nbibToBib converts an nbib reference file to a bibtex file.
Note updates I have made are at the bottom of this readme
- bibutils https://github.com/biodranik/bibutils
- At a glance, this looks like rather robust code, but it does not have the same level of flexiability for deciding which entries to print out. However, it does have a bit more flexibility in controlling if you prefere braces or "" and does have some white space settings.
- One upside is that this code was designed to be a general conversion tool, so it supports more than just bibtex conversion. The downside ofto this is that the user needs to run two programs. One to convert the nbib to xml and another to conver the xml to a bibtex file. Best to make a bash script for this.
- nbib2bib https://github.com/thiagonds65/nbib2bib
- This is a python script which converts all nbib files in a directory to bibtex files.
- This works, but has no flexibility for choosing the output format.
- It uses the doi entry to get the bibtex file, which allows it to get the month entry that both nbibToBib and bibutils miss for PMID 22078026. However, this also causes it to run slowly, 6 seconds for all eight of its test cases. In comparision both nbibToBib and bibutils takes less than a second for all eight test cases.
- Finally their are plenty of PMID to bibtex programs
that can convert a pmid to a bibtex entry for you.
- For example https://gist.github.com/tommycarstensen/ec3c57761f3846c339de925b66f4ac1b took 1 second to output the bibtex entries for all of nbib2bib's test cases.
Entries extracted from the pubmed file are: author names (full only), title, journal name, year, journal id, month, day, volume, issue, doi, page number (non-electronic prioritized), edition, PMID, PMC id, ISBN, ISSN (linking/paper prioritized), abstract, abbreviated journal name, article type, language, mesh terms, and keywords.
The default printed extracted entries are: author names (full only), title, journal name, year, journal id, month, day, volume, issue, doi, page number (non-electronic prioritized), edition, PMID, PMC id, ISSN (linking/paper prioritized), abstract, and keywords.
Other entries printed include the full doi address as the url and some of my own tags (file, tags, supplemental, and notesp)
I should note that this extracts the date of publication using the DP (date of publication) entry instead of the DEP (date of electronic publication) entry. There are times were the DP entry will have less information than the DEP entry.
The .nbib file can have multiple articles, so long as each articles entry is separated by a blank line. See test/test-case.nbib for an example. You can merge individual nbib files into a single nbib file with new lines at the end of each reference using sed.
sed '/SO/{a\
};' *.nbib > merged.nbib;
Nbib files can be downloaded from a pubmed search by save (under search box)->format->selecting the PubMed format->create file. The file will be saved as a .txt file, but is an .nbib file.
I suspect most people using this code will not need this
diagram, but I included it just in case someone might
find it useful.
Visual of downloading a .nbiv file from pubmed.
-----------------------------------------------------------
Step 1: Select save
-----------------------------------------------------------
+-----\\
| NIH ++ National Library of Medicine
+-----// National Center for Biotechnology information
+-------------------------------+
PubMed | Search term |
+-------------------------------+
Advanced Create alert Creat RSS
+------+
**CLICK**->| Save |
+------+
-----------------------------------------------------------
Step 2: Select PubMed (nbib) output
-----------------------------------------------------------
+-----\\
| NIH ++ National Library of Medicine
+-----// National Center for Biotechnology information
+-------------------------------+
PubMed | Search term |
+-------------------------------+
Advanced Create alert Creat RSS
+------+
| Save |
+------+
Save citations to file
+--------------------------+
Seletion: | All results on this page |
+--------------------------+
+-----------------+
Format: | Summary (text) |
**SELECT**---------->| PubMed |
| PMID |
| Abstract (text) |
| CSV |
+-----------------+
+-------------+
| Create file |
+-------------+
-----------------------------------------------------------
Step 3: Select create file
-----------------------------------------------------------
+-----\\
| NIH ++ National Library of Medicine
+-----// National Center for Biotechnology information
+-------------------------------+
PubMed | Search term |
+-------------------------------+
Advanced Create alert Creat RSS
+------+
| Save |
+------+
Save citations to file
+--------------------------+
Seletion: | All results on this page |
+--------------------------+
+-----------------+
Format: | PubMed |
+-----------------+
+-------------+
**CLICK**->| Create file |
+-------------+
NbibToBib depends on unac to remove accents from the first authors last name when building the citation key for the bibtex file. This feature can be removed by running make normaccent. However, this means that the citation key will only use the year and pubmed id.
The make file will download both unac and libiconv-1.9.1, which unac depends on with curl and git. The unac_version function is removed from the unac.c/h files (prevents errors). libiconv and unac are then used to build a static build of nbibToBib.
You can change the gcc complier used to build nbibToBi with make CC=newGCC. The default is cc.
Dependencies:
- unac (Downloaded during installation)
- https://github.com/QuickDict/unac
- Removed with make normaccent
- QuickDict is under the GLP 2 license, so use
make normaccentif you do not want an GLP 2 dependency. That is why I have this program download quickDict and not have it here. If I could get around quickDict I would.
- libconiv-1.9.1 (Download during installion)
- ftp://ftp.gnu.org/pub/gnu/libiconv/libiconv-1.9.1.tar.gz
- Removed with
make normaccent
- unix cmd programs used in build (removed by
make normaccent)
- curl: Used to download libiconv-1.9.1
- git: Used to download unac
- sed: Removes unac_version function from unac.c/h
- tar: Extract libiconv files
- Modern gcc compiler
# Building on Linux
make;
# Building on OpenBsd
make CC=egcc; # Changing the default compiler
# Linux or openbsd no unac support
make normaccent # for a no dependecy build
#install
make install cleanall
# - cleanall removes libiconv/unac files
# - make clean just removes the extracted libiconv files
# but still keeps the tar.gz file. This is nice if you
# plan to compile this program multiple times, since
# you will not have to redownload libiconv and unac.
# install to an alternate location
make PREFIX=/alternate/location/ install clean
nbibToBib -nbib file.nbib > file.bib
# Show the help message
nbibToBib -h | less
# Show the message to change entries printed to bibtex
nbibToBib -bib-help | less
The help message for nbibToBib can be printed with -h. A further help message showing the fine tuning options, which control what is printed out can be called with -bib-help.
Each of the parameters printed by -bib-help have two forms. The first is the -p form, which sets the entry to be printed out. The second is the -no form, which sets the entry to not be printed out. However, there are no entry parameters for entries required by the bibtex format (title, author, journal, and year). These required entries are always printed out.
These values can also be changed at compile time be editing nbibToBib-settings.h. To enable an option change defPXBl, were x is the entry to enable to 1. To disable an option change defPXBl, were x is the entry to disable to 0 (these entries are on lines 14 to 43). This will mess up the help message a bit, but works. The -bib-help message is printed out by Fun-06 in nbibToBib.c (this is at the bottom of the document).
The output line wrapping of the bibtex file can be set with -line-wrap. This variable only affects the title, author, journal name, keywords, mesh terms, article type, tags, and abstract entries.
The -tag option adds a entry into the tags={}, entry. This is my way of adding in key words into the file without adding to the authors keywords.
You have two options for how lines end. The unix option (-break-unix), which uses a new line only or the windows option (-break-win), which uses a carriage return and new line. It may be possible that the current windows may allow the unix option to be used. Unix can handle both options.
The input file is specified by -pubmed and the output file is specified by -bib.
- Fixed an error were the bibtex entry type was printed as
"@Type={key," instead of "@Type{key,". This error can
be fixed with
sed 's/\(^@.*\)={/\1{/' file.bib > tmp.bib; mv tmp.bib file.bib; - Added in support for getting the ppi entry "-p-ppi". This is off by default.
- Added in support for detecting doi's in the AID entry. I found out the LID entry is not always present when a doi is present.
- Made the help message check the default values before
printing. This means the help message will reflect the
changes you made to the default values in
nbibToBibSettings.h
- Also added in a default line break setting option to nbibToBibSettings (defBreakType). So, you can choose now if you linux ("\n" set to 0) or windows ("\r\n" set to 1) as the default. Any other value goes to linux.
- Fixed an error were a comma was not put after the
citation key
- This error can be corrected in older versions with
sed 's/^\(@.*={.*\)/\1,/' file.bib > tmp.bib; mv tmp.bib file.bib;
- This error can be corrected in older versions with