The requirement came from @cpaulik in #39 (comment):

It would be great if there would be an option during import to use the existing folder names from Chrome as tags in buku.

Greetings,
buku --sany http://www.something.com/
fails while
buku --sany http://www.something.com
does not, even if the former is the url as displayed by buku.

It would be useful to have the option to add one or more tags to a record, without having to write the complete set of tags (along with -u / update command)

Hello there,
I have some weird issue on VoidLinux distro:

~ [ buku -p | tail -3                                                 ] 9:59 PM
  File "/usr/bin/buku", line 1535
    print('\nInterrupted.', file=sys.stderr)
                                ^
SyntaxError: invalid syntax

I would like to know, if there is problem with database, or the package is somehow wrong in VoidLinux:

pkgver: buku-2.5_1
repository: http://repo.voidlinux.eu/current

Idea of saving data to mysql database? LIVE database not on local pc

Hi @jarun

This is just a question why don't you put this project inside a proper python file structure please see link below.

https://github.com/kennethreitz/samplemod

Simple and easy to read and not have 1 file consisting 2921 lines of code.

not working due to:
_arguments:comparguments:319: invalid option definition: (-f --format)-f[limit fields in print and Json output:value

fixed with #110

When shortening the following URL:
https://launchpad.net/~twodopeshaggy/+archive/ubuntu/jarun/
we receive:
http://tny.im/HntZ2
which expands to
https://launchpad.net/~twodopeshaggy/archive/ubuntu/jarun/

Note the missing +.

I've recently installed Buku via Homebrew on OS X (zsh version 5.0.8) and using the provided makefile on Fedora 24 (zsh version 5.2). On both systems I get an error when trying to tab-complete:

_arguments:comparguments:313: invalid argument: (--merge){--merge}[merge another buku database]:buku db file

Deleting the line starting with "(--merge)" only causes the next line in the completion file to be reported. I came across a similar error with SVN completions, but the solution provided there was not helpful in case of Buku. I wonder if anyone else is experiencing this issue. I'm quite sure that its not my particular zsh configuration that's at fault, the error keeps popping out even when running a fresh shell with zsh --no-rcs.

with the git version of buku writing of tags is not handled correctly

With version 2.0:

carnager@caprica ~ > buku -a http://www.catsthatlooklikehitler.com/cgi-bin/seigmiaow.pl "hitler, cats, fun"
Title: [Cats That Look Like Hitler!]
Added at index 106

106. http://www.catsthatlooklikehitler.com/cgi-bin/seigmiaow.pl
    Cats That Look Like Hitler!
    [TAGS] hitler, cats, fun

With checkout from git:

carnager@caprica ~ > /home/carnager/Buku/buku -a http://www.catsthatlooklikehitler.com/cgi-bin/seigmiaow.pl "hitler, cats, fun"
106. http://www.catsthatlooklikehitler.com/cgi-bin/seigmiaow.pl
   > Cats That Look Like Hitler!
   #  cats, fun,hitler

The following command fails with Buku:

buku -w 'macvim -f'

As a result macvim opens in the background and exits.

Continued from #78.

Notes

The list below is a growing one. While suggesting new features please consider contributing to Buku. The code is intentionally kept simple and easy to understand with comments. We'll be happy to help out any new contributor.

Some of the fresh-baked features may not have been released yet. Grab the master branch for those.

Identified tasks

• Ubuntu Unity scope
• A browser plugin (probably a new project; see #122)
• Add more tests
• API documentation
• Rest API for webapps
• Android app (using the same database)
• Text editor support (thanks @ZwodahS)
• Need a PyPI maintainer
• Make refreshdb faster using threads (record updates should be synchronized).
• Show usage count in lag list
• Proxy support
• Continuous search at prompt
• Add prompt help
• Specify custom DB file to class BukuDb (library usage, no exposed option)
• Move to urllib3
• Handle redirects using referrer masking. Example URL. Fixed with urllib3.
• Support URL shortening. This helps to share URLs. (see #92 for limitations)
• Make a bookmark title immutable via refreshdb()
• Markdown import/export
• Regex search
• Ubuntu PPA
• Export specific tags to HTML
• Exact word match using REGEX. Make substring match optional.
• Delete all records based on a search result
• Delete multiple items, support combination of indices and ranges
• Append tags
• Travis CI integration
• Ubuntu deb package generation on new tag
• Merge bookmark database files (for users who work on multiple systems)
• Export bookmarks in FF or Chrome html format.
• Option to add folder names as tags while importing HTML (see #80)
• Implement self-upgrade (see #83)
• Anything else which would add value (please discuss in this thread)

Can we have an option "add link to buku" when right clicked on a link in firefox? The option can fire up terminal with the url took from firefox and we just have to add tags there.

With the support for URL shortening, why not introduce an option to store URLs directly shortened with --add option. Another improvement would be to (un)shorten URLs stored in DB with --update.

Hi,

Some years ago, Opera has been refactored completely. I did not like how bookmarks were reorganised, so I went to use Buku for bookmarking. After feeding Opera's bookmarks into Buku, I attempted to use Buku with Opera. To force Buku to open bookmarks in Opera, I am exporting BROWSER=opera. There are two generations of Opera I tried this with: Opera v12.16 (before the refactoring) and Opera v42 & v43 (recent). With the latter, when opening the bookmark of https://github.com/jarun/Buku, I am obtaining:

• In the address bar, http://openurl%28https//github.com/jarun/Buku [sic] is displayed;
• instead of the web content, This site can't be reached / openurl%28https's server DNS address could not be found. shows up.

This makes it impossible for me to use Buku with a recent Opera. The %28 corresponds to the character (: chr(0x28) is (. So it looks like that Buku is trying to tell Opera to openurl(https://github.com/jarun/Buku), however, it fails to do so. Nota bene that the colon : is missing in the https//github.com/jarun/Buku substring of the address bar content shown. I am not knowledgeable about how Buku instructs the browser of choice to open an URL. I am also not sure if the behaviour described here pertains to all users of Opera, or if it has something to do with my export BROWSER=opera approach described above. Buku is working fine on my Arch Linux machine with both recent versions of Firefox as well as with the ancient Opera v12.16 when using the abovementioned BROWSER=[...] method. So there are alternatives for me to using Buku with recent versions of Opera. However, I still would like to see this problem fixed to not have to work around it.

Friedrich

Hi there !

First, thanks for your work, it's very promising !

I'm trying to make a small nodejs server that calls buku to retrieve my bookmarks. For simplicity, it would be great if buku could return the bookmarks formatted in json through a command line flag !

Thanks in advance

Accidentally, by omitting the bookmark number, I updated the titles to all my records :) Not fun..
I believe the update command should ask for confirmation if no index is provided.
It would be very frustrating if you have 10K records..
Thanks,

It would be useful to delete multiple records at once:

• delete all records based on a search result
• delete multiple items eg: 1,2,3,10-30,45 (to delete records 1,2,3, from 10 to 30.. etc)

Hi there,
great work! Was searching a bit to get a quick workflow going to view/search/open bookmarks. Was going with buku-run as advised but found it already scraping the boundary of what poor old bash should be tasked with. Trying to extend it I was quickly in need to parse buku output.

With the pretty format being detrimental to stream processing, as it should be, I turned to JSON output and found that the index does not feature as a key. Although I am not sure I will really need it in the future I expect other people might.

Cheers

I really like the concept, but keeping it only as a CLI-utility makes it hard to do interesting things.

It would be cool if you could abstract out some of the complexity (db connections, file encryption etc) so that you end up with a simple object that can then be called from other scripts. I'm thinking something like

import buku
db = buku.BukuDb()
db.addItem(url, title, description, keywords)
items = db.search(param=value)

This would make it easier to write add-ons / import mechanisms and so on in Python, rather than having to pipe through the CLI (slow) or decipher all your internals (error-prone, inflexible).

Keep up the good work! 👍 👍

It would be nice to have regexp search in titles/keywords
Also, an idea would be to have an option to fetch the page content (eg. wget/w3m/curl) and search inside the body

With version 2.0 I can't seem to import any bookmarks.

carnager@caprica ~/Documents > buku -i bookmarks_5_26_16.html 
usage: buku [-a URL [tags ...]] [-u [N [URL tags ...]]]
            [-t [...]] [-d [N]] [-h]
            [-s keyword [...]] [-S keyword [...]]
            [-k [N]] [-l [N]] [-p [N]] [-f N]
            [-r oldtag [newtag ...]] [-j] [-o N] [-z]
buku: error: unrecognized arguments: -i bookmarks_5_26_16.html

Continued from #14.

Notes

The list below is a growing one. While suggesting new features please consider contributing to Buku. The code is intentionally kept simple and easy to understand with comments. We'll be happy to help out any new contributor.

Some of the just-completed features may not have been released yet. Grab the master branch for those.

Currently identified ToDos

• Implement self-upgrade (refer to googler source)
• Move to REPL style (re-use the current Argparser, make sure DB is consistent after each run). Branch - repl
• Load DB in memory, use encryption by default (unlock for a complete REPL session, lock before program exit)
• Option to add folder names as tags while importing HTML (refer to #80)
• Search as you type at REPL prompt
• Fuzzy search at REPL prompt
• PyPI packaging (branch pypi-packaging)
• Android app (using the same database)
• Markdown import/export
• Regex search
• Ubuntu PPA
• Export specific tags to HTML
• Add more tests (incremental @wheresmyjetpack, @poikjhn)
• Exact word match instead of substring match as it works currently. (Hint: REGEXP)
• Delete all records based on a search result
• Delete multiple items eg: 1,2,3,10-30,45 (to delete records 1,2,3, from 10 to 30.. etc)
• Append tags
• Travis CI integration
• Ubuntu deb package generation on new tag
• Merge bookmark database files (for users who work on multiple systems)
• Export bookmarks in FF or Chrome html format.
• Anything else which would add value (please discuss in this thread)

Hello,
When I add bookmarks, sometimes I get this error, and the entry is invalid

$ buku -a https://medium.com/@Spurrya/20-pcb-tips-and-tricks-47b95a4ba706#.4ksvpq9hi
297. https://medium.com/@Spurrya/20-pcb-tips-and-tricks-47b95a4ba706#.4ksvpq9hi
EXCEPTION [add_bookmark]: (UnicodeEncodeError) 'ascii' codec can't encode character '\u2014' in position 37: ordinal not in range(128)
dstoica@pentoo ~ $

$ buku -s ''
297. https://medium.com/@Spurrya/20-pcb-tips-and-tricks-47b95a4ba706#.4ksvpq9hi [297]
Traceback (most recent call last):
File "/usr/local/bin/buku", line 1466, in
bdb.searchdb(args.sany, False, jsonOutput)
File "/usr/local/bin/buku", line 402, in searchdb
prompt(results)
File "/usr/local/bin/buku", line 891, in prompt
print_record(row, count)
File "/usr/local/bin/buku", line 931, in print_record
print(' \x1B[91m>\x1B[0m %s' % row[2])
UnicodeEncodeError: 'ascii' codec can't encode character '\u2014' in position 37: ordinal not in range(128)

When adding a https url (https://www.example.com), urllib3 emits a warning that an Unverified HTTPS request is being made.

Buku version: 2.6

How to test:

$ buku -a https://www.example.com

Result:

$ buku -a https://www.example.com
/usr/lib/python3.5/site-packages/urllib3/connectionpool.py:821: InsecureRequestWarning: Unverified HTTPS request is being made. Adding certificate verification is strongly advised. See: https://urllib3.readthedocs.org/en/latest/security.html
  InsecureRequestWarning)
1. https://www.example.com
   > Example Domain

Possible solutions:

• Suppress warnings with urllib3.disable_warnings()
• Alternatively, add certificate verification (see https://urllib3.readthedocs.io/en/latest/user-guide.html#ssl

More information: https://urllib3.readthedocs.io/en/latest/advanced-usage.html#ssl-warnings

Would it be beneficial to verify the certificate or is it unnecessary?

Is there any way to make this sync with Chrome/Firefox/etc without having to manually import/export after each change?

buku.py is an UTF-8 file but setup.py opened it in ASCII mode instead of UTF-8 mode.

I: pybuild base:184: python3.5 setup.py clean
Traceback (most recent call last):
File "setup.py", line 13, in
version = re.search('version = '([^\']+)'', f.read()).group(1)
File "/usr/lib/python3.5/encodings/ascii.py", line 26, in decode
return codecs.ascii_decode(input, self.errors)[0]
UnicodeDecodeError: 'ascii' codec can't decode byte 0xc2 in position 65988: ordinal not in range(128)
E: pybuild pybuild:276: clean: plugin distutils failed with: exit code=1: python3.5 setup.py clean

I already sent the pull request. #101

It's possible to do the following:

$ buku -a http://google.com
343. http://google.com
   > Google

$ buku -a http://GOOGLE.com
344. http://GOOGLE.com
   > Google

I would like to propose some sort of import functionality.

For those of us with a significant bookmark collection, it might be easier than starting from scratch. The chrome export format could be used, or something else which is easily accessible.

Thanks for this tool, very useful.

It would be great if it could filter entries as-you-typed, similar to peco, specially for those with very large collections.

Hi! I just installed buku from Homebrew on a mac running Mac OS 10.11.15, and tried to import a bookmarks file exported from Chrome. Buku threw the following error:

usage: buku [-a URL [tags ...]] [-u [N [URL tags ...]]]
            [-t [...]] [-d [N]] [-h]
            [-s keyword [...]] [-S keyword [...]]
            [-k [N]] [-l [N]] [-p [N]] [-f N]
            [-r oldtag [newtag ...]] [-j] [-o N] [-z]
buku: error: unrecognized arguments: --import bookmarks_5_26_16.html

I also tried using -i instead of --import, with the same result, and checked the man page and found no mention of 'import'. Is this feature still supported?

Please refer to issue #16 raised by @nick-s-b: Also, buku doesn't seem to have a "comment" field. Most bookmark managers have that.

Breaking out as a separate request.

Note: This is my first interaction with an Open Source Community. My apologies if I happen to make any mistakes. I was directed to this project from http://up-for-grabs.net/#/

Getting to the point:
Operating System: OS X El Capitan Version 10.11
Python Version: 3.4.2

I followed the installation steps and got buku v1.9 running on a virtual environment.

• The first thing I did is added a sample bookmark as follows:

sagarbhuta$ buku -a -m 'Up for Grabs' http://up-for-grabs.net/#/ oss, projects

Output:

Added at index 1
1. http://up-for-grabs.net/#/
    Up for Grabs
    [TAGS] oss,projects

• The next thing I did was tried to remove the added bookmark as follows:

sagarbhuta$ buku -d 1

Output:

Removed index 1
Traceback (most recent call last):
  File "/usr/local/bin/buku", line 1224, in <module>
    cleardb(conn, cur, entry)
  File "/usr/local/bin/buku", line 599, in cleardb
    compactDB(conn, cur, int(index))
  File "/usr/local/bin/buku", line 566, in compactDB
    if row[0] > index:
TypeError: unorderable types: NoneType() > int()

• Trying to delete at index 1 after adding more than 1 bookmarks. Found out that the error exists only if there is just one bookmark in the database.

sagarbhuta$ buku -a -m 'Bookmark Manager' https://github.com/jarun/Buku command line, bookmarks, manager
Added at index 1

1. https://github.com/jarun/Buku
    Bookmark Manager
    [TAGS] command line,bookmarks,manager

sagarbhuta$ buku -a -m 'Up for Grabs' http://up-for-grabs.net/#/ oss, projects
Added at index 2

2. http://up-for-grabs.net/#/
    Up for Grabs
    [TAGS] oss,projects

sagarbhuta$ buku -d 1
Removed index 1
Index 2 moved to 1

sagarbhuta$ buku -g
     1. oss
     2. projects

sagarbhuta$ buku -d 1
Removed index 1
Traceback (most recent call last):
  File "/usr/local/bin/buku", line 1224, in <module>
    cleardb(conn, cur, entry)
  File "/usr/local/bin/buku", line 599, in cleardb
    compactDB(conn, cur, int(index))
  File "/usr/local/bin/buku", line 566, in compactDB
    if row[0] > index:
TypeError: unorderable types: NoneType() > int()

• I could reproduce the same error on buku v1.8 as well.

please implement password protection

This is a ToDo list item to support in-place upgrade of the program.

Hi,

I was wondering if you would consider using subcommands rather than command line flags for the various operations that buku provides.

For now, I found the usage section too complicated to read : I need to read the description to know how to perform an action.

Instead, you could have

• `buku add``
• buku rm|remove|delete
• buku tags
• buku open
• buku ls
• buku show

What do you think ?

What a great little utility! Really like how buku works. But to make it useful to me, I'd need to import all of my bookmarks in it somehow. Now that Delicious is dead (it's basically unusable), I’ve been looking for alternatives so I'm glad I came across this.

Since I have 40,000+ links in my bookmarks.html, adding them is a problem. I wish there was an automated way. Also, buku doesn't seem to have a "comment" field. Most bookmark managers have that.

Great util! Thank you for releasing it.

It imports the bookmarks.html file as a single list. Exporting these bookmarks from buku doesn't preserve the layout that was used in the originally exported bookmarks.html file.

Is the lack of this feature intentional (part of the design) or could it be added in the future?

buku shows id, url and tags in -p output.
Problem is that tags can have spaces, which makes proper parsing a mess.

A very easy solution would be to use tabs instead of spaces:
ID\tURL\tTAGS

Hey guys,

is it possible to set the tags in the markdown-file? I just only discovered the syntax (title)[url]. Or maybe with html?

thanks for your help,
Hans

$ buku -a "https://groups.google.com/forum/#!forums/iamthecavalry" personal, security
-bash: !forums/iamthecavalry": event not found

I tried wrapping the URL in "" and not, both ways get the same error. I don't find anything about this in the help or readme.md.

Any ideas? Thanks.

short:
Could you implement an option to simply allow the user to update the title manually?

long explenation:
A few years ago, I remapped the bookmarking hotkey in my .vimperatorrc for some sites, to a javascript snippet that serves as a shim. It fetches and then simplifies the title of the current page, and automatically applies tags, before it bookmarks that page using firefox's internal service. Since vimperator makes it is possible to execute some shell commands, I might try to change that to bookmark using Buku.

thanks for this tool!

I often mistype the names of the tags and additional effort is required to correct them (if discovered)
It would be useful to have bash completion for tags

On OS X El Capitan, I installed buku via brew and core functionality seems to work: adding, deleting, searching, and opening web pages. However, I cannot lock or unlock the database.

$ buku -z -l
[DEBUG] Version 2.5
[ERROR] cryptography lib(s) missing
$ buku -z -k
[DEBUG] Version 2.5
[ERROR] cryptography lib(s) missing

Continued from #39.

Notes

The list below is a growing one. While suggesting new features please consider contributing to Buku. The code is intentionally kept simple and easy to understand with comments. We'll be happy to help out any new contributor.

Some of the just-completed features may not have been released yet. Grab the master branch for those.

Identified tasks

• Need a PyPI maintainer
• Make refreshdb faster using threads (record updates should be synchronized).
• API documentation
• Add more tests
• Android app (using the same database)
• Rest API for webapps
• Show usage count in lag list
• Proxy support
• Continuous search at prompt
• Add prompt help
• Specify custom DB file to class BukuDb (library usage, no exposed option)
• Move to urllib3
• Handle redirects using referrer masking. Example URL. Fixed with urllib3.
• Support URL shortening. This helps to share URLs. (see #92 for limitations)
• Make a bookmark title immutable via refreshdb()
• Markdown import/export
• Regex search
• Ubuntu PPA
• Export specific tags to HTML
• Exact word match using REGEX. Make substring match optional.
• Delete all records based on a search result
• Delete multiple items, support combination of indices and ranges
• Append tags
• Travis CI integration
• Ubuntu deb package generation on new tag
• Merge bookmark database files (for users who work on multiple systems)
• Export bookmarks in FF or Chrome html format.
• Option to add folder names as tags while importing HTML (see #80)
• Implement self-upgrade (see #83)
• Anything else which would add value (please discuss in this thread)

When there is more than one tag specified for deletion while deleting tags from all bookmarks, there is some (possibly) unintended behavior. The behavior is that not all tags specified are getting deleted.

I have only been able to replicate this when all of the following conditions are true:

• when the to-be-deleted string contains more than one tag
• when one of the tags in the to-be-deleted string matches a tag in only one bookmark
• one of the following:
  • when at least one bookmark contains multiple tags specified by the string
  • - or - when each specified tag matches a tag for a different bookmark

It looks like if the string meets these conditions, then only the last tag in the string is deleted (last tag being the last one alphabetically, after input has been parsed and sorted). If the tag second-to-last in the order is found for a bookmark which also contains the last tag, then this penultimate tag is also deleted. However, if that penultimate tag exists for other bookmarks, those instances of the tag are not deleted.

The following command deletes search results though it should delete the bookmark at DB index 5:

$ buku -s hello -d 5

Search results should be deleted only if -d is passed without a value, like:

$ buku -s hello -d

Otherwise, it should process delete option separately.

• Add more tests
• PyPI packaging
• Exact word match (instead of substring match as it works currently. Hint: REGEXP)
• Travis CI integration
• Ubuntu deb package generation on new tag
• Merge bookmark database files (for users who work on multiple systems)
• Export bookmarks in FF or Chrome html format.
• Anything else which would add value (please discuss)

This is the scratchpad I had with the envisioned endpoints:

get     api/bookmarks/firefox.html  # export to html
get     api/bookmarks/firefox.md  # export to markdown

get     api/bookmarks     #get all bookmarks
post    api/bookmarks     #new bookmark
delete  api/bookmarks     #delete table
post    api/bookmarks/refresh     #refresh all bookmarks

get     api/bookmarks/search   # get all bookmarks that match search
put     api/bookmarks/search   # update all bookmarks that match search
delete  api/bookmarks/search  # delete all bookmarks that match search

get     api/bookmarks/:id   # get bookmark with given id
put     api/bookmarks/:id   # update bookmark with given id
delete  api/bookmarks/:id   # delete bookmark with given id
post    api/bookmarks/:id/refresh   # refresh bookmark with given id
get     api/bookmarks/:id/tiny   # tinyurl of bookmark with given id
get     api/bookmarks/:id/long   # expanded url of bookmark with given id

get     api/bookmarks/:id/:id   # get bookmark in the given range
put     api/bookmarks/:id/:id   # update bookmark in the given range
delete  api/bookmarks/:id/:id   # delete bookmark in the given range
post    api/bookmarks/:id/:id/refresh   # refresh bookmark in the given range

get     api/tags/   # get tags
put     api/tags/:tag   # replace tag

Edit: There are still things that are not yet clear, but I was thinking this was a good starting subset:

get     api/bookmarks
post    api/bookmarks

get     api/bookmarks/:id
put     api/bookmarks/:id
delete  api/bookmarks/:id

get     api/tags/

All of the short arg options have a single hyphen if they support a long and short form. Example:

-a, --add URL [tags ...]

But currently the search tags option lists the following options:

--st, --stag [...] search bookmarks by tag list tags alphabetically, if no arguments

It appears to be the only option where the short option enforces two hyphens as opposed to 1 hyphen.

I attempted to use buku -st to list my tags and saw no output; while I expected the option to work. Perhaps this was a simple typo in the sourcecode? I can supply a PR later today.