commit 5e8333356b664fa7b394fbcc590d2d2ba91b0a30
parent fa0603a5aca02ab8cd255741805c9bcd6bd301c8
Author: Anders Damsgaard <anders@adamsgaard.dk>
Date: Wed, 10 Apr 2019 13:56:42 +0200
Rewrite scholarref functionality to UNIX style, update Vim bindings
Diffstat:
9 files changed, 150 insertions(+), 142 deletions(-)
diff --git a/.commands.sh b/.commands.sh
@@ -22,7 +22,6 @@ c() {
## ls
-
if [[ "$(uname)" != 'Darwin' ]]; then
alias ls='ls --color=auto -F'
else
diff --git a/.config/newsboat/config b/.config/newsboat/config
@@ -67,3 +67,4 @@ macro r set browser "rtv"; open-in-browser ; toggle-article-read ; set browser o
# macro p set browser "dmenuhandler"; open-in-browser ; set browser xdg-open
macro y set browser "echo -n %u | xclip -selection clipboard" ; open-in-browser ; set browser xdg-open
macro p set browser "wkhtmltopdf %u out.pdf && mv out.pdf ~/tmp/$(basename $(mktemp)).pdf" ; open-in-browser ; set browser xdg-open
+macro s set browser "cd ~/tmp && tsp shdl -t %u" ; open-in-browser ; set browser xdg-open
diff --git a/.vim/plugin/keybinds.vim b/.vim/plugin/keybinds.vim
@@ -54,9 +54,9 @@ nnoremap <leader>CA :e ~/.vim/plugin/appearance.vim<CR>
nnoremap <leader>CG :e ~/.vim/plugin/00-general.vim<CR>
nnoremap <leader>CP :e ~/.vim/plugins.vim<CR>
nnoremap <leader>T :e ~/doc/todo.md<CR>
-nnoremap <leader>B :e ~/articles/own/BIB.bib<CR>
+nnoremap <leader>B :e ~/articles/own/BIBnew.bib \| :windo normal G<CR>
-nnoremap <leader>r :ScholarRef
+nnoremap <leader>r :read !scholarref
" insert date and time
nnoremap <leader>D :read !date +"\# \%Y-\%m-\%d \%T \%Z (\%z UTC)"<CR>
diff --git a/.vim/plugins.vim b/.vim/plugins.vim
@@ -89,7 +89,6 @@ Plug 'vim-scripts/mutt-aliases', { 'for': 'mail' } " complete aliases w C-x C-u
" LaTeX
Plug 'lervag/vimtex'
-Plug 'anders-dc/vim-scholarref' " search bibtex refs from web
" Vimscript
Plug 'junegunn/vader.vim' " testing framework
diff --git a/bin/getcitation b/bin/getcitation
@@ -15,6 +15,7 @@ function show_help {
echo " -t, --tor-socks use torsocks for HTTP requests"
echo " -j, --full-journal return full journal name in citation"
echo " -a, --full-author do not shorten author names"
+ echo " -n, --no-newline suppress trailing newline but prepend with newline"
echo " -- do not consider any following args as options"
}
@@ -68,6 +69,9 @@ function abbreviate_journal_name {
s/Communications/Commun./
s/Geoscience/Geosci./
s/Landforms/Land./
+ s/ of / /
+ s/ and / /
+ s/ in / /
}'
}
@@ -89,12 +93,17 @@ function get_citation {
result="$(echo "$result" | format_bibtex_key)"
[ "$fulljournal" = 0 ] && result="$(echo "$result" | abbreviate_journal_name)"
[ "$fullauthor" = 0 ] && result="$(echo "$result" | abbreviate_author_name)"
- echo "$result"
+ if [ "$nonewline" = 1 ]; then
+ echo -en "\n$result"
+ else
+ echo "$result"
+ fi
}
verbose=0
fulljournal=0
fullauthor=0
+nonewline=0
prefix=""
while :; do
case "$1" in
@@ -112,6 +121,9 @@ while :; do
-a|--full-author)
fullauthor=1
;;
+ -n|--no-newline)
+ nonewline=1
+ ;;
-V|--verbose)
verbose=1
;;
diff --git a/bin/getdoi b/bin/getdoi
@@ -83,8 +83,8 @@ while :; do
done
if [ $# -lt 1 ]; then
- doi="$(cat)"
- get_doi "$doi"
+ query="$(cat)"
+ get_doi "$query"
else
get_doi "$@"
fi
diff --git a/bin/scholarref b/bin/scholarref
@@ -1,3 +1,3 @@
#!/usr/bin/env bash
set -e
-getdoi "$@" | getcitation
+getdoi "$@" | getcitation -n
diff --git a/bin/scholarref.py b/bin/scholarref.py
@@ -1,120 +0,0 @@
-#!/usr/bin/env python2
-"""
-scholarref.py:
-
-This script searches an online database and returns a full BibTeX reference to
-the command line for the publication most closely matching the query.
-
-Usage:
- ./scholarref.py <query>
-
-Example:
- ./scholarref.py water flow in glaciers
- @incollection{Hooke,
- doi = {10.1017/cbo9780511614231.012},
- publisher = {Cambridge University Press ({CUP})},
- pages = {197--251},
- author = {Roger LeB. Hooke},
- title = {Water flow in and under glaciers: geomorphic implications},
- booktitle = {Principles of Glacier Mechanics}
- }
-
-Requirements are the `requests` Python library which can be installed with `pip
-install requests`.
-
-If you use Vim as your text editor, you can directly insert the output into the
-current document with the following binding:
- nmap <leader>r :read !scholarref.py
-
-Written by Anders Damsgaard, derived from doi2bib.py by Andrew Ning.
-"""
-
-import requests
-import sys
-import re
-
-verbose = False
-
-query = ' '.join(sys.argv[1:])
-nhits = 1
-
-# Use crossref metadata search (beta) to get the DOI
-params = {'q': query, 'rows': str(nhits)}
-doi_search_url = 'http://search.crossref.org/dois'
-r = requests.get(doi_search_url, params=params)
-
-dois = []
-try:
- for j in r.json():
- dois.append(j['doi'].split('dx.doi.org/')[1])
-except Exception as e:
- print('Error: ' + str(e))
- print('Maybe DOI service at ' + doi_search_url + ' is unavailable?')
- print('Server response: ' + str(r))
- sys.exit(1)
-
-# get the DOI of the first result
-doi = dois[0]
-if verbose:
- print('Received DOI from ' + doi_search_url)
-
-# fix escaped chars
-doi = doi.replace('\\', '')
-
-
-# use REST API (see http://crosscite.org/cn/)
-headers = {'Accept': 'application/x-bibtex'}
-r = requests.post('http://dx.doi.org/' + doi, headers=headers)
-if verbose:
- print('Received bibtex entry from dx.doi.org')
-
-# extract bibtex
-r.encoding = 'utf-8'
-bibtex = r.text
-bibtex = bibtex.replace('&', '&')
-bibtex = bibtex.replace('\t', ' ')
-bibtex = bibtex.replace('_', '', 1) # strip _ from cite key
-bibtex = re.sub('.*url.*\n', '', bibtex) # remove url field
-bibtex = re.sub('.*month.*\n', '', bibtex) # remove month field
-
-# appreviate journal names
-bibtex = bibtex.replace('Journal', 'J.')
-#bibtex = bibtex.replace('of ', '')
-#bibtex = bibtex.replace('and ', '')
-#bibtex = bibtex.replace('the ', '')
-bibtex = bibtex.replace('Geophysical', 'Geophys.')
-bibtex = bibtex.replace('Geophysics', 'Geophys.')
-bibtex = bibtex.replace('Research', 'Res.')
-bibtex = bibtex.replace('Letters', 'Lett.')
-bibtex = bibtex.replace('Mechanics', 'Mech.')
-bibtex = bibtex.replace('Glaciology', 'Glaciol.')
-bibtex = bibtex.replace('Proceedings', 'Proc.')
-bibtex = bibtex.replace('Royal', 'R.')
-bibtex = bibtex.replace('Society', 'Soc.')
-bibtex = bibtex.replace('Annals', 'Ann.')
-bibtex = bibtex.replace('Resources', 'Resour.')
-bibtex = bibtex.replace('Surface', 'Surf.')
-bibtex = bibtex.replace('Processes', 'Proc.')
-bibtex = bibtex.replace('National', 'Nat.')
-bibtex = bibtex.replace('Computers', 'Comput.')
-bibtex = bibtex.replace('Geotechnics', 'Geotech.')
-bibtex = bibtex.replace('Academy', 'Acad.')
-bibtex = bibtex.replace('Sciences', 'Sci.')
-bibtex = bibtex.replace('Review', 'Rev.')
-bibtex = bibtex.replace('Quaternary', 'Quat.')
-bibtex = bibtex.replace('Physical', 'Phys.')
-bibtex = bibtex.replace('Planetary', 'Planet.')
-bibtex = bibtex.replace('Quarterly', 'Q.')
-bibtex = bibtex.replace('Geological', 'Geol.')
-bibtex = bibtex.replace('Applied', 'Appl.')
-bibtex = bibtex.replace('Physics', 'Phys.')
-bibtex = bibtex.replace('Communications', 'Commun.')
-bibtex = bibtex.replace('Geoscience', 'Geosci.')
-bibtex = bibtex.replace('Landforms', 'Land.')
-
-bibtex = bibtex.strip()
-
-if bibtex.find('<!DOCTYPE html>') != -1:
- sys.stderr.write('Entry not found\n')
-else:
- print(bibtex.encode('utf-8'))
diff --git a/bin/shdl b/bin/shdl
@@ -1,21 +1,138 @@
#!/usr/bin/env bash
-set -e
-read -p "use torsocks? [Y/n] " -n 1 -r
-echo
-if [[ $REPLY =~ ^[Nn]$ ]]; then
- prefix=""
-else
- prefix="torsocks"
-fi
+version=1.0
+shhost="http://sci-hub.tw"
-shdl_fetch() {
- url="$($prefix curl --connect-timeout 30 --max-time 60 --silent \
- "http://sci-hub.tw/$*" | \
+function show_help {
+ echo "usage: ${0##*/} [OPTIONS] DOI1 [DOI2...[DOI N]]"
+ echo "will fetch each DOI from $shhost"
+ echo "If no DOI is specified, this program will expect DOIs as stdin."
+ echo "The program extracts the DOI if a full doi.org URL is passed to it."
+ echo "If the PDF download fails, e.g. due to captias, a tor-browser or "
+ echo "alternatively the system default browser will open sci-hub."
+ echo
+ echo "OPTIONS are one or more of the following:"
+ echo " -h, --help show this message"
+ echo " -v, --version show version and license information"
+ echo " -V, --verbose show verbose information"
+ echo " -t, --tor-socks use torsocks for requests to sci-hub"
+ echo " -b, --browser open sci-hub in browser immediately"
+ echo " -r, --reference add reference to bibliography using scholarref.py"
+ echo " -- do not consider any following args as options"
+}
+
+function show_version {
+ echo "${0##*/} version $version"
+ echo "Licensed under the GNU Public License, v3+"
+ echo "written by Anders Damsgaard, anders@adamsgaard.dk"
+ echo "https://gitlab.com/admesg/dotfiles"
+}
+
+function strip_doi {
+ sed 's/^(http:\/\/|https:\/\/)?(dx\.)?(doi\.org\/)//'
+}
+
+function browser_open {
+ if command -v tor-browser 2>/dev/null; then
+ [ "$verbose" = 1 ] && echo "attempting to launch tor-browser"
+ tor-browser "$1"
+ elif command -v xdg-open 2>/dev/null; then
+ [ "$verbose" = 1 ] && echo "launching default browser"
+ xdg-open "$1"
+ elif command -v open 2>/dev/null; then
+ [ "$verbose" = 1 ] && echo "launching default browser"
+ open "$1"
+ else
+ die 'Error: could not open a browser'
+ fi
+}
+
+function shdl_fetch {
+ doi=$(echo "$1" | strip_doi)
+ shurl="$shhost/$doi"
+ [ "$verbose" = 1 ] && echo "processing $doi"
+ [ "$verbose" = 1 ] && echo "connecting to $shurl $prefix"
+ pdfurl="$($prefix curl --connect-timeout 30 --silent "$shurl" | \
grep location.href | \
grep -o '//.*pdf' | sed 's/^/http:/')"
- echo "$url"
- $prefix curl --connect-timeout 30 --max-time 60 -O "$url"
+ [ "$verbose" = 1 ] && echo "fetching pdf from $pdfurl"
+ if ! $prefix curl --connect-timeout 30 -O "$pdfurl"; then
+ if [ "$verbose" = 1 ]; then
+ (echo "Error: could not fetch $doi PDF from $pdfurl" >&2)
+ fi
+ browser_open "$shurl"
+ fi
+}
+
+function handle_doi {
+ if [ "$browser" = 1 ]; then
+ browser_open "$1"
+ else
+ shdl_fetch "$1"
+ fi
+ if [ "$reference" = 1 ]; then
+ if command -v scholarref.py 2>/dev/null; then
+ if [ "$verbose" = 1 ]; then
+ echo "adding $1 to bibliography using scholarref.py"
+ fi
+ scholarref.py "$1"
+ else
+ die 'Error: scholarref.py not found in PATH'
+ fi
+ fi
+}
+
+function die {
+ printf '%s\n' "$1" >&2
+ exit 1
}
-shdl_fetch "$@"
+returnstatus=0
+prefix=""
+verbose=0
+browser=0
+reference=0
+while :; do
+ case "$1" in
+ -h|-\?|--help)
+ show_help
+ exit 0
+ ;;
+ -v|--version)
+ show_version
+ exit 0
+ ;;
+ -V|--verbose)
+ verbose=1
+ ;;
+ -b|--browser)
+ browser=1
+ ;;
+ -t|--tor-socks)
+ prefix="torsocks"
+ ;;
+ -r|--reference)
+ reference=1
+ ;;
+ --) # end all options
+ shift
+ break
+ ;;
+ -?*)
+ die 'Error: Unknown option specified'
+ ;;
+ *) # No more options
+ break
+ esac
+ shift
+done
+
+if [ $# -lt 1 ]; then
+ doi="$(cat)"
+ handle_doi "$doi"
+else
+ for doi in "$@"; do
+ handle_doi "$doi"
+ done
+fi
+exit "$returnstatus"
