 
 # dis   10.6.1   10.6.2   10.6.3   10.6.4
 word diss 145-24-234-234
-match %[^_~]%<* dis c![s'’] 256
-match %[^_~]%<* dis he![dsv] 256
-match %[^_~]%<* dis h![ceiprtw'’] 256
-match %[^_~]%<* dis p!i 256
-match %[^_~]%<* dis [abdefgijlmnoqrstuvwxyz] 256
+empmatchafter match %[^_~]%<* dis [Cc]![Ss'’] 256
+empmatchafter match %[^_~]%<* dis [Hh][Ee]![DSVdsv] 256
+empmatchafter match %[^_~]%<* dis [Hh]![CEIPRTWceiprtw'’] 256
+empmatchafter match %[^_~]%<* dis [Pp]![Ii] 256
+empmatchafter match %[^_~]%<* dis [ABDEFGIJLMNOQRSTUVWXYZabdefgijlmnoqrstuvwxyz] 256
 
 nofor begword dis 256
 
 # ea   10.6.5   10.6.6   10.6.7
-match %a ea %a 2
+empmatchbefore empmatchafter match %a ea %a 2
 
 #   cheap fix, still fails with ,ea,ea,
 nofor always , 2
@@ -673,7 +659,7 @@ sufword toenail 2345-135-15-1345-1-24-123
 sufword turtleneck 2345-136-1235-2345-123-15-1345-15-14-13
 
 # ff   10.6.5   10.6.6
-match %a ff %a 235
+empmatchbefore empmatchafter match %a ff %a 235
 
 nofor always ff 235
 
@@ -684,7 +670,7 @@ sufword effort 15-124-123456-2345
 sufword stafford 34-1-124-123456-145
 
 # gg   10.6.5   10.6.6
-match %a gg %a 2356
+empmatchbefore empmatchafter match %a gg %a 2356
 
 nofor always gg 2356
 

tests/yaml/en-ueb-08-capitalization.yaml

@@ -138,7 +138,6 @@ tests:
   - un,,self,'i%
 - - XXIInd
   - ',,xxii,''nd'
-  - xfail: true
 - - VIIb
   - ',,vii,''b'
 

tests/yaml/en-ueb-10-07-contractions.yaml

@@ -798,7 +798,6 @@ tests:
   - dis.1turb
   - typeform:
       italic: '   ++++'
-    xfail: true
 
   # 10.6.3  page 118
 
@@ -810,7 +809,6 @@ tests:
   - ',dis,,cord'
 - - DISlike
   - ',,dis,''like'
-  - xfail: true
 
   # 10.6.4 page 118
 
@@ -1015,7 +1013,6 @@ tests:
   - /iff.1ly
   - typeform:
       italic: '     ++'
-    xfail: true
 - - subbasement
   - subbase;t
 - - tea
@@ -1027,19 +1024,14 @@ tests:
 
 - - CliffSide
   - ',cliff,side'
-  - xfail: true
 - - EggHead
   - ',egg,h1d'
-  - xfail: true
 - - MacEACHEN
   - ',mac,,ea*5'
-  - xfail: true
 - - SeaWorld
   - ',sea,_w'
-  - xfail: true
 - - TEAspoon
   - ',,tea,''spoon'
-  - xfail: true
 
   # 10.6.7 page 121
 
@@ -1232,7 +1224,6 @@ tests:
   - _!s
 - - SeaWorld
   - ',sea,_w'
-  - xfail: true
 - - underworld
   - '"u_w'
 - - world-wide