[Soros] fix right boundary inheritance in conditional text

Conditional text didn't support boundary inheritance, treated always as boundary (end = True). Apply boundary detection at the end of conditional text, fixing existing problems of Spanish etc. Reported by JuanmaSP in GitHub issue #70 and Antonio Bueno in #94.
Numbertext · Nov 3, 2022 · ecb3d68 · ecb3d68
1 parent e410849
commit ecb3d68
Show file tree

Hide file tree

Showing 4 changed files with 4 additions and 8 deletions.
diff --git a/java/src/main/org/numbertext/Soros.java b/java/src/main/org/numbertext/Soros.java
@@ -27,7 +27,7 @@ public class Soros {
 
 	private static Pattern func = Pattern.compile(translate("(?:\\|?(?:\\$\\()+)?" + // optional nested calls
 			"(\\|?\\$\\(([^\\(\\)]*)\\)\\|?)" + // inner call (2 subgroups)
-			"(?:\\)+\\|?)?", // optional nested calls
+			"(?:\uE00A?\\)+\\|?)?", // optional nested calls
 			m2.substring(0, c.length()), c, "\\")); // \$, \(, \), \| -> \uE000..\uE003
 
 	public Soros(String source, String lang) {
@@ -74,7 +74,6 @@ public Soros(String source, String lang) {
 				// call inner separator: [ ... $1 ... ] -> $(\uE00A ... \uE00A$1\uE00A ... )
 				s2 = s2.replaceAll("^\\[[$](\\d\\d?|\\([^\\)]+\\))", "\\$(\uE00A\uE00A|\\$$1\uE00A"); // add "|"
 				s2 = s2.replaceAll("\\[([^$\\[\\\\]*)[$](\\d\\d?|\\([^\\)]+\\))", "\\$(\uE00A$1\uE00A\\$$2\uE00A");
-				s2 = s2.replaceAll("\uE00A\\]$", "|\uE00A)"); // add "|" in terminating position
 				s2 = s2.replaceAll("\\]", ")");
 				s2 = s2.replaceAll("(\\$\\d|\\))\\|\\$", "$1||\\$"); // $()|$() -> $()||$()
 				s2 = translate(s2, c, m, ""); // \uE000..\uE003-> \, ", ;, #

diff --git a/src/Soros.cxx b/src/Soros.cxx
@@ -29,7 +29,7 @@ const std::wstring Soros::pipe = L"\uE003";
 const wregex Soros::func ( Soros::translate (
     L"(?:\\|?(?:\\$\\()+)?"           // optional nested calls
     "(\\|?\\$\\(([^\\(\\)]*)\\)\\|?)" // inner call (2 subgroups)
-    "(?:\\)+\\|?)?",                  // optional nested calls
+    "(?:\uE00A?\\)+\\|?)?",           // optional nested calls
     Soros::m2.substr(0, 4), Soros::c, L"\\"));  // \$, \(, \), \| -> \uE000..\uE003
 
 void Soros::replace(std::wstring& s, const std::wstring& search,
@@ -122,7 +122,6 @@ Soros::Soros(std::wstring program, std::wstring filtered_lang):
                             L"$$(" SEP SEP L"|$$$1" SEP); // add "|" in terminating position
             s2 = regex_replace(s2, wregex(L"\\[([^$\\[\\\\]*)[$](\\d\\d?|\\([^\\)]+\\))"),
                             L"$$(" SEP L"$1" SEP L"$$$2" SEP);
-            s2 = regex_replace(s2, wregex(SEP L"\\]$"), L"|" SEP L")"); // add "|" in terminating position
             s2 = translate(s2, L"]", L")", L"");
             s2 = regex_replace(s2, wregex(L"([$]\\d|\\))\\|[$]"), L"$1||$$"); // $()|$() -> $()||$()
             s2 = translate(s2, c, m, L"");   // \uE000..\uE003-> \, ", ;, #

diff --git a/src/Soros.js b/src/Soros.js
@@ -1,5 +1,5 @@
 function Soros(program, lang) {
-    this.funcpat = /(\|?(\uE008\()+)?(\|?\uE008\(([^\(\)]*)\)\|?)(\)+\|?)?/
+    this.funcpat = /(\|?(\uE008\()+)?(\|?\uE008\(([^\(\)]*)\)\|?)(\uE00A?\)+\|?)?/
     this.meta = "\\\"$()|#;[]"
     this.enc = "\uE000\uE001\uE002\uE003\uE004\uE005\uE006\uE007\uE008\uE009"
     this.lines = []
@@ -116,7 +116,6 @@ function Soros(program, lang) {
                     // call inner separator: [ ... $1 ... ] -> $(\uE00A ... \uE00A$1\uE00A ... )
                     .replace(/^[[]\$(\d\d?|\([^\)]+\))/g,"$(\uE00A\uE00A|$$$1\uE00A")
                     .replace(/[[]([^\$[\\]*)\$(\d\d?|\([^\)]+\))/g,"$(\uE00A$1\uE00A$$$2\uE00A")
-                    .replace(/\uE00A]$/, "|\uE00A)") // add "|" in terminating position
                     .replace(/]/g, ")")
                     .replace(/(\$\d|\))\|\$/g,"$1||$$") // $(..)|$(..) -> $(..)||$(..)
                     .replace(/\$/g, "\uE008")

diff --git a/src/Soros.py b/src/Soros.py
@@ -24,7 +24,7 @@ def _tr(text, chars, chars2, delim):
 # pattern to recognize function calls in the replacement string
 _func = re.compile(_tr(r"""(?:\|?(?:\$\()+)?  # optional nested calls
                 (\|?\$\(([^\(\)]*)\)\|?)      # inner call (2 subgroups)
-                (?:\)+\|?)?""",               # optional nested calls
+                (?:\uE00A?\)+\|?)?""",        # optional nested calls
                 _m[4:8], _c[:4], "\\"), re.X)  # \$, \(, \), \| -> \uE000..\uE003
 
 class _Soros:
@@ -72,7 +72,6 @@ def __init__(self, prg, lang):
                 # call inner separator: [ ... $1 ... ] -> $(\uE00A ... \uE00A$1\uE00A ... )
                 s2 = re.sub(r"[\[]\$(\d\d?|\([^\)]+\))",u"$(\uE00A\uE00A|$\\1\uE00A", s2)
                 s2 = re.sub(r"[\[]([^\$[\\]*)\$(\d\d?|\([^\)]+\))",u"$(\uE00A\\1\uE00A$\\2\uE00A", s2)
-                s2 = re.sub(r"\uE00A]$","|\uE00A)", s2) # add "|" in terminating position
                 s2 = re.sub(r"]",")", s2)
                 s2 = re.sub(r"(\$\d|\))\|\$", r"\1||$", s2) # $()|$() -> $()||$()
                 s2 = _tr(s2, _c[:4], _m[:4], "")   # \uE000..\uE003-> \, ", ;, #