Add language guessing

1 files changed, 37 insertions, 5 deletions

diff --git a/cpaste b/cpaste
index 5df0b86..dfaf00e 100755
--- a/cpaste
+++ b/cpaste
@@ -23,13 +23,24 @@
 #######################################################################
 #
 # Requires: python-requests
+# Optionally requires: python-Pygments
 #
 
 import sys
 import json
 import requests
 from optparse import OptionParser
+try:
+    from pygments.lexers import guess_lexer, guess_lexer_for_filename
+    guess_lang = True
+except ImportError:
+    guess_lang = False
 
+default_lang = 'text'
+
+# Map numpy to python because the numpy lexer gives false positives
+# when guessing.
+lexer_lang_map = {'numpy': 'python'}
 
 paste_url = 'https://cpaste.org'
 
@@ -37,8 +48,8 @@ paste_url = 'https://cpaste.org'
 parser = OptionParser()
 parser.add_option("-f", "--file", dest="filename",
                       help="Read from a file instead of stdin", metavar="FILE")
-parser.add_option("-l", "--lang", action="store", type="string", dest="lang", default="text",
-                  help="Specifiy the language (defaults to text).", metavar="LANG")
+parser.add_option("-l", "--lang", action="store", type="string", dest="lang",
+                  help="Specifiy the language.", metavar="LANG")
 parser.add_option("-p", "--private",
                   action="store_true", dest="private", default=False,
                   help="Create a private post")
@@ -52,19 +63,40 @@ if options.filename:
     if not f:
         print "Oops, could not open file!"
 
-    paste_data = f.readlines()
+    paste_data = f.read()
 else:
-    paste_data = sys.stdin.readlines()
+    paste_data = sys.stdin.read()
 
 if not paste_data:
     print "Oops, we have no data"
     sys.exit(1)
 
+if not options.lang and guess_lang:
+    # If we have a filename, try guessing using the more reliable
+    # guess_lexer_for_filename function.
+    # If that fails, try the guess_lexer function on the code.
+    if options.filename:
+        try:
+            options.lang = guess_lexer_for_filename(options.filename,
+                                                    paste_data).name.lower()
+        except:
+            print "No guess by filename"
+            pass
+    if not options.lang:
+        try:
+            options.lang = guess_lexer(paste_data).name.lower()
+            options.lang = lexer_lang_map.get(options.lang, options.lang)
+        except:
+            pass
+
+# Use default if no language specified or guessed
+options.lang = options.lang or default_lang
+
 if int(options.expire) != 86400:
         options.expire = int(options.expire) * 60
 
 # json payload
-payload = {'data':''.join(paste_data), 'language': options.lang, 'private': options.private, 'expire': options.expire}
+payload = {'data': paste_data, 'language': options.lang, 'private': options.private, 'expire': options.expire}
 
 # http content type
 headers = {'Content-Type': 'application/json'}