Hi guys,
It's my first post to this list, so I'm sorry I'm going to complain, but before I do that I want to thank the Wave team for this great product. It's a lot of fun using, very useful and fun to program bots. Now, I've been working on a bot for a few days. This is a Bible bot, found on https://launchpad.net/wavebiblebot and http://wave-samples-gallery.appspot.com/about_app?app_id=70035 . This bot is written in python. It is inspired by the Wikifier bot quite a bit. Its function is to detect and replace tags with verses or verse links. Now, to the point. Bear with me, this is long, and I really hope the Wave API developers are around reading this. === Issues with SetTextInRange === When I began writing the bot, I used SetText like the Wikifier python robot does. I soon realized it was ruining all the existing annotations of the blip, so I stopped using it and began playing with ranges and SetTextInRange, which is what I use now. Then I found something weird, that I haven't understood yet: when I make several replacements in a blip, positions are somehow slided by 1 character after each replacement. Let me take an example. Let's say I want to replace 3 tags that I have identified. For each tag, I make sure to get the position from blip.GetDocument().GetText() each time, in order to not use an obsolete text. The position returned for the first tag is correct, and I replace it. Then I switch to the second tag, and once I replace it, it gets one character to the left of the returned position, and I replace it (and it gets ugly). Then I switch to the third tag, and it gets slided two characters to the left, and so on. I've scratched my head on this and I don't get it. As a patch in my code, I have a "replaced" variable which I indent everytime I call SetTextInRange, and I add its value to the calculated position for every tag. This is the first issue I've found. === Stretching it further, the "replaced" counter mess === Now the problem is that I had two functions, placeVerses(blip, replaced) and placeVerseLinks(blip, replaced), called one after the other. The first one would replace all verse quotes, while the second one would replace all verse links. Easy! Well, no! The "replaced" counter got in the way when I began to mix "verse" and "verselink" tags in one blip, as SetTextInRange would get called to replace a tag that was previous to a tag already replaced, hence messing up with the "replaced" counter again... As a result, I fixed this by merging my two functions into one, with the "replaced" counter inside it only, to ensure that all tags would be replaced in order. This new function is called placeTags(blip) and is getting quite big now ;-) === Annotations, langs... === Once this was fixed^Wpatched, I began to work on a fun thing again : detecting the blip's language using annotations in order to adjust the default Bible version automatically. It's kind of easy, since while you edit your blip, Wave places "lang" annotations with the value of the detected language. Well, it turned out to work great with blips that use only one language. But then I tried mixing languages inside a blip, to see what would happen... and I lost my mind trying to understand how ranges evolve when you use SetTextInRange. After fighting with it quite a bit, it seems to me that when you call SetTextInRange, lang annotations' ranges are not modified, which makes it quite hard to find the language applied to one of my tags once I have replaced some other tags before. I came up with two options : * try to calculate all the languages for my tags before replacing them and use this list later on. This is nice and clean, except I might have several tags that are identical but are in different language paragraphs, which ruins this method. * keep a counter (again) called blipDelta that I fill with the delta generated by each replacement using SetTextInRange. Then I remove this delta every time I need to get the language of a tag, thus referring to the initial ranges of the blip, since they seem to not be modified when using SetTextInRange. This second solution works... not always! and I really can't figure out how the ranges I see in my logs relate to the ranges I see in the debug editor in the sandbox. I ended up adding some ugly conditions to make it work most of the time in these (far stretched, granted) situations. See http://www.youtube.com/watch?v=ZG9ZOSWjnMY for an example of case I fought with and fixed in an ugly way. Now if you're not afraid to scare yourself, my code is available on http://bazaar.launchpad.net/~raphink/wavebiblebot/trunk/files . All of the issues I listed here are found in biblebot.py. I'd be more than happy to have answers about these issues I've faced. Regards, Raphaël Pinson -- You received this message because you are subscribed to the Google Groups "Google Wave API" group. To post to this group, send email to google-wave-...@googlegroups.com. To unsubscribe from this group, send email to google-wave-api+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/google-wave-api?hl=.
