Jump to content
  • 0

(Archived) Search issues: hyphen and underscore


~lg

Idea

I have started clipping things with Evernote a few weeks ago and I quite like it. With one, big exception: the search feature drives me nuts!

In general, I've found myself in need to search for things that include some punctuation symbols. I can live with some of the limitations. But here's the biggest problem: one cannot search even for a hyphenated word such as "x-ray". It sort of works, because it searches for "x" and "ray", but one also gets back random notes that have "ray" and "x" somewhere, instead of just the intended hyphenated word. I suppose if I had something like "to-do" to search for I would get all the notes back! This is very unintuitive behavior!

Also, from what I can tell, the underscore doesn't work at all.

I, for one, would prefer a normal, textual search (where a search for "xc%ds:s" would find just such a string anywhere in the text) to the complex search we have right now (make that an "advanced" feature).

Link to comment

15 replies to this idea

Recommended Posts

Using Evernote for Mac v 3.0.2 (195738) on Mac OS X 10.6.8, searches using the underscore _ do not work nor do search including the dash in quotes e.g. "a-name" will not find a-name but will find notes with words including 'a' and/or with 'name' in.

When these same searches are done on the iOS version of Evernote or on the web version of Evernote, they do work as expected. So it appears that there is an issue with the Mac version of Evernote.

Link to comment

I've been using Evernote for about 2 years and love it. However, it seems a search problem has cropped up over the past few months which I've identified recently (since the OCR usually produced the desired result in spite of the problem).

My scanner format on file names for docs has been: YYYY_MM_DD_file_name.pdf

So for example, a doc might be called 2011_11_19_WFB_MC_state.pdf

This becomes the title of my Note when I drag it onto Evernote.

I used to be able to search on WFB and quickly find the doc, but that no longer seems to work.

I believe it is because you searches are all "begins with" and the date is now considered the start of the phrase.

Should this work? I never saw anything that suggested it wouldn't. There are thousands of notes this effects. Will it be fixed soon? Thanks.

Link to comment
  • Level 5

Evernote indexes to make on-the-fly searches (i.e, as you type) fast. I'm not a computer expert, but based on the comments in this forum, I believe basic indexing like this:

Evernote Index for the word Evernote creates 8 phrases

e, ev, eve, ever, evern, everno, Evernot, Evernote

If Evernote added internal indexing (much more computer intensive), it would create 32 phrases

e, v, e, r, n, o, t, e, ev, ve, er, rn, no, ot, te,eve, ver,ern, rno, not, ote, ever, vern, erno,rnot, note, evern, verno, ernot, everno, vernot, ernote, Evernote

The Evernote API has more detailed information on search grammar at:

According to Evernote's Dave Engberg a year ago (Nov 17, 2010)

"You can search for complete words and phrases, or search for prefix substrings at the start of words, but
not for arbitrary character sequences in the middle of words
or punctuation. We aren't currently planning to change this behavior."
Link to comment

The search capability of EN is awkward. Difficult and bizarr. This I have pointed out for long ago.

I am not using EN any more except for just occasional jots, since it has so many ANNOYANCES which will never be corrected.

I am using Word (Office); nothing can replace Word.

Link to comment

So for example, a doc might be called 2011_11_19_WFB_MC_state.pdf

This becomes the title of my Note when I drag it onto Evernote.

I used to be able to search on WFB and quickly find the doc, but that no longer seems to work.

I believe it is because you searches are all "begins with" and the date is now considered the start of the phrase.

Should this work? I never saw anything that suggested it wouldn't. There are thousands of notes this effects. Will it be fixed soon? Thanks.

In the two plus years I've been using Evernote, that search has never worked, nor should it have, because letters, numbers & underscore are considered "words". Everything else is a delimiter. And Evernote does not search for arbitrary character sequences. Please read this thread.

Link to comment

Why would you drag a Word document to EN ? you are just degrading Word - the two programs have different uses.

Word is of course the better program, with very good search capabilities and it can also be organized into long documents; I have Word documents with almost 100 pages and find quickly everything.

EN on the ohter hand has this sync to the web feature which is good.

But EN is not nice to work with long texts. I cant use it to publish a website even if it is only text.

And EN staff are not responsive to criticism (feedback) as you can see from the response of the Evangelist BNF; it is a nonsensic motivation why EN does not search arbitrary character sequences.

Link to comment

Just to be clear of my frustration with Evernote for Mac v.3.0.3 searching, if I search for the word a_name or intitle:a_name on a Mac, I find nothing. If I search for a_name or intitle:a_name or even a_na* on Evernote for Web or Evernote for iOS, I find all notes with a_name .

Evernote support deny there is a problem saying, "evernote does not support the special characters between words while searching. Arish Evernote Support" when if fact Evernote web and iOS *does* support searches with underscore and hyphen e.g. a_name and "a-name" . This poor level of official Evernote Support does not encourage me to renew my Evernote Premium subscription.

A work-around I currently use is with an app called DataLore by Xtralean available in the Mac App Store or as a demo from their site. DataLore uses Spotlight and will find Evernotes for example with a_name .pdf or part words such as _na* . Double clicking the found note will open the Evernote note. [i have no connection with Xtralean.] It's not a great solution when wanting to work on multiple notes at once but I have been able to find things that I could not from within Evernote for Mac search.

Link to comment

I'll report it as a bug.

Good luck. When I reported it, Evernote replied "evernote does not support the special characters between words while searching." -- in short officially this issue with the Mac client is not seen as a bug. If you search using Evernote on the web, hyphenated searches work e.g. "my-name" finds my-name. It's just on the Mac client that doesn't work. To me that makes this issue a bug.

With the Evernote limitation of the number of syncing notebook, I have a lot of notes offline and therefore using iOS or using the web version of Evernote are not a viable alternative for underscore and hyphenated searches for me.

Link to comment

neilboyd, it might accept *any* special character in between the two words, not just a hyphen, but it should not match just one character. That is, if I search on "foo-bar", it might match "foo!bar" "foo bar" and "foo-bar", but should not match "foo" alone. I think that would be the correct behavior.

Link to comment

@neilboyd, it finds all of them. My understanding of the search grammar is that a search on "foo-bar" (quoted) should find the two words "foo" and "bar" separated by any single punctuation or special character. It does that, including "foo-bar". But it also finds "foo" alone, regardless of what follows it.

I suppose in a perfect world I'd rather it do more with special characters, but I can totally understand the limited search grammar given the size of the index they must be building.

Link to comment

I cannot understand your reasoning: first you criticize EN for a strange search grammar, then you add that you "totally understand" the why this is so

It sounds as if you were a programmer and have insight into the technical details of EN. Or maybe you are just trying to excuse EN.

But I am convinced that EN staff just don't care about criticism from users ; nobody has such an insight as you seem to have....

I have some programming experience with C+, I see no technical problem at all to fix the annoyance you mention.

Link to comment

@minkowski,

There are three issues regarding search grammar that I've commented on. One is the use of special characters such as underscore, the second is using quotes, and the third is (outside of this thread) logic combinations such as A AND (B OR C). The quotes issue is a bug: a search on "foo bar" should not find "foo notbar". The hyphens/undersores issues is something I think I understand better on some further reflection. If I've softened my criticism on this, that's why. I don't have any special insight into how Evernote does anything, and I've been plenty critical of Evernote on other issues.

My belief is that the issue with special characters is indexing. Evernote creates an index of every word in each account for fast (constant time) lookup. Like any such database, the index can take up a lot of storage. For many heavily indexed databases, indexing can require much more storage than the primary data, though I have no idea what the requirements are for Evernote. But regardless, the first task in indexing is deciding what's a word (which gets indexed) and what isn't. Not all combinations of all possible words can be indexed, because that would present a unscalable storage requirement. So some choices have to be made. When I look up "foo_bar", I believe Evernote should look up all "foo" and all "bar" and then selects the subset where the position of "foo" is just before the position of "bar". If a quoted search of "foo bar" matches "foo notbar", my understanding is that that's a bug. But even a successful search can't tell "foo_bar" from "foo bar" unless both the "_" and " " are indexed, or unless it performs a second, slower, task of searching the unindexed source for the exact match of "foo_bar". That's a choice I can understand, because of the indexing problem.

So I am very sympathetic to the need to delimit words and make tradeoffs in building and index. The issue of more comprehensive search logic, however, seems like a query engine issue that is solvable.

Link to comment

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...