inital checkin: libpcre
This the initial commit of libpcre, the Perl Compatible Regular
Expression library (http://www.pcre.org/)
The files in the dist/ directory correspond to pcre version
8.36 and are unmodified.
$ sha256sum pcre-8.36.tar.bz2
ef833457de0c40e82f573e34528f43a751ff20257ad0e86d272ed5637eb845bb pcre-8.36.tar.bz2
The top level directory contains 3 symlinks pointing into dist,
and one Android.mk file.
The code compiles, and I've tested a small program to verify that
the functionality is correct. However, no extensive testing has
been done at this point.
Bug: 17682157
Change-Id: Ie304aaf09cc37eb8920b40c3a3f7db61a0fda205
diff --git a/dist/testdata/testinput14 b/dist/testdata/testinput14
new file mode 100644
index 0000000..325ed9e
--- /dev/null
+++ b/dist/testdata/testinput14
@@ -0,0 +1,343 @@
+/-- This set of tests is run only with the 8-bit library. They do not require
+ UTF-8 or Unicode property support. The file starts with all the tests of
+ the POSIX interface, because that is supported only with the 8-bit library.
+ --/
+
+< forbid 8W
+
+/abc/P
+ abc
+ *** Failers
+
+/^abc|def/P
+ abcdef
+ abcdef\B
+
+/.*((abc)$|(def))/P
+ defabc
+ \Zdefabc
+
+/the quick brown fox/P
+ the quick brown fox
+ *** Failers
+ The Quick Brown Fox
+
+/the quick brown fox/Pi
+ the quick brown fox
+ The Quick Brown Fox
+
+/abc.def/P
+ *** Failers
+ abc\ndef
+
+/abc$/P
+ abc
+ abc\n
+
+/(abc)\2/P
+
+/(abc\1)/P
+ abc
+
+/a*(b+)(z)(z)/P
+ aaaabbbbzzzz
+ aaaabbbbzzzz\O0
+ aaaabbbbzzzz\O1
+ aaaabbbbzzzz\O2
+ aaaabbbbzzzz\O3
+ aaaabbbbzzzz\O4
+ aaaabbbbzzzz\O5
+
+/ab.cd/P
+ ab-cd
+ ab=cd
+ ** Failers
+ ab\ncd
+
+/ab.cd/Ps
+ ab-cd
+ ab=cd
+ ab\ncd
+
+/a(b)c/PN
+ abc
+
+/a(?P<name>b)c/PN
+ abc
+
+/a?|b?/P
+ abc
+ ** Failers
+ ddd\N
+
+/\w+A/P
+ CDAAAAB
+
+/\w+A/PU
+ CDAAAAB
+
+/\Biss\B/I+P
+ Mississippi
+
+/abc/\P
+
+/-- End of POSIX tests --/
+
+/a\Cb/
+ aXb
+ a\nb
+ ** Failers (too big char)
+ A\x{123}B
+ A\o{443}B
+
+/\x{100}/I
+
+/\o{400}/I
+
+/ (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* # optional leading comment
+(?: (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+|
+" (?: # opening quote...
+[^\\\x80-\xff\n\015"] # Anything except backslash and quote
+| # or
+\\ [^\x80-\xff] # Escaped something (something != CR)
+)* " # closing quote
+) # initial word
+(?: (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* \. (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+|
+" (?: # opening quote...
+[^\\\x80-\xff\n\015"] # Anything except backslash and quote
+| # or
+\\ [^\x80-\xff] # Escaped something (something != CR)
+)* " # closing quote
+) )* # further okay, if led by a period
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* @ (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+| \[ # [
+(?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
+\] # ]
+) # initial subdomain
+(?: #
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* \. # if led by a period...
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+| \[ # [
+(?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
+\] # ]
+) # ...further okay
+)*
+# address
+| # or
+(?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+|
+" (?: # opening quote...
+[^\\\x80-\xff\n\015"] # Anything except backslash and quote
+| # or
+\\ [^\x80-\xff] # Escaped something (something != CR)
+)* " # closing quote
+) # one word, optionally followed by....
+(?:
+[^()<>@,;:".\\\[\]\x80-\xff\000-\010\012-\037] | # atom and space parts, or...
+\(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) | # comments, or...
+
+" (?: # opening quote...
+[^\\\x80-\xff\n\015"] # Anything except backslash and quote
+| # or
+\\ [^\x80-\xff] # Escaped something (something != CR)
+)* " # closing quote
+# quoted strings
+)*
+< (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* # leading <
+(?: @ (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+| \[ # [
+(?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
+\] # ]
+) # initial subdomain
+(?: #
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* \. # if led by a period...
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+| \[ # [
+(?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
+\] # ]
+) # ...further okay
+)*
+
+(?: (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* , (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* @ (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+| \[ # [
+(?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
+\] # ]
+) # initial subdomain
+(?: #
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* \. # if led by a period...
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+| \[ # [
+(?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
+\] # ]
+) # ...further okay
+)*
+)* # further okay, if led by comma
+: # closing colon
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* )? # optional route
+(?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+|
+" (?: # opening quote...
+[^\\\x80-\xff\n\015"] # Anything except backslash and quote
+| # or
+\\ [^\x80-\xff] # Escaped something (something != CR)
+)* " # closing quote
+) # initial word
+(?: (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* \. (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+|
+" (?: # opening quote...
+[^\\\x80-\xff\n\015"] # Anything except backslash and quote
+| # or
+\\ [^\x80-\xff] # Escaped something (something != CR)
+)* " # closing quote
+) )* # further okay, if led by a period
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* @ (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+| \[ # [
+(?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
+\] # ]
+) # initial subdomain
+(?: #
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* \. # if led by a period...
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+| \[ # [
+(?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
+\] # ]
+) # ...further okay
+)*
+# address spec
+(?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* > # trailing >
+# name and address
+) (?: [\040\t] | \(
+(?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
+\) )* # optional trailing comment
+/xSI
+
+/-- Although this saved pattern was compiled with link-size=2, it does no harm
+to run this test with other link sizes because it is going to generated a
+"compiled in wrong mode" error as soon as it is loaded, so the link size does
+not matter. --/
+
+<!testsaved16
+
+<!testsaved32
+
+/\h/SI
+
+/\H/SI
+
+/\v/SI
+
+/\V/SI
+
+/\R/SI
+
+/[\h]/BZ
+ >\x09<
+
+/[\h]+/BZ
+ >\x09\x20\xa0<
+
+/[\v]/BZ
+
+/[\H]/BZ
+
+/[^\h]/BZ
+
+/[\V]/BZ
+
+/[\x0a\V]/BZ
+
+/\777/I
+
+/(*:0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF)XX/K
+ XX
+
+/(*:0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDE)XX/K
+ XX
+
+/\u0100/<JS>
+
+/[\u0100-\u0200]/<JS>
+
+/[^\x00-a]{12,}[^b-\xff]*/BZ
+
+/[^\s]*\s* [^\W]+\W+ [^\d]*?\d0 [^\d\w]{4,6}?\w*A/BZ
+
+/-- End of testinput14 --/