From 5e1c75a632caeed5c53eda2963170a5713142091 Mon Sep 17 00:00:00 2001
From: Johann Jacobsohn <j.jacobsohn@satzmedia.de>
Date: Tue, 4 Aug 2020 09:49:08 +0200
Subject: [PATCH] add experimental export-to-bibtex support
---
papersurfer.py | 102 +++++++++++++++++++++++++++++++++++++++++++++----
1 file changed, 94 insertions(+), 8 deletions(-)
diff --git a/papersurfer.py b/papersurfer.py
index 691fda4..bd395c4 100644
--- a/papersurfer.py
+++ b/papersurfer.py
@@ -13,9 +13,12 @@ import subprocess
from dataclasses import dataclass
import re
from functools import partial
+import requests
from mattermostdriver import Driver
import urwid
import configargparse
+from pybtex.database import BibliographyData, Entry
+import xml.etree.ElementTree as ET
URL = "mattermost.cen.uni-hamburg.de"
CHANNEL = "n5myem9yc7fyzb9am7ym5o41ry"
@@ -29,11 +32,60 @@ class PostDTO:
reporter: str
doi: str
+@dataclass
+class PaperDTO:
+ """"Encapsulate Mattermost Posts."""
+ author: str
+ title: str
+ journal: str
+ year: int
+ doi: str
+
+
+class Bibtex:
+ def entry(self, author, title, journal, year):
+ return BibliographyData({
+ f'{author}{year}': Entry('article', [
+ ('author', author),
+ ('title', title),
+ ('journal', journal),
+ ('year', year),
+ ])
+ }).to_string('bibtex')
+
+ def entry_from_doi(self, doi):
+ paper = Doi().get_info(doi)
+ return self.entry(paper.author, paper.title, paper.journal, paper.year)
+
+ def bib_from_dois(self, dois):
+ return "\n".join([self.entry_from_doi(doi) for doi in dois])
+
class Doi:
"""Interface w/ the doi.org api"""
def get_doi_link(self, doi):
- return f"http://doi.org/$doi"
+ """Assemble doi link."""
+ return f"http://doi.org/{doi}"
+
+ def load_doi_data(self, doi):
+ headers = {
+ 'Accept': 'application/vnd.crossref.unixsd+xml',
+ }
+ return requests.get(f'http://dx.doi.org/{doi}', headers=headers).content
+
+ def parse_doi_xml(self, xml):
+ root = ET.fromstring(xml)
+ author = root.find(".//{http://www.crossref.org/xschema/1.1}surname").text # fixme
+ title = root.find(".//{http://www.crossref.org/xschema/1.1}title").text
+ journal = root.find(".//{http://www.crossref.org/xschema/1.1}full_title").text
+ year = root.find(".//{http://www.crossref.org/xschema/1.1}year").text
+ doi = root.find(".//{http://www.crossref.org/xschema/1.1}year").text
+
+ return PaperDTO(author, title, journal, year, doi)
+
+ def get_info(self, doi):
+ xml = self.load_doi_data(doi)
+ return self.parse_doi_xml(xml)
def extract_doi(self, hay):
"""Parse doi from string, or None if not found.
@@ -41,7 +93,8 @@ class Doi:
>>> Doi().extract_doi("https://doi.org/10.1093/petrology/egaa077")
'10.1093/petrology/egaa077'
"""
- matches = re.compile(r'\b10\.\d{4,9}/[-._;()/:A-Z0-9]+', re.I).search(hay)
+ pattern = r'\b10\.\d{4,9}/[-._;()/:A-Z0-9]+'
+ matches = re.compile(pattern, re.I).search(hay)
return matches.group() if matches else None
@@ -57,10 +110,11 @@ class Mattermost:
self.mattermost.login()
self.reporters = {}
-
def get_reporter(self, id):
+ """Load user from mattermost api and cache."""
if id not in self.reporters:
self.reporters[id] = self.mattermost.users.get_user(id)["username"]
+
return self.reporters[id]
def retrieve_all_messages(self):
@@ -96,30 +150,35 @@ class Papersurfer:
def __init__(self, username, password):
self._screen = urwid.raw_display.Screen()
self.size = self._screen.get_cols_rows()
+ self.filter = ""
palette = [
('I say', 'default,bold', 'default', 'bold'),
('needle', 'default, bold, underline', 'default', 'bold')]
ask = urwid.Edit(('I say', u"Filter?\n"))
exitbutton = urwid.Button(u'Exit')
+ self.exportbutton = urwid.Button(u'Export filtered list as bibtex')
div = urwid.Divider(u'-')
self.mtm = Mattermost(username, password)
- body = [self.listItem(paper) for paper in self.mtm.retrieve()]
+ body = [self.list_item(paper) for paper in self.mtm.retrieve()]
self.listcontent = urwid.SimpleFocusListWalker(body)
paperlist = urwid.BoxAdapter(urwid.ListBox(self.listcontent),
self.size[1] - 5)
- pile = urwid.Pile([ask, div, paperlist, div, exitbutton])
+ pile = urwid.Pile([ask, div, paperlist, div,
+ urwid.Columns([exitbutton, self.exportbutton])])
top = urwid.Filler(pile, valign='middle')
urwid.connect_signal(ask, 'change', self.onchange)
urwid.connect_signal(exitbutton, 'click', self.on_exit_clicked)
+ urwid.connect_signal(self.exportbutton, 'click', self.on_export_clicked)
- urwid.MainLoop(top, palette).run()
+ self.mainloop = urwid.MainLoop(top, palette)
+ self.mainloop.run()
- def listItem(self, paper, needle=""):
+ def list_item(self, paper, needle=""):
"""Create highlighted text entry."""
text_items = []
needle = needle or "ßß"
@@ -143,16 +202,43 @@ class Papersurfer:
pile = urwid.Pile([title, button_bar, urwid.Divider()])
return pile
+ def updscrn(self):
+ self.mainloop.draw_screen()
+
def onchange(self, _, needle):
"""Handle filter change."""
+ self.filter = needle
self.listcontent.clear()
- self.listcontent.extend([self.listItem(paper, needle)
+ self.listcontent.extend([self.list_item(paper, needle)
for paper in self.mtm.get_filtered(needle)])
+ def running_export(self, state):
+ btn = self.exportbutton
+ label = btn.get_label()
+ running_indicator = " (running...)"
+ if state:
+ btn.set_label(label + running_indicator)
+ else:
+ btn.set_label(label.replace(running_indicator, ""))
+ self.updscrn()
+
def on_exit_clicked(self, button):
"""Handle exitbutton click and exit."""
raise urwid.ExitMainLoop()
+ def on_export_clicked(self, _):
+ """Handle exitbutton click and exit."""
+ self.running_export(True)
+ self.export_to_bibtex()
+ self.running_export(False)
+
+ def export_to_bibtex(self):
+ papers = self.mtm.get_filtered(self.filter)
+ dois = [paper.doi for paper in papers]
+ string = Bibtex().bib_from_dois(dois)
+ with open("export.bib", 'w') as f:
+ f.write(string)
+
def h_open_discussion(self, post, _):
"""Handle click/enter on discussion button."""
self.open_discussion(post)
--
GitLab