Change enum classes to be capitalized (#107)

chardet/charsetgroupprober.py

@@ -63,22 +63,22 @@ def feed(self, byte_str):
             state = prober.feed(byte_str)
             if not state:
                 continue
-            if state == ProbingState.found_it:
+            if state == ProbingState.FOUND_IT:
                 self._best_guess_prober = prober
                 return self.state
-            elif state == ProbingState.not_me:
+            elif state == ProbingState.NOT_ME:
                 prober.active = False
                 self._active_num -= 1
                 if self._active_num <= 0:
-                    self._state = ProbingState.not_me
+                    self._state = ProbingState.NOT_ME
                     return self.state
         return self.state
 
     def get_confidence(self):
         state = self.state
-        if state == ProbingState.found_it:
+        if state == ProbingState.FOUND_IT:
             return 0.99
-        elif state == ProbingState.not_me:
+        elif state == ProbingState.NOT_ME:
             return 0.01
         best_conf = 0.0
         self._best_guess_prober = None

chardet/charsetprober.py

@@ -42,7 +42,7 @@ def __init__(self, lang_filter=None):
         self.logger = logging.getLogger(__name__)
 
     def reset(self):
-        self._state = ProbingState.detecting
+        self._state = ProbingState.DETECTING
 
     @property
     def charset_name(self):

chardet/codingstatemachine.py

@@ -61,13 +61,13 @@ def __init__(self, sm):
         self.reset()
 
     def reset(self):
-        self._curr_state = MachineState.start
+        self._curr_state = MachineState.START
 
     def next_state(self, c):
         # for each byte we get its class
         # if it is first byte, we also get byte length
         byte_class = self._model['class_table'][c]
-        if self._curr_state == MachineState.start:
+        if self._curr_state == MachineState.START:
             self._curr_byte_pos = 0
             self._curr_char_len = self._model['char_len_table'][byte_class]
         # from byte's class and state_table, we get its next state

chardet/enums.py

@@ -9,39 +9,68 @@ class InputState(object):
     """
     This enum represents the different states a universal detector can be in.
     """
-    pure_ascii = 0
-    esc_ascii = 1
-    high_byte = 2
+    PURE_ASCII = 0
+    ESC_ASCII = 1
+    HIGH_BYTE = 2
 
 
 class LanguageFilter(object):
     """
     This enum represents the different language filters we can apply to a
     ``UniversalDetector``.
     """
-    chinese_simplified = 0x01
-    chinese_traditional = 0x02
-    japanese = 0x04
-    korean = 0x08
-    non_cjk = 0x10
-    all = 0x1F
-    chinese = chinese_simplified | chinese_traditional
-    cjk = chinese | japanese | korean
+    CHINESE_SIMPLIFIED = 0x01
+    CHINESE_TRADITIONAL = 0x02
+    JAPANESE = 0x04
+    KOREAN = 0x08
+    NON_CJK = 0x10
+    ALL = 0x1F
+    CHINESE = CHINESE_SIMPLIFIED | CHINESE_TRADITIONAL
+    CJK = CHINESE | JAPANESE | KOREAN
 
 
 class ProbingState(object):
     """
     This enum represents the different states a prober can be in.
     """
-    detecting = 0
-    found_it = 1
-    not_me = 2
+    DETECTING = 0
+    FOUND_IT = 1
+    NOT_ME = 2
 
 
 class MachineState(object):
     """
     This enum represents the different states a state machine can be in.
     """
-    start = 0
-    error = 1
-    its_me = 2
+    START = 0
+    ERROR = 1
+    ITS_ME = 2
+
+
+class SequenceLikelihood(object):
+    """
+    This enum represents the likelihood of a character following the previous one.
+    """
+    NEGATIVE = 0
+    UNLIKELY = 1
+    LIKELY = 2
+    POSITIVE = 3
+
+    @classmethod
+    def get_num_categories(cls):
+        """:returns: The number of likelihood categories in the enum."""
+        return 4
+
+
+class CharacterCategory(object):
+    """
+    This enum represents the different categories language models for
+    ``SingleByteCharsetProber`` put characters into.
+
+    Anything less than CONTROL is considered a letter.
+    """
+    UNDEFINED = 255
+    LINE_BREAK = 254
+    SYMBOL = 253
+    DIGIT = 252
+    CONTROL = 251

chardet/escprober.py

@@ -42,12 +42,12 @@ class EscCharSetProber(CharSetProber):
     def __init__(self, lang_filter=None):
         super(EscCharSetProber, self).__init__(lang_filter=lang_filter)
         self.coding_sm = []
-        if self.lang_filter & LanguageFilter.chinese_simplified:
+        if self.lang_filter & LanguageFilter.CHINESE_SIMPLIFIED:
             self.coding_sm.append(CodingStateMachine(HZ_SM_MODEL))
             self.coding_sm.append(CodingStateMachine(ISO2022CN_SM_MODEL))
-        if self.lang_filter & LanguageFilter.japanese:
+        if self.lang_filter & LanguageFilter.JAPANESE:
             self.coding_sm.append(CodingStateMachine(ISO2022JP_SM_MODEL))
-        if self.lang_filter & LanguageFilter.korean:
+        if self.lang_filter & LanguageFilter.KOREAN:
             self.coding_sm.append(CodingStateMachine(ISO2022KR_SM_MODEL))
         self.active_sm_count = None
         self._detected_charset = None
@@ -80,14 +80,14 @@ def feed(self, byte_str):
                 if not coding_sm or not coding_sm.active:
                     continue
                 coding_state = coding_sm.next_state(c)
-                if coding_state == MachineState.error:
+                if coding_state == MachineState.ERROR:
                     coding_sm.active = False
                     self.active_sm_count -= 1
                     if self.active_sm_count <= 0:
-                        self._state = ProbingState.not_me
+                        self._state = ProbingState.NOT_ME
                         return self.state
-                elif coding_state == MachineState.its_me:
-                    self._state = ProbingState.found_it
+                elif coding_state == MachineState.ITS_ME:
+                    self._state = ProbingState.FOUND_IT
                     self._detected_charset = coding_sm.get_coding_state_machine()
                     return self.state
 

chardet/escsm.py

@@ -63,12 +63,12 @@
 )
 
 HZ_ST = (
-MachineState.start,MachineState.error,     3,MachineState.start,MachineState.start,MachineState.start,MachineState.error,MachineState.error,# 00-07
-MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.its_me,# 08-0f
-MachineState.its_me,MachineState.its_me,MachineState.error,MachineState.error,MachineState.start,MachineState.start,     4,MachineState.error,# 10-17
-     5,MachineState.error,     6,MachineState.error,     5,     5,     4,MachineState.error,# 18-1f
-     4,MachineState.error,     4,     4,     4,MachineState.error,     4,MachineState.error,# 20-27
-     4,MachineState.its_me,MachineState.start,MachineState.start,MachineState.start,MachineState.start,MachineState.start,MachineState.start,# 28-2f
+MachineState.START,MachineState.ERROR,     3,MachineState.START,MachineState.START,MachineState.START,MachineState.ERROR,MachineState.ERROR,# 00-07
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,# 08-0f
+MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ERROR,MachineState.ERROR,MachineState.START,MachineState.START,     4,MachineState.ERROR,# 10-17
+     5,MachineState.ERROR,     6,MachineState.ERROR,     5,     5,     4,MachineState.ERROR,# 18-1f
+     4,MachineState.ERROR,     4,     4,     4,MachineState.ERROR,     4,MachineState.ERROR,# 20-27
+     4,MachineState.ITS_ME,MachineState.START,MachineState.START,MachineState.START,MachineState.START,MachineState.START,MachineState.START,# 28-2f
 )
 
 HZ_CHAR_LEN_TABLE = (0, 0, 0, 0, 0, 0)
@@ -115,14 +115,14 @@
 )
 
 ISO2022CN_ST = (
-MachineState.start,     3,MachineState.error,MachineState.start,MachineState.start,MachineState.start,MachineState.start,MachineState.start,# 00-07
-MachineState.start,MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.error,# 08-0f
-MachineState.error,MachineState.error,MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.its_me,# 10-17
-MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.error,MachineState.error,MachineState.error,     4,MachineState.error,# 18-1f
-MachineState.error,MachineState.error,MachineState.error,MachineState.its_me,MachineState.error,MachineState.error,MachineState.error,MachineState.error,# 20-27
-     5,     6,MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.error,# 28-2f
-MachineState.error,MachineState.error,MachineState.error,MachineState.its_me,MachineState.error,MachineState.error,MachineState.error,MachineState.error,# 30-37
-MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.its_me,MachineState.error,MachineState.start,# 38-3f
+MachineState.START,     3,MachineState.ERROR,MachineState.START,MachineState.START,MachineState.START,MachineState.START,MachineState.START,# 00-07
+MachineState.START,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,# 08-0f
+MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,# 10-17
+MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,     4,MachineState.ERROR,# 18-1f
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,# 20-27
+     5,     6,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,# 28-2f
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,# 30-37
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.ERROR,MachineState.START,# 38-3f
 )
 
 ISO2022CN_CHAR_LEN_TABLE = (0, 0, 0, 0, 0, 0, 0, 0, 0)
@@ -169,15 +169,15 @@
 )
 
 ISO2022JP_ST = (
-MachineState.start,     3,MachineState.error,MachineState.start,MachineState.start,MachineState.start,MachineState.start,MachineState.start,# 00-07
-MachineState.start,MachineState.start,MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.error,# 08-0f
-MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.its_me,# 10-17
-MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.error,MachineState.error,# 18-1f
-MachineState.error,     5,MachineState.error,MachineState.error,MachineState.error,     4,MachineState.error,MachineState.error,# 20-27
-MachineState.error,MachineState.error,MachineState.error,     6,MachineState.its_me,MachineState.error,MachineState.its_me,MachineState.error,# 28-2f
-MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.its_me,MachineState.its_me,# 30-37
-MachineState.error,MachineState.error,MachineState.error,MachineState.its_me,MachineState.error,MachineState.error,MachineState.error,MachineState.error,# 38-3f
-MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.its_me,MachineState.error,MachineState.start,MachineState.start,# 40-47
+MachineState.START,     3,MachineState.ERROR,MachineState.START,MachineState.START,MachineState.START,MachineState.START,MachineState.START,# 00-07
+MachineState.START,MachineState.START,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,# 08-0f
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,# 10-17
+MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ERROR,MachineState.ERROR,# 18-1f
+MachineState.ERROR,     5,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,     4,MachineState.ERROR,MachineState.ERROR,# 20-27
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,     6,MachineState.ITS_ME,MachineState.ERROR,MachineState.ITS_ME,MachineState.ERROR,# 28-2f
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.ITS_ME,# 30-37
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,# 38-3f
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.ERROR,MachineState.START,MachineState.START,# 40-47
 )
 
 ISO2022JP_CHAR_LEN_TABLE = (0, 0, 0, 0, 0, 0, 0, 0, 0, 0)
@@ -224,11 +224,11 @@
 )
 
 ISO2022KR_ST = (
-MachineState.start,     3,MachineState.error,MachineState.start,MachineState.start,MachineState.start,MachineState.error,MachineState.error,# 00-07
-MachineState.error,MachineState.error,MachineState.error,MachineState.error,MachineState.its_me,MachineState.its_me,MachineState.its_me,MachineState.its_me,# 08-0f
-MachineState.its_me,MachineState.its_me,MachineState.error,MachineState.error,MachineState.error,     4,MachineState.error,MachineState.error,# 10-17
-MachineState.error,MachineState.error,MachineState.error,MachineState.error,     5,MachineState.error,MachineState.error,MachineState.error,# 18-1f
-MachineState.error,MachineState.error,MachineState.error,MachineState.its_me,MachineState.start,MachineState.start,MachineState.start,MachineState.start,# 20-27
+MachineState.START,     3,MachineState.ERROR,MachineState.START,MachineState.START,MachineState.START,MachineState.ERROR,MachineState.ERROR,# 00-07
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ITS_ME,# 08-0f
+MachineState.ITS_ME,MachineState.ITS_ME,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,     4,MachineState.ERROR,MachineState.ERROR,# 10-17
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,     5,MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,# 18-1f
+MachineState.ERROR,MachineState.ERROR,MachineState.ERROR,MachineState.ITS_ME,MachineState.START,MachineState.START,MachineState.START,MachineState.START,# 20-27
 )
 
 ISO2022KR_CHAR_LEN_TABLE = (0, 0, 0, 0, 0, 0)

chardet/eucjpprober.py

@@ -53,15 +53,15 @@ def feed(self, byte_str):
         for i in range(len(byte_str)):
             # PY3K: byte_str is a byte array, so byte_str[i] is an int, not a byte
             coding_state = self.coding_sm.next_state(byte_str[i])
-            if coding_state == MachineState.error:
+            if coding_state == MachineState.ERROR:
                 self.logger.debug('%s prober hit error at byte %s',
                                   self.charset_name, i)
-                self._state = ProbingState.not_me
+                self._state = ProbingState.NOT_ME
                 break
-            elif coding_state == MachineState.its_me:
-                self._state = ProbingState.found_it
+            elif coding_state == MachineState.ITS_ME:
+                self._state = ProbingState.FOUND_IT
                 break
-            elif coding_state == MachineState.start:
+            elif coding_state == MachineState.START:
                 char_len = self.coding_sm.get_current_charlen()
                 if i == 0:
                     self._last_char[1] = byte_str[0]
@@ -75,10 +75,10 @@ def feed(self, byte_str):
 
         self._last_char[0] = byte_str[-1]
 
-        if self.state == ProbingState.detecting:
+        if self.state == ProbingState.DETECTING:
             if (self.context_analyzer.got_enough_data() and
                (self.get_confidence() > self.SHORTCUT_THRESHOLD)):
-                self._state = ProbingState.found_it
+                self._state = ProbingState.FOUND_IT
 
         return self.state
 

chardet/hebrewprober.py

@@ -220,9 +220,9 @@ def feed(self, byte_str):
         # We automatically filter out all 7-bit characters (replace them with
         # spaces) so the word boundary detection works properly. [MAP]
 
-        if self.state == ProbingState.not_me:
+        if self.state == ProbingState.NOT_ME:
             # Both model probers say it's not them. No reason to continue.
-            return ProbingState.not_me
+            return ProbingState.NOT_ME
 
         byte_str = self.filter_high_byte_only(byte_str)
 
@@ -249,8 +249,8 @@ def feed(self, byte_str):
             self._prev = cur
 
         # Forever detecting, till the end or until both model probers return
-        # ProbingState.not_me (handled above)
-        return ProbingState.detecting
+        # ProbingState.NOT_ME (handled above)
+        return ProbingState.DETECTING
 
     @property
     def charset_name(self):
@@ -282,7 +282,7 @@ def charset_name(self):
     @property
     def state(self):
         # Remain active as long as any of the model probers are active.
-        if (self._logical_prober.state == ProbingState.not_me) and \
-           (self._visual_prober.state == ProbingState.not_me):
-            return ProbingState.not_me
-        return ProbingState.detecting
+        if (self._logical_prober.state == ProbingState.NOT_ME) and \
+           (self._visual_prober.state == ProbingState.NOT_ME):
+            return ProbingState.NOT_ME
+        return ProbingState.DETECTING

chardet/latin1prober.py

@@ -116,15 +116,15 @@ def feed(self, byte_str):
             freq = Latin1ClassModel[(self._last_char_class * CLASS_NUM)
                                     + char_class]
             if freq == 0:
-                self._state = ProbingState.not_me
+                self._state = ProbingState.NOT_ME
                 break
             self._freq_counter[freq] += 1
             self._last_char_class = char_class
 
         return self.state
 
     def get_confidence(self):
-        if self.state == ProbingState.not_me:
+        if self.state == ProbingState.NOT_ME:
             return 0.01
 
         total = sum(self._freq_counter)

chardet/mbcharsetprober.py

@@ -57,15 +57,15 @@ def charset_name(self):
     def feed(self, byte_str):
         for i in range(len(byte_str)):
             coding_state = self.coding_sm.next_state(byte_str[i])
-            if coding_state == MachineState.error:
+            if coding_state == MachineState.ERROR:
                 self.logger.debug('%s prober hit error at byte %s',
                                   self.charset_name, i)
-                self._state = ProbingState.not_me
+                self._state = ProbingState.NOT_ME
                 break
-            elif coding_state == MachineState.its_me:
-                self._state = ProbingState.found_it
+            elif coding_state == MachineState.ITS_ME:
+                self._state = ProbingState.FOUND_IT
                 break
-            elif coding_state == MachineState.start:
+            elif coding_state == MachineState.START:
                 char_len = self.coding_sm.get_current_charlen()
                 if i == 0:
                     self._last_char[1] = byte_str[0]
@@ -76,10 +76,10 @@ def feed(self, byte_str):
 
         self._last_char[0] = byte_str[-1]
 
-        if self.state == ProbingState.detecting:
+        if self.state == ProbingState.DETECTING:
             if (self.distribution_analyzer.got_enough_data() and
                     (self.get_confidence() > self.SHORTCUT_THRESHOLD)):
-                self._state = ProbingState.found_it
+                self._state = ProbingState.FOUND_IT
 
         return self.state