steve
/
factor
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

fix parsing of attributes for nofollows

db4
Doug Coleman 2009-04-01 12:44:38 -05:00
parent 7060a5905f
commit 826d9f18c5
2 changed files with 22 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
extra/html/parser/parser-tests.factor
View File

@ -42,6 +42,19 @@ V{
} }
] [ "<a href = \"http://factorcode.org/\" foo = bar baz='quux'a=pirsqd >" parse-html ] unit-test ] [ "<a href = \"http://factorcode.org/\" foo = bar baz='quux'a=pirsqd >" parse-html ] unit-test
[
V{
T{ tag f "a"
H{
{ "a" "pirsqd" }
{ "foo" "bar" }
{ "href" "http://factorcode.org/" }
{ "baz" "quux" }
{ "nofollow" f }
} f f }
}
] [ "<a href = \"http://factorcode.org/\" nofollow foo = bar baz='quux'a=pirsqd >" parse-html ] unit-test
[ [
V{ V{
T{ tag f "html" H{ } f f } T{ tag f "html" H{ } f f }

21
extra/html/parser/parser.factor
View File

@ -6,7 +6,6 @@ unicode.case unicode.categories combinators.short-circuit
quoting fry ; quoting fry ;
IN: html.parser IN: html.parser
TUPLE: tag name attributes text closing? ; TUPLE: tag name attributes text closing? ;
SINGLETON: text SINGLETON: text
@ -52,7 +51,7 @@ SYMBOL: tagstack
skip-whitespace skip-whitespace
[ current { [ CHAR: = = ] [ blank? ] } 1|| ] take-until ; [ current { [ CHAR: = = ] [ blank? ] } 1|| ] take-until ;
: read-= ( state-parser -- ) : read-=1 ( state-parser -- )
skip-whitespace skip-whitespace
[ [ current CHAR: = = ] take-until drop ] [ next drop ] bi ; [ [ current CHAR: = = ] take-until drop ] [ next drop ] bi ;
@ -71,12 +70,8 @@ SYMBOL: tagstack
">" take-until-sequence dtd new-tag push-tag ; ">" take-until-sequence dtd new-tag push-tag ;
: read-bang ( state-parser -- ) : read-bang ( state-parser -- )
next dup { [ current CHAR: - = ] [ peek-next CHAR: - = ] } 1&& [ next dup { [ current CHAR: - = ] [ peek-next CHAR: - = ] } 1&&
next next [ next next read-comment ] [ read-dtd ] if ;
read-comment
] [
read-dtd
] if ;
: read-tag ( state-parser -- string ) : read-tag ( state-parser -- string )
[ [ current "><" member? ] take-until ] [ [ current "><" member? ] take-until ]
@ -88,15 +83,17 @@ SYMBOL: tagstack
: parse-text ( state-parser -- ) : parse-text ( state-parser -- )
read-until-< [ text new-tag push-tag ] unless-empty ; read-until-< [ text new-tag push-tag ] unless-empty ;
: parse-key/value ( state-parser -- key value )
[ read-key >lower ]
[ skip-whitespace "=" take-sequence ]
[ swap [ read-value ] [ drop f ] if ] tri ;
: (parse-attributes) ( state-parser -- ) : (parse-attributes) ( state-parser -- )
skip-whitespace skip-whitespace
dup state-parse-end? [ dup state-parse-end? [
drop drop
] [ ] [
[ [ parse-key/value swap set ] [ (parse-attributes) ] bi
[ read-key >lower ] [ read-= ] [ read-value ] tri
swap set
] keep (parse-attributes)
] if ; ] if ;
: parse-attributes ( state-parser -- hashtable ) : parse-attributes ( state-parser -- hashtable )