class documentFormatter:
""" base class for formatting bibliographies """
def __init__(self, db=None, style="numbered", sortRefs=False):
self.citekeys = []
self.sortRefs = sortRefs
self.style = style
self.references = bibArray(style)
self.uniqueCitekeys = []
self.missingRefs = []
self.inTextRefs = {}
self.opening = "["
self.closing = "]"
if style == "parenthetical":
self.opening = "("
self.closing = ")"
if db:
self.db = db
else:
self.db = DbConnection()
self.db.guiPrompt()
def getUniqueCitekeys(self):
""" return just the unique citekeys """
if len(self.citekeys) > 1:
u = {}
for key in self.citekeys: u[key] = key
#self.uniqueCitekeys = u.keys()
for k in u.keys():
print "Key: %s" % k
# remove duplicates here
if (self.citekeys.count(k) > 1): # if ref occurs more than once
indices = []
print "Before: %i" % len(self.citekeys)
for i in range(0, len(self.citekeys)):
if self.citekeys[i] == k: # find where this ref occurs
indices.append(i)
counter = 0
for i in range(1, len(indices)): # remove all but first
print "indices: %s" % indices
self.citekeys.pop(indices[i] - counter)
counter += 1 # counter needed since size of array is changing
print "After: %i" % len(self.citekeys)
#else:
# self.uniqueCitekeys = self.citekeys
def getReferences(self):
""" attempts to retrieve all the references from the DB """
for key in self.citekeys:
try:
b = bibItem(db=self.db, key=key)
except invalidDBKeyException:
self.missingRefs.append(key)
else:
if key:
print "Inserting key %s" % key
self.references.append(b)
print "CITATION: %s" % self.references.inTextCitation(b["citekey"])
if self.sortRefs == True:
self.references.sort(refSorter)
self.references.changeInTextCounts()
print "Citations: %s" % self.references.getInTextCitations()
self.formatReferences()
for i in self.references: print i
print "IN TEXT REFS %s" % self.inTextRefs
def printMissingRefs(self):
""" prints a list of missing references to the screen """
print "Missing keys: "
for r in self.missingRefs:
print " %s" % r
def formatReferences(self):
""" creates dict of formatted in text references """
if self.style == "parenthetical":
self.references.modifyDuplicates()
for k in self.inTextRefs.keys():
formatted = ""
formattedRefs = []
refs = self.refsInCitation(k)
for r in refs:
try:
c = self.references.inTextCitation(r)
except KeyError:
formattedRefs.append("MISSING: %s" % r)
else:
print "THIS WAS C: %s" % c
formattedRefs.append("%s" % self.references.inTextCitation(r))
if len(refs) > 1:
print "FR: %s" % formattedRefs
s = ", ".join(formattedRefs)
formatted = "%s%s%s" % (self.opening, s, self.closing)
else:
formatted = "%s%s%s" % (self.opening, formattedRefs[0], self.closing)
self.inTextRefs[k] = formatted
def removeColons(self, reference):
"""quick hack to replace Doe:2001 with Doe2001"""
p = re.compile(r':')
return p.sub('', reference)
def refsInCitation(self, citation):
""" get the citekeys inside an unformatted inText citation """
refs = []
spaceRemover = re.compile(r'^\s+')
squiqqlyRemover = re.compile(r'[\{\}]')
citation = squiqqlyRemover.sub('', citation)
print "List: %s" % citation
for reference in citation.split(","):
reference = spaceRemover.sub('', reference)
reference = self.removeColons(reference)
refs.append(reference)
return(refs)
def openFile(self, fileName):
""" an empty method that needs to be subclassed """
pass
def getCitekeys(self):
""" an empty method that needs to be subclassed """
pass
def replaceCitekeys(self):
""" an empty method that needs to be subclassed """
pass
def formatBibliography(self):
""" an empty method that needs to be subclassed """
pass
def writeFile(self, output):
""" an empty method that needs to be subclassed """
pass
def formatDocument(self, input=None, output=None):
""" formats a document - replacing references and creating a ref list"""
if input:
self.filename = input
# open the file - this method needs to be subclassed.
self.openFile(self.filename)
# extract the citekeys from the document (handled by the subclass)
self.getCitekeys()
# get just the unique citekeys
self.getUniqueCitekeys()
print "Unique: %s" % self.uniqueCitekeys
# get the references from the database
self.getReferences()
# replace the citekeys inside the document (handled by subclass)
if self.style != "unformatted":
self.replaceCitekeys()
# format the bibliography (handled by subclass)
self.formatBibliography()
# write to file (handled by subclass)
if output:
self.writeFile(output)
self.printMissingRefs()
notequery.prepare(QtCore.QString("INSERT INTO note (citekey, note) VALUES (?, ?)"))
notequery.bindValue(0, QtCore.QVariant(currentRef))
notequery.bindValue(1, QtCore.QVariant(note))
notequery.exec_()
if notequery.lastError().type():
print "Error inserting note: [%s] %s" % (citekey, note)
if __name__ == "__main__":
app = QtGui.QApplication(sys.argv)
fileHandle = open("/micehome/jlerch/Documents/MICE/articles/morris-water-maze/lit_notes.txt", 'r')
fileContents = fileHandle.read()
db = DbConnection()
db.guiPrompt()
starts = []
ends = []
refStarter = re.compile(r'^====')
noteStarter = re.compile(r'^\*')
spaceStarter = re.compile(r'^\s')
lines = fileContents.split("\n")
counter = 0
note = None
currentRef = None
notecounter = 0
for l in lines: