How do I search for a Hash/Pound sign in a title?

[robe070 4]

I have added notes via email with non-existent tags and I want to find them all and add the corrrect tag.

If the title conatins #ABC, then intitle:#ABC will find it. But I want to find ALL tags still in the title. So I tried intitle:#*. I get nothing. Is there a way to do this?

[jefito 5,589]

Looks like '#' is ignored in intitle search. If I have '#stuff' in a title, then a search of intitle:stuff will find it. And using quotes around the arguments doesn't seem to help. I'm guessing that any punctuation characters in the search and the title are discarded; I think that's something from the Web Services API doc. I'll try to look it up later on; Sense and Sensibility is on now...

[Brendan 17]

I have added notes via email with non-existent tags and I want to find them all and add the corrrect tag.

I've had this problem too. I looked at the Evernote Formal Search Specification and I'm not sure if it's possible.

wordchar = ? any Unicode Letter or Number (Unicode character classes L and N) or underscore '_' ? ;

I'm not sure if # is a Unicode character classes L or N

I also tried searching for a quote using \" (as described in the formal search spec), but couldn't find a note with a title containing a quote.

[jefito 5,589]

I'm guessing '#' is treated as punctuation, which is then handled as described in: http://www.evernote.com/about/developer/api/evernote-api.htm#_Toc297053082

[JMichaelTX 4,117]

Seems like punctuation should be included when the search string is in quotes.

IOW, Evernote should search for the exact string when it is in quotes.

This is what I expect based on my experience with other search engines.

[robe070 4]

You've reminded me about something I read years ago. Its about the indexing method used by Evernote. Punctuation is thrown away and # is treated that way. Hence there is no way to search for it. It does not exist in the index. Unless there is a search method that allows you to search for absolutely anything, but I don't know of such a feature.

And @JMichael, do you have a link to a list of differences between EN Mac and EN Win?

[CalS 5,280]

Dump to HTML, search that file and fix EN? Yuck, but you might get there?

[Owyn 457]

I'm guessing '#' is treated as punctuation, which is then handled as described in: http://www.evernote....m#_Toc297053082

Yeah. Evernote's Lucene processing discards most punctuation when creating full text search word index.

The various clients replicate this function when updating local search indexes.

http://blog.evernote...cene-explainin/

[JMichaelTX 4,117]

So Evernote doesn't really have full text search, it has word search only, right?

[Owyn 457]

Basically correct, however it also has phrase search.

e.g. If a note contains the phrase "cat in the hat", then cat, hat, and cat hat are valid searches but "cat hat" (with the quotes) is not.

[JMichaelTX 4,117]

Basically correct, however it also has phrase search.

e.g. If a note contains the phrase "cat in the hat", then cat, hat, and cat hat are valid searches but "cat hat" (with the quotes) is not.

Is that actually a phrase search, or just a shortcut way of searching for multiple words?

IOW, do all words have to appear together in the order of the search expression, or can all of the words appear anywhere in Note whether or not they are adjacent to one another?

[Owyn 457]

As I said it is a proper phrase search. The words have to be adjacent.

[JMichaelTX 4,117]

That's great then! Not as good as a true full text search, but MUCH better than just a word search.

The Search engine should work like this on all EN clients, correct?

[Owyn 457]

If they don't then it is an bug in that client.

[helix 10]

All the non-digit, non-letter characters such as @, #, $, etc. are not searchable in EN, except one - the underscore _ . (btw why that?.. )

So if your title has _ABC, and you are trying intitle:_* , it will work for you.

Pls note that the underscore _ serves like any other letter or digit, so in this case intitle:_ABC will not find ABC and intitle:ABC will not find _ABC.

[BurgersNFries 2,407]

except one - the underscore _ . (btw why that?.. )

[helix 10]

except one - the underscore _ . (btw why that?.. )

http://discussion.ev...9921#entry59921

Engberg wrote (see that link):

This is actually intentional. We treat a "word" as any contiguous sequence of:

* letters (as defined by Unicode)

* numbers

* the underscore character

This allows you to use the underscore to make special "magic keywords" like "_foo" that you may want to search for again later.

Thanks for the link!

In EN2 all special characters were searchable, but they have had removed during the upgrade to EN3, except the underscore.

I was afraid that it was kind of a "left over", maybe due to its low profile , and then someday a diligent EN employee will discover that and will kill the poor underscore as well..

Knowing that it was left intentionally and it is here to stay is very relaxing!..

[CalS 5,280]

I have added notes via email with non-existent tags and I want to find them all and add the corrrect tag.

If the title conatins #ABC, then intitle:#ABC will find it. But I want to find ALL tags still in the title. So I tried intitle:#*. I get nothing. Is there a way to do this?

Workaround. For Windows, there is a free utility Sysexporter, http://www.nirsoft.n...ils/sysexp.html. Select all of your notes in list view, make sure you have the columns you want, then use this utility to export to .CSV, open in Xcel and you can search for '#' or any other invisible character, then search in EN and fix the note. Not pretty, but takes less than a minute to create the export to test and helps find those pesky notes.

[Owyn 457]

Also note that search is case and diacritical insensitive. Another part of the Lucene standard.

Finally, we perform a fairly elaborate pipeline of transformations on both note text and search expressions in order to normalize the representation of the text for correct comparisons. We have Lucene analysis filters that operate on the sequence of tokens to:

• Remove apostrophes and other intra-word punctuation
  
• Convert upper-case letters to lower case
  
• Remove English “stop words” like “the” and “and”
  
• Normalize letters with diacritics so that “ñ” becomes “n”
  
• Convert “narrow width” Japanese characters to “full width”
  
• Reorganize Chinese/Japanese/Korean text into pseudo “words”
  

[thebeefman 2]

I don't know why Evernote wants to be inconsistent with google and other search engines here. Search engines will search for punctuation if you put it in quotes. Try it #Paul yields all Paul's but "#Paul" yields just the #Paul's. They really should fix this, it is what people expect.

[MikhailV 2]

I, too, would appreciate the ability to (optionally) search for Unicode characters (in addition to underscore, "_") outside general classes L and N within note titles and/or bodies.  (My very limited ad hoc testing suggests such characters can be explicitly searched on when they appear in Tags.)  One way this would serve me (and perhaps others) is by providing "search value" to the appending of a non-letter/number character as an abbreviated, symbolic modifier for a class of terms of repeated interest to me.

 

This would allow for personalized classification with appended strings as short as 1 character, conceptually analogous to the pervasive use of pictographic icons in computer operating systems introduced about 3 decades ago.  I acknowledge that such systems don't (as far as I know) provide for visual-based search based on icon appearance to find associated information (e.g., apps), and that system icons of this type are too free-form in appearance for such computer-mediated (rather than human-vision-based) searching to do much good anyway.  However, Unicode-character-based symbology -- while not based on accepted languages -- can be much more rigorously and systematically applied, and is obviously potentially amenable to supporting computer-mediated, text-based searching.

 

'robe070's hashtag ("#") example is only one of many possible uses of this approach, albeit probably the most straightforward one given its ubiquity in Twitter.  To illustrate, consider Unicode v7.  I've inferred from trial and error that -- naturally -- not all of the ~27,000 characters I've imported from the applicable Unicode database into a Google Sheet will be accepted and properly displayed within the title, body, AND tag(s) of Evernote notes and across multiple Evernote platforms/interfaces.  Still, consider that only ~18,000 of the aforementioned 27,000 characters are (generally) classed as L or N -- that leaves ~9,000 characters, or ~1/3 of the entire database -- outside those general classes.

 

The "abbreviated, symbolic modifier" method I previously described might be esoteric, and I admit that only a tiny proportion of such characters (including "#") are explicitly accessible via typical physical and (QWERTY-based) touch keyboards.  Still, I suspect that 'robe070', 'thebeefman', and I are far from the only three Evernote users who'd tangibly benefit from Evernote modifying their search methodology to allow explicit searches on at least some of those ~9,000 characters.

 

Regarding the references in this topic to "Lucene"...I'm not even close to a software developer / programmer.  However, to the extent I understood my recent quick scan of the Apache Lucene Project's documentation, it appears to me that this omission of all non-L/N characters except "_" relates to EN's particular implementation of Lucene, not the inherent nature of Lucene itself.  If anyone more knowledgeable about that particular subject disagrees, please let me know.

 

Finally, I've seen reference elsewhere to EN's beta-testing of a natural-language-type search process in its Mac client.  If this is something that EN intends to ultimately pursue for Windows desktop, Web, and perhaps iOS clients as well, I recommend that its final implementation accommodate/append broader capabilities in explicit searching for non-L/N characters, even if that particular feature might be considered far from "natural".

[Niente99 0]

You can try with this string (in Explorer search field):

 

~=#

 

For other special chars, use the expression ~=? (where ? is the required character)

[Gee_Cee 0]

I too would like to be able to search for special characters.  

I was simply trying to find a specific percentage, e.g. I was searching for "5%" but the search returned every note that contained a 5; the % was ignored.

Given that searching for content in notes and images is one of EN's key selling points, I was astonished that this didn't work.

Please EverNote Santa, can we have special character searching for Xmas?