From 6b1b2be0f19bc44cc5994734dbb4f56a201a7acf Mon Sep 17 00:00:00 2001
From: Mathieu Loiseau <mathieu.loiseau@univ-grenoble-alpes.fr>
Date: Sun, 2 May 2021 00:27:25 +0200
Subject: [PATCH] marche presque
---
extract_notes.py | 154 +++++++++++++++++++++++++++++++++++++----------
1 file changed, 123 insertions(+), 31 deletions(-)
diff --git a/extract_notes.py b/extract_notes.py
index 3096a1c..6dabbb0 100755
--- a/extract_notes.py
+++ b/extract_notes.py
@@ -6,11 +6,12 @@ import os, sys, os.path
import getpass
import sqlite3
import json
+import bisect
import re #regular expression
parser = argparse.ArgumentParser(description="Extraire les notes d'une liseuse Vivlio")
parser.add_argument("-s", "--source", help="le dossier de la liseuse", type=str, default=None)
-parser.add_argument("-f","--format", help="le format d'export (JSON ou wiki)", type=str, default=None)
+parser.add_argument("-f","--format", help="le format d'export (JSON ou wiki)", type=str, default="json")
parser.add_argument("-d", "--destination", default=".", help="le chemin et nom du dossier d'export (un fichier par livre)", type=str)
args = parser.parse_args()
@@ -21,25 +22,26 @@ def getPage(anchor):
output = output.group(1)
return int(output)
-class Annotation:
- BM = "bookmark" #bookmark
- QUOTE = "citation" #citation
- NOTE = "note" #note
+class Note:
+ BM = 1 #bookmark
+ QUOTE = 2 #citation
+ NOTE = 3 #note
def __init__(self, dbString):
tmp = json.loads(dbString)
self.text = None
self.endP = None
+ self.note = None
if len(tmp)==3:
- self.type = Annotation.NOTE
+ self.type = Note.NOTE
tmp[0].update(tmp[1])
tmp[0]["note"] = tmp[2]["text"].replace("\n","")
self.note = tmp[0]["note"]
elif len(tmp[1]) == 1:
- self.type = Annotation.BM
+ self.type = Note.BM
else:
tmp[0].update(tmp[1])
- self.type = Annotation.QUOTE
+ self.type = Note.QUOTE
#TODO mistakes Crayon for a quote, quoting "crayon"
self.raw = tmp[0]
self.p = getPage(tmp[0]["anchor"])
@@ -50,14 +52,109 @@ class Annotation:
if "text" in tmp[0] :
self.text = tmp[0]["text"].replace("\n","")
+ def __lt__(self, note2):
+ if self.p > note2.p:
+ res = -1
+ elif self.p < note2.p:
+ res = 1
+ elif self.endP == note2.endP:
+ if self.type > note2.type:
+ res = -1
+ elif self.type < note2.type:
+ res = 1
+ else:
+ res = 0
+ elif self.endP == None:
+ res = 1
+ elif note2.endP == None:
+ res = -1
+ elif self.endP > note2.endP:
+ res = -1
+ else:
+ res = 1
+ return res
+
def toJSON(self):
- json_note = {"raw":self.raw,"type":self.type,"p":self.p}
- if self.text != None :
- json_note["text"] = self.text
+ json_note = {"type":self.type,"p":self.p}
if self.endP != None :
json_note["endP"] = self.endP
+ if self.text != None :
+ json_note["text"] = self.text
+ json_note["raw"] = self.raw
return json_note
+ def toWiki(self):
+ wiki_note = ""
+ if self.type != Note.BM:
+ wiki_note = "{{p|"+str(self.p)
+ if self.endP != None and self.p!=self.endP:
+ wiki_note += "–"+str(self.endP)
+ wiki_note+="}}\n"
+ if self.text != None:
+ wiki_note += "«"+self.text+"»\n"
+ if self.note != None:
+ wiki_note += "{{commentaire|"+self.note+"}}\n"
+ return wiki_note
+
+class Book:
+ def __init__(self, author=None, title=None):
+ self.author = author
+ self.title = title
+ self.data = []
+
+ def equals(self, author, title):
+ return self.author == author and self.title == title
+
+ def is_empty(self):
+ return len(self.data)==0
+
+ def empty(self):
+ self.data = []
+
+ def addNote(self, note):
+ bisect.insort_left(self.data,note)
+
+ def fileName(self, directory="."):
+ res = False
+ if self.author == None:
+ if self.title == None:
+ raise ValueError("Can't export title and author-less book")
+ else:
+ res = self.title
+ elif self.title == None:
+ res = self.author
+ else:
+ res = self.author+"_"+self.title
+ return directory+"/"+res
+
+ def store(self, directory=".", format="json"):
+ if format == "json":
+ self.JSON_store(self.fileName(directory)+".json")
+ elif format == "wiki":
+ self.wiki_store(self.fileName(directory)+".wiki")
+ else:
+ raise ValueError("Unexpected format: "+str(format))
+
+ def JSON_store(self, fileName):
+ content = {"author":self.author, "title":self.title, "data":[]}
+ for note in self.data:
+ content["data"].append(note.toJSON())
+ out_file = open(fileName, "w")
+ json.dump(content, out_file, indent = 2, ensure_ascii=False)
+ out_file.close()
+
+ def wiki_store(self, fileName):
+ out_file = open(fileName, "w")
+ out_file.write("="+self.author+" — "+self.title+"=\n")
+ for note in self.data:
+ out_file.write(note.toWiki())
+ out_file.write("\n")
+ out_file.close()
+
+
+#####
+# Main functions
+#####
def getSource():
print("Warning, this only works for linux systems")
basepath = "/media/"+getpass.getuser()
@@ -81,7 +178,7 @@ def connect(sourceDir):
path = sourceDir+'/system/config/books.db'
conn = sqlite3.connect("file://"+path+'?mode=ro', uri=True)
print("Connected to '"+path+"'.")
- except (ValueError,sqlite3.Error) as e:
+ except (sqlite3.Error) as e:
print (str(e))
conn = None
return conn
@@ -93,25 +190,17 @@ def retrieveData(conn):
return cur.fetchall()
#while conn.fetchone()
-def dataToJSON(data, directory):
- curfile = None
+def export_data(data, directory, format="json"):
+ curbook = Book()
for comment in data:
- fileName = directory+"/"+comment[1]+"_"+comment[0]+".json"
- if curfile != fileName:
- if curfile != None:
- JSON_store(curdata, curfile)
- curfile = fileName
- curdata = {"title":comment[0],"author":comment[1],"data":[]}
- jj = Annotation(comment[2])
- curdata["data"].append(jj.toJSON())
- fileName = None
- if curfile != None:
- JSON_store(curdata, curfile)
-
-def JSON_store(data, filename):
- out_file = open(filename, "w")
- json.dump(data, out_file, indent = 2, ensure_ascii=False)
- out_file.close()
+ if not curbook.equals(comment[1], comment[0]):
+ if not curbook.is_empty():
+ curbook.store(directory, format)
+ curbook.empty()
+ curbook = Book(comment[1],comment[0])
+ curbook.addNote(Note(comment[2]))
+ if curbook.is_empty():
+ curbook.store(directory, format)
#main
if __name__ == "__main__":
@@ -125,4 +214,7 @@ if __name__ == "__main__":
data = retrieveData(conn)
conn.close()
if data != None:
- dataToJSON(data, args.destination)
+ try:
+ export_data(data, args.destination, args.format)
+ except ValueError as e:
+ print(e)
--
GitLab