towards 1.6b1

1 files changed, 50 insertions, 22 deletions

diff --git a/Lib/sre_constants.py b/Lib/sre_constants.py
index 531dc31..c996960 100644
--- a/Lib/sre_constants.py
+++ b/Lib/sre_constants.py
@@ -15,6 +15,11 @@
 # other compatibility work.
 #
 
+# should this really be here?
+
+class error(Exception):
+    pass
+
 # operators
 
 FAILURE = "failure"
@@ -30,20 +35,20 @@ GROUP = "group"
 GROUP_IGNORE = "group_ignore"
 IN = "in"
 IN_IGNORE = "in_ignore"
+INFO = "info"
 JUMP = "jump"
 LITERAL = "literal"
 LITERAL_IGNORE = "literal_ignore"
 MARK = "mark"
 MAX_REPEAT = "max_repeat"
 MAX_REPEAT_ONE = "max_repeat_one"
-MAX_UNTIL = "max_until"
 MIN_REPEAT = "min_repeat"
-MIN_UNTIL = "min_until"
 NEGATE = "negate"
 NOT_LITERAL = "not_literal"
 NOT_LITERAL_IGNORE = "not_literal_ignore"
 RANGE = "range"
 REPEAT = "repeat"
+REPEAT_ONE = "repeat_one"
 SUBPATTERN = "subpattern"
 
 # positions
@@ -63,14 +68,16 @@ CATEGORY_WORD = "category_word"
 CATEGORY_NOT_WORD = "category_not_word"
 CATEGORY_LINEBREAK = "category_linebreak"
 CATEGORY_NOT_LINEBREAK = "category_not_linebreak"
-CATEGORY_LOC_DIGIT = "category_loc_digit"
-CATEGORY_LOC_NOT_DIGIT = "category_loc_not_digit"
-CATEGORY_LOC_SPACE = "category_loc_space"
-CATEGORY_LOC_NOT_SPACE = "category_loc_not_space"
 CATEGORY_LOC_WORD = "category_loc_word"
 CATEGORY_LOC_NOT_WORD = "category_loc_not_word"
-CATEGORY_LOC_LINEBREAK = "category_loc_linebreak"
-CATEGORY_LOC_NOT_LINEBREAK = "category_loc_not_linebreak"
+CATEGORY_UNI_DIGIT = "category_uni_digit"
+CATEGORY_UNI_NOT_DIGIT = "category_uni_not_digit"
+CATEGORY_UNI_SPACE = "category_uni_space"
+CATEGORY_UNI_NOT_SPACE = "category_uni_not_space"
+CATEGORY_UNI_WORD = "category_uni_word"
+CATEGORY_UNI_NOT_WORD = "category_uni_not_word"
+CATEGORY_UNI_LINEBREAK = "category_uni_linebreak"
+CATEGORY_UNI_NOT_LINEBREAK = "category_uni_not_linebreak"
 
 OPCODES = [
 
@@ -85,12 +92,13 @@ OPCODES = [
     CATEGORY,
     GROUP, GROUP_IGNORE,
     IN, IN_IGNORE,
+    INFO,
     JUMP,
     LITERAL, LITERAL_IGNORE,
     MARK,
-    MAX_REPEAT, MAX_UNTIL,
+    MAX_REPEAT,
     MAX_REPEAT_ONE,
-    MIN_REPEAT, MIN_UNTIL,
+    MIN_REPEAT,
     NOT_LITERAL, NOT_LITERAL_IGNORE,
     NEGATE,
     RANGE,
@@ -106,10 +114,11 @@ ATCODES = [
 CHCODES = [
     CATEGORY_DIGIT, CATEGORY_NOT_DIGIT, CATEGORY_SPACE,
     CATEGORY_NOT_SPACE, CATEGORY_WORD, CATEGORY_NOT_WORD,
-    CATEGORY_LINEBREAK, CATEGORY_NOT_LINEBREAK, CATEGORY_LOC_DIGIT,
-    CATEGORY_LOC_NOT_DIGIT, CATEGORY_LOC_SPACE,
-    CATEGORY_LOC_NOT_SPACE, CATEGORY_LOC_WORD, CATEGORY_LOC_NOT_WORD,
-    CATEGORY_LOC_LINEBREAK, CATEGORY_LOC_NOT_LINEBREAK
+    CATEGORY_LINEBREAK, CATEGORY_NOT_LINEBREAK, CATEGORY_LOC_WORD,
+    CATEGORY_LOC_NOT_WORD, CATEGORY_UNI_DIGIT, CATEGORY_UNI_NOT_DIGIT,
+    CATEGORY_UNI_SPACE, CATEGORY_UNI_NOT_SPACE, CATEGORY_UNI_WORD,
+    CATEGORY_UNI_NOT_WORD, CATEGORY_UNI_LINEBREAK,
+    CATEGORY_UNI_NOT_LINEBREAK
 ]
 
 def makedict(list):
@@ -138,23 +147,35 @@ AT_MULTILINE = {
 }
 
 CH_LOCALE = {
-    CATEGORY_DIGIT: CATEGORY_LOC_DIGIT,
-    CATEGORY_NOT_DIGIT: CATEGORY_LOC_NOT_DIGIT,
-    CATEGORY_SPACE: CATEGORY_LOC_SPACE,
-    CATEGORY_NOT_SPACE: CATEGORY_LOC_NOT_SPACE,
+    CATEGORY_DIGIT: CATEGORY_DIGIT,
+    CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT,
+    CATEGORY_SPACE: CATEGORY_SPACE,
+    CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE,
     CATEGORY_WORD: CATEGORY_LOC_WORD,
     CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD,
-    CATEGORY_LINEBREAK: CATEGORY_LOC_LINEBREAK,
-    CATEGORY_NOT_LINEBREAK: CATEGORY_LOC_NOT_LINEBREAK
+    CATEGORY_LINEBREAK: CATEGORY_LINEBREAK,
+    CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK
+}
+
+CH_UNICODE = {
+    CATEGORY_DIGIT: CATEGORY_UNI_DIGIT,
+    CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT,
+    CATEGORY_SPACE: CATEGORY_UNI_SPACE,
+    CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE,
+    CATEGORY_WORD: CATEGORY_UNI_WORD,
+    CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD,
+    CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK,
+    CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK
 }
 
 # flags
-SRE_FLAG_TEMPLATE = 1 # NYI
+SRE_FLAG_TEMPLATE = 1
 SRE_FLAG_IGNORECASE = 2
 SRE_FLAG_LOCALE = 4
 SRE_FLAG_MULTILINE = 8
 SRE_FLAG_DOTALL = 16
-SRE_FLAG_VERBOSE = 32
+SRE_FLAG_UNICODE = 32
+SRE_FLAG_VERBOSE = 64
 
 if __name__ == "__main__":
     import string
@@ -168,5 +189,12 @@ if __name__ == "__main__":
     dump(f, OPCODES, "SRE_OP")
     dump(f, ATCODES, "SRE")
     dump(f, CHCODES, "SRE")
+    f.write("#define SRE_FLAG_TEMPLATE %d\n" % SRE_FLAG_TEMPLATE)
+    f.write("#define SRE_FLAG_IGNORECASE %d\n" % SRE_FLAG_IGNORECASE)
+    f.write("#define SRE_FLAG_LOCALE %d\n" % SRE_FLAG_LOCALE)
+    f.write("#define SRE_FLAG_MULTILINE %d\n" % SRE_FLAG_MULTILINE)
+    f.write("#define SRE_FLAG_DOTALL %d\n" % SRE_FLAG_DOTALL)
+    f.write("#define SRE_FLAG_UNICODE %d\n" % SRE_FLAG_UNICODE)
+    f.write("#define SRE_FLAG_VERBOSE %d\n" % SRE_FLAG_VERBOSE)
     f.close()
     print "done"