hi all members,
i've started working on superopener to create a text links only version. i'll post in this thread all the changes i've made to discuss them with your help. this filter is really hard to study but i think here there are some filters guru that could help me.
at first, excuse me for the long (and probably hard to read) message. i know the jd5000 solution but understand his filter is more or less the same of understandig the bpm original, so i've started from bpm version, trying to make no errors.
this is the second part of the original filter set of three parts, in its unmodified form. all my consideration have to be referred to this part.
Name = "Links >^ SUPER-OPENER BETA 37 (aB) (bC)"
Active = TRUE
Multi = TRUE
URL = "$TYPE(htm)"
Bounds = "<as*(<(\|)/a>|(<as) )(^<!-- BPM_(W|A) -->)"
Limit = 450
Match = "<a"
"("
"([^>]++(shref=$AV(*))1[^>]+>)"
"&&"
"((shref=$AV(*)|starget=$AV(_blank|_new)"
"|(sclass=$AV(*))2|(sstyle=$AV(*))3|((s[^ ]+)|>)#))+"
")"
""
"( $NEST(<,(^(\|)/a(^?))*,> )+)4"
""
"("
"<(/a>|as)"
"$SET(6= class="BPM-supero-d")"
"$SET(7=◊)"
"|"
"(&[^; ]++; |[^<] )5"
"($NEST(<,(^(\|)/a(^?))*,> )+ )8"
"<((\|)/a>|as)$SET(7=<font size=-2>◊</font>)"
"|"
"(( (&[^; ]++;|[^<])"
"$NEST( <,(^(\|)/a(^?))*,>)+)++{1,2})5"
"( (&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+ )8"
"<((\|)/a>|as)"
"|"
"( (&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+"
"(&[^; ]++;|[^<])"
" (&[^; ]++;|[^<]|$NEST(<,(^(\|)/a(^?))*,> )+)++ )5"
"( (&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+"
"(&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+ )8"
"<((\|)/a>|as)"
")"
Replace = "<a123@45</a>"
"<a id=BPM-supero6"
" title=Open?in?new?window"
" target=_blank123>"
"78</a>"
"<!-- BPM_W --> "
changelog and explanations:
1. removed the filter "Comments >^ Remove temporary proxomitron comment tags (C)"
it's the third part of the superopener filter set. you don't really need it because it removes only this short comment "<!-- BPM_W -->" added by "Links >^ SUPER-OPENER BETA 37 (aB) (bC)" to each matching links. btw, you have to consider this balance: with "Comments" active you have additional checks of this filter for each character of the web page that non matches the previous filters (their number is equal to the non matching characters contained in the page); with "Comments" disabled, you have additional checks for each active filter, for each character of "<!-- BPM_W -->" and for each link matched by superopener. the number of additional checks in this case is (14 x N x M) where N is the number of active filters and M the number of links matched in the page. it's clear that you can take advantage from disabling "Comments" only disabling also the addition of "<!-- BPM_W -->" in "Links" filter.
2. removed (^<!-- BPM_(W|A) -->) from Bounds
3. removed <!-- BPM_W --> from Replace
4. removed Multi = TRUE
these changes are strictly related because superopener needs to add <!-- BPM_W --> to avoid double matching of the code replaced by itself. in fact, it needs the option Multi = TRUE to enable other filters to match the <a> tags, but needs also a protection for the <a> tags that itself adds to the end of the link to avoid an infinite loop (this is the reason of the ^ function in the Bounds). then, you can safely remove Multi = TRUE only if superopener is placed at the very end of your filter set. from this point i'll assume that you have placed superopener at the end of your filter set.
5. simplyfied Bounds in "<as*(/a>|(<as) )"
due to the wildcard * there is no need to have <(\|) in the first part of the substring. i've left (<as) in the second part because it makes superopener able to match also erroneously nested links in the form <a..<a.. correcting them in the replacement code. this is the first point where i don't agree with jd5000 solution, btw the discussion is open and suggestions are welcome.
- note about Bounds: it is possible add in bounds an exclusion for the images, something like (^<img*>) but my goal is remove from superopener all matching code related to images links so i can't add this exception because i have first to find and remove unnecessary code. i'll add the exception only at the end.
ok, here i stopped my first session. this is the actual result:
Name = "Super Opener - Text Links Only 2"
Active = TRUE
URL = "$TYPE(htm)"
Bounds = "<as*(/a>|(<as) )"
Limit = 450
Match = "<a"
"("
"([^>]++(shref=$AV(*))1[^>]+>)"
"&&"
"((shref=$AV(*)|starget=$AV(_blank|_new)"
"|(sclass=$AV(*))2|(sstyle=$AV(*))3|((s[^ ]+)|>)#))+"
")"
""
"( $NEST(<,(^(\|)/a(^?))*,> )+)4"
""
"("
"<(/a>|as)"
"$SET(6= class="BPM-supero-d")"
"$SET(7=◊)"
"|"
"(&[^; ]++; |[^<] )5"
"($NEST(<,(^(\|)/a(^?))*,> )+ )8"
"<((\|)/a>|as)$SET(7=<font size=-2>◊</font>)"
"|"
"(( (&[^; ]++;|[^<])"
"$NEST( <,(^(\|)/a(^?))*,>)+)++{1,2})5"
"( (&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+ )8"
"<((\|)/a>|as)"
"|"
"( (&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+"
"(&[^; ]++;|[^<])"
" (&[^; ]++;|[^<]|$NEST(<,(^(\|)/a(^?))*,> )+)++ )5"
"( (&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+"
"(&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+ )8"
"<((\|)/a>|as)"
")"
Replace = "<a123@45</a>"
"<a id=BPM-supero6"
" title=Open?in?new?window"
" target=_blank123>"
"78</a> "
now some comments to the code:
"([^>]++(shref=$AV(*))1[^>]+>)"
it matches <a this_content>, no problem.
---------------
"&&"
"((shref=$AV(*)|starget=$AV(_blank|_new)"
"|(sclass=$AV(*))2|(sstyle=$AV(*))3|((s[^ ]+)|>)#))+"
it matches <a this_content>, no problem. at this point the <a..> tag is entirely stored in the 1, 2, 3 variables and the stack
---------------
"( $NEST(<,(^(\|)/a(^?))*,> )+)4"
here it is!! this code matches images links, because it matches <a..><this_image><even_doubled></a>. we don't need this line of code neither the 4 variable in the replacement. but if this line is removed, we need something to match <a..>this_text</a>.
---------------
"<(/a>|as)"
"$SET(6= class="BPM-supero-d")"
"$SET(7=◊)"
if the link is an image link this code matches the closing tag </a> or an eventually erroneous nested <a> tag and set 6 for the style and 7 to make possible the addition of the lozange. in a text only version we need only the first line and can safely remove the class "BPM-supero-d" from the first of the 3 superopener filters. note also that the closing </a> tag, even when matches, is not stored anyway, but is expressely replaced.
----------------
"(&[^; ]++; |[^<] )5"
"($NEST(<,(^(\|)/a(^?))*,> )+ )8"
"<((\|)/a>|as)$SET(7=<font size=-2>◊</font>)"
this seems to be an error. i think the second line of this code is unnecessary, but i still haven't understood at all this block of code so i'll write nothing about my conclusions. btw, this could be the bug that sometimes add the lozenge to text links.
----------------
"(( (&[^; ]++;|[^<])"
"$NEST( <,(^(\|)/a(^?))*,>)+)++{1,2})5"
"( (&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+ )8"
"<((\|)/a>|as)"
"|"
"( (&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+"
"(&[^; ]++;|[^<])"
" (&[^; ]++;|[^<]|$NEST(<,(^(\|)/a(^?))*,> )+)++ )5"
"( (&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+"
"(&[^; ]++;|[^<])"
" $NEST(<,(^(\|)/a(^?))*,> )+ )8"
"<((\|)/a>|as)"
here is where i need your help. this code apply to text links but i'm not sure we need such complicated thing. do you agree? for example, i think we can remove (//|) from all $NEST commands.
that's all. thank you for your patience reading all this. i'm looking in advance for your comments and suggestions. now i'm still testing the first result, mainly its compatibility with the rest of the filter set. it's still a text+images version but as you all understand this word require to be done step by step.
see you later in this thread,
regards to all,
altosax.
Edited by - altosax on 25 May 2002 16:16:22
Edited by - altosax on 25 May 2002 21:36:51
