# scale to half a page width
r = mi.draw_y_proj(p,image=r,color=RGBPixel(100,0,0),fac=0.5)
r.save_PNG("%s/%s-03-words-and-word-projections.png"%(outputbase,basename))
logging.debug("03-words-and-word-projections.png")
# Lines
r = mi.to_rgb()
bsimg = mi.without_insidestaves_info()
p2 = p.rspikes(mi._image.ncols-1)
for myr in p2:
r.draw_hollow_rect(myr,RGBPixel(100,100,100))
lines = t._confirmed_text_lines(words)
for l in lines:
r.draw_hollow_rect(l.l,RGBPixel(255,0,0))
occs = mi.ccs_overall()
for c in occs["text"]:
r.highlight(c,RGBPixel(0,150,0))
r.save_PNG("%s/%s-04-lines-and-text.png"%(outputbase,basename))
logging.debug("04-lines-and-text.png")
r = mi.color_segment()
r.save_PNG("%s/%s-05-colorsegmented.png"%(outputbase,basename))
class Page:
def __init__(self,origfile,pagenumber,classifier=None):
self._pagenumber = pagenumber
self._origfile = origfile
self._noend = origfile[:-4] # shortcut
self._classifier = classifier
# internal logger
self._l = logging.getLogger(self.__class__.__name__)
# Internal params init.
self._file = None
self._mi = None
self._nostavesfile = None
def save(self,filename=None):
start = time.time()
if filename is None:
filename = self._genfilename()
cmd = ' '.join([gspath,
'-dNOPAUSE',
'-q',
'-r300',
'-sDEVICE=tiffg4',
'-dBATCH',
'-sOutputFile=%s'%filename,
'-sPAPERSIZE=a4',
'-dFirstPage=%d'%self._pagenumber,
'-dLastPage=%d'%self._pagenumber,
self._origfile
])
po = Popen(cmd,shell=True,stdout=PIPE,stderr=STDOUT).stdout
for l in po.readlines():
self._l.debug("GS Output:%s",l)
po.close()
self._file = filename
self._l.debug("Saving file %s (duration %f)",
filename,(time.time()-start))
return filename
def _init_mi(self):
""" Setup a MusicImage with our classifier """
self._mi = IllMusicImage(self._file,classifier=self._classifier)
def save_nostaves(self,filename=None):
start = time.time()
if self._file is None:
self._l.debug("No converted tif page. Forcing one now.")
self.save()
if filename is None:
filename = self._genfilename(postfix="-nostaves",extension=".png")
if self._mi is None:
self._init_mi()
self._mi.without_staves().save_PNG(filename)
self._nostavesfile = filename
self._l.debug("Saved file %s (duration %f)",filename,(time.time()-start))
return filename
def generate_gamera_script(self,dir=".",filename=None,openfile=None):
start = time.time()
if filename is None:
filename = self._genfilename(dir=dir,extension=".py")
#self._l.info("Filename not given. Using %s",filename)
if openfile is None and not self._nostavesfile is None:
openfile = self._nostavesfile
if openfile is None:
raise Exception,"No file to open"
with open(filename,'w') as gamscript:
gamscript.write("# Open %s in gamera with a classifier\n"%self._nostavesfile)
with open("gamscripthead.py") as gamscript_head:
gamscript.write(gamscript_head.read())
gamscript.write("\n####\n")
gamscript.write("image = load_image(\"%s\")\n"%self._nostavesfile)
gamscript.write("ccs = image.cc_analysis()\n")
gamscript.write("classifier.display(ccs,image)\n")
self._l.debug("Saved file %s (duration %f)",filename,(time.time()-start))
return filename
def save_color_segmented(self,filename=None):
start = time.time()
if self._mi is None:
self._init_mi()
if filename is None:
filename = self._genfilename(postfix="-colorseg",extension=".png")
color = self._mi.color_segment(classified_box=True)
color.save_PNG(filename)
self._l.debug("Saved file %s (duration %f)",
filename,(time.time()-start))
return filename
def save_without(self,filename=None,type="png"):
start = time.time()
if self._mi is None:
self._init_mi()
if filename is None:
filename = self._genfilename(postfix="-without",extension=(".%s"%type))
color = self._mi.without()
if type is "png":
color.save_PNG(filename)
elif type is "tif":
color.save_tiff(filename)
else:
raise Exception,"Unknown type %s"%type
self._l.debug("Saved file %s (duration %f)",
filename,(time.time()-start))
return filename
def gen_count_yaml(self,filename=None):
if filename is None:
filename = self._genfilename(postfix="-colorseg",extension=".yaml")
if self._mi is None:
self._init_mi()
c = self._mi.ccs_overall()
cl = dict([ (k,len(v)) for k,v in c.iteritems() ])
with open(filename,"w") as f:
data = {'text':{'found':cl['text'],'count':cl['text']},
'dynamics':{'found':cl['classified'], 'count':cl['classified']
}
}
yaml.dump(data,f)
logging.debug("Wrote %s",filename)
def _genfilename(self,dir=None,postfix="",extension=".tif"):
if dir is None:
dir = self._noend
filename = "%s/%s-page%02d%s%s"%\
(dir,self._noend,self._pagenumber,postfix,extension)
if not os.path.exists(dir):
self._l.debug("%s dir dint exits. Creating it.",dir)
os.mkdir(dir)
return filename
r.highlight(c, RGBPixel(0, 200, 0))
words = wordlist(t._words(image=bsimg, ccs=inccs), inccs,
t._min_wordlength)
for c in words:
r.draw_hollow_rect(c.w, RGBPixel(255, 0, 0))
p = t._word_projections(words=words)
# scale to half a page width
r = mi.draw_y_proj(p, image=r, color=RGBPixel(100, 0, 0), fac=0.5)
r.save_PNG("%s/%s-03-words-and-word-projections.png" %
(outputbase, basename))
logging.debug("03-words-and-word-projections.png")
# Lines
r = mi.to_rgb()
bsimg = mi.without_insidestaves_info()
p2 = p.rspikes(mi._image.ncols - 1)
for myr in p2:
r.draw_hollow_rect(myr, RGBPixel(100, 100, 100))
lines = t._confirmed_text_lines(words)
for l in lines:
r.draw_hollow_rect(l.l, RGBPixel(255, 0, 0))
occs = mi.ccs_overall()
for c in occs["text"]:
r.highlight(c, RGBPixel(0, 150, 0))
r.save_PNG("%s/%s-04-lines-and-text.png" % (outputbase, basename))
logging.debug("04-lines-and-text.png")
r = mi.color_segment()
r.save_PNG("%s/%s-05-colorsegmented.png" % (outputbase, basename))