phpBB • Free and Open Source Forum Software

Lumpy: phpBB does distinguish between keywords and phrases. The highlighting is not bound to the actual search results:

AmigoJack wrote:The highlighting in those displayed posts however is flawed - this is a technical limitation. You can test this if you just search with your internet browser thru the following search results and you'll always find at least one occurance of what you searched for - unbound to what is highlighted. Examples:

1. Phrase "Dunkin Donuts" only
2. Phrase "exact phrase search" only
3. Phrase "exact phrase search" plus keyword Donuts only

I get your point however, notice that the hilited words are not all there.

it is not hiliting the phrase, it did not actually search the exact phrase. it searched for each word in the phrase and brought you back the post with all of those words
to have been a true exact phrase search return it should have been like this:

... searching is thus; if you go to google and type in "big blue widgets" with the quotes. then your returns will only ... "exact phrase". in the phpbb search, you can not do the same thing. if ...

with the exact line you put between the quotes highlited and no other words highlited.


robert

davidbutler wrote:exact phrase search is a non-negotiable requirement

Can you explain why and what it is you're trying to achieve, there may be another way.

davidbutler wrote:Thread starter

phpBB doesn't have threads, only topics.

davidbutler wrote:exact phrase search

Such a thing does not exist: if you're searching for a phrase it's already something exact. The terms you and wilsonlogan are using are redundant like "IRC chat" or "ATM machine".

davidbutler wrote:(ver.3.0.14), when quotes are used, the search results ONLY include posts that include the exact phrase.

No, not the exact phrase - the phrase. Period. What would be a non-exact phrase anyway?

davidbutler wrote:However, all instances of the words contained within the quotes are highlighted. In your example, a comment with "My friend Dunkin likes Donuts" must have also contained the exact phrase "Dunkin Dounuts." I can see how that would be confusing.

Leaving apart that phpBB does not know comments either (those are posts) let me help distinguishing this: search results display only those posts (or topics) which have the phrase you searched for. The highlighting in those displayed posts however is flawed - this is a technical limitation. You can test this if you just search with your internet browser thru the following search results and you'll always find at least one occurance of what you searched for - unbound to what is highlighted. Examples:

1. Phrase "Dunkin Donuts" only
2. Phrase "exact phrase search" only
3. Phrase "exact phrase search" plus keyword Donuts only

Thread starter Wilson Logan is helping me stand up a professional/technical forum. He started this thread on my behalf. I need to settle on which software we're going to use, and exact phrase search is a non-negotiable requirement.

I originally planned to use VBulletin, which I recall has an "exact phrase search" checkbox on the advanced search page. I recently learned that feature wasn't implemented in version 5, and for reasons I can't begin to fathom, it was ripped out of version 4, which is still in wide use.

Wilson recommended phpBB, until this...

wilsonlogan wrote:..results... contain "Dunkin Donuts" but also results like

"My friend Dunkin likes Donuts"

which to me is not an exact phrase search. Unless I'm missing something...

Wilson, actually, what I think you're seeing is exact phrase search with a twist. Based on my testing (ver.3.0.14), when quotes are used, the search results ONLY include posts that include the exact phrase. However, all instances of the words contained within the quotes are highlighted. In your example, a comment with "My friend Dunkin likes Donuts" must have also contained the exact phrase "Dunkin Dounuts." I can see how that would be confusing.

I would consider that an annoyance bug but it doesn't break exact phrase search capability.

wilsonlogan wrote:My experience is that if you search for
"Dunkin Donuts"
you will indeed be returned results that contain
"Dunkin Donuts"
"My friend Dunkin likes Donuts"

Then you've experienced no phrase searches, but merely keyword searches that strip quotes from your input. A keyword search will naturally also find results where keywords appear in the same order without anything inbetween.

Either they accept phrases, or they don't - there is no "exact phrase search".

My experience is that if you search for

"Dunkin Donuts"

you will indeed be returned results that contain "Dunkin Donuts" but also results like

"My friend Dunkin likes Donuts"

which to me is not an exact phrase search. Unless I'm missing something and you can do something like

"Dunkin Donuts" -"Dunkin" -"Donuts"

Hi All,

Am I being obtuse or is there really no exact phrase search functionality in 3.0.x?

Thanks,

Wilson.

This has certainly been an interesting read. Took me a few days to work through it all. Couple of quick responses to prior comments before I interject my own opinions into the mix...

randomessence wrote:http://www.forum-software.org/forum-com ... machines-2

I know it's old, but for grins I did a compare between phpBB2 and phpBB3 using the same site linked above. They were woefully wrong on many points on phpBB2, which does not give me confidence that the rest of their evaluations are reliable. Take it for what it's worth.

SilentDeath1226 wrote:I mean with a simple on/off button you can just disable the feature.

I don't think you can really simplify things to that level. Features could be added that are peripheral and easily turned on / off as you say. But other features could be very complex and therefore permeate the entire body of code. It's not like the code isn't there, it's just bracketed with a big "if" statement that determines if the code has been enabled or not. In fact, it can make the code far more complex, difficult to maintain and extend, and frankly that's part of the cost that has to be evaluated before adding some new feature to the core. You could go back and look at the attachments MOD for phpBB2 (before it was added as a feature to phpBB3) and see an example. The attachments code touched the permissions system, the posting system, the private message system, the search system... basically almost the entire board. Yes, there is an "allow attachments: yes / no" option in the admin panel, but that doesn't change the fact that there is attachments code throughout the entire system. One of the more popular sub-forums MODs for phpBB2 nearly rewrote the entire codebase; now sub-forums are a core feature of phpBB3. You don't turn them on or off, but you do have the choice of using them or not.

Adding significant features to the core increases code complexity, so in my opinion simply saying "turn it off if you don't want it" is a bit unrealistic.

noth wrote:

this thread really does show phpBB2 for the obsolete lump it always was from 2008 onwards and yet we have a merry band of phpBB2 renegades who swear that it still has a use!!

Who would do that?

In the sake of full disclosure, yes, my main board (started in 2002 and now well over 60K users and 750K posts and approaching our tenth anniversary) is running phpBB2. But why? As has been stated previously in this topic, it's because I used MODs or my own custom code to adapt and extend the core board to do what I wanted it to do, including adding many features that ultimately appeared in phpBB3 in some form or another. Given that I manage a fairly large community using obsolete software, one might expect there would be complaints about missing features due to the "user expectations" mentioned earlier, right? After all of these years, the only request that I have seen with any regularity are to add an RSS feed and provide an exact phrase search. Of course RSS is something present in phpBB3 that was missing in phpBB2.

But nobody has asked about AJAX or WYSIWYG, and believe me I have some very vocal community members that would certainly ask if they thought it was important. Even if they didn't ask by name, nobody has complained about having to "go to another screen to edit a post" or some of the other usability features that have been mentioned here in this topic.

Do I think these features could be nice to have? Perhaps. From a usability perspective I think some of those improvements would be very interesting. But to be honest, I turn off the WYSIWYG editor in Wordpress because I hate it.

I believe that people that use this site (phpbb.com) are likely to be more technical and web-savvy than the general population. They know what trends are because they're leading them or at least very closely following them. The Internet population as a whole is probably several years behind the experience and expectation level of users that have post here (and have participated in this very discussion). You could say that the sheer number of folks still using IE6 is evidence of that!

My original decision to select phpBB2 over the competitors of the day was made because of the active community found here. I believe that this very topic continues to show that the community actively cares about (one might even say is passionate about) phpBB and wants it to continue to succeed. I think that bodes well for the future. A lot of talk in this topic seemed to revolve around the idea of contributing code, but there are far more ways for folks to contribute. What is the one thing that always seems to come up when explaining why things take so long? Not enough time. Do you have some extra time but don't know how to code? Then figure out another way to help, take over some other task that you can do. Or perhaps even make up a new one! I've not visited the dev board for a very long time, and don't have time to do so now. But in my experience in the IT world one of the most important steps in getting a new feature implemented is a full and complete requirements document.

If you're passionate about a particular feature, go out and do some research. Find out what's currently available, if anything, and document that. Find other people that have shown some interest in the same topic and get them involved, even if you only pester them with questions. Figure out what this new feature has to do, and how it interacts (in a process way, not a technical as in "write code this way" way) with the rest of the system.

Don't like that idea? Come up with something else that makes you feel good about helping the project move forward. That's the beauty of open source, right? In the old days of phpBB2 the code may have been open source but the development process was extremely closed. That has changed, but as has been observed not everyone is comfortable with or confident in their coding abilities. I'm just trying to say there are more ways to contribute than that, and that in my opinion it's that spirit and and passion that makes phpBB unique.

drathbun wrote:phpBB does not do phrase search, only word search.

I see.

drathbun wrote:The reason for the two options (use all terms, or use as entered) is you can put logic in between the words. You might search for "cake and ice cream" and the word and would be dropped because it's a stop word, or if you say use the search as entered then and becomes a boolean operator and will impact your query results.

Okay, thanks, that makes sense. But the wording of the description doesn't seem very explicit to me (maybe because, as I said, english is not my first language)...

drathbun wrote:That means I know which posts have "last" and which posts have "post" in them, but I can't tell you if both words appear next to each other as a phrase.

I understand, but IMHO this would be a very useful thing to implement, almost vital for large forums.
You see, I came here looking about a problem with the "last post" column in our forum index pointing all of a sudden to older posts. Before posting my question, I thought I might check for similar cases, so I tried to make a search... Woe is me!
This forum being what it is, 90% of posts contain the word "Post", and "Last" isn't rare either; Only a exact phrase search would had spared me the 17019 matches the search yielded...

I'm no programmer, but an additional search option allowing to check among the general results for cases where the two words are only separated by a space character shouldn't be too difficult to add, would it?
Oh well.
Thanks for explaining the search to me!

Its a private knowledge base for a yearly trivia contest - having our notes accessible to anyone doing a google search is pretty much a non-starter. A VPS is financially more doable than their corporate pay-version as well.

Ultimately we can get around this by yanking the mysql tables and just rolling our own on-site search/queries for the weekend we need this - shame we can't have our cake and eat it too - all for the want of two character search and exact phrase search. Stupid shared hosting and the tasty cheap cake they offer...

I searched my forum content in google and found 0 results.
Searched by TOPIC or post content using "whatever goes here" as exact phrase search.
NONE comes out, Why ?

I mean i have so much useful information there and its not searchable on the net ?

NominaSanctaMaria wrote:1) Ever since I have started working with phpBB (which in general I love) I noticed that the search function, in my opinion, has a major defect , and I later found out why. I am referring to the fact that it is impossible (as far as I can find) to do an 'exact phrase'-type search, such as you might use in Google - ie to find "phpbb forum" rather than phpbb+forum. Looking at the sort of volumes which you get on your impressive board I can understand why the restrictive methodology used by phpbb comes into its own.

However, I personally am convinced that not to be able to do an exact phrase search is a major no-no and am considering re-writing the search function to allow it - in other words it wouldn't use the search word indices. Although there would be some (probably fairly major) penalty in performance, I think users would be prepared to wait an extra 5 seconds for example to get the right result. After all, a fast search is only a good thing if it helps you find what you are looking for. Now with the volumes which you have this would almost certainly be unthinkable and people would have lived, paid taxes and died by the time the results were returned... However, the 'natural size' for my forum is unlikely to exceed 100,000 posts at any one time. The question: how insane would I be to try this out in your opinion? Has it a hope of working satisfactorily? No doubt it would work in the early stages, but I don't want to be running into major problems at say 50,000 posts...

Your post is exactly like this one, posted years earlier in this very topic: http://www.phpbb.com/community/viewtopi ... 45#p855333

And thus it was answered in the topic, so read it.

There are several different aspects to consider. First, one user might be willing to wait 5 extra seconds for their search results, but what if their search adds to the server load, making everyone else wait 1-2 extra seconds for their pages while the search is going on? That would not be popular.

That being said, there have been several discussions over the years about rewriting or extending the search to include an exact phrase search. The idea that I had (but have never completed) is to use the standard phpBB search first to isolate those topics with posts that contain all of the required words, and then do a brute-force text search on just those posts to find the exact phrase.

Some people with very large forums (Darth Wong, who doesn't come around much anymore, had 2M+ posts) that have experimented with switching completely to the MySQL fulltext search have reported that it slows down the search results. With enough memory and stopwords and other tweaks the phpBB search can perform quite well. I'm up to 400K+ posts (a far cry from 2 million, but I have dreams of getting there) and don't have any problems with searching on my board at all.

Here are some related topics that I've bookmarked over the years as people make attempts to play with the search process:

http://www.phpbb.com/community/viewtopi ... 6&t=531915
http://www.phpbb.com/community/viewtopi ... 5&t=601589
http://www.phpbb.com/community/viewtopi ... 8&t=208647

1) Ever since I have started working with phpBB (which in general I love) I noticed that the search function, in my opinion, has a major defect , and I later found out why. I am referring to the fact that it is impossible (as far as I can find) to do an 'exact phrase'-type search, such as you might use in Google - ie to find "phpbb forum" rather than phpbb+forum. Looking at the sort of volumes which you get on your impressive board I can understand why the restrictive methodology used by phpbb comes into its own.

However, I personally am convinced that not to be able to do an exact phrase search is a major no-no and am considering re-writing the search function to allow it - in other words it wouldn't use the search word indices. Although there would be some (probably fairly major) penalty in performance, I think users would be prepared to wait an extra 5 seconds for example to get the right result. After all, a fast search is only a good thing if it helps you find what you are looking for. Now with the volumes which you have this would almost certainly be unthinkable and people would have lived, paid taxes and died by the time the results were returned... However, the 'natural size' for my forum is unlikely to exceed 100,000 posts at any one time. The question: how insane would I be to try this out in your opinion? Has it a hope of working satisfactorily? No doubt it would work in the early stages, but I don't want to be running into major problems at say 50,000 posts...