def plot(self, file_name): self.debug.print_debug(self, u'Loading ' + file_name) textplot = Text.from_file(join(self.in_dir, file_name), self.debug, nostem=self.nostem) self.debug.print_debug(self, u'Plotting ' + file_name) if self.action == 'single': graph = textplot.plot_terms(self.terms, self.caption) elif self.action == 'group': graph = textplot.plot_terms_two_groups(self.terms, self.term_name, self.second_terms,self.second_term_name, self.caption) elif self.action == 'hist': graph = textplot.plot_terms_histogram(self.terms, self.caption, 5000) elif self.action == 'rawcount': graph = textplot.plot_terms_raw_count(self.terms, self.caption, 5000) self.debug.print_debug(self, u'Saving ' + file_name.replace('.txt', '.png')) graph.savefig(join(self.in_dir, file_name.replace('.txt', '.png'))) graph.close()
def plot(self, file_name): self.debug.print_debug(self, u'Loading ' + file_name) textplot = Text.from_file(join(self.in_dir, file_name), self.debug, nostem=self.nostem) self.debug.print_debug(self, u'Plotting ' + file_name) if self.action == 'single': graph = textplot.plot_terms(self.terms, self.caption) elif self.action == 'group': graph = textplot.plot_terms_two_groups(self.terms, self.term_name, self.second_terms,self.second_term_name, self.caption) elif self.action == 'hist': graph = textplot.plot_terms_histogram(self.terms, self.caption, self.words) elif self.action == 'rawcount': graph = textplot.plot_terms_raw_count(self.terms, self.caption, self.words) elif self.action == 'overlap': graph = textplot.plot_kde_overlap(self.terms) elif self.action == 'search': newterms = textplot.anchored_scores(self.terms[0]) count = 0 self.debug.print_(self, u'Top twenty correlated terms (with more than one occurrence) for {0}: '.format(self.terms[0])) for item in newterms: if len(textplot.terms[item]) > 1 and item != textplot.stem(self.terms[0]): if count > self.max: break self.debug.print_(self, item) count += 1 if self.action != 'search': self.debug.print_debug(self, u'Saving ' + file_name.replace('.txt', '.png')) graph.savefig(join(self.in_dir, file_name.replace('.txt', '.png'))) graph.close()
from graph import TextGraph from text import Text import sys, textwrap, urllib2 print >>sys.stderr, ">>> starting..." if "://" in sys.argv[1]: text = Text.from_url(sys.argv[1]) else: text = Text.from_file(sys.argv[1]) graph = TextGraph(text) summary = graph.summary() print "\n".join(textwrap.wrap(summary)) print graph = TextGraph(text) print "Tags:", ", ".join(graph.tags())